On Tue, 20 Apr 2010, Roman Chyla wrote:
> Trac is great! I have a question though. Is it really necessary to
> have a separate inspire instance?

It is not strictly speaking necessary, but it may be advantageous to
think of clustering things around our distinct source code repositories,
for example.  Since we have two git repositories, we can as well have a
dedicated Trac instance around one Invenio repo, and another dedicated
Trac instance around the other INSPIRE repo.  The setup with split Trac
instances also provides for a smoother continuation of the
currently-split Twiki and Savannah projects, as well as a cleaner
separation of blog announcements.  The latter may be important since we
would like to use the Trac wiki in order to replace the currently-static
software distribution site for Invenio and Indico
<http://cdsware.cern.ch/>.  (Hence its cdswaredev.cern.ch domain name.)

But I agree that the separation leads to a certain DID (Dissociative
Identity Disorder), where certain tickets or wiki topics that should
rather be living in the Invenio work space do live on the INSPIRE work
space, and vice versa.  This is a pity, and leads to troubles
where-is-this-or-that-thing, as you mentioned.  We also use mailing
lists in this DID-like manner.  So we may want to investigate the
feasibility of using the same Trac project, e.g. using dedicated
invenio/inspire prefixes or labels for wiki and tickets.  It may be
nicely feasible, but it may also create further confusion for non-CERN,
non-INSPIRE clients that are simply interested in generic Invenio.  I'd
tend to think the latter is more probable, which is why I replied
yesterday about the split Trac instances.  But we can definitely try to
evaluate that.

Until then, in order to avoid DIDs with the current split setup, we can
follow the good old generic principle: everything potentially
interesting to any Invenio instance should go to the Invenio
repo/tracker/wiki, while anything instance-specific should go to its own
instance-specific repo/tracker/wiki, be it the INSPIRE service, the CERN
Document Server service, and any other sizeable service.  Some examples:
the user tagging facility is for Invenio, the famous formats with dozens
of arXiv mirror links are for INSPIRE, the equally famous MediaArchive
handshaking or CMS batch importing is for CDS, etc.

Best regards
--
Tibor Simko

Reply via email to