[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PATCH 1/6] Avoid that autom4te complains about unexpanded macros
From: |
Eric Blake |
Subject: |
Re: [PATCH 1/6] Avoid that autom4te complains about unexpanded macros |
Date: |
Fri, 7 Nov 2008 21:36:31 +0000 (UTC) |
User-agent: |
Loom/3.14 (http://gmane.org/) |
Paolo Bonzini <bonzini <at> gnu.org> writes:
> In the meanwhile, the series cleans up some problematic usages of
> M4sh, so the preliminary patches can be applied independently.
A disclaimer - while I am a maintainer of autoconf (and hence m4sh), I am only
a contributor on libtool, and not the most prolific at that, so my reviews
don't hold quite as much weight on this list as on autoconf-patches. That
said, I still think I'm qualified to chime in with concurrence, whether or not
you wait for an additional reviewer.
> +++ b/libtoolize.m4sh
> @@ -1,5 +1,7 @@
> +m4_pattern_allow([dnl])
Forbidding dnl is a safety valve in m4sugar; by allowing it, you are no longer
able to catch typos in future edits that accidentally overquote and leave raw
dnl in the final output file. On the other hand, I think libtoolize.m4sh is
small enough to carefully audit (it is not like configure.ac, where catching
errors in third-party macros is nice, since libtoolize doesn't really depend on
third-party code). And since libtoolize's job is to scan configure.ac
(including ignoring dnl in configure.ac), there is a strong argument for having
literal dnl in the output without having to add quadrigraphs everywhere. So
I'm okay with this patch as-is.
That said, it would be nicer if libtoolize could use autom4te --trace rather
than sed scripts to learn some of what it scans from configure.ac, but that's
unrelated to this patch series...
--
Eric Blake