On Tue, Dec 1, 2009 at 9:27 PM, Joshua Facemyer
<jfacemyer@...400...> wrote:
The main things we need to be sure of (the reasons for switching to a CMS in the first place) are:
- Ease of use for contributors
Ease of use for contributors is something that I think we'll be able to get better with Django than with Drupal (partly as we decide much more what we want; a framework is generally more flexible than a rigid CMS like Drupal).
- Ease of maintenance (probably the biggest drawback to Django - it requires some programming knowledge, while Drupal doesn't)
A lot of the maintenance can be just updating a handful of
SVN/Mercurial/Git repositories (depending on the version control method
used for the Django application, svn and hg seem to be the most common)
from time to time, or using easy_install. Whereas with Drupal you've
constantly got to be going to the admin system, seeing what there is to
update, downloading the files, extracting, taking the site offline,
uploading, running the update script, putting the site back online...
(yes, I did put a very bad slant on Drupal's update process in that :P)
- Some ease of styling ability
The actual styling and template creation is simpler with Django than with Drupal. It's a lot simpler to turn a plain HTML page into a Django template than into a Drupal theme.
I should think it would be simple enough to get a script to pull it from a Bazaar repo. I think it's important to keep all of the Inkscape stuff together, that way people won't have to know and use different VCSes just for Inkscape. Is it possible to get bzr on OSUOSL?
easy_install bzr (of course it's best to check with OSUOSL to make sure they're happy with that). The server's got Python 2.5.4 which is fine for bzr and django.
* Content can then be updated through two or three methods:
o the website
o the version control system
We'd have to determine which method to use, wouldn't we? Otherwise we'd have two differing sets of data, unless any update by the Django backend would also commit to the repo. Hmm...is that possible?
Why ever would a Django site not be able to commit and push? I've had that in my mind right from the start. Especially as Bazaar is also written in Python which makes it easier (if it were possible).
I think it sounds fine, so long as we've got people who are able to manage the process. It would be good to have all (or most) of the website content managed by one backend which is really capable of doing it. How do you feel about it? I think you'd be the one to do most of the difficult stuff, unless someone else with appropriate knowledge steps up.
I think I could probably manage the process, but I'll have to see how much time I can give to it though. I'll probably slot it in with my own commercial Django project, borrowing bits of code each way when they overlap at all. And it'll all help me with familiarity and experience with Django (which I hadn't ever used three months ago).
I think that the first step should be version control for collaboration, and a basic port of the current functionality of the static file-based site at
www.inkscape.org (it's dynamic in that it's PHP, but it's only used as static content really). Should I start working in a subdirectory of
https://inkscape.svn.sourceforge.net/svnroot/inkscape/inkscape_web/trunk which is the current location, or should we get something bizarre set up?