Best practices for tagging on branches

[Greg Ward]

On Mon, Apr 27, 2009 at 2:56 PM, Dirkjan Ochtman <dirkjan at ochtman.nl> wrote:
> I've thought that it might be worthwhile to forbid tagging changesets
> that where ancestor(tagging, tagged) != tagged, i.e. the tagging
> changeset must be a child of the tagged changeset. Not doing so caused
> some problems for some guy a while back. Not sure that comes down to the
> same rules you're proposing, though.

Interesting idea, but it would conflict with my pie-in-the-sky idea of
"only tag on default branch".  (I might not have merged that release
branch to default yet, but I still want to tag a point on it.)  My
proposed policy might be a dumb idea, but I don't think Mercurial
should stop me from implementing dumb policies.

I think this just means that any strict tagging policies need to be
configurable, and the default policy should be the status quo.  (Hmmm:
would it be possible to implement such tagging policies in hooks?
That would be the ideal place for various projects to experiment and
come up with guidelines like "this policy works great" or "this is a
dumb policy -- don't do it".)

Greg