On Mon, Aug 24, 2009 at 12:57:58PM +0200, Marc Espie wrote:
> On Mon, Aug 24, 2009 at 11:44:15AM +0100, Stuart Henderson wrote:
> > On 2009/08/24 11:10, Mikolaj Kucharski wrote:
> > > Question about versioning. Current MPlayer pkgname is mplayer-1.0rc2p27
> > > and pkgname from this thread is mplayer-20090708.
> > >
> > > Would it be possible to have the name mplayer-1.0rc2.20090708? I think
> > > (hope?) that MPlayer team finally will release next rc, or 1.0 version of
> > > mplayer. With mplayer-20090708 that would require `v' bump in pkgname to
> > > go back to 1.0* version.
> >
> > Looking at PackageName.pm I think we might not want the . after rc2,
> > it's just using \d i.e. matching only digits.
> >
> > -- -- --
> > sub make_dewey
> > {
> > my $o = shift;
> > $o->{deweys} = [ split(/\./o, $o->{string}) ];
> > for my $suffix (qw(rc beta pre pl)) {
> > if ($o->{deweys}->[-1] =~ m/^(\d+)$suffix(\d*)$/) {
> > $o->{deweys}->[-1] = $1;
> > $o->{$suffix} = $2;
> > }
> > }
> > }
> > -- -- --
> >
> > Though this reminded me to have a play with check-name in the
> > regression tests and it isn't working quite how I'd expect anyway;
> > if I add these,
> >
> > ok(check_list([qw(foo-1.0rc2)],
> > check_pkgspec('foo-<1.0', @list)),
> > 'before 1.0 came 1.0rc2');
> > ok(check_list([qw(foo-1.0)],
> > check_pkgspec('foo-<1.0pl1', @list)),
> > 'before 1.0pl1 came 1.0');
> >
> > both of them fail and I /think/ that they need to pass (otherwise when
> > the package tools know not to replace an installed package with an older
> > version we could have a little trouble with ports using this syntax).
> > Marc, can you give me any insight into how this should work please?
> >
> > > ..or maybe I'm just caviling.
> >
> > Sometimes we need v0 but I'm not especially keen on adding it
> > so, personally, I try and avoid getting in a situation where I'll
> > need it in the future.
>
> I don't think we're ever gonna get an actual mplayer release, I think it's
> a case where it's logical to go to 20090708v0
we don't need v. 20090708 > 1. I've tested updates from ports
and packages.
> v is there to be used, it doesn't make any sense to whack our own weird naming
> scheme that doesn't match anything out there...
agreed. 1.0rc2.20090708 is more confusing than a simple date.
--
jake...@sdf.lonestar.org
SDF Public Access UNIX System - http://sdf.lonestar.org
