On Tue, Jun 12, 2001 at 10:21:39PM -0500, Craig A. Berry wrote:
> At 7:45 PM -0700 6/12/01, Peter Prymmer wrote:
> >One way to fix that for this version of memchr might be the following:
> >
> >--- util.c;1 Tue Jun 12 17:33:57 2001
> >+++ util.c Tue Jun 12 17:52:44 2001
>
> No, what we really need is to fess up to having memchr, which we
> would if we didn't have a missing semicolon in our try.c. util.c
> compiles fine after this patch:
>
> --- configure.com;-0 Mon Jun 11 16:48:35 2001
> +++ configure.com Tue Jun 12 18:26:49 2001
> @@ -3654,7 +3654,7 @@
> $ WS "int main()"
> $ WS "{"
> $ WS "char * place;"
> -$ WS "place = memchr(""foo"", 47, 3)"
> +$ WS "place = memchr(""foo"", 47, 3);"
> $ WS "exit(0);"
> $ WS "}"
> $ CS
Ack.
> [end of patch]
>
> The memchr redefinition for platforms that don't have the real thing
> may still need fixing, but it's not our problem anymore ;-).
Peter's patch retracted since finding out platforms without working
memchr is nice.
> >--- ext/attrs/attrs.xs;1 Tue Jun 12 17:30:47 2001
> >+++ ext/attrs/attrs.xs Tue Jun 12 18:11:41 2001
> >@@ -25,7 +25,7 @@
>
> I haven't looked at this one yet.
Fixed by Doug.
> I looked into putting the DEBUGGING option on the command line, but
> it's messy because:
>
> cc/define=foo/define=bar
>
> ignores the foo definition and only remembers bar; to get both you
> have combine them in one define qualifier to say:
>
> cc/define=(foo,bar)
>
> which is a bit icky to do if foo and bar are discovered a few
> thousand lines apart in the configuration script and may have to be
> combined with different defines for different subsets of the build.
> I think MM_VMS.pm does this sort of thing for extensions, so we
> probably need to hack the Makefile.PL instead of the xs code. I'll
> look into that.
[standing by]
--
$jhi++; # http://www.iki.fi/jhi/
# There is this special biologist word we use for 'stable'.
# It is 'dead'. -- Jack Cohen
