This is the mail archive of the guile@sources.redhat.com mailing list for the Guile project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: GUILE's GC - why we struggling to solve already solved pr oblems ? (was Re: finding the top of the stack in linux)


19-Jul-00 09:20 you wrote:
> jimb@savonarola.red-bean.com writes:
>> > > I'm very interested in seeing the results of an actual try.  Would you
>> > > be willing to try the experiment, and tell us how it goes?
>> >
>> > But wasn't that the very experiment han-wen conducted this morning?
>>
>> Han-wen says that was just a quick-and-dirty try.  It slowed down
>> Guile by 70%, but I'm interested in seeing the result when folks make
>> proper use of Boehm's stubborn allocation, precise marking, etc.

> folks should also remember that you don't need to know about GUILE's
> GC to do this: when you use Boehm, you can just junk most of gc.c

Unfrtunatelly this is not true: you can "just junk most of gc.c" if you
do not care about performance. You you DO care about performance then you
just can not do it. Even more: you must understood BOTH GUILE internals
AND Boehm's GC internals (to some extent - just enough to push knobs;
Boehm's GC has a lot of knobs internally but very few are available and
well-documented in public interface). Not easy task, I afraid :-/
Eventually I'll try to dig deep in this problem but do not hold you
breath - I'm not sure if I'll be able to do it this summer :-(




Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]