Jan Hendrik wrote:
> I suppose that it still would be necessary to have the SVN client on
> the server, right?
Usually, though I also wrote an Rsync method (instead of SSH), so if it
was a problem to install Subversion, you could use Rsync instead.
> A downside of all this automating however is that at times for
> logical or other reason a change may be committed in a number of
> consecutive commits, with the website not really being presentable
> until the final commit. No automated updating of the live server
> working copy could take care of this while it can be easily forgotten
> to be turned off in time. Or turned on again ...
That's actually not a problem. When I deployed this at my previous
company, I kept a test server in sync with trunk. The only time the
production server was updated is when a tag (matching a specific regular
expression) was created. That way you can confirm that the production
site is only updated at well defined locations. It also means you can
roll back to any previous tag (which should all be stable) at any time.
The neat thing about using a tool like this is that the web developers
don't even need login access to the production server; the mere act of
creating the tag does the right thing...
John
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe_at_subversion.tigris.org
For additional commands, e-mail: users-help_at_subversion.tigris.org
Received on 2008-06-24 18:15:22 CEST