There are cca 1 000 Pywikibot bugs and tasks to do, it would be great for a hackaton, but it's too much of them. We can work on some project if there will be much interest. One project is already in plan: T186208 will be a challenge for me (as integration team is too busy to help).
Anyway I would rather do a training/skillsharing/workshop - personal help and knowledge. From basics/essentials to hard/deep-coded stuff. A look into the opportunities of the library; from using the scripts, through writing own scripts, up to the development. I can mentor patches for bugs driving you crazy.
Finally the meetup group is still growing to talk about any wonders coming to our minds, Pywikibot future and cooperations (T222608), issues and development, possibly some games (I have one: who can reach bug Traceback most quickly? :D I can do in 6 seconds with cfd.py, who can beat me? :D )
First question: Wanted templates generator
Is it possible to get a generator of the list of Wanted:Templates/subtemplates in order to copy all the Wikipedia templates needed on a target wiki thanks to the script transfertbot.py?
Not yet, created a task for it: T223713
Issues with code reviews
It takes long to review patches. There are too many open issues and changesets. There should be a maximum of 5 patches open at a time.
Issues with scripts
Some are poorly written and should be removed from the framework.
Every script has header identifying script to server. With access to foundation logs one can identify unused scripts.
Perhaps the probably unused should be archived, marked as "archived in 2019". ( T223826 )
After a year without complaints about them being archived, they should be removed.
Also be aware of non-WM wikis, which can use the scripts we no longer need (interwiki.py)
Another Python bot for Wikidata. Should Wikidata integrator and Pywikibot cooperate or merge?
It is not always clear how to use the framework correctly and understand the layers.
Incomplete warning should be lowered (PS: done). Some methods still missing.
Too many patches to review
There are patches from 2015 (older were abandoned because of being old)
There are 1000+ tasks:
- https://phabricator.wikimedia.org/maniphest/report/burn/?project=PHID-PROJ-orw42whe2lepxc7gghdq says even 1,300+
When is something worth upstreaming?
Just submit it and it can be improved later.
Is Gerrit apropriate for Pywikibot?
GitHub/Gitlab atracks more attention, are easy to use, but Gerrit from Wikimedia has online interface too and the tutorials are also great written.
Anyway Gerrit will be moved to Polygerrit this summer perhaps, this experience could be much better after that.
Do review all historic patches!
Pywikibot & Commons
There are too few scripts. Also there is missing a support for many API queries.
Perhaps tommorrow we should work together on Structured Data support ( T223820 ).
Python 2 deprecation ( T213287 )
We should not break the existing behavior, but many dependencies will do this soon. We should find a good compromise in a level of deprecation.
There are several possibilities:
- Don't patch py2 patches anymore
- above + don't write py2 compatible code anymore (breaking)
- above + remove all py2 specific code (breaking)
- ... (perhaps this should be further discussed) There were suggestions to fix only hard breaking changes and not soft breaking changes or the other way around as well.
- @Xqt: Remarks from outside: These proposals are problematic due to syntax changes in Python 3 and new functionalities which can improve further developing (e.g. Enum, async, coroutine, Pathlib) You can't have the cake and eat it ;) Everyone can go back to the last tag supporting older Python releases on their own risk. Currently we activly propose to migrate to Python 3 in next few months til 2020. But we also can hold the current core release for a while and then mark it as read-only in a year or so. And deploy a next release PWB 4.0 likewise Requests does it with Requests III supporting Python 3.6+ only.