Fri, Mar 27
This would be a content handler for whatever contentmodel type wikidata is using for the Q42 pages.
@matmarex's analysis seems correct to me.
Making T230683 a subtask at the moment, since we're probably not going to fix this until the cure is less disruptive. (Right now you'd recursively need to convert the list and all parent lists to explicit HTML syntax.)
I believe this was fixed by @Tgr in the patch above.
Documentation now at https://www.mediawiki.org/wiki/Parsoid/PHP
Only thing that's changed recently is 90d0f45209175f8313540c15a5be37a658fcc0a1; I don't immediately see anything wrong with that, but it might be worth trying to revert it locally on scandium and see if that helps.
@Arlolra did rewrite/touch the redirectToOldId function recently, didn't he? I remember commenting on IRC that there were dragons there...
Thu, Mar 26
Wed, Mar 25
One of the issues flagged by the VE team was stashing/switching support. Needs investigation to see what needs to be done/could be done here. If switching doesn't work reliably w/o RESTBase, we should turn it off in the default configuration, so that whatever we ship in the LTS "just works", even if it doesn't have 100% of features.
This is @matmarex's configuration: https://github.com/MatmaRex/patchdemo/blob/master/LocalSettings.txt#L15-L27
Mon, Mar 23
I'm treating the subtasks here as "candidates" for now, we can always remove stuff as LTS nears but it's harder to find or remember stuff we might want to do.
I believe this issue was resolved sufficiently by https://gerrit.wikimedia.org/r/550548 but re-open (or open a new bug) if additional issues arise during T248340: Remove use of ParsoidBatchAPI from Parsoid/PHP.
This was fixed, I think? Can it be closed? Or merged with T247206?
Sat, Mar 21
Fri, Mar 20
Needs re-investigation now that we've ported to PHP, who knows how we're encoding URLs now. Mostly likely we're using the same mechanism that core uses.
Can we get the mobile app team to review this? Mobile app and kiwix are the only ones who use the Parsoid style output, I believe. (The parser team uses the styles only for internal debugging.)
In general we would like to separate "page content" (in the "page content language") from user interface (in the "user interface language"). Currently much of that user interface is expressed in wikitext on sites like commons, but we'd like to reduce that over time. As mentioned before, wikitext in the user interface language complicates caching.