This work is complete now.  subversion.tigris.org is now using it's own
Tigris-hosted Subversion repository, complete with instant auto-publishing
of the /trunk/www area.  (Man, have I missed having that feature...)  Commit
notifications from that repository will continue to go to s...@.  (If folks
think we should move to the Tigris-standard commits@ list instead, that's
fine, just say the word.)  All the files that were in our downloads/
directory remain there, but are now versioned in the repository's
/trunk/www/downloads directory, with a Subversion-versioned symlink
/trunk/www/tarballs -> 'downloads' like we had before.  It all looks good as
far as I can tell.

My intent with the subversion.tigris.org site is to not add any more pages,
period, but just slowly deprecate pages by turning them into redirects and
references to the Apache-hosted pages that contain the same information.  So
my recommendation is that we avoid tweaking the subversion.t.o website too
much until we know what we're going to do with the subversion.a.o site.  I'm
not as big a stickler for "gotta make URLs never go away" as some people
are, but it's those people that stand to lose out if we prematurely start
publishing subversion.a.o URLs that aren't well-thunk-out.  :-)


C. Michael Pilato wrote:
> Currently, the Tigris.org machine has an "instance set" (custom code) which
> causes that site to be published from the /trunk/www directory of our
> now-read-only svn.collab.net repository via some cron job.  Now that we have
> our CVS data converted to Subversion, there's no reason to keep the live CVS
> repository at subversion.tigris.org.  And now that we aren't using the
> svn.collab.net repos, the website on subversion.tigris.org is frozen in time.
> 
> I'm proposing that we do this:
> 
>    * Ask CollabNet Ops to convert subversion.tigris.org from a CVS-backed
>      project to a Subversion-backed one, giving us a clean Subversion
>      repository on the box to play with.
> 
>    * Ask CollabNet Ops to remove or disable the instance set that tells
>      the server to pull www bits from svn.collab.net, and let the default
>      behavior (of auto-publishing the /trunk/www directory of the on-site
>      project repository) kick in.  Of course, in the process, we'll be
>      careful *not* to lose our 'downloads'/'tarballs' folder (which isn't
>      versioned).
> 
>    * 'svn export' our ASF 'www' directory, and 'svn import' that into
>      the new subversion.t.o repository's /trunk/www directory, effective
>      forking the history of those files, but for what I think are obviously
>      good reasons.
> 
> At this point, we've preserved the current state of the subversion.t.o
> website, but we can now make changes to it in the same way every other
> project makes those changes.  And, of course, we're free to rethink,
> refactor, reorganize, and re-everything-else our subversion.a.o web presence.
> 
> And yes, I'm volunteering to manage this.
> 
> Thoughts?  (Silence is assent.)



-- 
C. Michael Pilato <cmpil...@collab.net>
CollabNet   <>   www.collab.net   <>   Distributed Development On Demand

Reply via email to