Files
akbasic/tests/trap_verbs.c

404 lines
14 KiB
C
Raw Permalink Normal View History

Finish the language: every remaining verb group, and the defects that blocked them Closes groups A, C, D, E, F, H and J of TODO.md section 4, plus RESTORE and RENUMBER, and closes section 6 -- all seventeen reference defects. Seven of those turned out to have been fixed or never ported and nobody had written it down; the audit records the evidence for each. Two of the seventeen were real. math_plus mutated its left operand when the operand was mutable, so A# + 1 could modify A#; it was gated on FOR/NEXT coverage because NEXT relied on the mutation, so tests/for_next.c came first and NEXT now writes the counter back itself. And the binary operators summed both numeric fields of their right operand, which no BASIC program can reach -- that one needed a test written against the value API. Writing the tests turned up eight defects nobody had listed. Seven are fixed: IF A = 2 THEN was a parse error; only == worked IF ... AND ... was a parse error, because a condition parsed as one relation IF A = 1 OR B = 2 THEN was silently always false, and so was IF A THEN EXIT before any NEXT restarted the program and exhausted the variable pool READ never found a DATA line above it, and swallowed the lines between PRINT 2 + 2 at the prompt was filed as program text instead of answering a short read discarded its bytes, so COPY produced empty files every verb taking an argument list said "peek() returned nil token!" on none The eighth is not fixed and cannot be quietly: a FOR whose step overshoots runs its body one extra time, and FOR I = 1 TO 1 runs it zero times. The two errors cancel for a step of 1, which is why neither was noticed. Correcting them changes the expected output of a checked-in acceptance file, and tests/reference/README.md forbids editing one to suit this interpreter. It is tests/for_semantics.c in AKBASIC_KNOWN_FAILING_TESTS, asserting the correct contract, and TODO.md items 19 and 20. Sprites are real libakgl actors with a renderfunc of their own, because akgl_actor_render draws every sprite square and an actor has no per-axis scale. Both are filed upstream. SPRSAV takes an image file, an SSHAPE handle or a 63-element integer array -- a string here cannot hold a zero byte. Verbs that need hardware that does not exist are refused by name with the reason rather than faked: SYS, HEADER, COLLECT, BACKUP, BOOT, FILTER, and DIRECTORY, which is refused for a missing libakstdlib wrapper filed upstream. 94 tests in the default build, 93 with SDL, 94 under ASan and UBSan, doxygen clean. The Go acceptance corpus stayed green throughout. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-31 21:50:37 -04:00
/**
* @file trap_verbs.c
* @brief Tests the group C verbs: TRAP, RESUME and ERR().
*
* The interesting half of error trapping is what *stops* happening: an armed
* TRAP suppresses the "? line : CLASS message" report and stops the run from
* ending, and neither is visible unless a test says the output is empty of one
* and the program kept going.
*
* `ER` and `EL` are `ER#` and `EL#` here -- ordinary globals, because this
* dialect has no bare variable names. See TODO.md section 5.
*/
Assert the log line a failed report leaves behind A mutation run over src/runtime.c found that deleting the `LOG_ERROR_WITH_MESSAGE` in `report_and_reraise()` left the whole suite green. That line *is* the visible half of the "a failed report must not swallow the program's own error" fix -- without it the secondary failure goes nowhere and the only evidence is the original error still being the one raised. Nothing looked at it, so nothing noticed. `tests/trap_verbs.c` now redirects `akerr_log_method` into a buffer and drives the path with a sink that refuses every write, which is the only way left to make reporting fail now that the `TRAP` dispatch no longer allocates. It asserts the log line and that `errclass` still records the program's own error, and it kills the mutant. TODO.md gains a section for what the run found and what it did not: it was cut off at 551 of 997 mutants after ninety minutes, so 446 are unexamined and a full run belongs in the release workflow rather than here. The three regions this work touched were all covered, and the survivors in the skipped-block guard are named and assessed -- three of the four are equivalent for any program that can actually be written, and the one that is not would need a test with a loop on line 1. Also recorded: src/variable.c at 68.2% with every mutant on the inline-storage lines killed, and src/runtime_trap.c at 82.5% with no survivors in `set_error_variables()`. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-02 02:17:17 -04:00
#include <stdarg.h>
#include <stdio.h>
Finish the language: every remaining verb group, and the defects that blocked them Closes groups A, C, D, E, F, H and J of TODO.md section 4, plus RESTORE and RENUMBER, and closes section 6 -- all seventeen reference defects. Seven of those turned out to have been fixed or never ported and nobody had written it down; the audit records the evidence for each. Two of the seventeen were real. math_plus mutated its left operand when the operand was mutable, so A# + 1 could modify A#; it was gated on FOR/NEXT coverage because NEXT relied on the mutation, so tests/for_next.c came first and NEXT now writes the counter back itself. And the binary operators summed both numeric fields of their right operand, which no BASIC program can reach -- that one needed a test written against the value API. Writing the tests turned up eight defects nobody had listed. Seven are fixed: IF A = 2 THEN was a parse error; only == worked IF ... AND ... was a parse error, because a condition parsed as one relation IF A = 1 OR B = 2 THEN was silently always false, and so was IF A THEN EXIT before any NEXT restarted the program and exhausted the variable pool READ never found a DATA line above it, and swallowed the lines between PRINT 2 + 2 at the prompt was filed as program text instead of answering a short read discarded its bytes, so COPY produced empty files every verb taking an argument list said "peek() returned nil token!" on none The eighth is not fixed and cannot be quietly: a FOR whose step overshoots runs its body one extra time, and FOR I = 1 TO 1 runs it zero times. The two errors cancel for a step of 1, which is why neither was noticed. Correcting them changes the expected output of a checked-in acceptance file, and tests/reference/README.md forbids editing one to suit this interpreter. It is tests/for_semantics.c in AKBASIC_KNOWN_FAILING_TESTS, asserting the correct contract, and TODO.md items 19 and 20. Sprites are real libakgl actors with a renderfunc of their own, because akgl_actor_render draws every sprite square and an actor has no per-axis scale. Both are filed upstream. SPRSAV takes an image file, an SSHAPE handle or a 63-element integer array -- a string here cannot hold a zero byte. Verbs that need hardware that does not exist are refused by name with the reason rather than faked: SYS, HEADER, COLLECT, BACKUP, BOOT, FILTER, and DIRECTORY, which is refused for a missing libakstdlib wrapper filed upstream. 94 tests in the default build, 93 with SDL, 94 under ASan and UBSan, doxygen clean. The Go acceptance corpus stayed green throughout. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-31 21:50:37 -04:00
#include <string.h>
#include <akbasic/error.h>
#include <akbasic/runtime.h>
#include "harness.h"
#include "testutil.h"
/** @brief Run a program to completion in RUN mode, from a string. */
static akerr_ErrorContext AKERR_NOIGNORE *run_program(const char *source)
{
PREPARE_ERROR(errctx);
PASS(errctx, harness_start(NULL));
PASS(errctx, akbasic_runtime_load(&HARNESS_RUNTIME, source));
PASS(errctx, akbasic_runtime_start(&HARNESS_RUNTIME, AKBASIC_MODE_RUN));
/* Bounded: a bare RESUME that never fixes anything is an infinite retry. */
PASS(errctx, akbasic_runtime_run(&HARNESS_RUNTIME, 4000));
SUCCEED_RETURN(errctx);
}
/** @brief An armed TRAP swallows the report and enters the handler instead. */
static void test_trap_catches(void)
{
TEST_REQUIRE_OK(run_program("10 TRAP 100\n"
"20 DIM Q#(2)\n"
"30 PRINT Q#(9)\n"
"40 PRINT \"AFTER\"\n"
"50 END\n"
"100 PRINT \"TRAPPED \" + ER# + \" ON LINE \" + EL#\n"
"110 RESUME NEXT\n"));
/*
* AKBASIC_ERR_BOUNDS is 515: base 512 plus SYNTAX, TYPE, UNDEFINED. Spelled
* out rather than referenced so that a renumbering of the error band shows up
* here as a failure rather than as a silently different program.
*/
TEST_REQUIRE_STR(HARNESS_OUTPUT, "TRAPPED 515 ON LINE 30\nAFTER\n");
TEST_REQUIRE(strstr(HARNESS_OUTPUT, "RUNTIME ERROR") == NULL,
"a trapped error must not also be reported, got \"%s\"", HARNESS_OUTPUT);
harness_stop();
}
/**
* @brief A *parse* error is trappable too, and carries its code into ER#.
*
* Found while writing the error-code appendix, and it was the worst kind of
* defect: arming a handler made errors disappear. `akbasic_runtime_error()`
* intercepts for the trap and leaves `errclass` clear, but the parse branch of
* `process_line_run()` set `run_finished_mode` regardless -- QUIT for a file --
* so the run ended before the next line boundary the handler would have been
* entered at. No error line, because the trap suppressed it; no handler, because
* the program had stopped. Nothing on stdout at all.
*
* The second half is that the same branch never recorded the status, so the
* handler that now runs used to be handed `ER# 0`.
*/
static void test_trap_catches_parse_error(void)
{
TEST_REQUIRE_OK(run_program("10 TRAP 100\n"
"20 SCALE\n"
"30 PRINT \"AFTER\"\n"
"40 END\n"
"100 PRINT \"TRAPPED \" + ER# + \" ON LINE \" + EL#\n"
"110 RESUME NEXT\n"));
/* 512 is AKBASIC_ERR_SYNTAX, the base of the band. Spelled out for the same
reason test_trap_catches() spells out 515. */
TEST_REQUIRE_STR(HARNESS_OUTPUT, "TRAPPED 512 ON LINE 20\nAFTER\n");
TEST_REQUIRE(strstr(HARNESS_OUTPUT, "PARSE ERROR") == NULL,
"a trapped parse error must not also be reported, got \"%s\"",
HARNESS_OUTPUT);
harness_stop();
/* And with no trap it still reports and still stops -- the half of the
behaviour that was never broken and must not become so. */
TEST_REQUIRE_OK(run_program("10 SCALE\n"
"20 PRINT \"AFTER\"\n"));
TEST_REQUIRE(strstr(HARNESS_OUTPUT, "? 10 : PARSE ERROR") != NULL,
"an untrapped parse error should be reported, got \"%s\"", HARNESS_OUTPUT);
TEST_REQUIRE(strstr(HARNESS_OUTPUT, "AFTER") == NULL,
"an untrapped parse error should stop the run, got \"%s\"", HARNESS_OUTPUT);
harness_stop();
}
Finish the language: every remaining verb group, and the defects that blocked them Closes groups A, C, D, E, F, H and J of TODO.md section 4, plus RESTORE and RENUMBER, and closes section 6 -- all seventeen reference defects. Seven of those turned out to have been fixed or never ported and nobody had written it down; the audit records the evidence for each. Two of the seventeen were real. math_plus mutated its left operand when the operand was mutable, so A# + 1 could modify A#; it was gated on FOR/NEXT coverage because NEXT relied on the mutation, so tests/for_next.c came first and NEXT now writes the counter back itself. And the binary operators summed both numeric fields of their right operand, which no BASIC program can reach -- that one needed a test written against the value API. Writing the tests turned up eight defects nobody had listed. Seven are fixed: IF A = 2 THEN was a parse error; only == worked IF ... AND ... was a parse error, because a condition parsed as one relation IF A = 1 OR B = 2 THEN was silently always false, and so was IF A THEN EXIT before any NEXT restarted the program and exhausted the variable pool READ never found a DATA line above it, and swallowed the lines between PRINT 2 + 2 at the prompt was filed as program text instead of answering a short read discarded its bytes, so COPY produced empty files every verb taking an argument list said "peek() returned nil token!" on none The eighth is not fixed and cannot be quietly: a FOR whose step overshoots runs its body one extra time, and FOR I = 1 TO 1 runs it zero times. The two errors cancel for a step of 1, which is why neither was noticed. Correcting them changes the expected output of a checked-in acceptance file, and tests/reference/README.md forbids editing one to suit this interpreter. It is tests/for_semantics.c in AKBASIC_KNOWN_FAILING_TESTS, asserting the correct contract, and TODO.md items 19 and 20. Sprites are real libakgl actors with a renderfunc of their own, because akgl_actor_render draws every sprite square and an actor has no per-axis scale. Both are filed upstream. SPRSAV takes an image file, an SSHAPE handle or a 63-element integer array -- a string here cannot hold a zero byte. Verbs that need hardware that does not exist are refused by name with the reason rather than faked: SYS, HEADER, COLLECT, BACKUP, BOOT, FILTER, and DIRECTORY, which is refused for a missing libakstdlib wrapper filed upstream. 94 tests in the default build, 93 with SDL, 94 under ASan and UBSan, doxygen clean. The Go acceptance corpus stayed green throughout. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-31 21:50:37 -04:00
/** @brief With no TRAP the error is reported and the program stops, as before. */
static void test_untrapped_still_reports(void)
{
TEST_REQUIRE_OK(run_program("10 DIM Q#(2)\n"
"20 PRINT Q#(9)\n"
"30 PRINT \"UNREACHED\"\n"));
TEST_REQUIRE(strstr(HARNESS_OUTPUT, "RUNTIME ERROR") != NULL,
"an untrapped error should be reported, got \"%s\"", HARNESS_OUTPUT);
TEST_REQUIRE(strstr(HARNESS_OUTPUT, "UNREACHED") == NULL,
"an untrapped error should stop the program, got \"%s\"", HARNESS_OUTPUT);
harness_stop();
}
/** @brief Bare TRAP disarms, and the next error is reported normally again. */
static void test_trap_disarms(void)
{
TEST_REQUIRE_OK(run_program("10 TRAP 100\n"
"20 TRAP\n"
"30 DIM Q#(2)\n"
"40 PRINT Q#(9)\n"
"50 PRINT \"UNREACHED\"\n"
"100 PRINT \"NOT SEEN\"\n"
"110 RESUME NEXT\n"));
TEST_REQUIRE(strstr(HARNESS_OUTPUT, "RUNTIME ERROR") != NULL,
"a disarmed TRAP should let the error through, got \"%s\"", HARNESS_OUTPUT);
TEST_REQUIRE(strstr(HARNESS_OUTPUT, "NOT SEEN") == NULL,
"the handler should not have run, got \"%s\"", HARNESS_OUTPUT);
harness_stop();
}
/** @brief A handler may be named by label, which is where a handler usually sits. */
static void test_trap_by_label(void)
{
TEST_REQUIRE_OK(run_program("10 TRAP HANDLER\n"
"20 DIM Q#(2)\n"
"30 PRINT Q#(9)\n"
"40 PRINT \"AFTER\"\n"
"50 END\n"
"100 LABEL HANDLER\n"
"110 PRINT \"CAUGHT\"\n"
"120 RESUME NEXT\n"));
TEST_REQUIRE_STR(HARNESS_OUTPUT, "CAUGHT\nAFTER\n");
harness_stop();
}
/** @brief RESUME's three forms land on three different lines. */
static void test_resume_forms(void)
{
/* RESUME <line> goes where it is told. */
TEST_REQUIRE_OK(run_program("10 TRAP 100\n"
"20 DIM Q#(2)\n"
"30 PRINT Q#(9)\n"
"40 PRINT \"NOT HERE\"\n"
"50 PRINT \"HERE\"\n"
"60 END\n"
"100 RESUME 50\n"));
TEST_REQUIRE_STR(HARNESS_OUTPUT, "HERE\n");
harness_stop();
/*
* Bare RESUME retries the failing line. The handler fixes the array first,
* so the retry succeeds -- which is the only way a bare RESUME terminates,
* and the reason the verb exists.
*/
TEST_REQUIRE_OK(run_program("10 TRAP 100\n"
"20 DIM Q#(2)\n"
"30 PRINT Q#(N#)\n"
"40 PRINT \"AFTER\"\n"
"50 END\n"
"100 N# = 1\n"
"110 RESUME\n"));
TEST_REQUIRE_STR(HARNESS_OUTPUT, "0\nAFTER\n");
harness_stop();
}
/** @brief RESUME outside a handler is refused rather than popping something else. */
static void test_resume_outside_handler(void)
{
TEST_REQUIRE_OK(run_program("10 RESUME NEXT\n"));
TEST_REQUIRE(strstr(HARNESS_OUTPUT, "outside the context of a TRAP handler") != NULL,
"a bare RESUME should be refused, got \"%s\"", HARNESS_OUTPUT);
harness_stop();
}
/**
* @brief An error inside a handler is reported rather than re-entering it.
*
* Otherwise a broken handler traps its own failure forever and the program can
* never be stopped. A C128 behaves the same way.
*/
static void test_error_inside_handler(void)
{
TEST_REQUIRE_OK(run_program("10 TRAP 100\n"
"20 DIM Q#(2)\n"
"30 PRINT Q#(9)\n"
"40 END\n"
"100 PRINT \"IN HANDLER\"\n"
"110 PRINT Q#(9)\n"
"120 RESUME NEXT\n"));
TEST_REQUIRE(strstr(HARNESS_OUTPUT, "IN HANDLER") != NULL,
"the handler should have run once, got \"%s\"", HARNESS_OUTPUT);
TEST_REQUIRE(strstr(HARNESS_OUTPUT, "RUNTIME ERROR") != NULL,
"an error inside the handler should be reported, got \"%s\"", HARNESS_OUTPUT);
harness_stop();
}
/** @brief ERR() names a status code, using the registry every code is named in. */
static void test_err_function(void)
{
TEST_REQUIRE_OK(run_program("10 TRAP 100\n"
"20 DIM Q#(2)\n"
"30 PRINT Q#(9)\n"
"40 END\n"
"100 PRINT ERR(ER#)\n"
"110 RESUME NEXT\n"));
TEST_REQUIRE(strstr(HARNESS_OUTPUT, "Out Of Bounds") != NULL,
"ERR should name the status, got \"%s\"", HARNESS_OUTPUT);
harness_stop();
/*
* A code nobody reserved reads back as libakerror's own "Unknown Error",
* which is the same string a stack trace would print for it.
*/
TEST_REQUIRE_OK(run_program("10 PRINT ERR(30000)\n"));
TEST_REQUIRE_STR(HARNESS_OUTPUT, "Unknown Error\n");
harness_stop();
}
Enter a TRAP handler when the variable table is full Two separable faults, both on the path a program takes when it is already in trouble. `akbasic_trap_set_error_variables()` reached `ER#` and `EL#` through `akbasic_runtime_global()`, which *creates* a name the program never used -- and creating one takes a variable slot. So a program that had filled the 128-slot table could not have its handler entered at all, and because the failure happened inside the error path rather than raising, nothing was reported and the program carried on with the failing statement's effect quietly missing. A wrong answer delivered as a right one, which is worse than an abort. `akbasic_runtime_reserve_globals()` now creates both at runtime init, where there is always room, and `clear_variables()` puts them back after `CLR` and `NEW` empty the table. Second: `report_and_reraise()` used a plain `PASS` around the report, so a failure while reporting *replaced* the error the program had actually made -- "Maximum runtime variables reached" in place of the subscript that was out of range. The secondary failure is now logged and the original is re-raised, which is what a user needs to hear. **The reduction in TODO.md no longer reproduces, and not because of this.** The value-pool fix in the previous commit made a scalar free, so the pool can no longer be emptied by creating names. The defect was still live through the variable table: 124 names and a TRAP armed, and the handler was silently skipped. That is what the new test in tests/trap_verbs.c pins, together with the invariant -- both globals present before a program runs. Verified by reverting the fix against the new tests: both fail, and the second prints the log line the swallow used to eat, "could not report a BASIC error 515 (Out Of Bounds): Maximum runtime variables reached". TODO.md section 6 item 33, struck. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-01 23:48:56 -04:00
/**
* @brief `ER#` and `EL#` exist before a program runs, so the dispatch cannot fail.
*
* The dispatch used to reach them through akbasic_runtime_global(), which
* *creates* a name the program never used -- and creating one takes a variable
* slot. So the handler could not be entered by a program that had filled the
* table, which is the one kind of program most likely to raise. Asserted here as
* the invariant rather than only through the symptom below, because the symptom
* needs 124 lines of setup and this needs none.
*/
static void test_error_globals_are_reserved(void)
{
bool haveer = false;
bool haveel = false;
int i = 0;
TEST_REQUIRE_OK(harness_start(NULL));
for ( i = 0; i < AKBASIC_MAX_VARIABLES; i++ ) {
if ( !HARNESS_RUNTIME.variables[i].used ) {
continue;
}
if ( strcmp(HARNESS_RUNTIME.variables[i].name, "ER#") == 0 ) {
haveer = true;
}
if ( strcmp(HARNESS_RUNTIME.variables[i].name, "EL#") == 0 ) {
haveel = true;
}
}
TEST_REQUIRE(haveer, "ER# must exist as soon as the runtime is initialised");
TEST_REQUIRE(haveel, "EL# must exist as soon as the runtime is initialised");
harness_stop();
}
/**
* @brief A handler is entered even by a program that has filled the variable table.
*
* TODO.md section 6 item 33, and much worse than an abort: the dispatch failed
* *inside* the error path rather than raising, so nothing was reported, the
* handler never ran, and the program carried on with the failing statement's
* effect quietly missing -- a wrong answer delivered as a right one.
*
* 124 names, which leaves the table with no room to invent `ER#` and `EL#` at
* the moment they are wanted. It is 124 rather than 128 because the runtime
* reserves those two and the loop counter is a third.
*/
static void test_handler_runs_with_a_full_variable_table(void)
{
char source[AKBASIC_MAX_VARIABLES * 32];
size_t used = 0;
int i = 0;
used = (size_t)snprintf(source, sizeof(source), "10 TRAP CAUGHT\n");
for ( i = 0; i < 124; i++ ) {
used += (size_t)snprintf(source + used, sizeof(source) - used,
"%d V%03d# = %d\n", 20 + i, i, i);
}
snprintf(source + used, sizeof(source) - used,
"800 DIM TOOBIG#(5000)\n"
"810 PRINT \"NOT TRAPPED\"\n"
"820 END\n"
"900 LABEL CAUGHT\n"
"910 PRINT \"TRAPPED\"\n"
"920 END\n");
TEST_REQUIRE_OK(run_program(source));
TEST_REQUIRE_STR(HARNESS_OUTPUT, "TRAPPED\n");
harness_stop();
}
Assert the log line a failed report leaves behind A mutation run over src/runtime.c found that deleting the `LOG_ERROR_WITH_MESSAGE` in `report_and_reraise()` left the whole suite green. That line *is* the visible half of the "a failed report must not swallow the program's own error" fix -- without it the secondary failure goes nowhere and the only evidence is the original error still being the one raised. Nothing looked at it, so nothing noticed. `tests/trap_verbs.c` now redirects `akerr_log_method` into a buffer and drives the path with a sink that refuses every write, which is the only way left to make reporting fail now that the `TRAP` dispatch no longer allocates. It asserts the log line and that `errclass` still records the program's own error, and it kills the mutant. TODO.md gains a section for what the run found and what it did not: it was cut off at 551 of 997 mutants after ninety minutes, so 446 are unexamined and a full run belongs in the release workflow rather than here. The three regions this work touched were all covered, and the survivors in the skipped-block guard are named and assessed -- three of the four are equivalent for any program that can actually be written, and the one that is not would need a test with a loop on line 1. Also recorded: src/variable.c at 68.2% with every mutant on the inline-storage lines killed, and src/runtime_trap.c at 82.5% with no survivors in `set_error_variables()`. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-02 02:17:17 -04:00
/*
* A sink whose writes fail, and a place to catch what libakerror logs.
*
* Between them these drive the one path that has no BASIC-visible output at all:
* reporting an error failing while an error is already being reported.
*/
static char LOGGED[2048];
static void capture_log(const char *format, ...)
{
va_list args;
size_t used = strlen(LOGGED);
va_start(args, format);
vsnprintf(LOGGED + used, sizeof(LOGGED) - used, format, args);
va_end(args);
}
static akerr_ErrorContext AKERR_NOIGNORE *failing_write(akbasic_TextSink *self, const char *text)
{
PREPARE_ERROR(errctx);
(void)self; (void)text;
FAIL_RETURN(errctx, AKERR_IO, "this sink always fails");
}
static akerr_ErrorContext AKERR_NOIGNORE *failing_clear(akbasic_TextSink *self)
{
PREPARE_ERROR(errctx);
(void)self;
FAIL_RETURN(errctx, AKERR_IO, "this sink always fails");
}
/**
* @brief A failure while *reporting* does not replace the program's own error.
*
* `report_and_reraise()` used a plain `PASS` around the report, so anything that
* went wrong while reporting became the error that propagated -- and the one the
* program actually made was gone. A user was told "could not write to the sink"
* about a subscript that was out of range.
*
* Provoked here by a sink that refuses every write, which is the only way left
* to make reporting fail now that the `TRAP` dispatch no longer allocates. The
* two assertions are the two halves of the fix: the secondary failure is
* *logged*, where a developer will see it, and the *original* status is what
* comes back out.
*
* The log assertion needs `akerr_log_method` redirected, and that is the point:
* a mutation run showed the log line could be deleted with the whole suite still
* passing, because nothing looked at it. TODO.md section 6 item 33.
*/
static void test_a_failed_report_keeps_the_original_error(void)
{
akerr_ErrorLogFunction saved = akerr_log_method;
akerr_ErrorContext *raised = NULL;
LOGGED[0] = '\0';
TEST_REQUIRE_OK(harness_start(NULL));
/* Out of bounds, which is what the program did wrong. */
TEST_REQUIRE_OK(akbasic_runtime_load(&HARNESS_RUNTIME,
"10 DIM Q#(2)\n"
"20 PRINT Q#(9)\n"));
TEST_REQUIRE_OK(akbasic_runtime_start(&HARNESS_RUNTIME, AKBASIC_MODE_RUN));
HARNESS_SINK.write = failing_write;
HARNESS_SINK.writeln = failing_write;
HARNESS_SINK.clear = failing_clear;
akerr_log_method = capture_log;
raised = akbasic_runtime_run(&HARNESS_RUNTIME, 0);
akerr_log_method = saved;
test_discard_error(raised);
/*
* The secondary failure is logged rather than swallowed, and it says which
* one it was: "could not report" names the reporting, and the status on the
* end of the line is the *sink's* AKERR_IO. A mutation run showed this line
* could be deleted with the whole suite still green, because nothing looked
* at it.
*/
TEST_REQUIRE(strstr(LOGGED, "could not report a BASIC error") != NULL,
"the failure to report must be logged, got \"%s\"", LOGGED);
/*
* And the program's own error is still what happened: the class was recorded
* before the write that failed, so the run ended as a runtime error rather
* than carrying on past a statement that did not work.
*/
TEST_REQUIRE_INT(HARNESS_RUNTIME.errclass, AKBASIC_ERRCLASS_RUNTIME);
harness_stop();
}
Finish the language: every remaining verb group, and the defects that blocked them Closes groups A, C, D, E, F, H and J of TODO.md section 4, plus RESTORE and RENUMBER, and closes section 6 -- all seventeen reference defects. Seven of those turned out to have been fixed or never ported and nobody had written it down; the audit records the evidence for each. Two of the seventeen were real. math_plus mutated its left operand when the operand was mutable, so A# + 1 could modify A#; it was gated on FOR/NEXT coverage because NEXT relied on the mutation, so tests/for_next.c came first and NEXT now writes the counter back itself. And the binary operators summed both numeric fields of their right operand, which no BASIC program can reach -- that one needed a test written against the value API. Writing the tests turned up eight defects nobody had listed. Seven are fixed: IF A = 2 THEN was a parse error; only == worked IF ... AND ... was a parse error, because a condition parsed as one relation IF A = 1 OR B = 2 THEN was silently always false, and so was IF A THEN EXIT before any NEXT restarted the program and exhausted the variable pool READ never found a DATA line above it, and swallowed the lines between PRINT 2 + 2 at the prompt was filed as program text instead of answering a short read discarded its bytes, so COPY produced empty files every verb taking an argument list said "peek() returned nil token!" on none The eighth is not fixed and cannot be quietly: a FOR whose step overshoots runs its body one extra time, and FOR I = 1 TO 1 runs it zero times. The two errors cancel for a step of 1, which is why neither was noticed. Correcting them changes the expected output of a checked-in acceptance file, and tests/reference/README.md forbids editing one to suit this interpreter. It is tests/for_semantics.c in AKBASIC_KNOWN_FAILING_TESTS, asserting the correct contract, and TODO.md items 19 and 20. Sprites are real libakgl actors with a renderfunc of their own, because akgl_actor_render draws every sprite square and an actor has no per-axis scale. Both are filed upstream. SPRSAV takes an image file, an SSHAPE handle or a 63-element integer array -- a string here cannot hold a zero byte. Verbs that need hardware that does not exist are refused by name with the reason rather than faked: SYS, HEADER, COLLECT, BACKUP, BOOT, FILTER, and DIRECTORY, which is refused for a missing libakstdlib wrapper filed upstream. 94 tests in the default build, 93 with SDL, 94 under ASan and UBSan, doxygen clean. The Go acceptance corpus stayed green throughout. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-31 21:50:37 -04:00
int main(void)
{
test_trap_catches();
test_trap_catches_parse_error();
Finish the language: every remaining verb group, and the defects that blocked them Closes groups A, C, D, E, F, H and J of TODO.md section 4, plus RESTORE and RENUMBER, and closes section 6 -- all seventeen reference defects. Seven of those turned out to have been fixed or never ported and nobody had written it down; the audit records the evidence for each. Two of the seventeen were real. math_plus mutated its left operand when the operand was mutable, so A# + 1 could modify A#; it was gated on FOR/NEXT coverage because NEXT relied on the mutation, so tests/for_next.c came first and NEXT now writes the counter back itself. And the binary operators summed both numeric fields of their right operand, which no BASIC program can reach -- that one needed a test written against the value API. Writing the tests turned up eight defects nobody had listed. Seven are fixed: IF A = 2 THEN was a parse error; only == worked IF ... AND ... was a parse error, because a condition parsed as one relation IF A = 1 OR B = 2 THEN was silently always false, and so was IF A THEN EXIT before any NEXT restarted the program and exhausted the variable pool READ never found a DATA line above it, and swallowed the lines between PRINT 2 + 2 at the prompt was filed as program text instead of answering a short read discarded its bytes, so COPY produced empty files every verb taking an argument list said "peek() returned nil token!" on none The eighth is not fixed and cannot be quietly: a FOR whose step overshoots runs its body one extra time, and FOR I = 1 TO 1 runs it zero times. The two errors cancel for a step of 1, which is why neither was noticed. Correcting them changes the expected output of a checked-in acceptance file, and tests/reference/README.md forbids editing one to suit this interpreter. It is tests/for_semantics.c in AKBASIC_KNOWN_FAILING_TESTS, asserting the correct contract, and TODO.md items 19 and 20. Sprites are real libakgl actors with a renderfunc of their own, because akgl_actor_render draws every sprite square and an actor has no per-axis scale. Both are filed upstream. SPRSAV takes an image file, an SSHAPE handle or a 63-element integer array -- a string here cannot hold a zero byte. Verbs that need hardware that does not exist are refused by name with the reason rather than faked: SYS, HEADER, COLLECT, BACKUP, BOOT, FILTER, and DIRECTORY, which is refused for a missing libakstdlib wrapper filed upstream. 94 tests in the default build, 93 with SDL, 94 under ASan and UBSan, doxygen clean. The Go acceptance corpus stayed green throughout. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-31 21:50:37 -04:00
test_untrapped_still_reports();
test_trap_disarms();
test_trap_by_label();
test_resume_forms();
test_resume_outside_handler();
test_error_inside_handler();
test_err_function();
Enter a TRAP handler when the variable table is full Two separable faults, both on the path a program takes when it is already in trouble. `akbasic_trap_set_error_variables()` reached `ER#` and `EL#` through `akbasic_runtime_global()`, which *creates* a name the program never used -- and creating one takes a variable slot. So a program that had filled the 128-slot table could not have its handler entered at all, and because the failure happened inside the error path rather than raising, nothing was reported and the program carried on with the failing statement's effect quietly missing. A wrong answer delivered as a right one, which is worse than an abort. `akbasic_runtime_reserve_globals()` now creates both at runtime init, where there is always room, and `clear_variables()` puts them back after `CLR` and `NEW` empty the table. Second: `report_and_reraise()` used a plain `PASS` around the report, so a failure while reporting *replaced* the error the program had actually made -- "Maximum runtime variables reached" in place of the subscript that was out of range. The secondary failure is now logged and the original is re-raised, which is what a user needs to hear. **The reduction in TODO.md no longer reproduces, and not because of this.** The value-pool fix in the previous commit made a scalar free, so the pool can no longer be emptied by creating names. The defect was still live through the variable table: 124 names and a TRAP armed, and the handler was silently skipped. That is what the new test in tests/trap_verbs.c pins, together with the invariant -- both globals present before a program runs. Verified by reverting the fix against the new tests: both fail, and the second prints the log line the swallow used to eat, "could not report a BASIC error 515 (Out Of Bounds): Maximum runtime variables reached". TODO.md section 6 item 33, struck. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-01 23:48:56 -04:00
test_error_globals_are_reserved();
test_handler_runs_with_a_full_variable_table();
Assert the log line a failed report leaves behind A mutation run over src/runtime.c found that deleting the `LOG_ERROR_WITH_MESSAGE` in `report_and_reraise()` left the whole suite green. That line *is* the visible half of the "a failed report must not swallow the program's own error" fix -- without it the secondary failure goes nowhere and the only evidence is the original error still being the one raised. Nothing looked at it, so nothing noticed. `tests/trap_verbs.c` now redirects `akerr_log_method` into a buffer and drives the path with a sink that refuses every write, which is the only way left to make reporting fail now that the `TRAP` dispatch no longer allocates. It asserts the log line and that `errclass` still records the program's own error, and it kills the mutant. TODO.md gains a section for what the run found and what it did not: it was cut off at 551 of 997 mutants after ninety minutes, so 446 are unexamined and a full run belongs in the release workflow rather than here. The three regions this work touched were all covered, and the survivors in the skipped-block guard are named and assessed -- three of the four are equivalent for any program that can actually be written, and the one that is not would need a test with a loop on line 1. Also recorded: src/variable.c at 68.2% with every mutant on the inline-storage lines killed, and src/runtime_trap.c at 82.5% with no survivors in `set_error_variables()`. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-02 02:17:17 -04:00
test_a_failed_report_keeps_the_original_error();
Finish the language: every remaining verb group, and the defects that blocked them Closes groups A, C, D, E, F, H and J of TODO.md section 4, plus RESTORE and RENUMBER, and closes section 6 -- all seventeen reference defects. Seven of those turned out to have been fixed or never ported and nobody had written it down; the audit records the evidence for each. Two of the seventeen were real. math_plus mutated its left operand when the operand was mutable, so A# + 1 could modify A#; it was gated on FOR/NEXT coverage because NEXT relied on the mutation, so tests/for_next.c came first and NEXT now writes the counter back itself. And the binary operators summed both numeric fields of their right operand, which no BASIC program can reach -- that one needed a test written against the value API. Writing the tests turned up eight defects nobody had listed. Seven are fixed: IF A = 2 THEN was a parse error; only == worked IF ... AND ... was a parse error, because a condition parsed as one relation IF A = 1 OR B = 2 THEN was silently always false, and so was IF A THEN EXIT before any NEXT restarted the program and exhausted the variable pool READ never found a DATA line above it, and swallowed the lines between PRINT 2 + 2 at the prompt was filed as program text instead of answering a short read discarded its bytes, so COPY produced empty files every verb taking an argument list said "peek() returned nil token!" on none The eighth is not fixed and cannot be quietly: a FOR whose step overshoots runs its body one extra time, and FOR I = 1 TO 1 runs it zero times. The two errors cancel for a step of 1, which is why neither was noticed. Correcting them changes the expected output of a checked-in acceptance file, and tests/reference/README.md forbids editing one to suit this interpreter. It is tests/for_semantics.c in AKBASIC_KNOWN_FAILING_TESTS, asserting the correct contract, and TODO.md items 19 and 20. Sprites are real libakgl actors with a renderfunc of their own, because akgl_actor_render draws every sprite square and an actor has no per-axis scale. Both are filed upstream. SPRSAV takes an image file, an SSHAPE handle or a 63-element integer array -- a string here cannot hold a zero byte. Verbs that need hardware that does not exist are refused by name with the reason rather than faked: SYS, HEADER, COLLECT, BACKUP, BOOT, FILTER, and DIRECTORY, which is refused for a missing libakstdlib wrapper filed upstream. 94 tests in the default build, 93 with SDL, 94 under ASan and UBSan, doxygen clean. The Go acceptance corpus stayed green throughout. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-31 21:50:37 -04:00
return akbasic_test_failures;
}