Topic on Project:Village Pump

Cleaning up empty Flow placeholders

10
Pppery (talkcontribs)

Flow is (sometime soon) going to be archived and undeployed from Wikipedia, which is going to involve a maintenance script moving every Flow page to a subpage and adding wikitext placeholder (phab:T371738). I'm thinking about, before that happens, running a bot to:

  1. Delete Flow pages that have an empty header and zero topics (like Talk:Wikimedia Release Engineering Team/SSD Sync Up/2019-03-05)
  2. Delete Flow pages that have no content other than a pointer to a /LQT Archive 1 page, and delete the /LQT Archive 1 page as well (like Extension talk:Vandal Brake and Extension talk:Vandal Brake/LQT Archive 1)

There are several hundred total of each type of page. I deleted a bunch manually earlier today.

Does this seem like a good idea.

Dinoguy1000 (talkcontribs)

and delete the /LQT Archive 1 page as well

...where the LQT archive page is empty as well, right?

Other than that, I agree, there's no reason to keep these empty pages around, especially if the maintenance script isn't going to attempt to delete them itself.

Pppery (talkcontribs)

In said cases the /LQT Archive 1 should always be empty since if it had content it would have been copied to the Flow page. But yes I should check explicitly.

Pppery (talkcontribs)

I reviewed the LQT archive 1 pages with substantive content. In a few places the Flow page was empty so I deleted the Flow page and moved the LQT archive over the Flow page (and removed the archive template). And there were a lot of other oddities.

Probably a lot of the /LQT Archive 1 pages can be deleted too (even if there is a Flow page with content it was based off of), since the vast majority of them have no content other than someone adding LiquidThreads, and then Flow replacing it.

Most of the ones that don't meet that description either contain only a pointer to old pre-anything archives (which are already in the Flow header). And there's a few cases where the LQT page contains wikitext because people mixed systems I guess ...

Pppery (talkcontribs)

I've continued reviewing a bunch of Flow pages. There are several thousands pages to delete per the above criteria. And I manually deleted a ton of old leftover crap of every imaginable kind

What should be done with pages like User_talk:Thennicke that contain nothing other than a pointer to a wikitext archive, in user talk namespace? These differ from the /LQT cases because the user explicitly enabled Flow. I'm inclined to delete the Flow board, move the wikitext archive back to the base title, and remove the {{Archive }} template. The other option would be to do nothing and let the migration script handle it.

Pppery (talkcontribs)

It also seems to be worthwhile to batch delete user talk pages that contain only the header and no Flow topics where the user in question has made few other edits. There are several thousand of them.

I manually reviewed the first ~100 cases of user talk pages that contain only the header and except for the above situation (repeated several times) every single one was either spam/out-of-scope/test edits that I deleted on-sight or some kind of useless platitude like "talk to me here", which I didn't bother pressing the delete button on but isn't worth letting the migration script move to a subpage on either.

Pppery (talkcontribs)

There are a total of 5031 empty flow boards, 297 boards with nothing other than /LQT placeholders, and 722 user talk pages containing only a header of users with <50 global edits.

TheDJ (talkcontribs)

Let's get rid of at least the empty stuff. Those 722 can go as well as far as I'm concerned, but we might want a bit more review and consensus on that ?

Quiddity (WMF) (talkcontribs)

It should be possible to update the maintenance script so that it automatically deletes Flow pages that have an empty header and zero topics. I see you (Pppery) commented about that at phab:T371738#10286448 already, so I'm just confirming that it is being looked into by the team.

For the completely-empty pages, I wonder if we might want to delete them without-log-entry, so that the empty wikitext page doesn't have the "this page was deleted" banner at the top as a cluttering-distraction? I believe that is technically possible. --- If folks here agree that this is a good idea, then we can ask the team to make it so.

For the "almost empty" pages, I agree those would need to be examined manually. Personally, I agree with both of you that those pages can probably just be deleted, once they've been checked for edge-cases.

Thanks again for all your associated efforts and thoughts on this, Pppery! It's deeply appreciated.

Pppery (talkcontribs)

I think deleting things without producing a log entry is downright scary and contrary to the wiki way of every action being logged.

On other issues, I assumed from the fact that my comment on phab:T371738#10215964 wasn't responded to and a maintenance script was merged and run without doing any of those things that the WMF was disclaiming interest in implementing them and thus the task fell to me to do it myself - I was prepared to run Project:Requests for permissions/Flow cleanup bot (and still am).

Even if the WMF does do that, though, I learned quite a bit from this experience, and the code I wrote there I also used to manually clean up a bunch of pages that really did need manual review.

Reply to "Cleaning up empty Flow placeholders"