Fix generator teardown leaks, add RETURN-in-GEN and LOOP conditions on DO EACH

Review findings and follow-ups from PR #61 review:

- runtime_generator.c: akbasic_runtime_release_generator() now releases the
  forGeneratorEnv of every scope it walks through. Abandoning a generator
  that was itself suspended inside a FOR EACH over another generator
  stranded the inner generator's pool slot; a loop doing so exhausted the
  twelve-slot pool and died far from the cause.
- runtime.c/runtime.h: new akbasic_runtime_unwind_to_environment(), the
  shared teardown for the error unwinds in pump_generator() and
  call_function() -- both previously bare prev_environment() loops with the
  same suspended-generator blindness.
- runtime_commands.c: bare RETURN standing in a GEN's own frame ends the
  generator exactly as END GEN does -- a GEN is a function at heart. RETURN
  with a value there is refused (values leave a GEN only through EMIT). The
  no-frame error message now says "GOSUB, DEF, or GEN".
- runtime_structure.c: LOOP WHILE/UNTIL composes with DO EACH -- checked
  after each trip with the loop variable still holding that trip's value; a
  condition that stops the loop abandons the generator exactly as EXIT
  does. Previously the condition was silently ignored, while the verb
  reference documented it as working.
- parser_commands.c: trailing tokens after the generator call on a FOR
  EACH/DO EACH line are refused at parse. Previously they sat unparsed and
  blew up only after the loop completed, when the parent scope resumed the
  line mid-statement -- an error at the loop's end pointing at its start.
- tests/generators.c: pool-exhaustion tests for the nested-abandonment and
  LOOP-condition paths, RETURN semantics tests, and a direct test of the
  unwind primitive. Three new golden pairs cover RETURN, LOOP conditions
  and the misplaced-condition parse error.
- docs: RETURN and LOOP-condition semantics in 04-control-flow.md and
  11-verb-reference.md; corrected the self-recursion analogy (functions
  are re-entrant here). TODO.md 1.10 records the generator design
  decisions the code comments were already citing, plus the zero-arg
  parameter-list limitation. MAINTENANCE.md gains the abandoned-generators
  invariant those comments also cited.

Co-Authored-By: Andrew Kesterson <andrew@aklabs.net>
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
Ishikawa
2026-08-06 11:29:17 -04:00
committed by Andrew Kesterson
parent f7e8d4b82b
commit a29c7f34fe
17 changed files with 426 additions and 13 deletions

View File

@@ -119,6 +119,137 @@ static void test_exit_partway_through(void)
harness_stop();
}
/**
* @brief Abandoning a generator that is itself suspended inside a FOR EACH
* over another generator releases the inner generator too.
*
* The inner generator's environment hangs off the *loop* scope inside OUTER's
* body as a child, off the parent chain -- the one place a bare parent walk
* never looks. Before akbasic_runtime_release_generator() recursed into
* `forGeneratorEnv`, every trip through this loop stranded one pool slot and
* the 13th trip died with "Environment pool exhausted".
*/
static void test_exit_releases_nested_generators(void)
{
TEST_REQUIRE_OK(run_program("10 GEN INNER(N#)\n"
"20 EMIT 1\n"
"30 EMIT 2\n"
"40 END GEN\n"
"50 GEN OUTER(N#)\n"
"60 FOR EACH I# IN INNER(0)\n"
"70 EMIT I#\n"
"80 NEXT I#\n"
"90 END GEN\n"
"100 FOR K# = 1 TO 40\n"
"110 FOR EACH V# IN OUTER(0)\n"
"120 EXIT\n"
"130 NEXT V#\n"
"140 NEXT K#\n"
"150 PRINT \"DONE\"\n"));
TEST_REQUIRE_STR(HARNESS_OUTPUT, "DONE\n");
harness_stop();
}
/**
* @brief A LOOP UNTIL that stops a DO EACH early releases the generator it
* abandons, every time.
*/
static void test_loop_condition_releases_generator(void)
{
TEST_REQUIRE_OK(run_program("10 GEN COUNTUP(N#)\n"
"20 FOR I# = 1 TO N#\n"
"30 EMIT I#\n"
"40 NEXT I#\n"
"50 END GEN\n"
"60 FOR K# = 1 TO 40\n"
"70 DO EACH V# IN COUNTUP(10)\n"
"80 LOOP UNTIL V# = 2\n"
"90 NEXT K#\n"
"100 PRINT \"DONE\"\n"));
TEST_REQUIRE_STR(HARNESS_OUTPUT, "DONE\n");
harness_stop();
}
/**
* @brief RETURN standing in a GEN's own frame ends the generator early,
* exactly as END GEN would -- a GEN is a function at heart.
*/
static void test_return_ends_generator(void)
{
TEST_REQUIRE_OK(run_program("10 GEN G(N#)\n"
"20 EMIT 1\n"
"30 RETURN\n"
"40 EMIT 2\n"
"50 END GEN\n"
"60 FOR EACH V# IN G(0)\n"
"70 PRINT V#\n"
"80 NEXT V#\n"
"90 PRINT \"DONE\"\n"));
TEST_REQUIRE_STR(HARNESS_OUTPUT, "1\nDONE\n");
harness_stop();
}
/**
* @brief RETURN with a value inside a GEN is refused: values leave a GEN one
* at a time, through EMIT, and there is no return slot waiting.
*/
static void test_return_value_in_generator_refused(void)
{
TEST_REQUIRE_OK(run_program_bounded("10 GEN G(N#)\n"
"20 EMIT 1\n"
"30 RETURN 99\n"
"40 END GEN\n"
"50 FOR EACH V# IN G(0)\n"
"60 PRINT V#\n"
"70 NEXT V#\n"
"80 PRINT \"UNREACHABLE\"\n", 2000));
TEST_REQUIRE(strstr(HARNESS_OUTPUT, "1\n") != NULL,
"expected the first EMIT in \"%s\"", HARNESS_OUTPUT);
TEST_REQUIRE(strstr(HARNESS_OUTPUT, "UNREACHABLE") == NULL,
"RETURN with a value must stop the run, got \"%s\"", HARNESS_OUTPUT);
harness_stop();
}
/**
* @brief The unwind primitive releases a popped scope's suspended generator.
*
* Built by hand rather than through BASIC because the paths that need this --
* the error unwinds in pump_generator() and call_function() -- only trigger
* on C-level failures a program cannot politely ask for. The shape is the
* one EMIT leaves behind: a loop scope holding a detached generator child,
* with a further scope active above it.
*/
static void test_unwind_releases_suspended_generators(void)
{
akbasic_Environment *root = NULL;
akbasic_Environment *loopenv = NULL;
akbasic_Environment *genenv = NULL;
akbasic_Environment *forenv = NULL;
TEST_REQUIRE_OK(harness_start(NULL));
root = HARNESS_RUNTIME.environment;
TEST_REQUIRE_OK(akbasic_runtime_new_environment(&HARNESS_RUNTIME));
loopenv = HARNESS_RUNTIME.environment;
loopenv->isEachLoop = true;
TEST_REQUIRE_OK(akbasic_runtime_new_environment(&HARNESS_RUNTIME));
genenv = HARNESS_RUNTIME.environment;
genenv->isGenerator = true;
TEST_REQUIRE_OK(akbasic_runtime_detach_environment(&HARNESS_RUNTIME));
loopenv->forGeneratorEnv = genenv;
TEST_REQUIRE_OK(akbasic_runtime_new_environment(&HARNESS_RUNTIME));
forenv = HARNESS_RUNTIME.environment;
TEST_REQUIRE_OK(akbasic_runtime_unwind_to_environment(&HARNESS_RUNTIME, root));
TEST_REQUIRE(HARNESS_RUNTIME.environment == root,
"unwind must land on the target scope");
TEST_REQUIRE(!forenv->used && !loopenv->used && !genenv->used,
"unwind must release the chain and the suspended generator");
harness_stop();
}
/**
* @brief A GEN invoked like an ordinary function, rather than through FOR
* EACH/DO EACH, fails cleanly.
@@ -216,6 +347,11 @@ int main(void)
test_exit_releases_generator_for_each();
test_exit_releases_generator_do_each();
test_exit_partway_through();
test_exit_releases_nested_generators();
test_loop_condition_releases_generator();
test_return_ends_generator();
test_return_value_in_generator_refused();
test_unwind_releases_suspended_generators();
test_called_like_a_function();
test_emit_outside_gen();
test_self_recursion_refused();

View File

@@ -0,0 +1,8 @@
10 REM A DO EACH takes its condition on the LOOP, not on the DO line.
20 GEN ONE(N#)
30 EMIT N#
40 END GEN
50 DO EACH V# IN ONE(1) WHILE V# < 9
60 PRINT V#
70 LOOP
80 PRINT "UNREACHABLE"

View File

@@ -0,0 +1,2 @@
? 50 : PARSE ERROR DO EACH takes its WHILE/UNTIL on the LOOP, and nothing else here

View File

@@ -0,0 +1,15 @@
10 REM A WHILE or UNTIL on the LOOP composes with DO EACH: it is checked
20 REM after each trip through the body, with the loop variable still
30 REM holding that trip's value. Stopping abandons the generator cleanly.
40 GEN COUNTUP(N#)
50 FOR I# = 1 TO N#
60 EMIT I#
70 NEXT I#
80 END GEN
90 DO EACH V# IN COUNTUP(10)
100 PRINT V#
110 LOOP UNTIL V# = 3
120 DO EACH W# IN COUNTUP(4)
130 PRINT W# * 10
140 LOOP WHILE W# < 3
150 PRINT "DONE"

View File

@@ -0,0 +1,7 @@
1
2
3
10
20
30
DONE

View File

@@ -0,0 +1,15 @@
10 REM RETURN ends a GEN early, exactly as END GEN would: a GEN is a
20 REM function at heart, and only EMIT is different about it. Like a
30 REM GOSUB's or DEF's RETURN, it must stand in the GEN's own scope,
40 REM not inside a FOR or DO the body opened.
50 GEN FIRSTFEW(N#)
60 EMIT 1
70 IF N# < 2 THEN RETURN
80 EMIT 2
90 IF N# < 3 THEN RETURN
100 EMIT 3
110 END GEN
120 FOR EACH V# IN FIRSTFEW(2)
130 PRINT V#
140 NEXT V#
150 PRINT "DONE"

View File

@@ -0,0 +1,3 @@
1
2
DONE