Files
libakgl/tests/actor.c

1084 lines
42 KiB
C
Raw Normal View History

#define UNHANDLED_ERROR_TERMINATION_BEHAVIOR \
handle_unhandled_error(errctx);
#include <akerror.h>
#include <akgl/error.h>
#define UNHANDLED_ERROR_EXIT 0
#define UNHANDLED_ERROR_SET 1
#include <SDL3/SDL.h>
#include <stdlib.h>
#include <akgl/iterator.h>
#include <akgl/registry.h>
#include <akgl/actor.h>
Add physics, heap, json_helpers, game, and actor test suites Raise line coverage from 39.6 to 61.8 percent with four new suites and an extension to the actor suite, and register every suite through a single CMake list so a new test file cannot be left out of the coverage fixture. Give the test targets a build-tree RPATH and prepend the build tree to LD_LIBRARY_PATH for CTest, so a developer with a previously installed libakgl.so exercises the library that was just compiled. Fix six defects the new tests exposed: - akgl_physics_simulate read self->gravity_time before its NULL check, so a NULL backend crashed instead of reporting AKERR_NULLPOINTER. - akgl_game_save transposed CLEANUP and PROCESS, which placed the fclose inside the PROCESS switch. An ordinary save never flushed or closed its stream and produced an empty file. - akgl_game_save_actors wrote each name table terminator from the address of a single char, emitting stack contents into the save file and a sentinel the loader could not recognize. - akgl_game_load_objectnamemap used CATCH directly inside while(1), where the break leaves the loop rather than propagating, so a truncated name table loaded as a successful game. - akgl_Actor_cmhf_up_on and _down_on dereferenced actor->basechar with no NULL check, unlike their left and right counterparts. - akgl_actor_logic_movement checked actor twice instead of checking actor->basechar before dereferencing it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-30 02:03:21 -04:00
#include <akgl/character.h>
#include <akgl/sprite.h>
#include <akgl/game.h>
#include <akgl/heap.h>
Draw actors at their sprite's height, and update each one once a frame Closes Defects item 26 and Performance item 32. akgl_actor_render set dest.h from curSprite->width, so every actor was drawn square and a non-square sprite was stretched or squashed. Invisible in the fixtures because they are all square, so tests/actor.c gets a 48x24 sprite and a render backend whose draw_texture records the rectangle it is handed instead of drawing it. That recording backend is the first coverage akgl_actor_render has had at all -- every other test in the file stubs renderfunc out. akgl_game_update looped over AKGL_TILEMAP_MAX_LAYERS with the actor sweep nested inside it and never compared an actor's layer to the layer it was on, so every live actor's updatefunc ran sixteen times a frame. The sweep is hoisted out: updating an actor is not a per-layer operation, and akgl_render_2d_draw_world already walks the layers for the half that is. AKGL_ITERATOR_OP_LAYERMASK is honoured rather than ignored now, so a caller who wants one layer can still ask, and gets each of those actors once. Counting is the assertion, deliberately. The defect was invisible in the frame total because the tilemap blits are three orders of magnitude larger, so a timing test would have measured the rasterizer. tests/game.c counts calls into a stub updatefunc and reports 16 against the old code. PERFORMANCE.md records the timing side as what it honestly is: the gap between the akgl_game_update and draw_world rows of the same run, 92 us before and noise in both directions after. The absolute table is not re-taken -- a later run on this machine read every row about 15% high, including rows nothing has touched. 25/25 pass, reindent --check clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-01 06:45:34 -04:00
#include <akgl/renderer.h>
Add physics, heap, json_helpers, game, and actor test suites Raise line coverage from 39.6 to 61.8 percent with four new suites and an extension to the actor suite, and register every suite through a single CMake list so a new test file cannot be left out of the coverage fixture. Give the test targets a build-tree RPATH and prepend the build tree to LD_LIBRARY_PATH for CTest, so a developer with a previously installed libakgl.so exercises the library that was just compiled. Fix six defects the new tests exposed: - akgl_physics_simulate read self->gravity_time before its NULL check, so a NULL backend crashed instead of reporting AKERR_NULLPOINTER. - akgl_game_save transposed CLEANUP and PROCESS, which placed the fclose inside the PROCESS switch. An ordinary save never flushed or closed its stream and produced an empty file. - akgl_game_save_actors wrote each name table terminator from the address of a single char, emitting stack contents into the save file and a sentinel the loader could not recognize. - akgl_game_load_objectnamemap used CATCH directly inside while(1), where the break leaves the loop rather than propagating, so a truncated name table loaded as a successful game. - akgl_Actor_cmhf_up_on and _down_on dereferenced actor->basechar with no NULL check, unlike their left and right counterparts. - akgl_actor_logic_movement checked actor twice instead of checking actor->basechar before dereferencing it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-30 02:03:21 -04:00
#include "testutil.h"
int UNHANDLED_ERROR_BEHAVIOR;
akerr_ErrorContext *unhandled_error_context;
void handle_unhandled_error_noexit(akerr_ErrorContext *errctx)
{
if ( errctx == NULL ) {
return;
}
if ( UNHANDLED_ERROR_BEHAVIOR == UNHANDLED_ERROR_EXIT ) {
exit(errctx->status);
}
if ( UNHANDLED_ERROR_BEHAVIOR == UNHANDLED_ERROR_SET ) {
unhandled_error_context = errctx;
errctx->refcount += 1;
return;
}
}
int akgl_actor_updated;
akerr_ErrorContext *akgl_actor_update_noop(akgl_Actor *obj)
{
PREPARE_ERROR(errctx);
akgl_actor_updated = 1;
SUCCEED_RETURN(errctx);
}
// Currently the renderer assumes there is a global variable named `renderer`
int akgl_actor_rendered;
akerr_ErrorContext *akgl_actor_render_noop(akgl_Actor *obj)
{
PREPARE_ERROR(errctx);
akgl_actor_rendered = 1;
SUCCEED_RETURN(errctx);
}
akerr_ErrorContext *test_registry_actor_iterator_nullpointers(void)
{
PREPARE_ERROR(errctx);
akerr_ErrorUnhandledErrorHandler defaulthandler = akerr_handler_unhandled_error;
akerr_handler_unhandled_error = handle_unhandled_error_noexit;
ATTEMPT {
UNHANDLED_ERROR_BEHAVIOR = UNHANDLED_ERROR_SET;
DETECT(unhandled_error_context, akgl_registry_iterate_actor(NULL, AKGL_REGISTRY_ACTOR, ""));
} CLEANUP {
UNHANDLED_ERROR_BEHAVIOR = UNHANDLED_ERROR_EXIT;
} PROCESS(unhandled_error_context) {
} HANDLE(unhandled_error_context, AKERR_NULLPOINTER) {
printf("Handled\n");
} FINISH(unhandled_error_context, true);
akerr_handler_unhandled_error = defaulthandler;
SUCCEED_RETURN(errctx);
}
akerr_ErrorContext *test_registry_actor_iterator_missingactor(void)
{
PREPARE_ERROR(errctx);
akerr_ErrorUnhandledErrorHandler defaulthandler = akerr_handler_unhandled_error;
akgl_Iterator iter = {
.layerid = 0,
.flags = 0
};
akerr_handler_unhandled_error = handle_unhandled_error_noexit;
ATTEMPT {
UNHANDLED_ERROR_BEHAVIOR = UNHANDLED_ERROR_SET;
DETECT(
unhandled_error_context,
akgl_registry_iterate_actor(
&iter,
AKGL_REGISTRY_ACTOR,
"Actor who doesn't exist")
);
} CLEANUP {
UNHANDLED_ERROR_BEHAVIOR = UNHANDLED_ERROR_EXIT;
} PROCESS(unhandled_error_context) {
} HANDLE(unhandled_error_context, AKERR_KEY) {
printf("Handled\n");
} FINISH(unhandled_error_context, true);
akerr_handler_unhandled_error = defaulthandler;
SUCCEED_RETURN(errctx);
}
akerr_ErrorContext *test_registry_actor_iterator_updaterender(void)
{
akgl_Actor *testactor;
akgl_Iterator iter = {
.layerid = 0,
.flags = AKGL_ITERATOR_OP_UPDATE | AKGL_ITERATOR_OP_RENDER
};
akerr_ErrorUnhandledErrorHandler defaulthandler = akerr_handler_unhandled_error;
PREPARE_ERROR(errctx);
akerr_handler_unhandled_error = handle_unhandled_error_noexit;
ATTEMPT {
UNHANDLED_ERROR_BEHAVIOR = UNHANDLED_ERROR_SET;
CATCH(unhandled_error_context, akgl_heap_next_actor(&testactor));
CATCH(unhandled_error_context, akgl_actor_initialize(testactor, "test"));
testactor->layer = 0;
testactor->updatefunc = &akgl_actor_update_noop;
testactor->renderfunc = &akgl_actor_render_noop;
DETECT(
unhandled_error_context,
akgl_registry_iterate_actor(
&iter,
AKGL_REGISTRY_ACTOR,
"test")
);
FAIL_ZERO_BREAK(
unhandled_error_context,
akgl_actor_updated,
AKGL_ERR_BEHAVIOR,
"actor->updatefunc not called by the iterator"
);
FAIL_ZERO_BREAK(
unhandled_error_context,
akgl_actor_rendered,
AKGL_ERR_BEHAVIOR,
"actor->renderfunc not called by the iterator"
);
} CLEANUP {
UNHANDLED_ERROR_BEHAVIOR = UNHANDLED_ERROR_EXIT;
IGNORE(akgl_heap_release_actor(testactor));
} PROCESS(unhandled_error_context) {
} FINISH(unhandled_error_context, true);
akerr_handler_unhandled_error = defaulthandler;
SUCCEED_RETURN(errctx);
}
akerr_ErrorContext *test_akgl_actor_set_character(void)
{
akgl_Actor *testactor = NULL;
akgl_Character *testchar = NULL;
PREPARE_ERROR(errctx);
ATTEMPT {
CATCH(errctx, akgl_actor_set_character(NULL, "test"));
} CLEANUP {
} PROCESS(errctx) {
} HANDLE(errctx, AKERR_NULLPOINTER) {
printf("Handled\n");
} FINISH(errctx, true);
ATTEMPT {
CATCH(errctx, akgl_heap_next_actor(&testactor));
CATCH(errctx, akgl_actor_initialize(testactor, "test"));
testactor->layer = 0;
testactor->updatefunc = &akgl_actor_update_noop;
testactor->renderfunc = &akgl_actor_render_noop;
CATCH(errctx, akgl_actor_set_character(testactor, "test"));
} CLEANUP {
IGNORE(akgl_heap_release_actor(testactor));
} PROCESS(errctx) {
} HANDLE(errctx, AKERR_NULLPOINTER) {
printf("Handled\n");
} FINISH(errctx, true);
ATTEMPT {
CATCH(errctx, akgl_heap_next_actor(&testactor));
CATCH(errctx, akgl_heap_next_character(&testchar));
CATCH(errctx, akgl_actor_initialize(testactor, "test"));
testactor->layer = 0;
testactor->updatefunc = &akgl_actor_update_noop;
testactor->renderfunc = &akgl_actor_render_noop;
CATCH(errctx, akgl_character_initialize(testchar, "test"));
CATCH(errctx, akgl_actor_set_character(testactor, "test"));
} CLEANUP {
IGNORE(akgl_heap_release_actor(testactor));
IGNORE(akgl_heap_release_character(testchar));
} PROCESS(errctx) {
} FINISH(errctx, true);
SUCCEED_RETURN(errctx);
}
akerr_ErrorContext *test_actor_manage_children(void)
{
akgl_Actor *parent = NULL;
akgl_Actor *child = NULL;
akgl_String *tmpstring = NULL;
int i = 0;
int j = 0;
PREPARE_ERROR(errctx);
ATTEMPT {
CATCH(errctx, akgl_heap_init());
CATCH(errctx, akgl_heap_next_string(&tmpstring));
CATCH(errctx, akgl_heap_next_actor(&parent));
CATCH(errctx, akgl_actor_initialize(parent, "parent"));
for ( i = 0 ; i < AKGL_ACTOR_MAX_CHILDREN; i++ ) {
sprintf((char *)&tmpstring->data, "child %d", i);
CATCH(errctx, akgl_heap_next_actor(&child));
CATCH(errctx, akgl_actor_initialize(child, (char *)&tmpstring->data));
CATCH(errctx, parent->addchild(parent, child));
// Release our claim on the actor so the parent can own the child and kill it
CATCH(errctx, akgl_heap_release_actor(child));
}
} CLEANUP {
IGNORE(akgl_heap_release_string(tmpstring));
} PROCESS(errctx) {
} FINISH(errctx, true);
ATTEMPT {
// Doesn't matter which child we use for this test
child = parent->children[1];
CATCH(errctx, parent->addchild(parent, child));
} CLEANUP {
if ( errctx == NULL ) {
FAIL(errctx, AKGL_ERR_BEHAVIOR, "addchild does not throw AKERR_RELATIONSHIP when child already has a parent");
}
} PROCESS(errctx) {
} HANDLE(errctx, AKERR_RELATIONSHIP) {
// Expected behavior
SDL_Log("addchild throws AKERR_RELATIONSHIP when child already has a parent");
} FINISH(errctx, true);
ATTEMPT {
CATCH(errctx, akgl_heap_next_actor(&child));
CATCH(errctx, parent->addchild(parent, child));
} CLEANUP {
if ( errctx == NULL ) {
FAIL(errctx, AKGL_ERR_BEHAVIOR, "addchild does not throw AKERR_OUTOFBOUNDS when all children already set");
}
} PROCESS(errctx) {
} HANDLE(errctx, AKERR_OUTOFBOUNDS) {
// Expected behavior
SDL_Log("addchild throws AKERR_OUTOFBOUNDS when all children already set");
} FINISH(errctx, true);
ATTEMPT {
CATCH(errctx, akgl_heap_release_actor(parent));
// All actor objects on the heap should be empty now
for ( i = 0; i < AKGL_MAX_HEAP_ACTOR; i++) {
Namespace every exported symbol, and bump to 0.5.0 Closes internal-consistency items 1 through 6, 12 and 13. Every include guard is _AKGL_<FILE>_H_, every in-project header include is angled, and every exported function, type and global carries the akgl_ prefix. This is an ABI break; the soname goes to libakgl.so.0.5. TODO.md carries the full rename table. The renames were driven by renaming each declaration and letting the compiler find the uses, not by pattern substitution: renderer, physics and camera are also parameter and struct-member names, and a sed would have rewritten map->physics and every akgl_RenderBackend *renderer parameter without a word. Item 4 turned out not to be cosmetic. The library exported a global called renderer and tests/character.c defined an SDL_Renderer *renderer of its own; the executable's definition preempted the library's, akgl_sprite_load_json read a SDL_Renderer * through an akgl_RenderBackend *, and every texture load in that suite failed. The suite reported success anyway, because libakerror's unhandled-error handler ends in exit(errctx->status), exit keeps only the low byte, and AKGL_ERR_SDL is exactly 256. So character had been green while running one of its four tests, and every suite in the tree was unable to fail on the most common status in a library built on SDL. Both are fixed. tests/testutil.h gains TEST_TRAP_UNHANDLED_ERRORS(), which collapses any status a byte cannot carry onto 1, and every suite installs it. character binds a real backend with akgl_render_2d_bind. Its fourth test then runs for the first time and fails on a defect it has asserted all along, so akgl_heap_release_character now walks state_sprites with AKGL_ITERATOR_OP_RELEASE and destroys the property set before zeroing the slot -- TODO.md Defects item 21 and half of Carried over item 1. AKGL_TIME_ONESEC_MS said "one second in milliseconds" and held 1000000, so akgl_game_state_lock waited roughly sixteen minutes rather than one second. It is AKGL_TIME_ONEMS_NS now, the budget is its own named constant, and tests/game.c holds the mutex from a second thread to assert the wait -- the contended path had no coverage at all. Headers are self-contained and it is enforced: AKGL_PUBLIC_HEADERS drives both install() and a generated translation unit per header, so a header that ships is a header that is checked. Writing that found registry.h, which used SDL_PropertiesID in eight declarations and included no SDL header. 23/23 suites pass, memcheck is clean, reindent --check is clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-31 23:32:21 -04:00
FAIL_NONZERO_BREAK(errctx, akgl_heap_actors[i].refcount, AKERR_VALUE, "Actor not properly cleared");
FAIL_NONZERO_BREAK(errctx, akgl_heap_actors[i].parent, AKERR_VALUE, "Actor not properly cleared");
for ( j = 0 ; j < AKGL_ACTOR_MAX_CHILDREN; j++) {
Namespace every exported symbol, and bump to 0.5.0 Closes internal-consistency items 1 through 6, 12 and 13. Every include guard is _AKGL_<FILE>_H_, every in-project header include is angled, and every exported function, type and global carries the akgl_ prefix. This is an ABI break; the soname goes to libakgl.so.0.5. TODO.md carries the full rename table. The renames were driven by renaming each declaration and letting the compiler find the uses, not by pattern substitution: renderer, physics and camera are also parameter and struct-member names, and a sed would have rewritten map->physics and every akgl_RenderBackend *renderer parameter without a word. Item 4 turned out not to be cosmetic. The library exported a global called renderer and tests/character.c defined an SDL_Renderer *renderer of its own; the executable's definition preempted the library's, akgl_sprite_load_json read a SDL_Renderer * through an akgl_RenderBackend *, and every texture load in that suite failed. The suite reported success anyway, because libakerror's unhandled-error handler ends in exit(errctx->status), exit keeps only the low byte, and AKGL_ERR_SDL is exactly 256. So character had been green while running one of its four tests, and every suite in the tree was unable to fail on the most common status in a library built on SDL. Both are fixed. tests/testutil.h gains TEST_TRAP_UNHANDLED_ERRORS(), which collapses any status a byte cannot carry onto 1, and every suite installs it. character binds a real backend with akgl_render_2d_bind. Its fourth test then runs for the first time and fails on a defect it has asserted all along, so akgl_heap_release_character now walks state_sprites with AKGL_ITERATOR_OP_RELEASE and destroys the property set before zeroing the slot -- TODO.md Defects item 21 and half of Carried over item 1. AKGL_TIME_ONESEC_MS said "one second in milliseconds" and held 1000000, so akgl_game_state_lock waited roughly sixteen minutes rather than one second. It is AKGL_TIME_ONEMS_NS now, the budget is its own named constant, and tests/game.c holds the mutex from a second thread to assert the wait -- the contended path had no coverage at all. Headers are self-contained and it is enforced: AKGL_PUBLIC_HEADERS drives both install() and a generated translation unit per header, so a header that ships is a header that is checked. Writing that found registry.h, which used SDL_PropertiesID in eight declarations and included no SDL header. 23/23 suites pass, memcheck is clean, reindent --check is clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-31 23:32:21 -04:00
if ( akgl_heap_actors[i].children[j] != NULL ) {
FAIL(errctx, AKERR_VALUE, "Actor not properly cleared");
goto _test_actor_addchild_heaprelease_cleanup;
}
}
}
for ( j = 0; j < AKGL_ACTOR_MAX_CHILDREN; j++) {
sprintf((char *)&tmpstring->data, "child %d", i);
FAIL_NONZERO_BREAK(
errctx,
SDL_GetPointerProperty(AKGL_REGISTRY_ACTOR, (char *)&tmpstring->data, NULL),
AKERR_KEY,
"Child %s was not removed from the registry",
(char *)&tmpstring->data);
}
_test_actor_addchild_heaprelease_cleanup:
} CLEANUP {
} PROCESS(errctx) {
} FINISH(errctx, true);
ATTEMPT {
CATCH(errctx, akgl_heap_next_actor(&parent));
CATCH(errctx, akgl_actor_initialize(parent, "parent"));
CATCH(errctx, akgl_heap_next_actor(&child));
CATCH(errctx, akgl_actor_initialize(child, "child"));
// Don't release our claim on the child. The child should not be reclaimed.
CATCH(errctx, akgl_heap_release_actor(parent));
FAIL_NONZERO_BREAK(errctx, child->parent, AKERR_VALUE, "Child still references released parent");
FAIL_ZERO_BREAK(errctx, child->refcount, AKERR_VALUE, "Child prematurely released");
FAIL_NONZERO_BREAK(errctx, strcmp(child->name, "child"), AKERR_VALUE, "Child had identity erased");
FAIL_ZERO_BREAK(
errctx,
(child == SDL_GetPointerProperty(AKGL_REGISTRY_ACTOR, child->name, NULL)),
AKERR_KEY,
"Child prematurely removed from the registry");
// Now we can release the child
CATCH(errctx, akgl_heap_release_actor(child));
} CLEANUP {
} PROCESS(errctx) {
} FINISH(errctx, true);
SUCCEED_RETURN(errctx);
}
Add physics, heap, json_helpers, game, and actor test suites Raise line coverage from 39.6 to 61.8 percent with four new suites and an extension to the actor suite, and register every suite through a single CMake list so a new test file cannot be left out of the coverage fixture. Give the test targets a build-tree RPATH and prepend the build tree to LD_LIBRARY_PATH for CTest, so a developer with a previously installed libakgl.so exercises the library that was just compiled. Fix six defects the new tests exposed: - akgl_physics_simulate read self->gravity_time before its NULL check, so a NULL backend crashed instead of reporting AKERR_NULLPOINTER. - akgl_game_save transposed CLEANUP and PROCESS, which placed the fclose inside the PROCESS switch. An ordinary save never flushed or closed its stream and produced an empty file. - akgl_game_save_actors wrote each name table terminator from the address of a single char, emitting stack contents into the save file and a sentinel the loader could not recognize. - akgl_game_load_objectnamemap used CATCH directly inside while(1), where the break leaves the loop rather than propagating, so a truncated name table loaded as a successful game. - akgl_Actor_cmhf_up_on and _down_on dereferenced actor->basechar with no NULL check, unlike their left and right counterparts. - akgl_actor_logic_movement checked actor twice instead of checking actor->basechar before dereferencing it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-30 02:03:21 -04:00
/**
* @brief Build an actor bound to a character, without touching the renderer.
*
* The sprite and spritesheet are taken straight off the heap and populated by
* hand, because akgl_sprite_load_json() would need a live renderer to build a
* texture and none of the logic under test reads one.
*/
static akerr_ErrorContext *make_bound_actor(
akgl_Actor **actor,
akgl_Character **basechar,
akgl_Sprite **sprite,
char *name,
int state)
{
PREPARE_ERROR(e);
akgl_SpriteSheet *sheet = NULL;
char spritename[AKGL_SPRITE_MAX_NAME_LENGTH];
ATTEMPT {
CATCH(e, akgl_heap_next_character(basechar));
CATCH(e, akgl_character_initialize(*basechar, name));
CATCH(e, akgl_heap_next_spritesheet(&sheet));
sheet->refcount += 1;
snprintf((char *)&spritename, AKGL_SPRITE_MAX_NAME_LENGTH, "%s_sprite", name);
CATCH(e, akgl_heap_next_sprite(sprite));
CATCH(e, akgl_sprite_initialize(*sprite, (char *)&spritename, sheet));
(*sprite)->width = 32;
(*sprite)->height = 32;
(*sprite)->frames = 4;
(*sprite)->speed = 100;
CATCH(e, akgl_character_sprite_add(*basechar, *sprite, state));
CATCH(e, akgl_heap_next_actor(actor));
CATCH(e, akgl_actor_initialize(*actor, name));
(*actor)->basechar = *basechar;
(*actor)->state = state;
} CLEANUP {
} PROCESS(e) {
} FINISH(e, true);
SUCCEED_RETURN(e);
}
akerr_ErrorContext *test_actor_control_handlers_on(void)
{
PREPARE_ERROR(e);
akgl_Actor actor;
akgl_Character basechar;
SDL_Event event;
ATTEMPT {
memset(&event, 0x00, sizeof(SDL_Event));
memset(&basechar, 0x00, sizeof(akgl_Character));
basechar.ax = 7.0f;
basechar.ay = 11.0f;
// Left: clears every facing and movement bit, then sets its own pair and
// takes acceleration from the base character with the sign reversed.
memset(&actor, 0x00, sizeof(akgl_Actor));
actor.basechar = &basechar;
actor.state = (AKGL_ACTOR_STATE_FACE_RIGHT | AKGL_ACTOR_STATE_MOVING_RIGHT | AKGL_ACTOR_STATE_ALIVE);
Namespace every exported symbol, and bump to 0.5.0 Closes internal-consistency items 1 through 6, 12 and 13. Every include guard is _AKGL_<FILE>_H_, every in-project header include is angled, and every exported function, type and global carries the akgl_ prefix. This is an ABI break; the soname goes to libakgl.so.0.5. TODO.md carries the full rename table. The renames were driven by renaming each declaration and letting the compiler find the uses, not by pattern substitution: renderer, physics and camera are also parameter and struct-member names, and a sed would have rewritten map->physics and every akgl_RenderBackend *renderer parameter without a word. Item 4 turned out not to be cosmetic. The library exported a global called renderer and tests/character.c defined an SDL_Renderer *renderer of its own; the executable's definition preempted the library's, akgl_sprite_load_json read a SDL_Renderer * through an akgl_RenderBackend *, and every texture load in that suite failed. The suite reported success anyway, because libakerror's unhandled-error handler ends in exit(errctx->status), exit keeps only the low byte, and AKGL_ERR_SDL is exactly 256. So character had been green while running one of its four tests, and every suite in the tree was unable to fail on the most common status in a library built on SDL. Both are fixed. tests/testutil.h gains TEST_TRAP_UNHANDLED_ERRORS(), which collapses any status a byte cannot carry onto 1, and every suite installs it. character binds a real backend with akgl_render_2d_bind. Its fourth test then runs for the first time and fails on a defect it has asserted all along, so akgl_heap_release_character now walks state_sprites with AKGL_ITERATOR_OP_RELEASE and destroys the property set before zeroing the slot -- TODO.md Defects item 21 and half of Carried over item 1. AKGL_TIME_ONESEC_MS said "one second in milliseconds" and held 1000000, so akgl_game_state_lock waited roughly sixteen minutes rather than one second. It is AKGL_TIME_ONEMS_NS now, the budget is its own named constant, and tests/game.c holds the mutex from a second thread to assert the wait -- the contended path had no coverage at all. Headers are self-contained and it is enforced: AKGL_PUBLIC_HEADERS drives both install() and a generated translation unit per header, so a header that ships is a header that is checked. Writing that found registry.h, which used SDL_PropertiesID in eight declarations and included no SDL header. 23/23 suites pass, memcheck is clean, reindent --check is clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-31 23:32:21 -04:00
TEST_EXPECT_OK(e, akgl_actor_cmhf_left_on(&actor, &event), "left on");
Add physics, heap, json_helpers, game, and actor test suites Raise line coverage from 39.6 to 61.8 percent with four new suites and an extension to the actor suite, and register every suite through a single CMake list so a new test file cannot be left out of the coverage fixture. Give the test targets a build-tree RPATH and prepend the build tree to LD_LIBRARY_PATH for CTest, so a developer with a previously installed libakgl.so exercises the library that was just compiled. Fix six defects the new tests exposed: - akgl_physics_simulate read self->gravity_time before its NULL check, so a NULL backend crashed instead of reporting AKERR_NULLPOINTER. - akgl_game_save transposed CLEANUP and PROCESS, which placed the fclose inside the PROCESS switch. An ordinary save never flushed or closed its stream and produced an empty file. - akgl_game_save_actors wrote each name table terminator from the address of a single char, emitting stack contents into the save file and a sentinel the loader could not recognize. - akgl_game_load_objectnamemap used CATCH directly inside while(1), where the break leaves the loop rather than propagating, so a truncated name table loaded as a successful game. - akgl_Actor_cmhf_up_on and _down_on dereferenced actor->basechar with no NULL check, unlike their left and right counterparts. - akgl_actor_logic_movement checked actor twice instead of checking actor->basechar before dereferencing it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-30 02:03:21 -04:00
TEST_ASSERT(e, AKGL_BITMASK_HAS(actor.state, AKGL_ACTOR_STATE_MOVING_LEFT),
"left on did not set MOVING_LEFT (state %d)", actor.state);
TEST_ASSERT(e, AKGL_BITMASK_HAS(actor.state, AKGL_ACTOR_STATE_FACE_LEFT),
"left on did not set FACE_LEFT (state %d)", actor.state);
TEST_ASSERT(e, AKGL_BITMASK_HASNOT(actor.state, AKGL_ACTOR_STATE_MOVING_RIGHT),
"left on left MOVING_RIGHT set (state %d)", actor.state);
TEST_ASSERT(e, AKGL_BITMASK_HASNOT(actor.state, AKGL_ACTOR_STATE_FACE_RIGHT),
"left on left FACE_RIGHT set (state %d)", actor.state);
TEST_ASSERT(e, AKGL_BITMASK_HAS(actor.state, AKGL_ACTOR_STATE_ALIVE),
"left on cleared an unrelated state bit (state %d)", actor.state);
TEST_ASSERT_FEQ(e, actor.ax, -7.0f, "left on set ax to %f, expected -7", actor.ax);
// Right: same, with the sign preserved.
memset(&actor, 0x00, sizeof(akgl_Actor));
actor.basechar = &basechar;
Namespace every exported symbol, and bump to 0.5.0 Closes internal-consistency items 1 through 6, 12 and 13. Every include guard is _AKGL_<FILE>_H_, every in-project header include is angled, and every exported function, type and global carries the akgl_ prefix. This is an ABI break; the soname goes to libakgl.so.0.5. TODO.md carries the full rename table. The renames were driven by renaming each declaration and letting the compiler find the uses, not by pattern substitution: renderer, physics and camera are also parameter and struct-member names, and a sed would have rewritten map->physics and every akgl_RenderBackend *renderer parameter without a word. Item 4 turned out not to be cosmetic. The library exported a global called renderer and tests/character.c defined an SDL_Renderer *renderer of its own; the executable's definition preempted the library's, akgl_sprite_load_json read a SDL_Renderer * through an akgl_RenderBackend *, and every texture load in that suite failed. The suite reported success anyway, because libakerror's unhandled-error handler ends in exit(errctx->status), exit keeps only the low byte, and AKGL_ERR_SDL is exactly 256. So character had been green while running one of its four tests, and every suite in the tree was unable to fail on the most common status in a library built on SDL. Both are fixed. tests/testutil.h gains TEST_TRAP_UNHANDLED_ERRORS(), which collapses any status a byte cannot carry onto 1, and every suite installs it. character binds a real backend with akgl_render_2d_bind. Its fourth test then runs for the first time and fails on a defect it has asserted all along, so akgl_heap_release_character now walks state_sprites with AKGL_ITERATOR_OP_RELEASE and destroys the property set before zeroing the slot -- TODO.md Defects item 21 and half of Carried over item 1. AKGL_TIME_ONESEC_MS said "one second in milliseconds" and held 1000000, so akgl_game_state_lock waited roughly sixteen minutes rather than one second. It is AKGL_TIME_ONEMS_NS now, the budget is its own named constant, and tests/game.c holds the mutex from a second thread to assert the wait -- the contended path had no coverage at all. Headers are self-contained and it is enforced: AKGL_PUBLIC_HEADERS drives both install() and a generated translation unit per header, so a header that ships is a header that is checked. Writing that found registry.h, which used SDL_PropertiesID in eight declarations and included no SDL header. 23/23 suites pass, memcheck is clean, reindent --check is clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-31 23:32:21 -04:00
TEST_EXPECT_OK(e, akgl_actor_cmhf_right_on(&actor, &event), "right on");
Add physics, heap, json_helpers, game, and actor test suites Raise line coverage from 39.6 to 61.8 percent with four new suites and an extension to the actor suite, and register every suite through a single CMake list so a new test file cannot be left out of the coverage fixture. Give the test targets a build-tree RPATH and prepend the build tree to LD_LIBRARY_PATH for CTest, so a developer with a previously installed libakgl.so exercises the library that was just compiled. Fix six defects the new tests exposed: - akgl_physics_simulate read self->gravity_time before its NULL check, so a NULL backend crashed instead of reporting AKERR_NULLPOINTER. - akgl_game_save transposed CLEANUP and PROCESS, which placed the fclose inside the PROCESS switch. An ordinary save never flushed or closed its stream and produced an empty file. - akgl_game_save_actors wrote each name table terminator from the address of a single char, emitting stack contents into the save file and a sentinel the loader could not recognize. - akgl_game_load_objectnamemap used CATCH directly inside while(1), where the break leaves the loop rather than propagating, so a truncated name table loaded as a successful game. - akgl_Actor_cmhf_up_on and _down_on dereferenced actor->basechar with no NULL check, unlike their left and right counterparts. - akgl_actor_logic_movement checked actor twice instead of checking actor->basechar before dereferencing it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-30 02:03:21 -04:00
TEST_ASSERT(e, AKGL_BITMASK_HAS(actor.state, AKGL_ACTOR_STATE_MOVING_RIGHT),
"right on did not set MOVING_RIGHT (state %d)", actor.state);
TEST_ASSERT(e, AKGL_BITMASK_HAS(actor.state, AKGL_ACTOR_STATE_FACE_RIGHT),
"right on did not set FACE_RIGHT (state %d)", actor.state);
TEST_ASSERT_FEQ(e, actor.ax, 7.0f, "right on set ax to %f, expected 7", actor.ax);
// Up reverses the Y acceleration, because Y grows downward.
memset(&actor, 0x00, sizeof(akgl_Actor));
actor.basechar = &basechar;
Namespace every exported symbol, and bump to 0.5.0 Closes internal-consistency items 1 through 6, 12 and 13. Every include guard is _AKGL_<FILE>_H_, every in-project header include is angled, and every exported function, type and global carries the akgl_ prefix. This is an ABI break; the soname goes to libakgl.so.0.5. TODO.md carries the full rename table. The renames were driven by renaming each declaration and letting the compiler find the uses, not by pattern substitution: renderer, physics and camera are also parameter and struct-member names, and a sed would have rewritten map->physics and every akgl_RenderBackend *renderer parameter without a word. Item 4 turned out not to be cosmetic. The library exported a global called renderer and tests/character.c defined an SDL_Renderer *renderer of its own; the executable's definition preempted the library's, akgl_sprite_load_json read a SDL_Renderer * through an akgl_RenderBackend *, and every texture load in that suite failed. The suite reported success anyway, because libakerror's unhandled-error handler ends in exit(errctx->status), exit keeps only the low byte, and AKGL_ERR_SDL is exactly 256. So character had been green while running one of its four tests, and every suite in the tree was unable to fail on the most common status in a library built on SDL. Both are fixed. tests/testutil.h gains TEST_TRAP_UNHANDLED_ERRORS(), which collapses any status a byte cannot carry onto 1, and every suite installs it. character binds a real backend with akgl_render_2d_bind. Its fourth test then runs for the first time and fails on a defect it has asserted all along, so akgl_heap_release_character now walks state_sprites with AKGL_ITERATOR_OP_RELEASE and destroys the property set before zeroing the slot -- TODO.md Defects item 21 and half of Carried over item 1. AKGL_TIME_ONESEC_MS said "one second in milliseconds" and held 1000000, so akgl_game_state_lock waited roughly sixteen minutes rather than one second. It is AKGL_TIME_ONEMS_NS now, the budget is its own named constant, and tests/game.c holds the mutex from a second thread to assert the wait -- the contended path had no coverage at all. Headers are self-contained and it is enforced: AKGL_PUBLIC_HEADERS drives both install() and a generated translation unit per header, so a header that ships is a header that is checked. Writing that found registry.h, which used SDL_PropertiesID in eight declarations and included no SDL header. 23/23 suites pass, memcheck is clean, reindent --check is clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-31 23:32:21 -04:00
TEST_EXPECT_OK(e, akgl_actor_cmhf_up_on(&actor, &event), "up on");
Add physics, heap, json_helpers, game, and actor test suites Raise line coverage from 39.6 to 61.8 percent with four new suites and an extension to the actor suite, and register every suite through a single CMake list so a new test file cannot be left out of the coverage fixture. Give the test targets a build-tree RPATH and prepend the build tree to LD_LIBRARY_PATH for CTest, so a developer with a previously installed libakgl.so exercises the library that was just compiled. Fix six defects the new tests exposed: - akgl_physics_simulate read self->gravity_time before its NULL check, so a NULL backend crashed instead of reporting AKERR_NULLPOINTER. - akgl_game_save transposed CLEANUP and PROCESS, which placed the fclose inside the PROCESS switch. An ordinary save never flushed or closed its stream and produced an empty file. - akgl_game_save_actors wrote each name table terminator from the address of a single char, emitting stack contents into the save file and a sentinel the loader could not recognize. - akgl_game_load_objectnamemap used CATCH directly inside while(1), where the break leaves the loop rather than propagating, so a truncated name table loaded as a successful game. - akgl_Actor_cmhf_up_on and _down_on dereferenced actor->basechar with no NULL check, unlike their left and right counterparts. - akgl_actor_logic_movement checked actor twice instead of checking actor->basechar before dereferencing it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-30 02:03:21 -04:00
TEST_ASSERT(e, AKGL_BITMASK_HAS(actor.state, AKGL_ACTOR_STATE_MOVING_UP),
"up on did not set MOVING_UP (state %d)", actor.state);
TEST_ASSERT(e, AKGL_BITMASK_HAS(actor.state, AKGL_ACTOR_STATE_FACE_UP),
"up on did not set FACE_UP (state %d)", actor.state);
TEST_ASSERT_FEQ(e, actor.ay, -11.0f, "up on set ay to %f, expected -11", actor.ay);
memset(&actor, 0x00, sizeof(akgl_Actor));
actor.basechar = &basechar;
Namespace every exported symbol, and bump to 0.5.0 Closes internal-consistency items 1 through 6, 12 and 13. Every include guard is _AKGL_<FILE>_H_, every in-project header include is angled, and every exported function, type and global carries the akgl_ prefix. This is an ABI break; the soname goes to libakgl.so.0.5. TODO.md carries the full rename table. The renames were driven by renaming each declaration and letting the compiler find the uses, not by pattern substitution: renderer, physics and camera are also parameter and struct-member names, and a sed would have rewritten map->physics and every akgl_RenderBackend *renderer parameter without a word. Item 4 turned out not to be cosmetic. The library exported a global called renderer and tests/character.c defined an SDL_Renderer *renderer of its own; the executable's definition preempted the library's, akgl_sprite_load_json read a SDL_Renderer * through an akgl_RenderBackend *, and every texture load in that suite failed. The suite reported success anyway, because libakerror's unhandled-error handler ends in exit(errctx->status), exit keeps only the low byte, and AKGL_ERR_SDL is exactly 256. So character had been green while running one of its four tests, and every suite in the tree was unable to fail on the most common status in a library built on SDL. Both are fixed. tests/testutil.h gains TEST_TRAP_UNHANDLED_ERRORS(), which collapses any status a byte cannot carry onto 1, and every suite installs it. character binds a real backend with akgl_render_2d_bind. Its fourth test then runs for the first time and fails on a defect it has asserted all along, so akgl_heap_release_character now walks state_sprites with AKGL_ITERATOR_OP_RELEASE and destroys the property set before zeroing the slot -- TODO.md Defects item 21 and half of Carried over item 1. AKGL_TIME_ONESEC_MS said "one second in milliseconds" and held 1000000, so akgl_game_state_lock waited roughly sixteen minutes rather than one second. It is AKGL_TIME_ONEMS_NS now, the budget is its own named constant, and tests/game.c holds the mutex from a second thread to assert the wait -- the contended path had no coverage at all. Headers are self-contained and it is enforced: AKGL_PUBLIC_HEADERS drives both install() and a generated translation unit per header, so a header that ships is a header that is checked. Writing that found registry.h, which used SDL_PropertiesID in eight declarations and included no SDL header. 23/23 suites pass, memcheck is clean, reindent --check is clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-31 23:32:21 -04:00
TEST_EXPECT_OK(e, akgl_actor_cmhf_down_on(&actor, &event), "down on");
Add physics, heap, json_helpers, game, and actor test suites Raise line coverage from 39.6 to 61.8 percent with four new suites and an extension to the actor suite, and register every suite through a single CMake list so a new test file cannot be left out of the coverage fixture. Give the test targets a build-tree RPATH and prepend the build tree to LD_LIBRARY_PATH for CTest, so a developer with a previously installed libakgl.so exercises the library that was just compiled. Fix six defects the new tests exposed: - akgl_physics_simulate read self->gravity_time before its NULL check, so a NULL backend crashed instead of reporting AKERR_NULLPOINTER. - akgl_game_save transposed CLEANUP and PROCESS, which placed the fclose inside the PROCESS switch. An ordinary save never flushed or closed its stream and produced an empty file. - akgl_game_save_actors wrote each name table terminator from the address of a single char, emitting stack contents into the save file and a sentinel the loader could not recognize. - akgl_game_load_objectnamemap used CATCH directly inside while(1), where the break leaves the loop rather than propagating, so a truncated name table loaded as a successful game. - akgl_Actor_cmhf_up_on and _down_on dereferenced actor->basechar with no NULL check, unlike their left and right counterparts. - akgl_actor_logic_movement checked actor twice instead of checking actor->basechar before dereferencing it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-30 02:03:21 -04:00
TEST_ASSERT(e, AKGL_BITMASK_HAS(actor.state, AKGL_ACTOR_STATE_MOVING_DOWN),
"down on did not set MOVING_DOWN (state %d)", actor.state);
TEST_ASSERT(e, AKGL_BITMASK_HAS(actor.state, AKGL_ACTOR_STATE_FACE_DOWN),
"down on did not set FACE_DOWN (state %d)", actor.state);
TEST_ASSERT_FEQ(e, actor.ay, 11.0f, "down on set ay to %f, expected 11", actor.ay);
// Each direction is exclusive: turning right after left leaves only right.
memset(&actor, 0x00, sizeof(akgl_Actor));
actor.basechar = &basechar;
Namespace every exported symbol, and bump to 0.5.0 Closes internal-consistency items 1 through 6, 12 and 13. Every include guard is _AKGL_<FILE>_H_, every in-project header include is angled, and every exported function, type and global carries the akgl_ prefix. This is an ABI break; the soname goes to libakgl.so.0.5. TODO.md carries the full rename table. The renames were driven by renaming each declaration and letting the compiler find the uses, not by pattern substitution: renderer, physics and camera are also parameter and struct-member names, and a sed would have rewritten map->physics and every akgl_RenderBackend *renderer parameter without a word. Item 4 turned out not to be cosmetic. The library exported a global called renderer and tests/character.c defined an SDL_Renderer *renderer of its own; the executable's definition preempted the library's, akgl_sprite_load_json read a SDL_Renderer * through an akgl_RenderBackend *, and every texture load in that suite failed. The suite reported success anyway, because libakerror's unhandled-error handler ends in exit(errctx->status), exit keeps only the low byte, and AKGL_ERR_SDL is exactly 256. So character had been green while running one of its four tests, and every suite in the tree was unable to fail on the most common status in a library built on SDL. Both are fixed. tests/testutil.h gains TEST_TRAP_UNHANDLED_ERRORS(), which collapses any status a byte cannot carry onto 1, and every suite installs it. character binds a real backend with akgl_render_2d_bind. Its fourth test then runs for the first time and fails on a defect it has asserted all along, so akgl_heap_release_character now walks state_sprites with AKGL_ITERATOR_OP_RELEASE and destroys the property set before zeroing the slot -- TODO.md Defects item 21 and half of Carried over item 1. AKGL_TIME_ONESEC_MS said "one second in milliseconds" and held 1000000, so akgl_game_state_lock waited roughly sixteen minutes rather than one second. It is AKGL_TIME_ONEMS_NS now, the budget is its own named constant, and tests/game.c holds the mutex from a second thread to assert the wait -- the contended path had no coverage at all. Headers are self-contained and it is enforced: AKGL_PUBLIC_HEADERS drives both install() and a generated translation unit per header, so a header that ships is a header that is checked. Writing that found registry.h, which used SDL_PropertiesID in eight declarations and included no SDL header. 23/23 suites pass, memcheck is clean, reindent --check is clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-31 23:32:21 -04:00
TEST_EXPECT_OK(e, akgl_actor_cmhf_left_on(&actor, &event), "left on before turning");
TEST_EXPECT_OK(e, akgl_actor_cmhf_right_on(&actor, &event), "right on after left");
Add physics, heap, json_helpers, game, and actor test suites Raise line coverage from 39.6 to 61.8 percent with four new suites and an extension to the actor suite, and register every suite through a single CMake list so a new test file cannot be left out of the coverage fixture. Give the test targets a build-tree RPATH and prepend the build tree to LD_LIBRARY_PATH for CTest, so a developer with a previously installed libakgl.so exercises the library that was just compiled. Fix six defects the new tests exposed: - akgl_physics_simulate read self->gravity_time before its NULL check, so a NULL backend crashed instead of reporting AKERR_NULLPOINTER. - akgl_game_save transposed CLEANUP and PROCESS, which placed the fclose inside the PROCESS switch. An ordinary save never flushed or closed its stream and produced an empty file. - akgl_game_save_actors wrote each name table terminator from the address of a single char, emitting stack contents into the save file and a sentinel the loader could not recognize. - akgl_game_load_objectnamemap used CATCH directly inside while(1), where the break leaves the loop rather than propagating, so a truncated name table loaded as a successful game. - akgl_Actor_cmhf_up_on and _down_on dereferenced actor->basechar with no NULL check, unlike their left and right counterparts. - akgl_actor_logic_movement checked actor twice instead of checking actor->basechar before dereferencing it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-30 02:03:21 -04:00
TEST_ASSERT(e, AKGL_BITMASK_HASNOT(actor.state, AKGL_ACTOR_STATE_MOVING_LEFT),
"turning right left MOVING_LEFT set (state %d)", actor.state);
} CLEANUP {
} PROCESS(e) {
} FINISH(e, true);
SUCCEED_RETURN(e);
}
akerr_ErrorContext *test_actor_control_handlers_off(void)
{
PREPARE_ERROR(e);
akgl_Actor actor;
akgl_Character basechar;
SDL_Event event;
ATTEMPT {
memset(&event, 0x00, sizeof(SDL_Event));
memset(&basechar, 0x00, sizeof(akgl_Character));
basechar.ax = 7.0f;
basechar.ay = 11.0f;
Fix three arcade physics defects the unit tests could not see tests/physics_sim.c runs the arcade backend the way a game does -- a Mario-esque jump and fall, Zelda-style top-down walking, and a run reversed at full speed -- and prints what the actor actually did. tests/physics.c already checked that akgl_physics_simulate does the arithmetic it claims. Every one of these got past it. The first step was however long the level took to load. gravity_time was never initialized and dt was unbounded, so a 250 ms load produced a 250 ms step: 100 px of fall where a 60 Hz frame under the same gravity is 0.44 px, straight through whatever was underneath. Both initializers seed the clock, and simulate bounds dt to max_timestep -- a new physics.max_timestep property, default 0.05 s, read like gravity and drag. Zero disables the bound. The field replaces the dead timer_gravity, so the struct is the same size. Releasing a vertical key cancelled gravity. The _off handlers zeroed ay, ey, ty and vy together, and ey is where the arcade backend accumulates gravity -- tapping down mid-jump stopped the character in the air. They clear ax/tx (or ay/ty) and nothing else now. Velocity was never theirs to clear either: simulate recomputes v as e + t every step. Diagonal movement was 41% too fast. Thrust was capped per axis, so an actor holding two directions got both caps at once and travelled their diagonal. It is capped as a vector against the sx/sy/sz ellipse, which also keeps a character whose horizontal and vertical top speeds differ moving at the ratio it asked for. An axis with a zero max speed stays out of the magnitude and is forced to zero, as the old clamp did. Each fix was checked by reverting it and confirming the simulation goes red: 100.1 px, vy 0.0 after a down tap, and 141% diagonal respectively. tests/actor.c and tests/physics.c pinned the old behaviour in both places and now assert the new contract. Bumped to 0.6.0: akgl_PhysicsBackend changed. TODO.md records the four things the simulations found and this does not fix -- no terminal velocity, no deceleration on release, Euler's frame-rate dependence, and a drag coefficient large enough to invert velocity. 26/26 ctest, memcheck clean, warning-clean at -Wall -Werror. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01B8T5FAYXE8HEJqFLCYwNNc
2026-08-01 12:10:53 -04:00
// Releasing a direction stops the actor pushing: acceleration and thrust
// go to zero. It does *not* touch the environmental force on that axis.
//
// It used to zero `e` and `v` as well, and that is a real bug rather
// than a style difference: `ey` is where the arcade backend accumulates
// gravity, so tapping down while falling zeroed the fall and the
// character stopped dead in mid-air. `tests/physics_sim.c` measures it
// (`vertical release keeps gravity`). Velocity is not the actor's to
// clear either -- akgl_physics_simulate recomputes `vx` as `ex + tx`
// every step, so whatever a handler writes there is overwritten before
// anything can read it.
Add physics, heap, json_helpers, game, and actor test suites Raise line coverage from 39.6 to 61.8 percent with four new suites and an extension to the actor suite, and register every suite through a single CMake list so a new test file cannot be left out of the coverage fixture. Give the test targets a build-tree RPATH and prepend the build tree to LD_LIBRARY_PATH for CTest, so a developer with a previously installed libakgl.so exercises the library that was just compiled. Fix six defects the new tests exposed: - akgl_physics_simulate read self->gravity_time before its NULL check, so a NULL backend crashed instead of reporting AKERR_NULLPOINTER. - akgl_game_save transposed CLEANUP and PROCESS, which placed the fclose inside the PROCESS switch. An ordinary save never flushed or closed its stream and produced an empty file. - akgl_game_save_actors wrote each name table terminator from the address of a single char, emitting stack contents into the save file and a sentinel the loader could not recognize. - akgl_game_load_objectnamemap used CATCH directly inside while(1), where the break leaves the loop rather than propagating, so a truncated name table loaded as a successful game. - akgl_Actor_cmhf_up_on and _down_on dereferenced actor->basechar with no NULL check, unlike their left and right counterparts. - akgl_actor_logic_movement checked actor twice instead of checking actor->basechar before dereferencing it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-30 02:03:21 -04:00
memset(&actor, 0x00, sizeof(akgl_Actor));
actor.basechar = &basechar;
actor.ax = 5; actor.ex = 5; actor.tx = 5; actor.vx = 5;
actor.ay = 5; actor.ey = 5; actor.ty = 5; actor.vy = 5;
actor.state = (AKGL_ACTOR_STATE_MOVING_LEFT | AKGL_ACTOR_STATE_FACE_LEFT);
Namespace every exported symbol, and bump to 0.5.0 Closes internal-consistency items 1 through 6, 12 and 13. Every include guard is _AKGL_<FILE>_H_, every in-project header include is angled, and every exported function, type and global carries the akgl_ prefix. This is an ABI break; the soname goes to libakgl.so.0.5. TODO.md carries the full rename table. The renames were driven by renaming each declaration and letting the compiler find the uses, not by pattern substitution: renderer, physics and camera are also parameter and struct-member names, and a sed would have rewritten map->physics and every akgl_RenderBackend *renderer parameter without a word. Item 4 turned out not to be cosmetic. The library exported a global called renderer and tests/character.c defined an SDL_Renderer *renderer of its own; the executable's definition preempted the library's, akgl_sprite_load_json read a SDL_Renderer * through an akgl_RenderBackend *, and every texture load in that suite failed. The suite reported success anyway, because libakerror's unhandled-error handler ends in exit(errctx->status), exit keeps only the low byte, and AKGL_ERR_SDL is exactly 256. So character had been green while running one of its four tests, and every suite in the tree was unable to fail on the most common status in a library built on SDL. Both are fixed. tests/testutil.h gains TEST_TRAP_UNHANDLED_ERRORS(), which collapses any status a byte cannot carry onto 1, and every suite installs it. character binds a real backend with akgl_render_2d_bind. Its fourth test then runs for the first time and fails on a defect it has asserted all along, so akgl_heap_release_character now walks state_sprites with AKGL_ITERATOR_OP_RELEASE and destroys the property set before zeroing the slot -- TODO.md Defects item 21 and half of Carried over item 1. AKGL_TIME_ONESEC_MS said "one second in milliseconds" and held 1000000, so akgl_game_state_lock waited roughly sixteen minutes rather than one second. It is AKGL_TIME_ONEMS_NS now, the budget is its own named constant, and tests/game.c holds the mutex from a second thread to assert the wait -- the contended path had no coverage at all. Headers are self-contained and it is enforced: AKGL_PUBLIC_HEADERS drives both install() and a generated translation unit per header, so a header that ships is a header that is checked. Writing that found registry.h, which used SDL_PropertiesID in eight declarations and included no SDL header. 23/23 suites pass, memcheck is clean, reindent --check is clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-31 23:32:21 -04:00
TEST_EXPECT_OK(e, akgl_actor_cmhf_left_off(&actor, &event), "left off");
Add physics, heap, json_helpers, game, and actor test suites Raise line coverage from 39.6 to 61.8 percent with four new suites and an extension to the actor suite, and register every suite through a single CMake list so a new test file cannot be left out of the coverage fixture. Give the test targets a build-tree RPATH and prepend the build tree to LD_LIBRARY_PATH for CTest, so a developer with a previously installed libakgl.so exercises the library that was just compiled. Fix six defects the new tests exposed: - akgl_physics_simulate read self->gravity_time before its NULL check, so a NULL backend crashed instead of reporting AKERR_NULLPOINTER. - akgl_game_save transposed CLEANUP and PROCESS, which placed the fclose inside the PROCESS switch. An ordinary save never flushed or closed its stream and produced an empty file. - akgl_game_save_actors wrote each name table terminator from the address of a single char, emitting stack contents into the save file and a sentinel the loader could not recognize. - akgl_game_load_objectnamemap used CATCH directly inside while(1), where the break leaves the loop rather than propagating, so a truncated name table loaded as a successful game. - akgl_Actor_cmhf_up_on and _down_on dereferenced actor->basechar with no NULL check, unlike their left and right counterparts. - akgl_actor_logic_movement checked actor twice instead of checking actor->basechar before dereferencing it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-30 02:03:21 -04:00
TEST_ASSERT_FEQ(e, actor.ax, 0.0f, "left off left ax at %f", actor.ax);
TEST_ASSERT_FEQ(e, actor.tx, 0.0f, "left off left tx at %f", actor.tx);
Fix three arcade physics defects the unit tests could not see tests/physics_sim.c runs the arcade backend the way a game does -- a Mario-esque jump and fall, Zelda-style top-down walking, and a run reversed at full speed -- and prints what the actor actually did. tests/physics.c already checked that akgl_physics_simulate does the arithmetic it claims. Every one of these got past it. The first step was however long the level took to load. gravity_time was never initialized and dt was unbounded, so a 250 ms load produced a 250 ms step: 100 px of fall where a 60 Hz frame under the same gravity is 0.44 px, straight through whatever was underneath. Both initializers seed the clock, and simulate bounds dt to max_timestep -- a new physics.max_timestep property, default 0.05 s, read like gravity and drag. Zero disables the bound. The field replaces the dead timer_gravity, so the struct is the same size. Releasing a vertical key cancelled gravity. The _off handlers zeroed ay, ey, ty and vy together, and ey is where the arcade backend accumulates gravity -- tapping down mid-jump stopped the character in the air. They clear ax/tx (or ay/ty) and nothing else now. Velocity was never theirs to clear either: simulate recomputes v as e + t every step. Diagonal movement was 41% too fast. Thrust was capped per axis, so an actor holding two directions got both caps at once and travelled their diagonal. It is capped as a vector against the sx/sy/sz ellipse, which also keeps a character whose horizontal and vertical top speeds differ moving at the ratio it asked for. An axis with a zero max speed stays out of the magnitude and is forced to zero, as the old clamp did. Each fix was checked by reverting it and confirming the simulation goes red: 100.1 px, vy 0.0 after a down tap, and 141% diagonal respectively. tests/actor.c and tests/physics.c pinned the old behaviour in both places and now assert the new contract. Bumped to 0.6.0: akgl_PhysicsBackend changed. TODO.md records the four things the simulations found and this does not fix -- no terminal velocity, no deceleration on release, Euler's frame-rate dependence, and a drag coefficient large enough to invert velocity. 26/26 ctest, memcheck clean, warning-clean at -Wall -Werror. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01B8T5FAYXE8HEJqFLCYwNNc
2026-08-01 12:10:53 -04:00
TEST_ASSERT_FEQ(e, actor.ex, 5.0f, "left off cleared ex (%f), which is the world's, not the actor's", actor.ex);
Add physics, heap, json_helpers, game, and actor test suites Raise line coverage from 39.6 to 61.8 percent with four new suites and an extension to the actor suite, and register every suite through a single CMake list so a new test file cannot be left out of the coverage fixture. Give the test targets a build-tree RPATH and prepend the build tree to LD_LIBRARY_PATH for CTest, so a developer with a previously installed libakgl.so exercises the library that was just compiled. Fix six defects the new tests exposed: - akgl_physics_simulate read self->gravity_time before its NULL check, so a NULL backend crashed instead of reporting AKERR_NULLPOINTER. - akgl_game_save transposed CLEANUP and PROCESS, which placed the fclose inside the PROCESS switch. An ordinary save never flushed or closed its stream and produced an empty file. - akgl_game_save_actors wrote each name table terminator from the address of a single char, emitting stack contents into the save file and a sentinel the loader could not recognize. - akgl_game_load_objectnamemap used CATCH directly inside while(1), where the break leaves the loop rather than propagating, so a truncated name table loaded as a successful game. - akgl_Actor_cmhf_up_on and _down_on dereferenced actor->basechar with no NULL check, unlike their left and right counterparts. - akgl_actor_logic_movement checked actor twice instead of checking actor->basechar before dereferencing it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-30 02:03:21 -04:00
TEST_ASSERT(e, AKGL_BITMASK_HASNOT(actor.state, AKGL_ACTOR_STATE_MOVING_LEFT),
"left off did not clear MOVING_LEFT (state %d)", actor.state);
// Facing is deliberately sticky, so an idle actor keeps looking where it was.
TEST_ASSERT(e, AKGL_BITMASK_HAS(actor.state, AKGL_ACTOR_STATE_FACE_LEFT),
"left off cleared FACE_LEFT, which should persist (state %d)", actor.state);
// The Y axis is untouched by a horizontal release.
Fix three arcade physics defects the unit tests could not see tests/physics_sim.c runs the arcade backend the way a game does -- a Mario-esque jump and fall, Zelda-style top-down walking, and a run reversed at full speed -- and prints what the actor actually did. tests/physics.c already checked that akgl_physics_simulate does the arithmetic it claims. Every one of these got past it. The first step was however long the level took to load. gravity_time was never initialized and dt was unbounded, so a 250 ms load produced a 250 ms step: 100 px of fall where a 60 Hz frame under the same gravity is 0.44 px, straight through whatever was underneath. Both initializers seed the clock, and simulate bounds dt to max_timestep -- a new physics.max_timestep property, default 0.05 s, read like gravity and drag. Zero disables the bound. The field replaces the dead timer_gravity, so the struct is the same size. Releasing a vertical key cancelled gravity. The _off handlers zeroed ay, ey, ty and vy together, and ey is where the arcade backend accumulates gravity -- tapping down mid-jump stopped the character in the air. They clear ax/tx (or ay/ty) and nothing else now. Velocity was never theirs to clear either: simulate recomputes v as e + t every step. Diagonal movement was 41% too fast. Thrust was capped per axis, so an actor holding two directions got both caps at once and travelled their diagonal. It is capped as a vector against the sx/sy/sz ellipse, which also keeps a character whose horizontal and vertical top speeds differ moving at the ratio it asked for. An axis with a zero max speed stays out of the magnitude and is forced to zero, as the old clamp did. Each fix was checked by reverting it and confirming the simulation goes red: 100.1 px, vy 0.0 after a down tap, and 141% diagonal respectively. tests/actor.c and tests/physics.c pinned the old behaviour in both places and now assert the new contract. Bumped to 0.6.0: akgl_PhysicsBackend changed. TODO.md records the four things the simulations found and this does not fix -- no terminal velocity, no deceleration on release, Euler's frame-rate dependence, and a drag coefficient large enough to invert velocity. 26/26 ctest, memcheck clean, warning-clean at -Wall -Werror. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01B8T5FAYXE8HEJqFLCYwNNc
2026-08-01 12:10:53 -04:00
TEST_ASSERT_FEQ(e, actor.ay, 5.0f, "left off disturbed ay (%f)", actor.ay);
TEST_ASSERT_FEQ(e, actor.ey, 5.0f, "left off disturbed ey (%f)", actor.ey);
TEST_ASSERT_FEQ(e, actor.ty, 5.0f, "left off disturbed ty (%f)", actor.ty);
Add physics, heap, json_helpers, game, and actor test suites Raise line coverage from 39.6 to 61.8 percent with four new suites and an extension to the actor suite, and register every suite through a single CMake list so a new test file cannot be left out of the coverage fixture. Give the test targets a build-tree RPATH and prepend the build tree to LD_LIBRARY_PATH for CTest, so a developer with a previously installed libakgl.so exercises the library that was just compiled. Fix six defects the new tests exposed: - akgl_physics_simulate read self->gravity_time before its NULL check, so a NULL backend crashed instead of reporting AKERR_NULLPOINTER. - akgl_game_save transposed CLEANUP and PROCESS, which placed the fclose inside the PROCESS switch. An ordinary save never flushed or closed its stream and produced an empty file. - akgl_game_save_actors wrote each name table terminator from the address of a single char, emitting stack contents into the save file and a sentinel the loader could not recognize. - akgl_game_load_objectnamemap used CATCH directly inside while(1), where the break leaves the loop rather than propagating, so a truncated name table loaded as a successful game. - akgl_Actor_cmhf_up_on and _down_on dereferenced actor->basechar with no NULL check, unlike their left and right counterparts. - akgl_actor_logic_movement checked actor twice instead of checking actor->basechar before dereferencing it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-30 02:03:21 -04:00
memset(&actor, 0x00, sizeof(akgl_Actor));
actor.ax = 5; actor.ex = 5; actor.tx = 5; actor.vx = 5;
actor.state = AKGL_ACTOR_STATE_MOVING_RIGHT;
Namespace every exported symbol, and bump to 0.5.0 Closes internal-consistency items 1 through 6, 12 and 13. Every include guard is _AKGL_<FILE>_H_, every in-project header include is angled, and every exported function, type and global carries the akgl_ prefix. This is an ABI break; the soname goes to libakgl.so.0.5. TODO.md carries the full rename table. The renames were driven by renaming each declaration and letting the compiler find the uses, not by pattern substitution: renderer, physics and camera are also parameter and struct-member names, and a sed would have rewritten map->physics and every akgl_RenderBackend *renderer parameter without a word. Item 4 turned out not to be cosmetic. The library exported a global called renderer and tests/character.c defined an SDL_Renderer *renderer of its own; the executable's definition preempted the library's, akgl_sprite_load_json read a SDL_Renderer * through an akgl_RenderBackend *, and every texture load in that suite failed. The suite reported success anyway, because libakerror's unhandled-error handler ends in exit(errctx->status), exit keeps only the low byte, and AKGL_ERR_SDL is exactly 256. So character had been green while running one of its four tests, and every suite in the tree was unable to fail on the most common status in a library built on SDL. Both are fixed. tests/testutil.h gains TEST_TRAP_UNHANDLED_ERRORS(), which collapses any status a byte cannot carry onto 1, and every suite installs it. character binds a real backend with akgl_render_2d_bind. Its fourth test then runs for the first time and fails on a defect it has asserted all along, so akgl_heap_release_character now walks state_sprites with AKGL_ITERATOR_OP_RELEASE and destroys the property set before zeroing the slot -- TODO.md Defects item 21 and half of Carried over item 1. AKGL_TIME_ONESEC_MS said "one second in milliseconds" and held 1000000, so akgl_game_state_lock waited roughly sixteen minutes rather than one second. It is AKGL_TIME_ONEMS_NS now, the budget is its own named constant, and tests/game.c holds the mutex from a second thread to assert the wait -- the contended path had no coverage at all. Headers are self-contained and it is enforced: AKGL_PUBLIC_HEADERS drives both install() and a generated translation unit per header, so a header that ships is a header that is checked. Writing that found registry.h, which used SDL_PropertiesID in eight declarations and included no SDL header. 23/23 suites pass, memcheck is clean, reindent --check is clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-31 23:32:21 -04:00
TEST_EXPECT_OK(e, akgl_actor_cmhf_right_off(&actor, &event), "right off");
Fix three arcade physics defects the unit tests could not see tests/physics_sim.c runs the arcade backend the way a game does -- a Mario-esque jump and fall, Zelda-style top-down walking, and a run reversed at full speed -- and prints what the actor actually did. tests/physics.c already checked that akgl_physics_simulate does the arithmetic it claims. Every one of these got past it. The first step was however long the level took to load. gravity_time was never initialized and dt was unbounded, so a 250 ms load produced a 250 ms step: 100 px of fall where a 60 Hz frame under the same gravity is 0.44 px, straight through whatever was underneath. Both initializers seed the clock, and simulate bounds dt to max_timestep -- a new physics.max_timestep property, default 0.05 s, read like gravity and drag. Zero disables the bound. The field replaces the dead timer_gravity, so the struct is the same size. Releasing a vertical key cancelled gravity. The _off handlers zeroed ay, ey, ty and vy together, and ey is where the arcade backend accumulates gravity -- tapping down mid-jump stopped the character in the air. They clear ax/tx (or ay/ty) and nothing else now. Velocity was never theirs to clear either: simulate recomputes v as e + t every step. Diagonal movement was 41% too fast. Thrust was capped per axis, so an actor holding two directions got both caps at once and travelled their diagonal. It is capped as a vector against the sx/sy/sz ellipse, which also keeps a character whose horizontal and vertical top speeds differ moving at the ratio it asked for. An axis with a zero max speed stays out of the magnitude and is forced to zero, as the old clamp did. Each fix was checked by reverting it and confirming the simulation goes red: 100.1 px, vy 0.0 after a down tap, and 141% diagonal respectively. tests/actor.c and tests/physics.c pinned the old behaviour in both places and now assert the new contract. Bumped to 0.6.0: akgl_PhysicsBackend changed. TODO.md records the four things the simulations found and this does not fix -- no terminal velocity, no deceleration on release, Euler's frame-rate dependence, and a drag coefficient large enough to invert velocity. 26/26 ctest, memcheck clean, warning-clean at -Wall -Werror. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01B8T5FAYXE8HEJqFLCYwNNc
2026-08-01 12:10:53 -04:00
TEST_ASSERT_FEQ(e, actor.ax, 0.0f, "right off left ax at %f", actor.ax);
TEST_ASSERT_FEQ(e, actor.tx, 0.0f, "right off left tx at %f", actor.tx);
TEST_ASSERT_FEQ(e, actor.ex, 5.0f, "right off cleared ex (%f)", actor.ex);
Add physics, heap, json_helpers, game, and actor test suites Raise line coverage from 39.6 to 61.8 percent with four new suites and an extension to the actor suite, and register every suite through a single CMake list so a new test file cannot be left out of the coverage fixture. Give the test targets a build-tree RPATH and prepend the build tree to LD_LIBRARY_PATH for CTest, so a developer with a previously installed libakgl.so exercises the library that was just compiled. Fix six defects the new tests exposed: - akgl_physics_simulate read self->gravity_time before its NULL check, so a NULL backend crashed instead of reporting AKERR_NULLPOINTER. - akgl_game_save transposed CLEANUP and PROCESS, which placed the fclose inside the PROCESS switch. An ordinary save never flushed or closed its stream and produced an empty file. - akgl_game_save_actors wrote each name table terminator from the address of a single char, emitting stack contents into the save file and a sentinel the loader could not recognize. - akgl_game_load_objectnamemap used CATCH directly inside while(1), where the break leaves the loop rather than propagating, so a truncated name table loaded as a successful game. - akgl_Actor_cmhf_up_on and _down_on dereferenced actor->basechar with no NULL check, unlike their left and right counterparts. - akgl_actor_logic_movement checked actor twice instead of checking actor->basechar before dereferencing it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-30 02:03:21 -04:00
TEST_ASSERT(e, AKGL_BITMASK_HASNOT(actor.state, AKGL_ACTOR_STATE_MOVING_RIGHT),
"right off did not clear MOVING_RIGHT (state %d)", actor.state);
memset(&actor, 0x00, sizeof(akgl_Actor));
actor.ay = 5; actor.ey = 5; actor.ty = 5; actor.vy = 5;
actor.state = AKGL_ACTOR_STATE_MOVING_UP;
Namespace every exported symbol, and bump to 0.5.0 Closes internal-consistency items 1 through 6, 12 and 13. Every include guard is _AKGL_<FILE>_H_, every in-project header include is angled, and every exported function, type and global carries the akgl_ prefix. This is an ABI break; the soname goes to libakgl.so.0.5. TODO.md carries the full rename table. The renames were driven by renaming each declaration and letting the compiler find the uses, not by pattern substitution: renderer, physics and camera are also parameter and struct-member names, and a sed would have rewritten map->physics and every akgl_RenderBackend *renderer parameter without a word. Item 4 turned out not to be cosmetic. The library exported a global called renderer and tests/character.c defined an SDL_Renderer *renderer of its own; the executable's definition preempted the library's, akgl_sprite_load_json read a SDL_Renderer * through an akgl_RenderBackend *, and every texture load in that suite failed. The suite reported success anyway, because libakerror's unhandled-error handler ends in exit(errctx->status), exit keeps only the low byte, and AKGL_ERR_SDL is exactly 256. So character had been green while running one of its four tests, and every suite in the tree was unable to fail on the most common status in a library built on SDL. Both are fixed. tests/testutil.h gains TEST_TRAP_UNHANDLED_ERRORS(), which collapses any status a byte cannot carry onto 1, and every suite installs it. character binds a real backend with akgl_render_2d_bind. Its fourth test then runs for the first time and fails on a defect it has asserted all along, so akgl_heap_release_character now walks state_sprites with AKGL_ITERATOR_OP_RELEASE and destroys the property set before zeroing the slot -- TODO.md Defects item 21 and half of Carried over item 1. AKGL_TIME_ONESEC_MS said "one second in milliseconds" and held 1000000, so akgl_game_state_lock waited roughly sixteen minutes rather than one second. It is AKGL_TIME_ONEMS_NS now, the budget is its own named constant, and tests/game.c holds the mutex from a second thread to assert the wait -- the contended path had no coverage at all. Headers are self-contained and it is enforced: AKGL_PUBLIC_HEADERS drives both install() and a generated translation unit per header, so a header that ships is a header that is checked. Writing that found registry.h, which used SDL_PropertiesID in eight declarations and included no SDL header. 23/23 suites pass, memcheck is clean, reindent --check is clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-31 23:32:21 -04:00
TEST_EXPECT_OK(e, akgl_actor_cmhf_up_off(&actor, &event), "up off");
Fix three arcade physics defects the unit tests could not see tests/physics_sim.c runs the arcade backend the way a game does -- a Mario-esque jump and fall, Zelda-style top-down walking, and a run reversed at full speed -- and prints what the actor actually did. tests/physics.c already checked that akgl_physics_simulate does the arithmetic it claims. Every one of these got past it. The first step was however long the level took to load. gravity_time was never initialized and dt was unbounded, so a 250 ms load produced a 250 ms step: 100 px of fall where a 60 Hz frame under the same gravity is 0.44 px, straight through whatever was underneath. Both initializers seed the clock, and simulate bounds dt to max_timestep -- a new physics.max_timestep property, default 0.05 s, read like gravity and drag. Zero disables the bound. The field replaces the dead timer_gravity, so the struct is the same size. Releasing a vertical key cancelled gravity. The _off handlers zeroed ay, ey, ty and vy together, and ey is where the arcade backend accumulates gravity -- tapping down mid-jump stopped the character in the air. They clear ax/tx (or ay/ty) and nothing else now. Velocity was never theirs to clear either: simulate recomputes v as e + t every step. Diagonal movement was 41% too fast. Thrust was capped per axis, so an actor holding two directions got both caps at once and travelled their diagonal. It is capped as a vector against the sx/sy/sz ellipse, which also keeps a character whose horizontal and vertical top speeds differ moving at the ratio it asked for. An axis with a zero max speed stays out of the magnitude and is forced to zero, as the old clamp did. Each fix was checked by reverting it and confirming the simulation goes red: 100.1 px, vy 0.0 after a down tap, and 141% diagonal respectively. tests/actor.c and tests/physics.c pinned the old behaviour in both places and now assert the new contract. Bumped to 0.6.0: akgl_PhysicsBackend changed. TODO.md records the four things the simulations found and this does not fix -- no terminal velocity, no deceleration on release, Euler's frame-rate dependence, and a drag coefficient large enough to invert velocity. 26/26 ctest, memcheck clean, warning-clean at -Wall -Werror. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01B8T5FAYXE8HEJqFLCYwNNc
2026-08-01 12:10:53 -04:00
TEST_ASSERT_FEQ(e, actor.ay, 0.0f, "up off left ay at %f", actor.ay);
TEST_ASSERT_FEQ(e, actor.ty, 0.0f, "up off left ty at %f", actor.ty);
TEST_ASSERT_FEQ(e, actor.ey, 5.0f, "up off cleared ey (%f), which is where gravity accumulates", actor.ey);
Add physics, heap, json_helpers, game, and actor test suites Raise line coverage from 39.6 to 61.8 percent with four new suites and an extension to the actor suite, and register every suite through a single CMake list so a new test file cannot be left out of the coverage fixture. Give the test targets a build-tree RPATH and prepend the build tree to LD_LIBRARY_PATH for CTest, so a developer with a previously installed libakgl.so exercises the library that was just compiled. Fix six defects the new tests exposed: - akgl_physics_simulate read self->gravity_time before its NULL check, so a NULL backend crashed instead of reporting AKERR_NULLPOINTER. - akgl_game_save transposed CLEANUP and PROCESS, which placed the fclose inside the PROCESS switch. An ordinary save never flushed or closed its stream and produced an empty file. - akgl_game_save_actors wrote each name table terminator from the address of a single char, emitting stack contents into the save file and a sentinel the loader could not recognize. - akgl_game_load_objectnamemap used CATCH directly inside while(1), where the break leaves the loop rather than propagating, so a truncated name table loaded as a successful game. - akgl_Actor_cmhf_up_on and _down_on dereferenced actor->basechar with no NULL check, unlike their left and right counterparts. - akgl_actor_logic_movement checked actor twice instead of checking actor->basechar before dereferencing it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-30 02:03:21 -04:00
TEST_ASSERT(e, AKGL_BITMASK_HASNOT(actor.state, AKGL_ACTOR_STATE_MOVING_UP),
"up off did not clear MOVING_UP (state %d)", actor.state);
memset(&actor, 0x00, sizeof(akgl_Actor));
actor.ay = 5; actor.ey = 5; actor.ty = 5; actor.vy = 5;
actor.state = AKGL_ACTOR_STATE_MOVING_DOWN;
Namespace every exported symbol, and bump to 0.5.0 Closes internal-consistency items 1 through 6, 12 and 13. Every include guard is _AKGL_<FILE>_H_, every in-project header include is angled, and every exported function, type and global carries the akgl_ prefix. This is an ABI break; the soname goes to libakgl.so.0.5. TODO.md carries the full rename table. The renames were driven by renaming each declaration and letting the compiler find the uses, not by pattern substitution: renderer, physics and camera are also parameter and struct-member names, and a sed would have rewritten map->physics and every akgl_RenderBackend *renderer parameter without a word. Item 4 turned out not to be cosmetic. The library exported a global called renderer and tests/character.c defined an SDL_Renderer *renderer of its own; the executable's definition preempted the library's, akgl_sprite_load_json read a SDL_Renderer * through an akgl_RenderBackend *, and every texture load in that suite failed. The suite reported success anyway, because libakerror's unhandled-error handler ends in exit(errctx->status), exit keeps only the low byte, and AKGL_ERR_SDL is exactly 256. So character had been green while running one of its four tests, and every suite in the tree was unable to fail on the most common status in a library built on SDL. Both are fixed. tests/testutil.h gains TEST_TRAP_UNHANDLED_ERRORS(), which collapses any status a byte cannot carry onto 1, and every suite installs it. character binds a real backend with akgl_render_2d_bind. Its fourth test then runs for the first time and fails on a defect it has asserted all along, so akgl_heap_release_character now walks state_sprites with AKGL_ITERATOR_OP_RELEASE and destroys the property set before zeroing the slot -- TODO.md Defects item 21 and half of Carried over item 1. AKGL_TIME_ONESEC_MS said "one second in milliseconds" and held 1000000, so akgl_game_state_lock waited roughly sixteen minutes rather than one second. It is AKGL_TIME_ONEMS_NS now, the budget is its own named constant, and tests/game.c holds the mutex from a second thread to assert the wait -- the contended path had no coverage at all. Headers are self-contained and it is enforced: AKGL_PUBLIC_HEADERS drives both install() and a generated translation unit per header, so a header that ships is a header that is checked. Writing that found registry.h, which used SDL_PropertiesID in eight declarations and included no SDL header. 23/23 suites pass, memcheck is clean, reindent --check is clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-31 23:32:21 -04:00
TEST_EXPECT_OK(e, akgl_actor_cmhf_down_off(&actor, &event), "down off");
Fix three arcade physics defects the unit tests could not see tests/physics_sim.c runs the arcade backend the way a game does -- a Mario-esque jump and fall, Zelda-style top-down walking, and a run reversed at full speed -- and prints what the actor actually did. tests/physics.c already checked that akgl_physics_simulate does the arithmetic it claims. Every one of these got past it. The first step was however long the level took to load. gravity_time was never initialized and dt was unbounded, so a 250 ms load produced a 250 ms step: 100 px of fall where a 60 Hz frame under the same gravity is 0.44 px, straight through whatever was underneath. Both initializers seed the clock, and simulate bounds dt to max_timestep -- a new physics.max_timestep property, default 0.05 s, read like gravity and drag. Zero disables the bound. The field replaces the dead timer_gravity, so the struct is the same size. Releasing a vertical key cancelled gravity. The _off handlers zeroed ay, ey, ty and vy together, and ey is where the arcade backend accumulates gravity -- tapping down mid-jump stopped the character in the air. They clear ax/tx (or ay/ty) and nothing else now. Velocity was never theirs to clear either: simulate recomputes v as e + t every step. Diagonal movement was 41% too fast. Thrust was capped per axis, so an actor holding two directions got both caps at once and travelled their diagonal. It is capped as a vector against the sx/sy/sz ellipse, which also keeps a character whose horizontal and vertical top speeds differ moving at the ratio it asked for. An axis with a zero max speed stays out of the magnitude and is forced to zero, as the old clamp did. Each fix was checked by reverting it and confirming the simulation goes red: 100.1 px, vy 0.0 after a down tap, and 141% diagonal respectively. tests/actor.c and tests/physics.c pinned the old behaviour in both places and now assert the new contract. Bumped to 0.6.0: akgl_PhysicsBackend changed. TODO.md records the four things the simulations found and this does not fix -- no terminal velocity, no deceleration on release, Euler's frame-rate dependence, and a drag coefficient large enough to invert velocity. 26/26 ctest, memcheck clean, warning-clean at -Wall -Werror. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01B8T5FAYXE8HEJqFLCYwNNc
2026-08-01 12:10:53 -04:00
TEST_ASSERT_FEQ(e, actor.ay, 0.0f, "down off left ay at %f", actor.ay);
TEST_ASSERT_FEQ(e, actor.ty, 0.0f, "down off left ty at %f", actor.ty);
TEST_ASSERT_FEQ(e, actor.ey, 5.0f, "down off cleared ey (%f), which is where gravity accumulates", actor.ey);
Add physics, heap, json_helpers, game, and actor test suites Raise line coverage from 39.6 to 61.8 percent with four new suites and an extension to the actor suite, and register every suite through a single CMake list so a new test file cannot be left out of the coverage fixture. Give the test targets a build-tree RPATH and prepend the build tree to LD_LIBRARY_PATH for CTest, so a developer with a previously installed libakgl.so exercises the library that was just compiled. Fix six defects the new tests exposed: - akgl_physics_simulate read self->gravity_time before its NULL check, so a NULL backend crashed instead of reporting AKERR_NULLPOINTER. - akgl_game_save transposed CLEANUP and PROCESS, which placed the fclose inside the PROCESS switch. An ordinary save never flushed or closed its stream and produced an empty file. - akgl_game_save_actors wrote each name table terminator from the address of a single char, emitting stack contents into the save file and a sentinel the loader could not recognize. - akgl_game_load_objectnamemap used CATCH directly inside while(1), where the break leaves the loop rather than propagating, so a truncated name table loaded as a successful game. - akgl_Actor_cmhf_up_on and _down_on dereferenced actor->basechar with no NULL check, unlike their left and right counterparts. - akgl_actor_logic_movement checked actor twice instead of checking actor->basechar before dereferencing it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-30 02:03:21 -04:00
TEST_ASSERT(e, AKGL_BITMASK_HASNOT(actor.state, AKGL_ACTOR_STATE_MOVING_DOWN),
"down off did not clear MOVING_DOWN (state %d)", actor.state);
} CLEANUP {
} PROCESS(e) {
} FINISH(e, true);
SUCCEED_RETURN(e);
}
akerr_ErrorContext *test_actor_control_handlers_nullpointers(void)
{
PREPARE_ERROR(e);
akgl_Actor actor;
SDL_Event event;
ATTEMPT {
memset(&event, 0x00, sizeof(SDL_Event));
memset(&actor, 0x00, sizeof(akgl_Actor));
Namespace every exported symbol, and bump to 0.5.0 Closes internal-consistency items 1 through 6, 12 and 13. Every include guard is _AKGL_<FILE>_H_, every in-project header include is angled, and every exported function, type and global carries the akgl_ prefix. This is an ABI break; the soname goes to libakgl.so.0.5. TODO.md carries the full rename table. The renames were driven by renaming each declaration and letting the compiler find the uses, not by pattern substitution: renderer, physics and camera are also parameter and struct-member names, and a sed would have rewritten map->physics and every akgl_RenderBackend *renderer parameter without a word. Item 4 turned out not to be cosmetic. The library exported a global called renderer and tests/character.c defined an SDL_Renderer *renderer of its own; the executable's definition preempted the library's, akgl_sprite_load_json read a SDL_Renderer * through an akgl_RenderBackend *, and every texture load in that suite failed. The suite reported success anyway, because libakerror's unhandled-error handler ends in exit(errctx->status), exit keeps only the low byte, and AKGL_ERR_SDL is exactly 256. So character had been green while running one of its four tests, and every suite in the tree was unable to fail on the most common status in a library built on SDL. Both are fixed. tests/testutil.h gains TEST_TRAP_UNHANDLED_ERRORS(), which collapses any status a byte cannot carry onto 1, and every suite installs it. character binds a real backend with akgl_render_2d_bind. Its fourth test then runs for the first time and fails on a defect it has asserted all along, so akgl_heap_release_character now walks state_sprites with AKGL_ITERATOR_OP_RELEASE and destroys the property set before zeroing the slot -- TODO.md Defects item 21 and half of Carried over item 1. AKGL_TIME_ONESEC_MS said "one second in milliseconds" and held 1000000, so akgl_game_state_lock waited roughly sixteen minutes rather than one second. It is AKGL_TIME_ONEMS_NS now, the budget is its own named constant, and tests/game.c holds the mutex from a second thread to assert the wait -- the contended path had no coverage at all. Headers are self-contained and it is enforced: AKGL_PUBLIC_HEADERS drives both install() and a generated translation unit per header, so a header that ships is a header that is checked. Writing that found registry.h, which used SDL_PropertiesID in eight declarations and included no SDL header. 23/23 suites pass, memcheck is clean, reindent --check is clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-31 23:32:21 -04:00
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_actor_cmhf_left_on(NULL, &event), "left on, NULL actor");
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_actor_cmhf_left_off(NULL, &event), "left off, NULL actor");
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_actor_cmhf_right_on(NULL, &event), "right on, NULL actor");
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_actor_cmhf_right_off(NULL, &event), "right off, NULL actor");
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_actor_cmhf_up_on(NULL, &event), "up on, NULL actor");
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_actor_cmhf_up_off(NULL, &event), "up off, NULL actor");
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_actor_cmhf_down_on(NULL, &event), "down on, NULL actor");
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_actor_cmhf_down_off(NULL, &event), "down off, NULL actor");
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_actor_cmhf_left_on(&actor, NULL), "left on, NULL event");
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_actor_cmhf_left_off(&actor, NULL), "left off, NULL event");
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_actor_cmhf_right_on(&actor, NULL), "right on, NULL event");
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_actor_cmhf_right_off(&actor, NULL), "right off, NULL event");
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_actor_cmhf_up_on(&actor, NULL), "up on, NULL event");
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_actor_cmhf_up_off(&actor, NULL), "up off, NULL event");
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_actor_cmhf_down_on(&actor, NULL), "down on, NULL event");
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_actor_cmhf_down_off(&actor, NULL), "down off, NULL event");
Add physics, heap, json_helpers, game, and actor test suites Raise line coverage from 39.6 to 61.8 percent with four new suites and an extension to the actor suite, and register every suite through a single CMake list so a new test file cannot be left out of the coverage fixture. Give the test targets a build-tree RPATH and prepend the build tree to LD_LIBRARY_PATH for CTest, so a developer with a previously installed libakgl.so exercises the library that was just compiled. Fix six defects the new tests exposed: - akgl_physics_simulate read self->gravity_time before its NULL check, so a NULL backend crashed instead of reporting AKERR_NULLPOINTER. - akgl_game_save transposed CLEANUP and PROCESS, which placed the fclose inside the PROCESS switch. An ordinary save never flushed or closed its stream and produced an empty file. - akgl_game_save_actors wrote each name table terminator from the address of a single char, emitting stack contents into the save file and a sentinel the loader could not recognize. - akgl_game_load_objectnamemap used CATCH directly inside while(1), where the break leaves the loop rather than propagating, so a truncated name table loaded as a successful game. - akgl_Actor_cmhf_up_on and _down_on dereferenced actor->basechar with no NULL check, unlike their left and right counterparts. - akgl_actor_logic_movement checked actor twice instead of checking actor->basechar before dereferencing it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-30 02:03:21 -04:00
// A movement handler reads acceleration off the base character, so an
// actor without one has to be reported rather than dereferenced.
Namespace every exported symbol, and bump to 0.5.0 Closes internal-consistency items 1 through 6, 12 and 13. Every include guard is _AKGL_<FILE>_H_, every in-project header include is angled, and every exported function, type and global carries the akgl_ prefix. This is an ABI break; the soname goes to libakgl.so.0.5. TODO.md carries the full rename table. The renames were driven by renaming each declaration and letting the compiler find the uses, not by pattern substitution: renderer, physics and camera are also parameter and struct-member names, and a sed would have rewritten map->physics and every akgl_RenderBackend *renderer parameter without a word. Item 4 turned out not to be cosmetic. The library exported a global called renderer and tests/character.c defined an SDL_Renderer *renderer of its own; the executable's definition preempted the library's, akgl_sprite_load_json read a SDL_Renderer * through an akgl_RenderBackend *, and every texture load in that suite failed. The suite reported success anyway, because libakerror's unhandled-error handler ends in exit(errctx->status), exit keeps only the low byte, and AKGL_ERR_SDL is exactly 256. So character had been green while running one of its four tests, and every suite in the tree was unable to fail on the most common status in a library built on SDL. Both are fixed. tests/testutil.h gains TEST_TRAP_UNHANDLED_ERRORS(), which collapses any status a byte cannot carry onto 1, and every suite installs it. character binds a real backend with akgl_render_2d_bind. Its fourth test then runs for the first time and fails on a defect it has asserted all along, so akgl_heap_release_character now walks state_sprites with AKGL_ITERATOR_OP_RELEASE and destroys the property set before zeroing the slot -- TODO.md Defects item 21 and half of Carried over item 1. AKGL_TIME_ONESEC_MS said "one second in milliseconds" and held 1000000, so akgl_game_state_lock waited roughly sixteen minutes rather than one second. It is AKGL_TIME_ONEMS_NS now, the budget is its own named constant, and tests/game.c holds the mutex from a second thread to assert the wait -- the contended path had no coverage at all. Headers are self-contained and it is enforced: AKGL_PUBLIC_HEADERS drives both install() and a generated translation unit per header, so a header that ships is a header that is checked. Writing that found registry.h, which used SDL_PropertiesID in eight declarations and included no SDL header. 23/23 suites pass, memcheck is clean, reindent --check is clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-31 23:32:21 -04:00
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_actor_cmhf_left_on(&actor, &event),
Add physics, heap, json_helpers, game, and actor test suites Raise line coverage from 39.6 to 61.8 percent with four new suites and an extension to the actor suite, and register every suite through a single CMake list so a new test file cannot be left out of the coverage fixture. Give the test targets a build-tree RPATH and prepend the build tree to LD_LIBRARY_PATH for CTest, so a developer with a previously installed libakgl.so exercises the library that was just compiled. Fix six defects the new tests exposed: - akgl_physics_simulate read self->gravity_time before its NULL check, so a NULL backend crashed instead of reporting AKERR_NULLPOINTER. - akgl_game_save transposed CLEANUP and PROCESS, which placed the fclose inside the PROCESS switch. An ordinary save never flushed or closed its stream and produced an empty file. - akgl_game_save_actors wrote each name table terminator from the address of a single char, emitting stack contents into the save file and a sentinel the loader could not recognize. - akgl_game_load_objectnamemap used CATCH directly inside while(1), where the break leaves the loop rather than propagating, so a truncated name table loaded as a successful game. - akgl_Actor_cmhf_up_on and _down_on dereferenced actor->basechar with no NULL check, unlike their left and right counterparts. - akgl_actor_logic_movement checked actor twice instead of checking actor->basechar before dereferencing it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-30 02:03:21 -04:00
"left on, actor with no base character");
Namespace every exported symbol, and bump to 0.5.0 Closes internal-consistency items 1 through 6, 12 and 13. Every include guard is _AKGL_<FILE>_H_, every in-project header include is angled, and every exported function, type and global carries the akgl_ prefix. This is an ABI break; the soname goes to libakgl.so.0.5. TODO.md carries the full rename table. The renames were driven by renaming each declaration and letting the compiler find the uses, not by pattern substitution: renderer, physics and camera are also parameter and struct-member names, and a sed would have rewritten map->physics and every akgl_RenderBackend *renderer parameter without a word. Item 4 turned out not to be cosmetic. The library exported a global called renderer and tests/character.c defined an SDL_Renderer *renderer of its own; the executable's definition preempted the library's, akgl_sprite_load_json read a SDL_Renderer * through an akgl_RenderBackend *, and every texture load in that suite failed. The suite reported success anyway, because libakerror's unhandled-error handler ends in exit(errctx->status), exit keeps only the low byte, and AKGL_ERR_SDL is exactly 256. So character had been green while running one of its four tests, and every suite in the tree was unable to fail on the most common status in a library built on SDL. Both are fixed. tests/testutil.h gains TEST_TRAP_UNHANDLED_ERRORS(), which collapses any status a byte cannot carry onto 1, and every suite installs it. character binds a real backend with akgl_render_2d_bind. Its fourth test then runs for the first time and fails on a defect it has asserted all along, so akgl_heap_release_character now walks state_sprites with AKGL_ITERATOR_OP_RELEASE and destroys the property set before zeroing the slot -- TODO.md Defects item 21 and half of Carried over item 1. AKGL_TIME_ONESEC_MS said "one second in milliseconds" and held 1000000, so akgl_game_state_lock waited roughly sixteen minutes rather than one second. It is AKGL_TIME_ONEMS_NS now, the budget is its own named constant, and tests/game.c holds the mutex from a second thread to assert the wait -- the contended path had no coverage at all. Headers are self-contained and it is enforced: AKGL_PUBLIC_HEADERS drives both install() and a generated translation unit per header, so a header that ships is a header that is checked. Writing that found registry.h, which used SDL_PropertiesID in eight declarations and included no SDL header. 23/23 suites pass, memcheck is clean, reindent --check is clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-31 23:32:21 -04:00
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_actor_cmhf_right_on(&actor, &event),
Add physics, heap, json_helpers, game, and actor test suites Raise line coverage from 39.6 to 61.8 percent with four new suites and an extension to the actor suite, and register every suite through a single CMake list so a new test file cannot be left out of the coverage fixture. Give the test targets a build-tree RPATH and prepend the build tree to LD_LIBRARY_PATH for CTest, so a developer with a previously installed libakgl.so exercises the library that was just compiled. Fix six defects the new tests exposed: - akgl_physics_simulate read self->gravity_time before its NULL check, so a NULL backend crashed instead of reporting AKERR_NULLPOINTER. - akgl_game_save transposed CLEANUP and PROCESS, which placed the fclose inside the PROCESS switch. An ordinary save never flushed or closed its stream and produced an empty file. - akgl_game_save_actors wrote each name table terminator from the address of a single char, emitting stack contents into the save file and a sentinel the loader could not recognize. - akgl_game_load_objectnamemap used CATCH directly inside while(1), where the break leaves the loop rather than propagating, so a truncated name table loaded as a successful game. - akgl_Actor_cmhf_up_on and _down_on dereferenced actor->basechar with no NULL check, unlike their left and right counterparts. - akgl_actor_logic_movement checked actor twice instead of checking actor->basechar before dereferencing it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-30 02:03:21 -04:00
"right on, actor with no base character");
Namespace every exported symbol, and bump to 0.5.0 Closes internal-consistency items 1 through 6, 12 and 13. Every include guard is _AKGL_<FILE>_H_, every in-project header include is angled, and every exported function, type and global carries the akgl_ prefix. This is an ABI break; the soname goes to libakgl.so.0.5. TODO.md carries the full rename table. The renames were driven by renaming each declaration and letting the compiler find the uses, not by pattern substitution: renderer, physics and camera are also parameter and struct-member names, and a sed would have rewritten map->physics and every akgl_RenderBackend *renderer parameter without a word. Item 4 turned out not to be cosmetic. The library exported a global called renderer and tests/character.c defined an SDL_Renderer *renderer of its own; the executable's definition preempted the library's, akgl_sprite_load_json read a SDL_Renderer * through an akgl_RenderBackend *, and every texture load in that suite failed. The suite reported success anyway, because libakerror's unhandled-error handler ends in exit(errctx->status), exit keeps only the low byte, and AKGL_ERR_SDL is exactly 256. So character had been green while running one of its four tests, and every suite in the tree was unable to fail on the most common status in a library built on SDL. Both are fixed. tests/testutil.h gains TEST_TRAP_UNHANDLED_ERRORS(), which collapses any status a byte cannot carry onto 1, and every suite installs it. character binds a real backend with akgl_render_2d_bind. Its fourth test then runs for the first time and fails on a defect it has asserted all along, so akgl_heap_release_character now walks state_sprites with AKGL_ITERATOR_OP_RELEASE and destroys the property set before zeroing the slot -- TODO.md Defects item 21 and half of Carried over item 1. AKGL_TIME_ONESEC_MS said "one second in milliseconds" and held 1000000, so akgl_game_state_lock waited roughly sixteen minutes rather than one second. It is AKGL_TIME_ONEMS_NS now, the budget is its own named constant, and tests/game.c holds the mutex from a second thread to assert the wait -- the contended path had no coverage at all. Headers are self-contained and it is enforced: AKGL_PUBLIC_HEADERS drives both install() and a generated translation unit per header, so a header that ships is a header that is checked. Writing that found registry.h, which used SDL_PropertiesID in eight declarations and included no SDL header. 23/23 suites pass, memcheck is clean, reindent --check is clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-31 23:32:21 -04:00
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_actor_cmhf_up_on(&actor, &event),
Add physics, heap, json_helpers, game, and actor test suites Raise line coverage from 39.6 to 61.8 percent with four new suites and an extension to the actor suite, and register every suite through a single CMake list so a new test file cannot be left out of the coverage fixture. Give the test targets a build-tree RPATH and prepend the build tree to LD_LIBRARY_PATH for CTest, so a developer with a previously installed libakgl.so exercises the library that was just compiled. Fix six defects the new tests exposed: - akgl_physics_simulate read self->gravity_time before its NULL check, so a NULL backend crashed instead of reporting AKERR_NULLPOINTER. - akgl_game_save transposed CLEANUP and PROCESS, which placed the fclose inside the PROCESS switch. An ordinary save never flushed or closed its stream and produced an empty file. - akgl_game_save_actors wrote each name table terminator from the address of a single char, emitting stack contents into the save file and a sentinel the loader could not recognize. - akgl_game_load_objectnamemap used CATCH directly inside while(1), where the break leaves the loop rather than propagating, so a truncated name table loaded as a successful game. - akgl_Actor_cmhf_up_on and _down_on dereferenced actor->basechar with no NULL check, unlike their left and right counterparts. - akgl_actor_logic_movement checked actor twice instead of checking actor->basechar before dereferencing it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-30 02:03:21 -04:00
"up on, actor with no base character");
Namespace every exported symbol, and bump to 0.5.0 Closes internal-consistency items 1 through 6, 12 and 13. Every include guard is _AKGL_<FILE>_H_, every in-project header include is angled, and every exported function, type and global carries the akgl_ prefix. This is an ABI break; the soname goes to libakgl.so.0.5. TODO.md carries the full rename table. The renames were driven by renaming each declaration and letting the compiler find the uses, not by pattern substitution: renderer, physics and camera are also parameter and struct-member names, and a sed would have rewritten map->physics and every akgl_RenderBackend *renderer parameter without a word. Item 4 turned out not to be cosmetic. The library exported a global called renderer and tests/character.c defined an SDL_Renderer *renderer of its own; the executable's definition preempted the library's, akgl_sprite_load_json read a SDL_Renderer * through an akgl_RenderBackend *, and every texture load in that suite failed. The suite reported success anyway, because libakerror's unhandled-error handler ends in exit(errctx->status), exit keeps only the low byte, and AKGL_ERR_SDL is exactly 256. So character had been green while running one of its four tests, and every suite in the tree was unable to fail on the most common status in a library built on SDL. Both are fixed. tests/testutil.h gains TEST_TRAP_UNHANDLED_ERRORS(), which collapses any status a byte cannot carry onto 1, and every suite installs it. character binds a real backend with akgl_render_2d_bind. Its fourth test then runs for the first time and fails on a defect it has asserted all along, so akgl_heap_release_character now walks state_sprites with AKGL_ITERATOR_OP_RELEASE and destroys the property set before zeroing the slot -- TODO.md Defects item 21 and half of Carried over item 1. AKGL_TIME_ONESEC_MS said "one second in milliseconds" and held 1000000, so akgl_game_state_lock waited roughly sixteen minutes rather than one second. It is AKGL_TIME_ONEMS_NS now, the budget is its own named constant, and tests/game.c holds the mutex from a second thread to assert the wait -- the contended path had no coverage at all. Headers are self-contained and it is enforced: AKGL_PUBLIC_HEADERS drives both install() and a generated translation unit per header, so a header that ships is a header that is checked. Writing that found registry.h, which used SDL_PropertiesID in eight declarations and included no SDL header. 23/23 suites pass, memcheck is clean, reindent --check is clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-31 23:32:21 -04:00
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_actor_cmhf_down_on(&actor, &event),
Add physics, heap, json_helpers, game, and actor test suites Raise line coverage from 39.6 to 61.8 percent with four new suites and an extension to the actor suite, and register every suite through a single CMake list so a new test file cannot be left out of the coverage fixture. Give the test targets a build-tree RPATH and prepend the build tree to LD_LIBRARY_PATH for CTest, so a developer with a previously installed libakgl.so exercises the library that was just compiled. Fix six defects the new tests exposed: - akgl_physics_simulate read self->gravity_time before its NULL check, so a NULL backend crashed instead of reporting AKERR_NULLPOINTER. - akgl_game_save transposed CLEANUP and PROCESS, which placed the fclose inside the PROCESS switch. An ordinary save never flushed or closed its stream and produced an empty file. - akgl_game_save_actors wrote each name table terminator from the address of a single char, emitting stack contents into the save file and a sentinel the loader could not recognize. - akgl_game_load_objectnamemap used CATCH directly inside while(1), where the break leaves the loop rather than propagating, so a truncated name table loaded as a successful game. - akgl_Actor_cmhf_up_on and _down_on dereferenced actor->basechar with no NULL check, unlike their left and right counterparts. - akgl_actor_logic_movement checked actor twice instead of checking actor->basechar before dereferencing it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-30 02:03:21 -04:00
"down on, actor with no base character");
} CLEANUP {
} PROCESS(e) {
} FINISH(e, true);
SUCCEED_RETURN(e);
}
akerr_ErrorContext *test_actor_automatic_face(void)
{
PREPARE_ERROR(e);
akgl_Actor actor;
ATTEMPT {
// With the flag off, facing is left entirely to the caller.
memset(&actor, 0x00, sizeof(akgl_Actor));
actor.movement_controls_face = false;
actor.state = (AKGL_ACTOR_STATE_MOVING_LEFT | AKGL_ACTOR_STATE_FACE_RIGHT);
TEST_EXPECT_OK(e, akgl_actor_automatic_face(&actor), "automatic face with the flag off");
TEST_ASSERT(e, AKGL_BITMASK_HAS(actor.state, AKGL_ACTOR_STATE_FACE_RIGHT),
"automatic face changed facing while disabled (state %d)", actor.state);
// With the flag on, facing follows movement.
memset(&actor, 0x00, sizeof(akgl_Actor));
actor.movement_controls_face = true;
actor.state = (AKGL_ACTOR_STATE_MOVING_LEFT | AKGL_ACTOR_STATE_FACE_RIGHT);
TEST_EXPECT_OK(e, akgl_actor_automatic_face(&actor), "automatic face while moving left");
TEST_ASSERT(e, AKGL_BITMASK_HAS(actor.state, AKGL_ACTOR_STATE_FACE_LEFT),
"automatic face did not turn the actor left (state %d)", actor.state);
TEST_ASSERT(e, AKGL_BITMASK_HASNOT(actor.state, AKGL_ACTOR_STATE_FACE_RIGHT),
"automatic face left the stale facing set (state %d)", actor.state);
memset(&actor, 0x00, sizeof(akgl_Actor));
actor.movement_controls_face = true;
actor.state = AKGL_ACTOR_STATE_MOVING_RIGHT;
TEST_EXPECT_OK(e, akgl_actor_automatic_face(&actor), "automatic face while moving right");
TEST_ASSERT(e, AKGL_BITMASK_HAS(actor.state, AKGL_ACTOR_STATE_FACE_RIGHT),
"automatic face did not turn the actor right (state %d)", actor.state);
memset(&actor, 0x00, sizeof(akgl_Actor));
actor.movement_controls_face = true;
actor.state = AKGL_ACTOR_STATE_MOVING_UP;
TEST_EXPECT_OK(e, akgl_actor_automatic_face(&actor), "automatic face while moving up");
TEST_ASSERT(e, AKGL_BITMASK_HAS(actor.state, AKGL_ACTOR_STATE_FACE_UP),
"automatic face did not turn the actor up (state %d)", actor.state);
memset(&actor, 0x00, sizeof(akgl_Actor));
actor.movement_controls_face = true;
actor.state = AKGL_ACTOR_STATE_MOVING_DOWN;
TEST_EXPECT_OK(e, akgl_actor_automatic_face(&actor), "automatic face while moving down");
TEST_ASSERT(e, AKGL_BITMASK_HAS(actor.state, AKGL_ACTOR_STATE_FACE_DOWN),
"automatic face did not turn the actor down (state %d)", actor.state);
// Left wins over up when both are set, per the order of the checks.
memset(&actor, 0x00, sizeof(akgl_Actor));
actor.movement_controls_face = true;
actor.state = (AKGL_ACTOR_STATE_MOVING_LEFT | AKGL_ACTOR_STATE_MOVING_UP);
TEST_EXPECT_OK(e, akgl_actor_automatic_face(&actor), "automatic face while moving diagonally");
TEST_ASSERT(e, AKGL_BITMASK_HAS(actor.state, AKGL_ACTOR_STATE_FACE_LEFT),
"diagonal movement did not resolve to the left facing (state %d)", actor.state);
TEST_ASSERT(e, AKGL_BITMASK_HASNOT(actor.state, AKGL_ACTOR_STATE_FACE_UP),
"diagonal movement set two facings at once (state %d)", actor.state);
// A stationary actor ends up facing nowhere.
memset(&actor, 0x00, sizeof(akgl_Actor));
actor.movement_controls_face = true;
actor.state = (AKGL_ACTOR_STATE_ALIVE | AKGL_ACTOR_STATE_FACE_UP);
TEST_EXPECT_OK(e, akgl_actor_automatic_face(&actor), "automatic face while stationary");
TEST_ASSERT(e, AKGL_BITMASK_HASNOT(actor.state, AKGL_ACTOR_STATE_FACE_ALL),
"a stationary actor kept a facing bit (state %d)", actor.state);
TEST_ASSERT(e, AKGL_BITMASK_HAS(actor.state, AKGL_ACTOR_STATE_ALIVE),
"automatic face cleared an unrelated state bit (state %d)", actor.state);
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_actor_automatic_face(NULL),
"automatic face with a NULL actor");
} CLEANUP {
} PROCESS(e) {
} FINISH(e, true);
SUCCEED_RETURN(e);
}
akerr_ErrorContext *test_actor_logic_movement(void)
{
PREPARE_ERROR(e);
akgl_Actor actor;
akgl_Character basechar;
ATTEMPT {
memset(&basechar, 0x00, sizeof(akgl_Character));
basechar.ax = 3.0f; basechar.ay = 4.0f;
basechar.sx = 30.0f; basechar.sy = 40.0f; basechar.sz = 50.0f;
// Max speed is always refreshed from the base character.
memset(&actor, 0x00, sizeof(akgl_Actor));
actor.basechar = &basechar;
TEST_EXPECT_OK(e, akgl_actor_logic_movement(&actor, 1.0f), "movement logic while idle");
TEST_ASSERT_FEQ(e, actor.sx, 30.0f, "sx copied as %f, expected 30", actor.sx);
TEST_ASSERT_FEQ(e, actor.sy, 40.0f, "sy copied as %f, expected 40", actor.sy);
TEST_ASSERT_FEQ(e, actor.sz, 50.0f, "sz copied as %f, expected 50", actor.sz);
actor.state = AKGL_ACTOR_STATE_MOVING_LEFT;
TEST_EXPECT_OK(e, akgl_actor_logic_movement(&actor, 1.0f), "movement logic while moving left");
TEST_ASSERT_FEQ(e, actor.ax, -3.0f, "moving left set ax to %f, expected -3", actor.ax);
actor.state = AKGL_ACTOR_STATE_MOVING_RIGHT;
TEST_EXPECT_OK(e, akgl_actor_logic_movement(&actor, 1.0f), "movement logic while moving right");
TEST_ASSERT_FEQ(e, actor.ax, 3.0f, "moving right set ax to %f, expected 3", actor.ax);
actor.state = AKGL_ACTOR_STATE_MOVING_UP;
TEST_EXPECT_OK(e, akgl_actor_logic_movement(&actor, 1.0f), "movement logic while moving up");
TEST_ASSERT_FEQ(e, actor.ay, -4.0f, "moving up set ay to %f, expected -4", actor.ay);
actor.state = AKGL_ACTOR_STATE_MOVING_DOWN;
TEST_EXPECT_OK(e, akgl_actor_logic_movement(&actor, 1.0f), "movement logic while moving down");
TEST_ASSERT_FEQ(e, actor.ay, 4.0f, "moving down set ay to %f, expected 4", actor.ay);
// Both axes at once.
actor.state = (AKGL_ACTOR_STATE_MOVING_LEFT | AKGL_ACTOR_STATE_MOVING_DOWN);
TEST_EXPECT_OK(e, akgl_actor_logic_movement(&actor, 1.0f), "movement logic while moving diagonally");
TEST_ASSERT_FEQ(e, actor.ax, -3.0f, "diagonal movement set ax to %f, expected -3", actor.ax);
TEST_ASSERT_FEQ(e, actor.ay, 4.0f, "diagonal movement set ay to %f, expected 4", actor.ay);
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_actor_logic_movement(NULL, 1.0f),
"movement logic with a NULL actor");
// Every value the movement logic writes is read off the base character,
// so an actor without one has to be reported rather than dereferenced.
memset(&actor, 0x00, sizeof(akgl_Actor));
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_actor_logic_movement(&actor, 1.0f),
"movement logic with no base character");
} CLEANUP {
} PROCESS(e) {
} FINISH(e, true);
SUCCEED_RETURN(e);
}
akerr_ErrorContext *test_actor_logic_changeframe(void)
{
PREPARE_ERROR(e);
akgl_Actor actor;
akgl_Sprite sprite;
ATTEMPT {
memset(&sprite, 0x00, sizeof(akgl_Sprite));
sprite.frames = 4;
// Mid-animation, the frame simply advances.
memset(&actor, 0x00, sizeof(akgl_Actor));
actor.curSpriteFrameId = 1;
TEST_EXPECT_OK(e, akgl_actor_logic_changeframe(&actor, &sprite, 0), "advancing mid-animation");
TEST_ASSERT(e, actor.curSpriteFrameId == 2,
"frame advanced to %d, expected 2", actor.curSpriteFrameId);
// At the last frame without looping, it wraps to the start.
memset(&actor, 0x00, sizeof(akgl_Actor));
actor.curSpriteFrameId = 3;
sprite.loop = false;
sprite.loopReverse = false;
TEST_EXPECT_OK(e, akgl_actor_logic_changeframe(&actor, &sprite, 0), "wrapping a non-looping sprite");
TEST_ASSERT(e, actor.curSpriteFrameId == 0,
"the last frame wrapped to %d, expected 0", actor.curSpriteFrameId);
// A forward-looping sprite behaves the same way at the end.
memset(&actor, 0x00, sizeof(akgl_Actor));
actor.curSpriteFrameId = 3;
sprite.loop = true;
sprite.loopReverse = false;
TEST_EXPECT_OK(e, akgl_actor_logic_changeframe(&actor, &sprite, 0), "wrapping a forward-looping sprite");
TEST_ASSERT(e, actor.curSpriteFrameId == 0,
"the forward loop wrapped to %d, expected 0", actor.curSpriteFrameId);
// A ping-pong sprite turns around at the end instead of wrapping.
memset(&actor, 0x00, sizeof(akgl_Actor));
actor.curSpriteFrameId = 3;
sprite.loop = true;
sprite.loopReverse = true;
TEST_EXPECT_OK(e, akgl_actor_logic_changeframe(&actor, &sprite, 0), "reversing at the end");
TEST_ASSERT(e, actor.curSpriteReversing == true,
"the sprite did not enter its reverse phase at the last frame");
TEST_ASSERT(e, actor.curSpriteFrameId == 2,
"reversing stepped to frame %d, expected 2", actor.curSpriteFrameId);
// While reversing, the frame counts back down.
TEST_EXPECT_OK(e, akgl_actor_logic_changeframe(&actor, &sprite, 0), "stepping back while reversing");
TEST_ASSERT(e, actor.curSpriteFrameId == 1,
"reversing stepped to frame %d, expected 1", actor.curSpriteFrameId);
// At frame zero it turns around again and resumes going forward.
actor.curSpriteFrameId = 0;
actor.curSpriteReversing = true;
TEST_EXPECT_OK(e, akgl_actor_logic_changeframe(&actor, &sprite, 0), "turning around at frame zero");
TEST_ASSERT(e, actor.curSpriteReversing == false,
"the sprite stayed in its reverse phase at frame zero");
TEST_ASSERT(e, actor.curSpriteFrameId == 1,
"turning around stepped to frame %d, expected 1", actor.curSpriteFrameId);
// A single-frame sprite has nowhere to advance to.
memset(&actor, 0x00, sizeof(akgl_Actor));
sprite.frames = 1;
sprite.loop = false;
sprite.loopReverse = false;
TEST_EXPECT_OK(e, akgl_actor_logic_changeframe(&actor, &sprite, 0), "advancing a single-frame sprite");
TEST_ASSERT(e, actor.curSpriteFrameId == 0,
"a single-frame sprite moved to frame %d, expected 0", actor.curSpriteFrameId);
sprite.frames = 4;
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_actor_logic_changeframe(NULL, &sprite, 0),
"changeframe with a NULL actor");
} CLEANUP {
} PROCESS(e) {
} FINISH(e, true);
SUCCEED_RETURN(e);
}
/** @brief Records calls made to the changeframe stub. */
static int changeframe_calls = 0;
/** @brief Changeframe stub that records its invocation without advancing. */
static akerr_ErrorContext *stub_changeframe(akgl_Actor *obj, akgl_Sprite *curSprite, SDL_Time curtimems)
{
PREPARE_ERROR(e);
changeframe_calls += 1;
SUCCEED_RETURN(e);
}
akerr_ErrorContext *test_actor_update(void)
{
PREPARE_ERROR(e);
akgl_Actor *actor = NULL;
akgl_Character *basechar = NULL;
akgl_Sprite *sprite = NULL;
SDL_Time now = 0;
ATTEMPT {
CATCH(e, akgl_registry_init_actor());
CATCH(e, akgl_registry_init_sprite());
CATCH(e, akgl_registry_init_spritesheet());
CATCH(e, akgl_registry_init_character());
CATCH(e, akgl_heap_init());
CATCH(e, make_bound_actor(&actor, &basechar, &sprite, "updatable", AKGL_ACTOR_STATE_ALIVE));
actor->changeframefunc = &stub_changeframe;
// Long enough since the last frame change, so the sprite advances.
SDL_GetCurrentTime(&now);
sprite->speed = 1;
actor->curSpriteFrameTimer = 0;
changeframe_calls = 0;
TEST_EXPECT_OK(e, akgl_actor_update(actor), "updating an actor whose frame is due");
TEST_ASSERT(e, changeframe_calls == 1,
"a due frame change fired %d times, expected 1", changeframe_calls);
TEST_ASSERT(e, actor->curSpriteFrameTimer != 0,
"the frame timer was not restamped after a frame change");
// Too soon since the last change, so nothing happens.
SDL_GetCurrentTime(&now);
sprite->speed = 1000000000;
actor->curSpriteFrameTimer = now;
changeframe_calls = 0;
TEST_EXPECT_OK(e, akgl_actor_update(actor), "updating an actor whose frame is not due");
TEST_ASSERT(e, changeframe_calls == 0,
"an early frame change fired %d times, expected 0", changeframe_calls);
// An actor in a state with no sprite bound is skipped, not failed: the
// missing binding is reported as AKERR_KEY and swallowed by update.
actor->state = AKGL_ACTOR_STATE_DEAD;
changeframe_calls = 0;
TEST_EXPECT_OK(e, akgl_actor_update(actor),
"updating an actor whose state has no sprite");
TEST_ASSERT(e, changeframe_calls == 0,
"an actor with no sprite for its state still changed frames");
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_actor_update(NULL),
"updating a NULL actor");
// An actor with no base character cannot resolve a sprite at all.
actor->basechar = NULL;
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_actor_update(actor),
"updating an actor with no base character");
} CLEANUP {
} PROCESS(e) {
} FINISH(e, true);
SUCCEED_RETURN(e);
}
akerr_ErrorContext *test_actor_character_sprite_binding(void)
{
PREPARE_ERROR(e);
akgl_Actor *actor = NULL;
akgl_Character *basechar = NULL;
akgl_Sprite *sprite = NULL;
akgl_Sprite *found = NULL;
ATTEMPT {
CATCH(e, akgl_registry_init_actor());
CATCH(e, akgl_registry_init_sprite());
CATCH(e, akgl_registry_init_spritesheet());
CATCH(e, akgl_registry_init_character());
CATCH(e, akgl_heap_init());
CATCH(e, make_bound_actor(&actor, &basechar, &sprite, "bound", AKGL_ACTOR_STATE_ALIVE));
TEST_EXPECT_OK(e, akgl_character_sprite_get(basechar, AKGL_ACTOR_STATE_ALIVE, &found),
"reading back a bound sprite");
TEST_ASSERT(e, found == sprite, "the bound sprite came back as a different object");
// A composite state is a distinct key from either of its components.
TEST_EXPECT_STATUS(e, AKERR_KEY,
akgl_character_sprite_get(basechar,
(AKGL_ACTOR_STATE_ALIVE | AKGL_ACTOR_STATE_FACE_UP),
&found),
"reading a composite state that was never bound");
TEST_EXPECT_STATUS(e, AKERR_KEY, akgl_character_sprite_get(basechar, 0, &found),
"reading state zero, which was never bound");
// Binding a composite state makes it resolvable.
TEST_EXPECT_OK(e, akgl_character_sprite_add(basechar, sprite,
(AKGL_ACTOR_STATE_ALIVE | AKGL_ACTOR_STATE_FACE_UP)),
"binding a sprite to a composite state");
TEST_EXPECT_OK(e, akgl_character_sprite_get(basechar,
(AKGL_ACTOR_STATE_ALIVE | AKGL_ACTOR_STATE_FACE_UP),
&found),
"reading back the composite binding");
TEST_ASSERT(e, found == sprite, "the composite binding resolved to a different sprite");
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_character_sprite_get(NULL, 0, &found),
"sprite_get with a NULL character");
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER,
akgl_character_sprite_get(basechar, AKGL_ACTOR_STATE_ALIVE, NULL),
"sprite_get with a NULL destination");
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_character_sprite_add(NULL, sprite, 0),
"sprite_add with a NULL character");
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_character_sprite_add(basechar, NULL, 0),
"sprite_add with a NULL sprite");
} CLEANUP {
} PROCESS(e) {
} FINISH(e, true);
SUCCEED_RETURN(e);
}
akerr_ErrorContext *test_actor_sprite_sheet_coords(void)
{
PREPARE_ERROR(e);
akgl_Sprite sprite;
SDL_FRect coords;
ATTEMPT {
memset(&sprite, 0x00, sizeof(akgl_Sprite));
sprite.width = 32;
sprite.height = 32;
Namespace every exported symbol, and bump to 0.5.0 Closes internal-consistency items 1 through 6, 12 and 13. Every include guard is _AKGL_<FILE>_H_, every in-project header include is angled, and every exported function, type and global carries the akgl_ prefix. This is an ABI break; the soname goes to libakgl.so.0.5. TODO.md carries the full rename table. The renames were driven by renaming each declaration and letting the compiler find the uses, not by pattern substitution: renderer, physics and camera are also parameter and struct-member names, and a sed would have rewritten map->physics and every akgl_RenderBackend *renderer parameter without a word. Item 4 turned out not to be cosmetic. The library exported a global called renderer and tests/character.c defined an SDL_Renderer *renderer of its own; the executable's definition preempted the library's, akgl_sprite_load_json read a SDL_Renderer * through an akgl_RenderBackend *, and every texture load in that suite failed. The suite reported success anyway, because libakerror's unhandled-error handler ends in exit(errctx->status), exit keeps only the low byte, and AKGL_ERR_SDL is exactly 256. So character had been green while running one of its four tests, and every suite in the tree was unable to fail on the most common status in a library built on SDL. Both are fixed. tests/testutil.h gains TEST_TRAP_UNHANDLED_ERRORS(), which collapses any status a byte cannot carry onto 1, and every suite installs it. character binds a real backend with akgl_render_2d_bind. Its fourth test then runs for the first time and fails on a defect it has asserted all along, so akgl_heap_release_character now walks state_sprites with AKGL_ITERATOR_OP_RELEASE and destroys the property set before zeroing the slot -- TODO.md Defects item 21 and half of Carried over item 1. AKGL_TIME_ONESEC_MS said "one second in milliseconds" and held 1000000, so akgl_game_state_lock waited roughly sixteen minutes rather than one second. It is AKGL_TIME_ONEMS_NS now, the budget is its own named constant, and tests/game.c holds the mutex from a second thread to assert the wait -- the contended path had no coverage at all. Headers are self-contained and it is enforced: AKGL_PUBLIC_HEADERS drives both install() and a generated translation unit per header, so a header that ships is a header that is checked. Writing that found registry.h, which used SDL_PropertiesID in eight declarations and included no SDL header. 23/23 suites pass, memcheck is clean, reindent --check is clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-31 23:32:21 -04:00
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_spritesheet_coords_for_frame(NULL, &coords, 0),
Add physics, heap, json_helpers, game, and actor test suites Raise line coverage from 39.6 to 61.8 percent with four new suites and an extension to the actor suite, and register every suite through a single CMake list so a new test file cannot be left out of the coverage fixture. Give the test targets a build-tree RPATH and prepend the build tree to LD_LIBRARY_PATH for CTest, so a developer with a previously installed libakgl.so exercises the library that was just compiled. Fix six defects the new tests exposed: - akgl_physics_simulate read self->gravity_time before its NULL check, so a NULL backend crashed instead of reporting AKERR_NULLPOINTER. - akgl_game_save transposed CLEANUP and PROCESS, which placed the fclose inside the PROCESS switch. An ordinary save never flushed or closed its stream and produced an empty file. - akgl_game_save_actors wrote each name table terminator from the address of a single char, emitting stack contents into the save file and a sentinel the loader could not recognize. - akgl_game_load_objectnamemap used CATCH directly inside while(1), where the break leaves the loop rather than propagating, so a truncated name table loaded as a successful game. - akgl_Actor_cmhf_up_on and _down_on dereferenced actor->basechar with no NULL check, unlike their left and right counterparts. - akgl_actor_logic_movement checked actor twice instead of checking actor->basechar before dereferencing it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-30 02:03:21 -04:00
"sheet coords with a NULL sprite");
Namespace every exported symbol, and bump to 0.5.0 Closes internal-consistency items 1 through 6, 12 and 13. Every include guard is _AKGL_<FILE>_H_, every in-project header include is angled, and every exported function, type and global carries the akgl_ prefix. This is an ABI break; the soname goes to libakgl.so.0.5. TODO.md carries the full rename table. The renames were driven by renaming each declaration and letting the compiler find the uses, not by pattern substitution: renderer, physics and camera are also parameter and struct-member names, and a sed would have rewritten map->physics and every akgl_RenderBackend *renderer parameter without a word. Item 4 turned out not to be cosmetic. The library exported a global called renderer and tests/character.c defined an SDL_Renderer *renderer of its own; the executable's definition preempted the library's, akgl_sprite_load_json read a SDL_Renderer * through an akgl_RenderBackend *, and every texture load in that suite failed. The suite reported success anyway, because libakerror's unhandled-error handler ends in exit(errctx->status), exit keeps only the low byte, and AKGL_ERR_SDL is exactly 256. So character had been green while running one of its four tests, and every suite in the tree was unable to fail on the most common status in a library built on SDL. Both are fixed. tests/testutil.h gains TEST_TRAP_UNHANDLED_ERRORS(), which collapses any status a byte cannot carry onto 1, and every suite installs it. character binds a real backend with akgl_render_2d_bind. Its fourth test then runs for the first time and fails on a defect it has asserted all along, so akgl_heap_release_character now walks state_sprites with AKGL_ITERATOR_OP_RELEASE and destroys the property set before zeroing the slot -- TODO.md Defects item 21 and half of Carried over item 1. AKGL_TIME_ONESEC_MS said "one second in milliseconds" and held 1000000, so akgl_game_state_lock waited roughly sixteen minutes rather than one second. It is AKGL_TIME_ONEMS_NS now, the budget is its own named constant, and tests/game.c holds the mutex from a second thread to assert the wait -- the contended path had no coverage at all. Headers are self-contained and it is enforced: AKGL_PUBLIC_HEADERS drives both install() and a generated translation unit per header, so a header that ships is a header that is checked. Writing that found registry.h, which used SDL_PropertiesID in eight declarations and included no SDL header. 23/23 suites pass, memcheck is clean, reindent --check is clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-31 23:32:21 -04:00
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_spritesheet_coords_for_frame(&sprite, NULL, 0),
Add physics, heap, json_helpers, game, and actor test suites Raise line coverage from 39.6 to 61.8 percent with four new suites and an extension to the actor suite, and register every suite through a single CMake list so a new test file cannot be left out of the coverage fixture. Give the test targets a build-tree RPATH and prepend the build tree to LD_LIBRARY_PATH for CTest, so a developer with a previously installed libakgl.so exercises the library that was just compiled. Fix six defects the new tests exposed: - akgl_physics_simulate read self->gravity_time before its NULL check, so a NULL backend crashed instead of reporting AKERR_NULLPOINTER. - akgl_game_save transposed CLEANUP and PROCESS, which placed the fclose inside the PROCESS switch. An ordinary save never flushed or closed its stream and produced an empty file. - akgl_game_save_actors wrote each name table terminator from the address of a single char, emitting stack contents into the save file and a sentinel the loader could not recognize. - akgl_game_load_objectnamemap used CATCH directly inside while(1), where the break leaves the loop rather than propagating, so a truncated name table loaded as a successful game. - akgl_Actor_cmhf_up_on and _down_on dereferenced actor->basechar with no NULL check, unlike their left and right counterparts. - akgl_actor_logic_movement checked actor twice instead of checking actor->basechar before dereferencing it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-30 02:03:21 -04:00
"sheet coords with a NULL rectangle");
// The sprite has no sheet attached, so there is no texture to measure against.
Namespace every exported symbol, and bump to 0.5.0 Closes internal-consistency items 1 through 6, 12 and 13. Every include guard is _AKGL_<FILE>_H_, every in-project header include is angled, and every exported function, type and global carries the akgl_ prefix. This is an ABI break; the soname goes to libakgl.so.0.5. TODO.md carries the full rename table. The renames were driven by renaming each declaration and letting the compiler find the uses, not by pattern substitution: renderer, physics and camera are also parameter and struct-member names, and a sed would have rewritten map->physics and every akgl_RenderBackend *renderer parameter without a word. Item 4 turned out not to be cosmetic. The library exported a global called renderer and tests/character.c defined an SDL_Renderer *renderer of its own; the executable's definition preempted the library's, akgl_sprite_load_json read a SDL_Renderer * through an akgl_RenderBackend *, and every texture load in that suite failed. The suite reported success anyway, because libakerror's unhandled-error handler ends in exit(errctx->status), exit keeps only the low byte, and AKGL_ERR_SDL is exactly 256. So character had been green while running one of its four tests, and every suite in the tree was unable to fail on the most common status in a library built on SDL. Both are fixed. tests/testutil.h gains TEST_TRAP_UNHANDLED_ERRORS(), which collapses any status a byte cannot carry onto 1, and every suite installs it. character binds a real backend with akgl_render_2d_bind. Its fourth test then runs for the first time and fails on a defect it has asserted all along, so akgl_heap_release_character now walks state_sprites with AKGL_ITERATOR_OP_RELEASE and destroys the property set before zeroing the slot -- TODO.md Defects item 21 and half of Carried over item 1. AKGL_TIME_ONESEC_MS said "one second in milliseconds" and held 1000000, so akgl_game_state_lock waited roughly sixteen minutes rather than one second. It is AKGL_TIME_ONEMS_NS now, the budget is its own named constant, and tests/game.c holds the mutex from a second thread to assert the wait -- the contended path had no coverage at all. Headers are self-contained and it is enforced: AKGL_PUBLIC_HEADERS drives both install() and a generated translation unit per header, so a header that ships is a header that is checked. Writing that found registry.h, which used SDL_PropertiesID in eight declarations and included no SDL header. 23/23 suites pass, memcheck is clean, reindent --check is clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-31 23:32:21 -04:00
TEST_EXPECT_STATUS(e, AKERR_NULLPOINTER, akgl_spritesheet_coords_for_frame(&sprite, &coords, 0),
Add physics, heap, json_helpers, game, and actor test suites Raise line coverage from 39.6 to 61.8 percent with four new suites and an extension to the actor suite, and register every suite through a single CMake list so a new test file cannot be left out of the coverage fixture. Give the test targets a build-tree RPATH and prepend the build tree to LD_LIBRARY_PATH for CTest, so a developer with a previously installed libakgl.so exercises the library that was just compiled. Fix six defects the new tests exposed: - akgl_physics_simulate read self->gravity_time before its NULL check, so a NULL backend crashed instead of reporting AKERR_NULLPOINTER. - akgl_game_save transposed CLEANUP and PROCESS, which placed the fclose inside the PROCESS switch. An ordinary save never flushed or closed its stream and produced an empty file. - akgl_game_save_actors wrote each name table terminator from the address of a single char, emitting stack contents into the save file and a sentinel the loader could not recognize. - akgl_game_load_objectnamemap used CATCH directly inside while(1), where the break leaves the loop rather than propagating, so a truncated name table loaded as a successful game. - akgl_Actor_cmhf_up_on and _down_on dereferenced actor->basechar with no NULL check, unlike their left and right counterparts. - akgl_actor_logic_movement checked actor twice instead of checking actor->basechar before dereferencing it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-30 02:03:21 -04:00
"sheet coords with a NULL spritesheet");
} CLEANUP {
} PROCESS(e) {
} FINISH(e, true);
SUCCEED_RETURN(e);
}
Draw actors at their sprite's height, and update each one once a frame Closes Defects item 26 and Performance item 32. akgl_actor_render set dest.h from curSprite->width, so every actor was drawn square and a non-square sprite was stretched or squashed. Invisible in the fixtures because they are all square, so tests/actor.c gets a 48x24 sprite and a render backend whose draw_texture records the rectangle it is handed instead of drawing it. That recording backend is the first coverage akgl_actor_render has had at all -- every other test in the file stubs renderfunc out. akgl_game_update looped over AKGL_TILEMAP_MAX_LAYERS with the actor sweep nested inside it and never compared an actor's layer to the layer it was on, so every live actor's updatefunc ran sixteen times a frame. The sweep is hoisted out: updating an actor is not a per-layer operation, and akgl_render_2d_draw_world already walks the layers for the half that is. AKGL_ITERATOR_OP_LAYERMASK is honoured rather than ignored now, so a caller who wants one layer can still ask, and gets each of those actors once. Counting is the assertion, deliberately. The defect was invisible in the frame total because the tilemap blits are three orders of magnitude larger, so a timing test would have measured the rasterizer. tests/game.c counts calls into a stub updatefunc and reports 16 against the old code. PERFORMANCE.md records the timing side as what it honestly is: the gap between the akgl_game_update and draw_world rows of the same run, 92 us before and noise in both directions after. The absolute table is not re-taken -- a later run on this machine read every row about 15% high, including rows nothing has touched. 25/25 pass, reindent --check clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-01 06:45:34 -04:00
/*
* A render backend whose draw_texture records what it was asked to draw instead
* of drawing it. akgl_actor_render's whole output is the rectangle it hands to
* draw_texture, so that rectangle is the only thing there is to assert on.
*/
static akgl_RenderBackend testbackend;
static SDL_FRect lastdrawdest;
static SDL_FRect lastdrawsrc;
static int drawcalls = 0;
static akerr_ErrorContext *record_draw_texture(
akgl_RenderBackend *self,
SDL_Texture *texture,
SDL_FRect *src,
SDL_FRect *dest,
double angle,
SDL_FPoint *center,
SDL_FlipMode flip)
{
PREPARE_ERROR(errctx);
if ( src != NULL ) {
lastdrawsrc = *src;
}
if ( dest != NULL ) {
lastdrawdest = *dest;
}
drawcalls += 1;
SUCCEED_RETURN(errctx);
}
/**
* @brief akgl_actor_render must take the drawn height from the sprite's height.
*
* It read `curSprite->width` for both, so every actor was drawn square and a
* non-square sprite was stretched or squashed. Invisible in the fixtures
* because they are square, which is exactly why this test uses a sprite that
* is not: 48 wide by 24 high, so a regression puts 48 in both.
*
* This is also the first test of akgl_actor_render at all -- everything else in
* this file stubs renderfunc out.
*/
akerr_ErrorContext *test_actor_render_uses_sprite_height(void)
{
PREPARE_ERROR(e);
akgl_Actor *actor = NULL;
akgl_Character *basechar = NULL;
akgl_Sprite *sprite = NULL;
SDL_Texture *sheettexture = NULL;
ATTEMPT {
CATCH(e, make_bound_actor(&actor, &basechar, &sprite, "renderactor", AKGL_ACTOR_STATE_ALIVE));
// akgl_spritesheet_coords_for_frame reads sheet->texture->w, so the
// sheet needs a real one even though nothing is rasterized.
sheettexture = SDL_CreateTexture(
testbackend.sdl_renderer,
SDL_PIXELFORMAT_RGBA8888,
SDL_TEXTUREACCESS_TARGET,
96, 48);
FAIL_ZERO_BREAK(e, sheettexture, AKGL_ERR_SDL, "%s", SDL_GetError());
sprite->sheet->texture = sheettexture;
sprite->width = 48;
sprite->height = 24;
sprite->frames = 2;
actor->curSpriteFrameId = 0;
actor->visible = true;
actor->scale = 1.0f;
actor->x = 0;
actor->y = 0;
akgl_camera->x = 0;
akgl_camera->y = 0;
akgl_camera->w = 640;
akgl_camera->h = 480;
drawcalls = 0;
TEST_EXPECT_OK(e, akgl_actor_render(actor), "rendering a non-square actor");
TEST_ASSERT(e, drawcalls == 1, "actor render made %d draw calls, expected 1", drawcalls);
TEST_ASSERT_FEQ(e, lastdrawdest.w, 48.0f,
"drawn width is %f, expected the sprite's 48", lastdrawdest.w);
TEST_ASSERT_FEQ(e, lastdrawdest.h, 24.0f,
"drawn height is %f, expected the sprite's 24 -- it used to take the width",
lastdrawdest.h);
// The source rectangle has always come from the sprite properly; assert
// it too, so a regression cannot move the bug one line up.
TEST_ASSERT_FEQ(e, lastdrawsrc.w, 48.0f, "source width is %f, expected 48", lastdrawsrc.w);
TEST_ASSERT_FEQ(e, lastdrawsrc.h, 24.0f, "source height is %f, expected 24", lastdrawsrc.h);
// Scale multiplies both, and must not collapse them back together.
actor->scale = 2.0f;
drawcalls = 0;
TEST_EXPECT_OK(e, akgl_actor_render(actor), "rendering a scaled non-square actor");
TEST_ASSERT_FEQ(e, lastdrawdest.w, 96.0f, "scaled width is %f, expected 96", lastdrawdest.w);
TEST_ASSERT_FEQ(e, lastdrawdest.h, 48.0f, "scaled height is %f, expected 48", lastdrawdest.h);
} CLEANUP {
if ( sprite != NULL && sprite->sheet != NULL ) {
sprite->sheet->texture = NULL;
}
if ( sheettexture != NULL ) {
SDL_DestroyTexture(sheettexture);
}
if ( actor != NULL ) {
IGNORE(akgl_heap_release_actor(actor));
}
if ( basechar != NULL ) {
IGNORE(akgl_heap_release_character(basechar));
}
} PROCESS(e) {
} FINISH(e, true);
SUCCEED_RETURN(e);
}
int main(void)
{
akgl_actor_updated = 0;
akgl_actor_rendered = 0;
UNHANDLED_ERROR_BEHAVIOR = UNHANDLED_ERROR_EXIT;
PREPARE_ERROR(errctx);
Add physics, heap, json_helpers, game, and actor test suites Raise line coverage from 39.6 to 61.8 percent with four new suites and an extension to the actor suite, and register every suite through a single CMake list so a new test file cannot be left out of the coverage fixture. Give the test targets a build-tree RPATH and prepend the build tree to LD_LIBRARY_PATH for CTest, so a developer with a previously installed libakgl.so exercises the library that was just compiled. Fix six defects the new tests exposed: - akgl_physics_simulate read self->gravity_time before its NULL check, so a NULL backend crashed instead of reporting AKERR_NULLPOINTER. - akgl_game_save transposed CLEANUP and PROCESS, which placed the fclose inside the PROCESS switch. An ordinary save never flushed or closed its stream and produced an empty file. - akgl_game_save_actors wrote each name table terminator from the address of a single char, emitting stack contents into the save file and a sentinel the loader could not recognize. - akgl_game_load_objectnamemap used CATCH directly inside while(1), where the break leaves the loop rather than propagating, so a truncated name table loaded as a successful game. - akgl_Actor_cmhf_up_on and _down_on dereferenced actor->basechar with no NULL check, unlike their left and right counterparts. - akgl_actor_logic_movement checked actor twice instead of checking actor->basechar before dereferencing it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-30 02:03:21 -04:00
SDL_SetHint(SDL_HINT_VIDEO_DRIVER, "dummy");
SDL_SetHint(SDL_HINT_AUDIO_DRIVER, "dummy");
ATTEMPT {
Migrate to the libakerror 1.0.0 status registry libakerror 1.0.0 replaced the consumer-sized status-name array with a private registry and made status-code ownership explicit and enforced. AKERR_MAX_ERR_VALUE and __AKERR_ERROR_NAMES are gone, and the registry entry points raise akerr_ErrorContext * instead of returning int. See deps/libakerror/UPGRADING.md. The break was not only source-level. libakgl's codes sat at AKERR_LAST_ERRNO_VALUE + 18 through + 22, and 1.0.0 claimed exactly those five offsets for its own AKERR_STATUS_* registry codes, so every AKGL_ERR_* was aliasing a libakerror status. HANDLE(e, AKGL_ERR_LOGICINTERRUPT) at physics.c:222 would have swallowed a foreign-name refusal. - Move the band to AKERR_FIRST_CONSUMER_STATUS (256) as fixed offsets, so a libc that grows an errno cannot move the codes, and add AKGL_ERR_OWNER, AKGL_ERR_LIMIT and AKGL_ERR_COUNT to describe it. - Reserve the range and register the names through the owned entry points, PASS-ing each: these are AKERR_NOIGNORE, and the old akerr_name_for_status calls discarded failure silently. - Drop the AKERR_MAX_ERR_VALUE=256 compile definition. - Guard on AKERR_FIRST_CONSUMER_STATUS in include/akgl/error.h, which now includes <akerror.h> so the guard is reliable. The embedded build is fine, but the find_package path can pick up a stale installed header, and 1.0.0 has an soname, so that pairing is an ABI mismatch rather than a compile problem. Same guard libakstdlib already carries. Registration also moves out of akgl_heap_init into a new akgl_error_init in src/error.c. It was in the heap pool's initializer only because that was the first thing akgl_game_init called, and the upgrade turned five fire-and-forget name calls into a library-wide ownership claim that can fail. That placement was hiding a defect: game.c raises AKGL_ERR_SDL when SDL_CreateMutex fails, five lines before akgl_heap_init ran, so the earliest error path in the library was guaranteed to print "Unknown Error". akgl_error_init is now the first statement in akgl_game_init. Callers that drive subsystems directly must call akgl_error_init first; it is idempotent, so ordering it precisely is not required. The eleven test suites that relied on akgl_heap_init to name their statuses now call it explicitly, or their failure messages would have degraded to "Unknown Error". Add tests/error.c: assert every code reads back its registered name, that the name table and AKGL_ERR_COUNT agree, that a foreign owner is refused with AKERR_STATUS_NAME_FOREIGN and AKERR_STATUS_RANGE_OVERLAP, and that repeating the init is a no-op. That last one is a live constraint, not a triviality -- libakerror treats only an identical reservation as a repeat, so a subset or superset raises. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-30 22:20:28 -04:00
CATCH(errctx, akgl_error_init());
Namespace every exported symbol, and bump to 0.5.0 Closes internal-consistency items 1 through 6, 12 and 13. Every include guard is _AKGL_<FILE>_H_, every in-project header include is angled, and every exported function, type and global carries the akgl_ prefix. This is an ABI break; the soname goes to libakgl.so.0.5. TODO.md carries the full rename table. The renames were driven by renaming each declaration and letting the compiler find the uses, not by pattern substitution: renderer, physics and camera are also parameter and struct-member names, and a sed would have rewritten map->physics and every akgl_RenderBackend *renderer parameter without a word. Item 4 turned out not to be cosmetic. The library exported a global called renderer and tests/character.c defined an SDL_Renderer *renderer of its own; the executable's definition preempted the library's, akgl_sprite_load_json read a SDL_Renderer * through an akgl_RenderBackend *, and every texture load in that suite failed. The suite reported success anyway, because libakerror's unhandled-error handler ends in exit(errctx->status), exit keeps only the low byte, and AKGL_ERR_SDL is exactly 256. So character had been green while running one of its four tests, and every suite in the tree was unable to fail on the most common status in a library built on SDL. Both are fixed. tests/testutil.h gains TEST_TRAP_UNHANDLED_ERRORS(), which collapses any status a byte cannot carry onto 1, and every suite installs it. character binds a real backend with akgl_render_2d_bind. Its fourth test then runs for the first time and fails on a defect it has asserted all along, so akgl_heap_release_character now walks state_sprites with AKGL_ITERATOR_OP_RELEASE and destroys the property set before zeroing the slot -- TODO.md Defects item 21 and half of Carried over item 1. AKGL_TIME_ONESEC_MS said "one second in milliseconds" and held 1000000, so akgl_game_state_lock waited roughly sixteen minutes rather than one second. It is AKGL_TIME_ONEMS_NS now, the budget is its own named constant, and tests/game.c holds the mutex from a second thread to assert the wait -- the contended path had no coverage at all. Headers are self-contained and it is enforced: AKGL_PUBLIC_HEADERS drives both install() and a generated translation unit per header, so a header that ships is a header that is checked. Writing that found registry.h, which used SDL_PropertiesID in eight declarations and included no SDL header. 23/23 suites pass, memcheck is clean, reindent --check is clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-31 23:32:21 -04:00
TEST_TRAP_UNHANDLED_ERRORS();
CATCH(errctx, akgl_registry_init_actor());
CATCH(errctx, akgl_registry_init_sprite());
CATCH(errctx, akgl_registry_init_spritesheet());
CATCH(errctx, akgl_registry_init_character());
Draw actors at their sprite's height, and update each one once a frame Closes Defects item 26 and Performance item 32. akgl_actor_render set dest.h from curSprite->width, so every actor was drawn square and a non-square sprite was stretched or squashed. Invisible in the fixtures because they are all square, so tests/actor.c gets a 48x24 sprite and a render backend whose draw_texture records the rectangle it is handed instead of drawing it. That recording backend is the first coverage akgl_actor_render has had at all -- every other test in the file stubs renderfunc out. akgl_game_update looped over AKGL_TILEMAP_MAX_LAYERS with the actor sweep nested inside it and never compared an actor's layer to the layer it was on, so every live actor's updatefunc ran sixteen times a frame. The sweep is hoisted out: updating an actor is not a per-layer operation, and akgl_render_2d_draw_world already walks the layers for the half that is. AKGL_ITERATOR_OP_LAYERMASK is honoured rather than ignored now, so a caller who wants one layer can still ask, and gets each of those actors once. Counting is the assertion, deliberately. The defect was invisible in the frame total because the tilemap blits are three orders of magnitude larger, so a timing test would have measured the rasterizer. tests/game.c counts calls into a stub updatefunc and reports 16 against the old code. PERFORMANCE.md records the timing side as what it honestly is: the gap between the akgl_game_update and draw_world rows of the same run, 92 us before and noise in both directions after. The absolute table is not re-taken -- a later run on this machine read every row about 15% high, including rows nothing has touched. 25/25 pass, reindent --check clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-01 06:45:34 -04:00
// A renderer, so akgl_actor_render can be driven at all. The backend is
// bound and then has its draw_texture replaced, which is the point:
// what the function computes is only visible in what it hands over.
if ( !SDL_Init(SDL_INIT_VIDEO) ) {
FAIL_BREAK(errctx, AKGL_ERR_SDL, "Couldn't initialize SDL: %s", SDL_GetError());
}
memset(&testbackend, 0x00, sizeof(akgl_RenderBackend));
if ( !SDL_CreateWindowAndRenderer("net/aklabs/libakgl/test_actor", 640, 480,
SDL_WINDOW_HIDDEN, &akgl_window, &testbackend.sdl_renderer) ) {
FAIL_BREAK(errctx, AKGL_ERR_SDL, "Couldn't create window/renderer: %s", SDL_GetError());
}
CATCH(errctx, akgl_render_2d_bind(&testbackend));
testbackend.draw_texture = &record_draw_texture;
akgl_renderer = &testbackend;
akgl_camera = &akgl_default_camera;
CATCH(errctx, test_registry_actor_iterator_nullpointers());
CATCH(errctx, test_registry_actor_iterator_missingactor());
CATCH(errctx, test_registry_actor_iterator_updaterender());
CATCH(errctx, test_akgl_actor_set_character());
CATCH(errctx, test_actor_manage_children());
Add physics, heap, json_helpers, game, and actor test suites Raise line coverage from 39.6 to 61.8 percent with four new suites and an extension to the actor suite, and register every suite through a single CMake list so a new test file cannot be left out of the coverage fixture. Give the test targets a build-tree RPATH and prepend the build tree to LD_LIBRARY_PATH for CTest, so a developer with a previously installed libakgl.so exercises the library that was just compiled. Fix six defects the new tests exposed: - akgl_physics_simulate read self->gravity_time before its NULL check, so a NULL backend crashed instead of reporting AKERR_NULLPOINTER. - akgl_game_save transposed CLEANUP and PROCESS, which placed the fclose inside the PROCESS switch. An ordinary save never flushed or closed its stream and produced an empty file. - akgl_game_save_actors wrote each name table terminator from the address of a single char, emitting stack contents into the save file and a sentinel the loader could not recognize. - akgl_game_load_objectnamemap used CATCH directly inside while(1), where the break leaves the loop rather than propagating, so a truncated name table loaded as a successful game. - akgl_Actor_cmhf_up_on and _down_on dereferenced actor->basechar with no NULL check, unlike their left and right counterparts. - akgl_actor_logic_movement checked actor twice instead of checking actor->basechar before dereferencing it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-30 02:03:21 -04:00
CATCH(errctx, test_actor_control_handlers_on());
CATCH(errctx, test_actor_control_handlers_off());
CATCH(errctx, test_actor_control_handlers_nullpointers());
CATCH(errctx, test_actor_automatic_face());
CATCH(errctx, test_actor_logic_movement());
CATCH(errctx, test_actor_logic_changeframe());
CATCH(errctx, test_actor_update());
CATCH(errctx, test_actor_character_sprite_binding());
CATCH(errctx, test_actor_sprite_sheet_coords());
Draw actors at their sprite's height, and update each one once a frame Closes Defects item 26 and Performance item 32. akgl_actor_render set dest.h from curSprite->width, so every actor was drawn square and a non-square sprite was stretched or squashed. Invisible in the fixtures because they are all square, so tests/actor.c gets a 48x24 sprite and a render backend whose draw_texture records the rectangle it is handed instead of drawing it. That recording backend is the first coverage akgl_actor_render has had at all -- every other test in the file stubs renderfunc out. akgl_game_update looped over AKGL_TILEMAP_MAX_LAYERS with the actor sweep nested inside it and never compared an actor's layer to the layer it was on, so every live actor's updatefunc ran sixteen times a frame. The sweep is hoisted out: updating an actor is not a per-layer operation, and akgl_render_2d_draw_world already walks the layers for the half that is. AKGL_ITERATOR_OP_LAYERMASK is honoured rather than ignored now, so a caller who wants one layer can still ask, and gets each of those actors once. Counting is the assertion, deliberately. The defect was invisible in the frame total because the tilemap blits are three orders of magnitude larger, so a timing test would have measured the rasterizer. tests/game.c counts calls into a stub updatefunc and reports 16 against the old code. PERFORMANCE.md records the timing side as what it honestly is: the gap between the akgl_game_update and draw_world rows of the same run, 92 us before and noise in both directions after. The absolute table is not re-taken -- a later run on this machine read every row about 15% high, including rows nothing has touched. 25/25 pass, reindent --check clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-01 06:45:34 -04:00
CATCH(errctx, test_actor_render_uses_sprite_height());
} CLEANUP {
} PROCESS(errctx) {
} FINISH_NORETURN(errctx);
return 0;
}