Ar an t-aonú lá déag de mí Eanair, scríobh Stephen J. Turnbull:
Aidan> There is a effective memory leak somewhere.
Buffers are usually a pretty small part of my XEmacs, less than 1/4.
So unless you're regularly creating and destroying 5MB buffers, I
doubt you'd notice much effect from the realloc code.
I am, that’s what reading mailboxes serially involves.
Also, it looks to me like on Linux and Mac OS X/X11 much of
XEmacs's
VSZ and probably RSS is out of its control. The X client libraries
are pigs, and the Xft client libraries maintain some middling large
data structures.
I’m seeing this effective leakage on 21.4.17 with just TTY support--that is,
no X libraries, no Xft libraries. LDAP and TLS support, though, for some
reason.
I don’t trust the memory usage stats in 21.5. I routinely have resident set
sizes of 500 MB on my desktop, and the total memory usage XEmacs tells me
about is a ridiculously tiny fraction of that.
I think that with MC_ALLOC we should be doing a better job of
tracking
the memory we use, just by regularizing the memory description stuff
and using it for more and more objects. Maybe it's time to put some
effort into looking for leaks.
Is anyone else seeing the incremental garbage collector tie up their process
at least as much as the old one? And is the reason no-one else has posted a
problem report on that crash of mine the other day because everyone’s turned
off incremental GC? Because I got the same crash again yesterday evening, so
it doesn’t appear to be anything atypical in my configuration.
--
I AM IN JAIL AND ALLOWED SEND ONLY ONE CABLE SINCE WAS ARRESTED WHILE
MEASURING FIFTEEN FOOT WALL OUTSIDE PALACE AND HAVE JUST FINISHED COUNTING
THIRTY EIGHT THOUSAND FIVE HUNDERED TWENTY TWO NAMES WHOS WHO IN MIDEAST.