dequeue

Primarily for my own convenience I've created a CVS module in my own
repository with my current set of patches in it.  Should Schwern
decide that CVS is a good enough solution I'll happily make a tarball
of the state of things and hand it off.

cvs -z3 -d :pserver:[email protected]:/home/richardc/cvs-repository co perl1
http://mirth.unixbeard.net/cgi-bin/viewcvs.cgi/perl1/

I also ran the full testsuite through valgrind, the results[0] of
which seem to indicate that Randals setjmp/longjmp hunch holds, but
also explains why perls malloc is flaky on ppc.  Using perls malloc
also seems to paper over some different cracks[1]

On that box I still see one test failing (t/op.flip#8), or 3 scripts
with a system malloc, so I can't claim the same success as Glenn.
This is making me wonder how easily the smoke suite for perl5 can be
used for smoking perl1, is anyone familiar with it?

[0] http://unixbeard.net/~richardc/lab/grinder/perlmalloc/
../Configure -Dcc="gcc-3.1 -g -fwritable-strings" -Dlibs="-lcrypt" -des 

[1] http://unixbeard.net/~richardc/lab/grinder/sysmalloc/
../Configure -Uusemymalloc -Dcc="gcc-3.1 -g -fwritable-strings" -Dlibs="-lcrypt" -des 

-- 
Richard Clamp <[email protected]>