Changes to the OpenHatch website code from July and August: 0.11.07 release notes
by Asheesh •
September 7th, 2011
I did a lot of traveling during 0.11.07, so that release covers two months: July and August.
The big highlights are the first implementation of the Buildhelper; new contributors fixing template issues; and a project-wide improvement in understanding of how to build reasonable web app front-ends. As usual, there were some code cleanups that make the backend more maintainable.
The openhatch-0.11.07 tag in git covers this work, which includes:
- Creating Puppet manifests and Vagrant configuration so that new contributors could use a virtual machine to develop the site
- Hadrien Dorio fixed some templates in the training missions that were confusing users
- When editing your personal list of projects, sometimes saving would fail. That was fixed.
- We wrote a new management command for downloading bug data, and made it parallel with gevent.
- Spent some time discussing CSS+JavaScript style with johnkpaul, and wrote a front-end style guide based on his advice.
- Jessica McKellar fixed a bug in our embedded copy of a patch parser that was causing patch validation crashes for users of the training missions.
- We now more-aggressively cache the JSON data that we use to render the map at /people/.
- We changed the training missions to have a big yellow box indicating where to get help if you are stuck.
- We disabled the automatic profile importer. This used to be one of the big features of the site, but recently we haven’t spent time maintaining it. It would be great to make it work again, but it’s not a priority given our current focus on new contributors rather than experienced open source contributors.
- We changed the front page to load faster, by avoiding calculating “recommended bugs” immediately. Now that is done in a separate HTTP GET, in JavaScript.
- Karen Rustad fixed a bug in urls.py that was preventing the admin interface from working.
- We were depending on a custom fork of django-kombu, but we now use a stock version from upstream. In this release, we also stopped using rabbitmq and store all Celery tasks in the database, through the Django ORM. In this release, we upgraded to Celery 2.x; due to inertia, we had been stuck on an old version for a while.
- Asheesh wrote a bare-bones web interface for adding bug trackers, which we documented on the Bug Trackers wiki page.
- Mark Freeeman made a style fix in profile/models.py.
- Karen built the first implementation of the buildhelper, which you can see in action on the OpenHatch project page.
- Vivek Shrivastava removed our use of the database for HitCountCache (see issue310).