Getting "Server Error (500)" from Django Admin

[elinaru]

When doing the following actions (and perhaps others), I'm getting a "Server Error (500)"
when logged into http://admin.caciviclab.org/admin/ -->

Viewing: Ballot item selections
Viewing: Ballot items
Viewing: Referendums
Deleting item: Ballot
Adding: Candidate
Deleting item: City

I'm almost certain I'm not to blame.

[bcipolli]

Hard to tell; the master branch is far behind the develop branch. The solution is to wrap up modifications on the develop branch and release to master. Hope to have that ready soon!

[elinaru]

CC @tdooner

[elinaru]

Hey @bcipolli any progress in this? The "Server Error (500)" thing is still happening --

[bcipolli]

Hi @elinaru , I will take a look at this over the weekend to validate the develop branch, and let y'all know if it's ready for master.

[jeremyad]

Accidentally posted this in the wrong issue before.

The problem appears to be with line 28 of disclosure/urls.py:
url(r'^docs/', include('rest_framework_swagger.urls')),
Simply removing this line eliminates the 500 error, but also gets rid of the (currently broken) documentation.

In a series of changes on June 7 (starting with this commit), the rest_framework_swagger module was totally rewritten, and there is no longer a rest_framework_swagger.urls module.

The new way to use django-rest-swagger is documented here. I should be able to take a look at this this evening or tomorrow, but if someone else wants to do it, they're welcome to.

@KyleW 's response there:

Good find! That pr is tagged with v2.0.0a0 which the release notes says has breaking changes. We probably didn't mean to pull that version in.

The easiest fix might be to point at an earlier, stable version and lock down our dependency so we don't accidentally grab latest.

[bcipolli]

Sounds like an easy fix? Just update the requirements.txt file to an appropriate version?

[bcipolli]

In the last build I used (worked), I had:

• rest_framework v3.3.2
• rest_framework_swagger v0.3.4

[KyleW]

@jeremyad just pushed the change. Although it looks like it inadvertently went straight to master (8f676fd). It's a pretty small change. Is everyone good with leaving the commit and closing the issue?

If we really want to keep the merge history clean we can revert it, cut a new branch, revert the revert on the new branch and open a pr from there. But I'm not sure it's worth it for this one.

We should also probably update the permissions to prevent pushing directly to master.