I recently cleaned up a trac (1.0 branch) install that still had the
gitplugin.  This was fairly simple, as I just enabled builtin git and
disabled the plugin.  However, I have some sync lines in trac.ini left
over, and it seems I'm not set up for best practices.  This message is
really a request to spiff up the userguide to answer the questions I
have, which I think are pretty universal.

After reading:

  http://trac.edgewall.org/wiki/TracRepositoryAdmin

I do not know:

* What does repository_sync_per_request do, exactly?  Is it about having
  trac keep a copy of repo state in the database, to make queries
  faster?  Is the notion that on a new request, trac reads enough vc
  bits to get new commits since the last read?  Is it believed that
  because it just had to read refs/heads/* or something that it's fast
  if there aren't any?

* What do cached_repository and persistent_cache (in the git section)
  mean?  I get it that one might have to set up calls in post-receive
  hooks, and to run an explicit sync, but what values should they be set
  to, and why?  How does this interact with repository_sync_per_request?

It seems there are 8 combinations of the above.  Which of those 8 make
sense?


When I set cached_repository=true, I found that git browsing from trac
completely broke (could not find revision) even if I had done a sync.



Thanks,
Greg

Attachment: pgpz5A3FdaVtK.pgp
Description: PGP signature

Reply via email to