[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: aclocal.m4 incorrectly remove parts of python-cflags output
From: |
Colin Hall |
Subject: |
Re: aclocal.m4 incorrectly remove parts of python-cflags output |
Date: |
Mon, 2 Jul 2012 10:45:50 +0100 |
User-agent: |
Mutt/1.5.20 (2009-06-14) |
On Mon, Jul 02, 2012 at 10:35:04AM +0200, Cyprien Nicolas wrote:
> Hi,
>
> At Gentoo, we've got a bug [1] about lilypond-2.14.2 configure's process.
>
> To summary the bug report:
>
> Line 978 of aclocal.m4, a sed rule is applied to remove "junk" in
> python-cflags output, according to python issue 3290 [2].
>
> It turns out that this sed expression is not correct, and strips parts
> of some cflags. The remaining part is no longer a valid CFLAGS, making
> the configure fails, saying it can't find Python.h.
>
> The sed rule's purpose is to remove unnecessary CFLAGS like -march,
> warning flags, optimization flags or definition flags. The sed rule
> matches strings that start with -[WDOm].
>
> We observed that if some flags contains the above described pattern, the
> matching part will be removed, event if it is part of another flag. For
> instance: -floop-strip-mime or -fvector-cost-model. -mime and -model
> will match the sed rule, and will get removed. Thus -floop-strip and
> -fvector-cost are passed to the C compiler which does not know about
> those flags. It then exit on failure, making the configure check failing.
>
> We propose a corrected sed rule [3] to match only -[WDOm] at the
> beginning of a string element, in order to not match substrings.
>
>
> 1. https://bugs.gentoo.org/show_bug.cgi?id=415793
> 2. http://bugs.python.org/issue3290
> 3, https://415793.bugs.gentoo.org/attachment.cgi?id=316865
>
Thanks for the detailed bug report, Cyprien.
I have created an issue tracker for your report here:
http://code.google.com/p/lilypond/issues/detail?id=2634
Cheers,
Colin.
--
Colin Hall