Fwd: Re: Subversion - Proposal

Sean Wheller sean at inwords.co.za
Sun Dec 26 07:58:20 UTC 2004


On Sunday 26 December 2004 04:43, Enrico Zini wrote:
> so that it is possible to branch and tag separately the various
>    different projects, but in this case we don't necessarily have
>    different subprojects, so it also makes sense to keep everything
>    together.

Yes we don't have sub projects. One trunk will be a good start.

>
>    If I don't hear anything bad, I'll proceed to move everything inside
>    trunk tomorrow morning (Taiwan time) while you all are sleeping.
>    
>  - Implement svn hooks
>    We've already been thinking about this, but we never managed to do
>    it.  The possibility of migrating the repository into Canonical
>    machines also gave the idea that it may have been more efficient to
>    spend time on this after the migration, to avoid doing things twice.

Well, you can do it and change it as many times as you like. Once you have the 
hook its just a case of moving it to the new machine. Then there is sendmail 
to configure and mailman and update to dns.

>
>  - Migration to Canonical machines
>    I asked Elmo about the status of svn migration to Canonical webfarm
>    and chatted a bit with him about it, trying to sync with the current
>    status of the situation.  We won't have the svn repo in Canonical
>    machines before beginning of next year, as Elmo is in vacation for a
>    week until new year's day.

OK so use current location, but lets create proper structure. Before you do 
anything please apply my patch 
https://bugzilla.ubuntu.com/show_bug.cgi?id=4978

>
>  - Hot backup
>    I don't know much about it: is it a feature of subversion?

Use the hot-backup.py a hook-template should already be created when you do 
svnadmin create

>  
>  - Setup stdin  and stdout (svn://)
>  - Spawn svn as service from inetd
>    I don't know too much about server-side subversion, so I need more
>    clues here: at the moment we already have full access to the
>    repository using HTTP and DAV, and there is not much load on the
>    server as committers are rather few.  What would these two changes
>    mean?

DAV is not a good idea. DAV is a direct commit. Use HTTP just for browsing. I 
prefer to use svn:// as the main interface since it has less dependencies of 
other technologies, just rely on SVN.

OK Step 1.
On Debian you should already have something like this in /etc/services

svnserv           3690/tcp   # Subversion
svnserv           3690/udp   # Subversion

Note the Debian guys have a mistake in the name but leave it as this it will 
work.

Fro inetd place the following line in your /etc/inetd.conf

svnserve -i -r /usr/local/repositories

This will cause any repos under repositories/ to be exported by svn when inetd 
spawn a svnserv.

So under repositories/ you will have a folder something like udpdocs/ that 
will be exported. 

The -r will export only the repos, so people will do 

svn co svn://xxx.xxx.xxx.xxx/udpdocs/trunk

or svn cat svn://xxx.xxx.xxx.xxx/udpdocs

-- 
Sean Wheller
Technical Author
sean at inwords.co.za
http://www.inwords.co.za
Registered Linux User #375355
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <https://lists.ubuntu.com/archives/ubuntu-doc/attachments/20041226/659c2c55/attachment.pgp>


More information about the ubuntu-doc mailing list