2015-06-03 9:56 GMT+02:00 Marcus Denker <[email protected]>: > > On 03 Jun 2015, at 09:52, Thierry Goubier <[email protected]> > wrote: > > > > 2015-06-03 8:22 GMT+02:00 Marcus Denker <[email protected]>: > >> Indeed... this has to do something with caching: when looking a .mcz the >> system just takes the one it finds, which might be >> one with the same name but different. >> > > Unless the mcz searched for is a dependency, in which case it checks the > uuid as well (not only the name). > > >> >> Maybe we could for real put (part of) the UID in the filename? >> > > Would that be backward compatible with existing repositories? > > > > I would just put it instead of the Name of the submitter... >
I'm not that worried about that problem, but the fact that in MC, when you query a repo, you build a kind of a match between your target version info and a "possible" file name... If you start creating names with UUIDs in them, how will the system distinguish between version infos where you should use the UUID and version infos (for example, in the history) where you shouldn't. Give a higher priority to matching with an UUID-based name ? Go through Gofer twice, first with a UUID version of the name, then with a non-UUID version of the name if the first one fails? I believe there are places (MC itself, Gofer?) where the true version info is decoupled from the search and only the package name is used (i.e. name + author + version number). Thierry > > Marcus > >
