On Tue, Nov 29, 2011 at 05:31:26PM +0100, Jan Zelený wrote:
> Dne 21.9.2011 13:32, Stephen Gallagher napsal(a):
> > On Wed, 2011-09-21 at 08:40 +0200, Jan Zelený wrote:
> >> I don't quite understand. Could you please elaborate what's wrong
with
> >> the original approach you suggested? That means one configure argument
> >> allowing to specify the version with a default version string if no
> >> such argument is provided.
> >
> > Well, for one thing what I asked for in the first place was
> > impossible :-/
> >
> > We can't have multi-line version strings in a configure option; autoconf
> > can't handle it. (All that happens is that the config.h is broken)
>
> Actually, it can handle it. Although not directly (you have to put \n on
> the end of line):
>
> $ ./configure --with-distro-version="-3.fc15\n\
>
> > Built: 2011-09-07" --enable-silent-rules
>
> $ ./sssd --version
> 1.7.0-3.fc15
> Built: 2011-09-07
I can confirm this works as described by Pavel.
> > So I was trying to think of an alternative that would serve our needs
> > better. Though admittedly, at this point I'm starting to wonder at the
> > usefulness of this feature at all.
> >
> > So my new thought was that we should default to displaying a version
> > that includes build-time information. Then if we wanted to display just
> > an official version (or a distro-specific build) we could do so by using
> > a --release argument.
I still prefer --version over --release. It's the standard switch
everyone uses.
> >
> > This way at least, we would be able to identify work-in-progress builds
> > easily as compared to official release builds.
>
I don't know if that will do any good. The only thing it will do is make it
possible to distinguish between work-in-progress build and official release.
This can be achieved by current patch as well, you can add a value to
PRERELEASE_VERSION when releasing new tarball (of course, the semantic of the
PRERELEASE_VERSION would be a bit different at that point). Due to this, I'd
like to consider this patch ACKed.
Any other opinions?
I thought the real value was identifying distribution patchlevels
easily?