Re: [gclist] Experience with conservative GC sought

[Frederick Smith <fms@CS.Cornell.EDU>]

Hi Mike,

I don't have experience with large long-running programs that relied on a
conservative collector. However, if the main problem you face with a
conservative collector is fragmentation then you might consider a
mostly-copying collector.  This could potentially help deal with
fragmentation.  We have a tech-report on it which you can find off my
homepage.

http://simon.cs.cornell.edu/Info/People/fms/fms.html

Let me know if you find this useful.

-Fred

Michael Sperber [Mr. Preprocessor] wrote:

> I'm looking for accounts, if part of research or just anecdotal, of
> experience with using conservative GC in long-running applications
> with large heaps and big changes in the size of the live data.