[MERGE][Trivial] invalid version tuple attempt to raise AssertionError fails with TypeError

[Mark Hammond]

> > Mark Hammond wrote:
> >> This is fairly trivial, but bzrlib._format_version_tuple() raises an
> >> unintended exception when passed an invalid version tuple (so I'm
> reusing
> >> that function even though it has an underscore - sue me ;)
> >
> > Since you admit that you're using a private function in an extension,
> > you should propose turning it into a public function.
> 
> I'm happy to rename it to be public if you want to.

That would probably be nice, especially if plugins or extensions to bzr
choose to adopt a similar version numbering scheme - which seems to make
sense.  On the other hand, QBzr uses a similarly formatted tuple, but
formats it differently:

  version_info = (0, 9, 5, 'dev', 0)
  __version__ = '.'.join(map(str, version_info)) 

and a quick grep of bzr-svn and bzrtools show they don't use it.  So I'm
really not sure how widespread usage of _format_version_tuple() is.

On the other hand, a simple grep shows bzr-svn using:

 from bzrlib.plugins.svn.transport import _url_escape_uri
 from bzrlib.plugins.svn.commit import _revision_id_to_svk_feature

bzrtools:

 from bzrlib import revision as _mod_revision
 from bzrlib.diff import _patch_header_date  (in a test)

QBzr:

 from bzrlib.annotate import _annotate_file
 from bzrlib.diff import _get_trees_to_diff
 bzrlib.builtins.cmd_merge._do_preview(self, merger)

So it looks like there are a few such candidates.  Maybe we should just
pretend we never had this conversation :)

Cheers,

Mark