Wikipedia:Bots/Requests for approval/VWBot 8
- The following discussion is an archived debate. Please do not modify it. To request review of this BRFA, please start a new section at WT:BRFA. The result of the discussion was Approved.
Operator: VernoWhitney (talk · contribs)
Automatic or Manually assisted: Automatic
Programming language(s): perl
Source code available: http://www.uberbox.org/~marc/csb.pl
Function overview: Duplicate CorenSearchBot. See Wikipedia:Bots/Requests for approval/CorenSearchBot
Links to relevant discussions (where appropriate): Wikipedia:Bot owners' noticeboard#CorenSearchBot; back-up?
Edit period(s): Continuous
Estimated number of pages affected: Many
Exclusion compliant (Y/N): N
Already has a bot flag (Y/N): Y
Function details: See Wikipedia:Bots/Requests for approval/CorenSearchBot
Discussion
[edit]I downloaded the code and required dependencies earlier and had it running a trial this morning. I am at present insufficiently familiar with perl to have completely tweaked it, so it made both the userspace edit I was looking for here, as well as a mainspace edit here. As such I have currently suspended execution. VernoWhitney (talk) 13:38, 20 August 2010 (UTC)[reply]
- As this is a cloned task, an essential task, and an experienced operator, speedily Approved for trial (100 edits). Please provide a link to the relevant contributions and/or diffs when the trial is complete. –xenotalk 13:45, 20 August 2010 (UTC)[reply]
- I should note that all of the templates still refer to CorenSearchBot, but that shouldn't (hopefully) cause too much of a problem during the trial as Coren's talk page is heavily watchlisted by those who work the results. I don't have time to tweak them just now I'm afraid since I have to run off for a few hours. VernoWhitney (talk) 13:54, 20 August 2010 (UTC)[reply]
- I made some tweaks [1]. –xenotalk 14:16, 20 August 2010 (UTC)[reply]
- Thanks. I went ahead and made the same changes to the rest of the templates it uses, so it should be all good and any annoyed customers should now be headed my way for the duration of the trial. VernoWhitney (talk) 17:31, 20 August 2010 (UTC)[reply]
- So now CorenSearchBot is back and running, [2], I've suspended the trial so articles don't get double-tagged (and I think they would). VernoWhitney (talk) 19:11, 20 August 2010 (UTC)[reply]
- Okay, so since the trial has resumed following the conversation (and pointing me at part of the code I had missed) at Wikipedia:Bot owners' noticeboard#CorenSearchBot; back-up?, I haven't noticed any double-tagging, but there is this odd edit when the bots were racing to notify the user. VernoWhitney (talk) 17:41, 23 August 2010 (UTC)[reply]
- As long as it does happen too often, it shouldn't be a problem. Maybe build in some delay so VWBot is truly a backup rather than a competitor... –xenotalk 17:44, 23 August 2010 (UTC)[reply]
- I've been stalking both bots' contributions and that's the only issue I've seen so far which has resulted from both of them running at once. There are some other bugs in the code which annoy me (like this notification where the page was never listed on the SCV daily page, and as I understand it that happens on occasion with CorenSearchBot too) but they all appear to be unrelated to the bot overlap. I can look into adding some delay, but I don't think I'll have time to do that before this weekend. VernoWhitney (talk) 17:53, 23 August 2010 (UTC)[reply]
- Update: One double notification occured - and I'm not sure why since (if I'm reading the code right) it should only notify the user after successfully tagging a page, which hasn't happened. VernoWhitney (talk) 23:28, 23 August 2010 (UTC)[reply]
Trial complete. While I haven't run 100 edits for this task, it's run ~90+ and I don't have access to the computer running it during the day, so I stopped it before I left this morning so it wouldn't run over. VernoWhitney (talk) 13:24, 24 August 2010 (UTC)[reply]
{{BAG assistance needed}} Please forgive the template, but as CorenSearchBot has made no edits in the last 4.5 hours, it is likely that it's is down again. VernoWhitney (talk) 14:14, 27 August 2010 (UTC)[reply]
- Have you made any progress in allowing these to work in parallel? Please go ahead and run the bot again for now, while we wait for another BAG member to give this a second look. –xenotalk 14:16, 27 August 2010 (UTC)[reply]
- I haven't had the chance to sit down and be sure I can delay it without breaking it. I should have plenty of time to do that tomorrow. VernoWhitney (talk) 14:33, 27 August 2010 (UTC)[reply]
- A fairly simple suggestion, IMO, is to just delay when looking for new pages. It uses the revision number as a bookmark so you won't miss any, and it doesn't tag anything that's already tagged so that being "laggy" on purpose won't generally cause problems. Adding a sleep() before the NewPages call should do the trick. However, there is a possible race with running two instances: (a) bot searches, tags. (b) User remove tag because it's a false positive or attributed or fixed, etc. (c) backup bot gets around to checking the same pages, and tags again. Probably, (d) user gets mighty annoyed. — Coren (talk) 16:25, 27 August 2010 (UTC)[reply]
- I haven't had the chance to sit down and be sure I can delay it without breaking it. I should have plenty of time to do that tomorrow. VernoWhitney (talk) 14:33, 27 August 2010 (UTC)[reply]
Okay, I have now modified the code so that when it determines that an article needs tagging, it will first wait 120 seconds to see if CSBot go first (this can be updated at User:VWBot/config if a longer wait is preferred, it'll just be a tradeoff between only being a backup vs. quick article tagging), and then it will check to see if CorenSearchBot has ever edited the article before tagging it, so as to avoid the race condition Coren pointed out above. VernoWhitney (talk) 19:38, 28 August 2010 (UTC)[reply]
- Approved for trial (1 week). Please provide a link to the relevant contributions and/or diffs when the trial is complete. Go ahead and run it for a while and see how it goes. –xenotalk 22:13, 28 August 2010 (UTC)[reply]
- Wouldn't it be worthwhile for me to shut CSBot down at some planified time for a bit so we see if VWBot takes over as expected? — Coren (talk) 17:55, 31 August 2010 (UTC)[reply]
- Probably. Regarding [3]; wasn't there supposed checks to avoid double-tagging? –xenotalk 17:59, 31 August 2010 (UTC)[reply]
- First, you could shut it down just to be sure; as you can see from the recent contribs it's still tagged a handful before CSBot, so I keep upping the MinEditDelay at User:VWBot/config, and it should work just fine as the only tagger. Second, that wasn't actually a double-tagging, although I can see why you'd think that. The article was previously deleted (at Quran with tafseer, but we got OTRS verification. For some reason RHaworth didn't want to simply restore the article so he emailed me the source and I created the article anew at its current location. I removed the CSB tag it came with so when VWBot looked at it, it didn't see one and the new article had never been edited by CorenSearchBot. Only later did RHaworth restore the history and make it look it had. VernoWhitney (talk) 19:17, 31 August 2010 (UTC)[reply]
- Ah ok- I didn't notice the deletion log. Thanks for explaining. Are there any lingering issues? This can probably be marked 'approved' and you can continue to work with Coren to ensure these two bots play nice together. –xenotalk 13:33, 3 September 2010 (UTC)[reply]
- I haven't noticed any lingering issues. It's still been tagging a couple a day and it's been reporting some as clean in my log files which CSBot has tagged as copyvios, I'm not sure why there's a difference (maybe Yahoo's feeding us different results?) but still almost all of the attempted taggings match up. I'm fairly certain it's no longer beating CSBot as it's been waiting the 5 minutes I have it set to and Coren's generally tags 0-3 minutes after article creation, so we shouldn't have to worry about CSBot coming along after VWBot's tag has been removed. If that does start happening I can give Coren my code for checking article history for previous taggings, but that's a bunch of extra queries to worry about that shouldn't be needed so long as my delay is large enough. VernoWhitney (talk) 14:03, 3 September 2010 (UTC)[reply]
- Cool. Thanks to both you and Coren for your efforts thus far. Keep it up. Approved. –xenotalk 14:12, 3 September 2010 (UTC)[reply]
- I haven't noticed any lingering issues. It's still been tagging a couple a day and it's been reporting some as clean in my log files which CSBot has tagged as copyvios, I'm not sure why there's a difference (maybe Yahoo's feeding us different results?) but still almost all of the attempted taggings match up. I'm fairly certain it's no longer beating CSBot as it's been waiting the 5 minutes I have it set to and Coren's generally tags 0-3 minutes after article creation, so we shouldn't have to worry about CSBot coming along after VWBot's tag has been removed. If that does start happening I can give Coren my code for checking article history for previous taggings, but that's a bunch of extra queries to worry about that shouldn't be needed so long as my delay is large enough. VernoWhitney (talk) 14:03, 3 September 2010 (UTC)[reply]
- Ah ok- I didn't notice the deletion log. Thanks for explaining. Are there any lingering issues? This can probably be marked 'approved' and you can continue to work with Coren to ensure these two bots play nice together. –xenotalk 13:33, 3 September 2010 (UTC)[reply]
- First, you could shut it down just to be sure; as you can see from the recent contribs it's still tagged a handful before CSBot, so I keep upping the MinEditDelay at User:VWBot/config, and it should work just fine as the only tagger. Second, that wasn't actually a double-tagging, although I can see why you'd think that. The article was previously deleted (at Quran with tafseer, but we got OTRS verification. For some reason RHaworth didn't want to simply restore the article so he emailed me the source and I created the article anew at its current location. I removed the CSB tag it came with so when VWBot looked at it, it didn't see one and the new article had never been edited by CorenSearchBot. Only later did RHaworth restore the history and make it look it had. VernoWhitney (talk) 19:17, 31 August 2010 (UTC)[reply]
- Probably. Regarding [3]; wasn't there supposed checks to avoid double-tagging? –xenotalk 17:59, 31 August 2010 (UTC)[reply]
- Wouldn't it be worthwhile for me to shut CSBot down at some planified time for a bit so we see if VWBot takes over as expected? — Coren (talk) 17:55, 31 August 2010 (UTC)[reply]
- The above discussion is preserved as an archive of the debate. Please do not modify it. To request review of this BRFA, please start a new section at WT:BRFA.