[Gllug] Practical release and deployment - how?

[Simon Wilcox]

Ziya Suzen wrote:
> I am managing a bunch of RHEL 4 boxes (both production &
> test/development) and developing websites for several clients (order
> of ten). Using SVN for source control. Currently what I have is
> basically SVN HEAD is what should be deployed/installed (basically
> copied over) on all the test/staging, development and production
> servers. However this is proving to be a little burden since I can't
> really track my changes during development since I can't commit
> anything until it's ready to be deployed. I should probably use
> svn/branches and so on to achieve this and possibly use RPM to deploy
> code. But it sounds like too much of an overhead sometimes, I can't
> convince myself doing that. What is your experience?

Definitely use branches. You'll probably find it simpler to use HEAD as 
the production branch and then create dev branches as required for each 
change set. These can then be tested in isolation and merged back into 
the trunk for release.

Managing the merges can get interesting if you're doing a lot of 
releases but in general it's fairly easy to deal with. Just make sure 
that you merge trunk changes into your development branch(es) at 
frequent intervals. That makes merging your final dev changes back into 
trunk a much simpler exercise.

However, if you need to maintain multiple releases, then having the 
trunk as the main development branch with multiple release branches is 
probably the way to go. Managing the merges between those branches can 
get a bit hairy though !

Simon.
-- 
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug