Re: GC + Java finalization

[Jonas Hahnfeld]

Am Freitag, dem 19.11.2021 um 13:17 +0000 schrieb Maxime Devos:
> Jonas Hahnfeld schreef op do 15-07-2021 om 20:44 [+0200]:
> > +  /* For guardians, we use unordered finalization `a la Java. */
> 
> Maybe add the reasons why this is only done when a guardian is created?
> E.g.,
> 
> /* Don't use unordered finalization when not using guardians,
>     because Java finalization prevents fast collection of chains of
> unreachable objects */

I think this is a misunderstanding: bdwgc still defaults to Java-style
finalization, my patches only make it possible to disable it while
maintaining a working Guile.

> Not 100% about the exact purpose of avoiding Java-style finalization,
> you might want to adjust the wording somewhat ...

As discussed in July, it's a possible optimization because it allows
the garbage collector to reclaim more memory at once. I'm not sure if
it's worth it in real applications, but making Guile ready seemed like
a good idea.

Jonas

signature.asc
Description: This is a digitally signed message part