segfault on rfc2047-like subject

[Matthias Andree]

David Relson <relson at osagesoftware.com> writes:

> Looking in string.h, I noticed a prototype for memmem() and tried using
> it in.  However the compiler complained, so I wrong a quick/dirty
> version, which may be buggy.  With your changes, it doesn't seen
> necessary, but I've attached the patch for your amusement.

Check the manpage, you need to #define _GNU_SOURCE early for the
function, and it's GLIBC specific. Given that GNU libc has several hefty
design flaws (NIS integration my pet peeve this week
http://sources.redhat.com/bugzilla/show_bug.cgi?id=430), I'm not too
inclined to use this.

I have a debugged memstr ready to commit, interface:

/** find needle in haystack (which is treated as unsigned char *). */
void *memstr(const void *haystack, size_t n, const unsigned char *needle);

This is a bit lower than word_t-compared-against-str. As this doesn't
make sense in word_t (which is opaque, but we're fiddling with its
internals), I've dubbed it memstr not word_str.

> In a bit, I'm going to go listen to the presidential debate.  It's
> kind of amusing to hear Bush and Kerry together.  The way they dodge
> questions makes you think they could walk through a hurricane and not
> get wet.

Let's not take politics onto this developer list unless they affect
bogofilter directly.

-- 
Matthias Andree

Encrypted mail welcome: my GnuPG key ID is 0x052E7D95 (PGP/MIME preferred)