Working with eZ publish and Subversion

Tuesday 15 June 2004 12:37:42 am

Hi

I work often with Subversion and i would like use it with eZ publish too.
But i have problems when working with SVN and eZ publish:
The configuration files can be modified by changing the local copy and commiting, but also can be modified from the web interface. The changes made from web interface aren't commited to repository, then i can't get them to my local copy with "update" command.
But i have problems with the templates also. When i define an override from web interface, it creates an empty template. If i want to commit my own personalized template (that will have the same name), i get a conflict (the file does exist because eZ publish created it, but i can't see it in my local copy).

The question it's... someone work with eZ publish and version control systems? How do you work with them?

Thanks in advance (and sorry for my poor english)

Cristian

Tuesday 15 June 2004 1:45:08 am

Hi Tony

I don't work with SVN repository versions of eZ publish, but that i downloaded release packages. This isn't the problem.
I am talking about develop web sites with eZ publish using version control systems, not develop eZ publish itself (sorry if i explained it confusedly).
The problem is the versioning of templates and some configuration files like <i>override.ini.append</i>, because i change them from my local copy and from web interface.
Should i modify files only from local copy (and commit to repository) and forget the web interface facilities?
Always i make changes (that affect to some file, not database) from web interface i get SVN conflicts that cause a lost of time. This is frequent when i define overrides from web interface (eZ publish modify a configuration file with changes that i can't see in my local copy and creates an empty template that i can't overwrite from my local copy without get a SVN conflict).
I am interesed in know how people develop sites with eZ publish using version control systems like SVN.

Bye

Cristian

Tuesday 15 June 2004 8:26:07 am

Hi Paul

The web server with eZ publish and the SVN repository are in a Linux machine. The SVN clients are running on Windows 2003 Server.
About the symbolic links... i can't understand how you can avoid the conflicts between changes from the web interface and changes commited from local copies.
Subversion data is stored in a database (not in a directory), therefore you can't configure the repository so that the changes are commited in the <i>/var/www</i> directory. Then, i know two ways of work with SVN and eZ publish:
- In the <i>/var/www</i> i can put an local copy (and configure SVN so that it do an <i>update</i> every time somebody do a <i>commit</i>).
- Configure SVN so that do an <i>export</i> to the <i>/var/www</i> directory every time somebody do a <i>commit</i>.
With both methods i can't get the changes made from web interface in my local copies. With the second method i overwrite all changes made from web interface when i do a commit. With the first method, if i change some configuration file or i define an override from web interface, i will get a conflict when i commit some change in those files.
How i should configure SVN for work without problems with eZ publish? Should i forget some facilities of web interface like override management and <i>ini</i> files modification?

Thanks (and sorry for my bad english)

Cristian

Thursday 17 June 2004 2:36:26 am

The problem that have Cristian and me is that we do not know how could subversion synchronizes the changes that the Web server does in "his local working copy" (using the interface of administration) on "our local working copies".

We have setup subversion like I say: If we made commit on "our local working copy", for example, from our directory design subversion (using a post-commit script) made update to directory design that there is in /apache/www/directory. But, for expale, if apache creates a template we will not have this template.

Thursday 17 June 2004 3:10:06 am

The problem of this solution is that you sopouse that we have access to the server computer (using ftp or telnet) to catch the files once in a while but, what can we do if we only have access to the server computer using svn for security reasons and for use only one method to access to this file (if you can get the files using svn why you need ftp)?.

The idea is try to use Apache like an "other subversion user" and them, when this "user" create/modify any file of his local working copy. can make an svn update.

I know it's really complicate but the question is... it will be possible to be done?

Thursday 17 June 2004 4:46:44 am

Well... the question:

* are the users on the apache server updating the templates via eZ publsh editor or via FTP or vi etc?
* could you install SVN client on the apache server?

The users "may" update the templates via commit her working copies or via editor... in this moment can do via ftp too... but I'm trying to use only one access method (I reconise...I am the administrator :-) )
Yes the server computer have an svn client.

THHAANKKSS REAAALLYY THAAANNNKSSS BY THE ANSWER but..."If you are able to install SVN client on the apache server then you could write a workflow or a shell script that ran svn commit /your directory/to/templates/ and updated the repository everytime a change was made." How can I do this?... how can i writ a workflow? this is exclaty my problem :-(.

Thursday 17 June 2004 6:01:36 am

Only one think...THANKS.