[mythtv] Re: LIBVERSION = 0.15.0.99?

[Isaac Richards]

On Saturday 29 May 2004 03:49 am, Axel Thimm wrote:
> On Sat, May 29, 2004 at 03:03:23AM -0400, Isaac Richards wrote:
> > On Saturday 29 May 2004 02:39 am, Henk Poley wrote:
> > > Op zaterdag 29 mei 2004 08:16, schreef Isaac Richards:
> > > > On Saturday 29 May 2004 02:08 am, Axel Thimm wrote:
> > > > > Hi,
> > > > >
> > > > > would it be possible to have an interim versioning scheme? This is
> > > > > useful for packaging CVS versions w/o possibly breaking upgrade
> > > > > paths. For packaging CVS versions one must either choose the latest
> > > > > release (e.g. 0.15) or guess what the next could be (0.15.1?
> > > > > 0.16?).
> > > > >
> > > > > Both methods have its drawbacks, the former is version-safe, but
> > > > > may indicate CVS _before_ 0.15 and not after, and the latter is
> > > > > guesswork which will break at some time.
> > > > >
> > > > > So it would be nice to have an interim versioning (e.g. 0.15.0.99
> > > > > leaving space for a .1 release, or right to 0.15.99 perhaps), if
> > > > > this is not disturbing in any way.
> > > >
> > > > The libversion changes when the library changes in an incompatible
> > > > way, not before.
> > > >
> > > > Also, I'm having second thoughts about you packing CVS versions --
> > > > check the users list about people insisting they're running 0.15 when
> > > > they're using a CVS package of yours from 3 weeks ago.
> > >
> > > Uhm, maybe that's exactly why he asked for extra versioning?
> >
> > No, try reading his email.  He's asking for extra versioning to make
> > making up a package name easier for his CVS rpms.
>
> Well, to be honest Henk is quite right. ;)
> http://lists.atrpms.net/pipermail/atrpms-users/2004-May/000486.html

That email _still_ says the only reason you want to reuse the libversion for 
non-binary compatability purposes is to make choosing names for your cvs rpms 
easier.

> A 0.14.99 version would have helped, so I'd still propose, if there is
> no drawback, to use interim versioning. It doesn't need to be bumbed at
> every cvs commit or similar. Probably only need attention at the
> beginning and end of a developmeent cycle (e.g. 0.15.0.9 now and
> shortly before the next release 0.15.99)

I don't see how it will help.  People will then say 'I'm running the 0.14.99 
release', and I made no such release.  If they say anything with the word 
'release' in there, that shows that they have no understanding that they're 
running a CVS snapshot that could be unstable, broken, corrupt their DB, etc.

Why don't you just put 'cvs' in the package name, say they conflict with the 
normally named packages, and use the date you generated the packages as the 
version number?

Isaac