[Gc] Re: Testing GC on IA-64

Ludovic Courtès ludo at gnu.org
Mon Sep 14 05:14:16 PDT 2009


Ivan Maidanski <ivmai at mail.ru> writes:

>> > (Also, GC_INITIAL_HEAP_SIZE might be specified thus resulting in such abort.)
>>
>> To avoid this kind of problem,  tests.am' should have something like
>> (info "(automake) Simple Tests"):
>>
>>  TESTS_ENVIRONMENT = GC_INITIAL_HEAP_SIZE= GC_ENABLE_INCREMENTAL=no
>
> May be, but for GC_INITIAL_HEAP_SIZE, this should (IMHO) be better handled in test.c code (eg: GC_get_heap_size() > max(max_heap_sz*n_tests, initial_heap_size_value)).

Sure.  What I meant is that, to improve reproducibility and ease of
analysis, the outcome of “make check” should not depend on parts of the
environment not captured by ‘configure’, most notably GC-specific
environment variables.

Thanks,
Ludo’.



More information about the Gc mailing list