On Thu, 2014-04-24 at 23:20 -0500, Gunnar Wolf wrote:
> Manuel A. Fernandez Montecelo dijo [Fri, Apr 25, 2014 at 01:27:02AM +0100]:
> > To both things above, I don't think that this is different to my example of
> > 'configure' script without corresponding .ac/.in; and I don't think that 
> > anybody
> > is thinking about adding lintian errors for that or considering those 
> > scripts
> > non-free (??).
> 
> Just FWIW, I didn't reply to this point because it's been many years
> since I last packaged a project using configure scripts (I just work
> with languages I am comfortable with, and C is not one of them). As
> others have pointed, shipping configure without compiling it from the
> .ac/.in is bad and should be seen as a warning, if not directly as a
> true bug.

Following this thread reveals that nobody? mention compiling Makefile.am
into Makefile.in and later on to Makefile with configure. One recent
example of troubles is gcc-4.9-pre-releases (probably 4.9-1 too). Since
automake is no longer used: Makefile.in are used for building from
sourec. It is very frustrating and error prone to manually patch
Makefile.in when modifying Makefile.am. Regenerating with autoreconf2.64
and automake1.11 is currently broken and will not create any Makefile.in
from the Makefile.am files.

According to upstream it is not sufficient with autoconf 2.64, automake
1.11 first in PATH, libtool 2.2.7a is also needed, and that package is
not available from the main repository, not even at snapshot.debian.org.


-- 
To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: https://lists.debian.org/1398415210.568.95.ca...@g3620.my.own.domain

Reply via email to