On Wed, 06 Mar 2013, Cristian Morales Vega wrote:

> Should we patch this? I didn't check if the version has any real
> impact on anything.

it depends if you use the 'make distcheck' to build packages, as certain
debian maintainers do, then it affects the naming of the final tarball.

in any case, yes please, patch it to 1.4. thanks.

> > I don't understand well this problem. can you suggest a way to avoid
> > it in future? a sequence of touch commands?
>
> In the tarball the timestamp of aclocal.m4 is "Feb 24 13:44" and the
> timestamp of include/config.h.in is "Feb 24 13:43". Since aclocal.m4
> is a dependency of include/config.h.in through a rule created by
> automake, a "make" call rebuilds include/config.h.in. The problem is
> that if the user hasn't autoheader installed the build will fail. The
> workaround is just "touch include/config.h.in". But it shouldn't be
> needed.
>
> Why the timestamp of aclocal.m4 is more recent than the one of
> include/config.h.in? No idea. But I guess something "touched"
> aclocal.m4.

I have used the relatively new 'autoreconf' script included in autoconf
to generate those files, maybe this is something to report to developers
over there, that the script should run autoheader before aclocal.

there is some smartypants comment in my autoreconf script (v2.69)

# We prefer running autoconf before autoheader, because (i) the
# latter runs the former, and (ii) autoconf is stricter than
# autoheader. So all in all, autoconf should give better error
# messages.


ciao