jhi at iki
Nov 27, 2006, 9:25 AM
Views: 324
Permalink
|
|
[PATCH] Re: [PATCH] abstract mempool header testing
|
|
Jarkko Hietaniemi wrote:
> I still don't know whether the mempool header testing for
> being in the same mempool (in perlio.c) is the right thing
> to do but to combat cut-and-pastitis (same code in op.c)
> here's a patch that abstracts this testing a bit.
The attached patch replaces the earlier one and it adds not doing
the PERL_TRACK_MEMPOOL if the PerlMemShared* are just plain malloc
et alia. What this means is that the PERL_TRACK_MEMPOOL is *NOT*
done in non-Win32/NetWare. I didn't tie the logic to
PERL_IMPLICIT_SYS as such, though I was much tempted to
(see
http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2006-11/msg00994.html
for additional pointers)
It is very possible that this patch is not quite right, but at least
it removes the valgrind memcheck hit, but I really still do not quite
get the PERL_TRACK_MEMPOOL stuff, and I really do hope Jan Dubois will
chime in at some point to resolve this mess.
|
mempool2.pat
[PATCH] Re: [PATCH] abstract mempool header testing
