Re: GC + Java finalization

[Maxime Devos]

Jonas Hahnfeld schreef op vr 19-11-2021 om 14:53 [+0100]:
> [...]
> The straight-forward solution is to statically tie the lifetime of
> regex_t to that of the smob. Which we cannot do with GC, but simply
> with malloc+free - as implemented in the patch.

OK, but for clarity I recommend adding a comment like

/* When not using Java-style finalisation, we must make sure
   the memory for the regex_t is only freed after regfree.
   To do so, use scm_malloc+free instead of scm_gc_free such
   that boehmgc will only free the regex_t when we ask it to
   in regex_free, and it won't automatically free the regex_t
   too early. */

Otherwise, it would be easy to think ‘Hmm, this code would be
simpler if we just use scm_gc_malloc and remove the unnecessary
free’ and accidentally introduce a bug ...

Greetings,
Maxime