This is the mail archive of the java@gcc.gnu.org mailing list for the Java project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: _Jv_SetMaximumHeapSize


On Sun, 21 Sep 2003, Erik Poupaert wrote:
> They seem to garbage-collect but make sure as well that reference counting
> de-allocates memory as early as possible. (I understand that reference counting in
> itself carries both a memory as well as a performance penalty).
>
> On the other side, it would increase the sense of control that is lacking now when
> one observes evergrowing memory usage (even when it is justified) without the ability
> to do something about it... Or maybe the whole idea is silly for some other reason I
> am not aware of?

It doesn't follow that deallocating memory sooner as with reference
counting avoids unbounded heap growth.  Faulty application code is still
the primary cause of such failures.

The GC used in libgcj is tunable and can be made to collect as often as
you like (look at GC_enable_incremental, GC_free_space_divisor in gc.h).

I'm well aware myself of the tradeoffs between reference counting and
mark/sweep collection.  There is extensive research available.  I could
point you to materials and/or other mailing lists, as could others on
this list.  This isn't really the place to debate it, unless we are
talking about libgcj implementation specifics.

Jeff


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