%{?dist}, recommended or optional?

[Michael A. Peters]

On Sat, 2006-01-07 at 19:30 +0330, Roozbeh Pournader wrote:
> On Sat, 2006-01-07 at 15:38 +0100, Michael Schwendt wrote:
> > Where is the part you find "somehow contradicting"? In all three quotes,
> > using %{?dist} is either recommended or described as being optional.
> > This dist tag macro exists in order to aid you. "recommended" is not
> > equal to "mandatory".
> 
> I'm not talking about mandatory. I'm talking about the difference
> between "optional", as in "use it if you think you should", and
> "recommended", as in "use it if you are not sure you may need it or
> not".
> 
> In some cases, when the packager has no strong feeling either way and he
> is tempted to not use it to avoid making the tag more complex, should he
> use it or not? For an example, see:
> https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=177096
> where if I add the tag, the version will be something like
> "2.0-0.1.20060103cvs.fc4", which I call ugly.

When I package development snapshots - I like to do

0%{?dist}.n.%{cvs_release}

where n is an integer number.

I increment the n with every spec file (in case I need to change to an
older checkout) - and when a final release is made, everything after
%{?dist} is dropped - and the 0 I bump to a 1.

See http://mpeters.us/silgraphite/silgraphite.spec

I don't know that that is the best way to do it - but it works well for
me.