Upgrade Python to 3.8.2 and Django to 3.0.7 #1507
Conversation
Also a bunch of other Django packages
Also, a system check failed because some of the emoji are 3 characters long, so this changes the max_length for the Team.emoji field to 3.
|
So… first roadblock! The django-jet package isn't compatible with Django 3. There's a nice loud message on their README: https://github.com/geex-arts/django-jet Basically, their reincarnation is the new Jet Bridge, which seems way overkill for what we want. It also requires a Jet account, which presumably means that everyone who installs Tabbycat will have to create a Jet account, and it won't just add itself seamlessly by being included in Tabbycat. Barukimang has patched django-jet for Django 3.0: geex-arts/django-jet#430 Of course, we could just remove Jet altogether and revert back to the original Django admin site for the Edit Database area. But there's a good reason we had Jet in the first place—the Django admin site isn't that friendly (and isn't meant to be). Or we could find a substitute. Thoughts? |
|
… Otherwise, everything seems fine. Our Travis PR hook is broken, but it happened on the branch: https://travis-ci.org/github/TabbycatDebate/tabbycat/builds/698352467 |
|
Not the first time changing DB admin UI: #259 We could try the default; it may have improved. |
Just did. It has not. 😅 But on a quick search, there are countless others! |
|
django-suit has a v2.0 in development that according to this issue is more or less functional, but not polished. Other packages here: https://djangopackages.org/grids/g/admin-styling/ |
|
|
Only a few of those other packages are themes, and most of them are relatively minimal. I'd be inclined to stick with the (patched) Jet and push the question back to whenever a future breaking change occurs. By then Suit 2.0 may be released, a Jet fork may be solid, or there may be other options. |
|
I see. My main concern about running with the patched Jet is that I don't think it's officially released on pip (understandably so, given that it's a fork). We could reference Barukimang's repository directly from our requirements file, but I think I'm just worried about the dependency reference potentially breaking one day, since it's not "official", so to speak. For a small project I wouldn't bat an eyelid, but I don't know how many people deploy Tabbycat sites per day these days 🙈 |
The original django-jet is no longer maintained and isn't compatible with Django 3. Another user, Barukimang, has made a fork to patch the parts that aren't Django 3-compatible. See https://github.com/Barukimang/django-jet/
config/requirements_core.txt
Outdated
| # Django admin (Jet) | ||
| # This is an unofficial fork of no-longer-maintained geex-arts/django-jet, | ||
| # patched to support Django 3. | ||
| git+https://github.com/Barukimang/[email protected] |
There was a problem hiding this comment.
@philipbelesky Just drawing attention to this line, which is the patched django-jet
There was a problem hiding this comment.
Ok. Also to address your comment above, we could always do a fork ourselves if we were concerned about it vanishing or having a breaking update.
There was a problem hiding this comment.
Yeah, we could. I'd describe my concern as more about the consequence than the probability of it happening—I doubt it'll disappear, but if it does, Tabbycat will become undeployable until we release a patch with our own fork.
|
More digging:
It seems that there are mostly just two changes to the original django-jet repository:
Do you have thoughts on picking an existing fork vs making our own? Having figured out the above, I'm now kind of divided about it. Again, for a personal project I'd just pick one and not worry too much about it, but I guess this isn't really a personal project anymore. |
|
Using r-django-jet sounds good if it is up pypi. Altered css files should be no problem. |
Closes #1501
Mostly making a PR some there's somewhere to discuss with reference to code, if needed.