Re: Check out best practices

Steve Finkelstein wrote:
> Does anyone have any best practices they use that they can share to
> allow web developers to update web directories with repository stored
> code? Anything besides giving them root on the servers would work for
> me.

I can't tell from your description exactly what your use case is. However, I
wrote SVN::Notify::Mirror[1] specifically to handle keeping multiple web servers
up to date with repository changes.

The workflow we used specifically was this:

1) All normal work was performed on trunk and all trunk commits were immediately
synced to a test web server using a post-commit script;

2) Large-scale reorg changes were done on a branch, and the test server was
switched to the branch for the duration of that development process;

3) When code was ready to go to the production website, the developer would
merely make a tag of trunk and the production server would switch to the new tag
(which had to match a predefined regex).

This has worked extremely well at my previous job (I escaped) for the last
several years with very little manual intervention. The working copies on the
test and production server do not need to be accessible by the developers at
all, since the post-commit can use an SSH key to transfer files to the web user
account.

John

1) http://search.cpan.org/search?query=SVN::Notify::Mirror

You'll also want to use SVN::Notify::Config for compact post-commit files (where
the post-commit literally *is* the configuration data for the hook:

http://search.cpan.org/search?query=SVN::Notify::Config

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe_at_subversion.tigris.org
For additional commands, e-mail: users-help_at_subversion.tigris.org
Received on 2008-03-13 02:47:01 CET