David Champion wrote:
> I tend to avoid perl/python/etc when more basic utilities will do.  It's
> true that virtually every system has perl preinstalled now, but mutt
> is also used on very bare-bones platforms (which is why the C version
> exists; there's a C compiler by definition).

Okay, sounds reasonable to me.  I was a little wary of introducing a
perl dependency too.

> 1. We can support BSD's sed for txt2c.sh. The difference is just that
> BSD sed only supports \n a an ascii escape.  (Will need literals in the
> makefile.)  But -

I barely know GNU sed, if you have a moment I would appreciate if you
could suggest a patch.  :-)

> 2. Why does the compiler fail? That needs to be addressed too.

I believe this is because of the $< in
  txt2c: txt2c.c
    -$${NATIVECC-$(CC)} -o $@ $<

However it looks like we are using this variable in a couple other
Makefiles too: intl/Makefile.in and po/Makefile.in.in.

The autoconf manual mentions $< can't be portably used except in
"inference rules and in the '.DEFAULT' rule".  I'll have to check the
other usages later, but for now we can probably just list txt2c.c
instead of $<.

-- 
Kevin J. McCarthy
GPG Fingerprint: 8975 A9B3 3AA3 7910 385C  5308 ADEF 7684 8031 6BDA
http://www.8t8.us/configs/gpg-key-transition-statement.txt

Attachment: signature.asc
Description: PGP signature

Reply via email to