Page MenuHomePhabricator

MatchBot: Determine what Flow - newtopic considers a duplicate post
Closed, ResolvedPublic

Description

This will have implications for allowing Co-op members to be matched with a new mentor, as well as making it easier to avoid duplicate posts.

Event Timeline

Fhocutt claimed this task.
Fhocutt raised the priority of this task from to Medium.
Fhocutt updated the task description. (Show Details)
Fhocutt added a project: Community-IdeaLab.
Fhocutt renamed this task from Determine what Flow - newtopic considers a duplicate post to MatchBot: Determine what Flow - newtopic considers a duplicate post.Jan 10 2015, 5:05 AM
Fhocutt set Security to None.
Capt_Swing raised the priority of this task from Medium to High.Jan 12 2015, 10:00 PM

I was able to spam https://test.wikipedia.org/wiki/Wikipedia_talk:Co-op/Flowtest2 with duplicate messages/duplicate topics via the Flow API (submodule=new-topic), even when making two consecutive posts with the same parameters.

The problem turned out to be https://phabricator.wikimedia.org/T86612 ; bot-created Flow-enabled pages are not being listed as such, effectively leading to blank API requests and nochange.

We don't currently deduplicate repeated submissions, some ideas for solving that are at T76559 but have not had a chance to implement yet. It seems like here the problem is more just that the bot repeats the query, due to not identifying the page as already converted? Following up in the other bug.

@EBernhardson, correct, that was the problem. It would be nice to detect duplicates, but probably not worth the effort for this.