Commons:Bots/Requests/タチコマ robot 2

From Wikimedia Commons, the free media repository
Jump to navigation Jump to search

タチコマ robot (talk · contribs)

Operator: とある白い猫 (talk · contributions · Statistics · Recent activity · block log · User rights log · uploads · Global account information)

Bot's tasks for which permission is being sought: Bot will tag featured pictures from other wikis (main concern is ar.wikipedia) using {{Assessments}}.

Automatic or manually assisted: Automatic, unsupervised after the generation of the list of pages to be edited.

Edit type (e.g. Continuous, daily, one time run): infrequent runs to tag new promotions/demotions

Maximum edit rate (eg edits per minute): 60

Programming language(s): AWB, bot already has a flag

-- とある白い猫 ちぃ? 07:36, 24 September 2012 (UTC)

Discussion

  • Please make a test run. --EugeneZelenko (talk) 14:55, 24 September 2012 (UTC)[reply]
  • I don't quite understand this request. Is this about extending the scope of a previously approved bot request? For the description it sounds like this is a small one time task. Does this warrant a request? --Dschwen (talk) 15:23, 2 November 2012 (UTC)[reply]
    • I doubt it would be one-time, since other wikis are going to be periodically featuring pictures and not marking them here. But it does seem like it would be infrequent. – Philosopher Let us reason together. 12:41, 1 December 2012 (UTC)[reply]
      • @Dschwen: It's quite common sense that small one time tasks (or small few times task) don't warrant a request, and I think such request basically give more work to bureaucrats, who have already a lot of work - judging by this page. Anyway, as far as I can see, bot policy is very strict about bots running without requesting, and there are no written exceptions in policy. IMO, providing some clear exceptions for low volume tasks would benefit bureaucrats, bot users and the project.--Pere prlpz (talk) 16:05, 11 December 2012 (UTC)[reply]
  • Four months and still no report of a test run. I will close this as stale within the next few days. --Dschwen (talk) 17:19, 27 February 2013 (UTC)[reply]
    • What exactly am I supposed to test? I am merely trying to understand the nature you want me to run. How many edits would be sufficient? My plan is to grab a category off a wiki that has featured pictures and use a regex to apply it to the assessments template (or add the template if it isn't present). -- とある白い猫 ちぃ? 23:58, 2 April 2013 (UTC)
      • 30-50 edits per step III of Commons:Bots/Requests. --99of9 (talk) 00:08, 3 April 2013 (UTC)[reply]
        • I might not be able to generate 30 edits due to people manually tagging which is why I was asking. I am processing all 3,345 files from and hopefully I have enough to reach that mark. -- とある白い猫 ちぃ? 16:08, 6 April 2013 (UTC)
          • Yeah I only have 17 edits (edit summary: "adding enwiki fa assessment, replaced: {{Assessments| → {{Assessments|enwiki=1|)"). It is a simple regex in AWB. Is this sufficient? -- とある白い猫 ちぃ? 19:09, 6 April 2013 (UTC)
  • Also, please note that 60 edits per minute is much higher than the normal maximum. --99of9 (talk) 00:09, 3 April 2013 (UTC)[reply]
    • This has been discussed before. Is there any reason why we have to worry about such an arbitrary number bot operators disregard? It artificially creates a massive backlog as 1000 edits would take 16.66 hours to complete. I have placed a remark on the talk page of the policy to avoid duplication. -- とある白い猫 ちぃ? 15:30, 6 April 2013 (UTC)
1000 edits would take 100 minutes with recommended maximum bot speed. I think that is a reasonable speed fot this kind of task. Bots can make errors and if they do they should not do it at to high speed. This bot needs to be fixed so that it links to correctly named nomination pages at English Wikipedia, like this. It also needs to make sure that the pictures it tags really are featured pictures, eg File:Eastern Screech Owl.jpg does not seem to ever have been featured and File:Ebony Bones backup performer.jpg seems to have been delisted as featured picture. /Ö 19:59, 6 April 2013 (UTC)[reply]
FA Page ghosts on en.wikipedia
  Eastern_Screech_Owl.jpg
  Ebony Bones-01.jpg
Linking to nomination pages is a later issue. Not all of them are marked on en.wikipedia or on commons and crawling for them is a non-trivial task. That is a future task I hope to tackle.
The idea is also to sync enwiki (and other wikis) delistings with commons. The bot has no way of verifying if a file is actually featured or if it tagged without even being nominated. It can however check if files are in the en:Category:Featured pictures or not. The two mentioned files were in the category or at least en.wikipedia reported them as such (Eastern_Screech_Owl.jpg doesn't exist as a page on en.wikipedia despite appearing on category, Ebony Bones-01.jpg is a redirect on en.wikipedia to Ebony Bones backup performer.jpg which is the matching image which again shouldn't appear on category). On my second sweep I'd run a similar regex to en:Category:Wikipedia former featured pictures. Also not everything is nicely marked so catching problematic images is again a future development task.
The restriction is no faster than 1 edit/10sec according to the linked policy page. Simple math is 10sec*1000=10,000secs, 10,000/60=166.66... minutes, 166.66.../60=2.77... hours, provided bot does not spend any time doing anything. I do not want to add a 10 second counter between edits which would serve no purpose than waste my time. I do not see the point of a speed limit for bots. It was thrown in as an idea that servers couldn't handle such speeds. Developers disagreed and did not see a problem. The quicker the bot edits the better as edits would be rolled out quicker. That way I can change the parameters to work on the next wiki. This script would probably check for about 20 wikis that have featured pictures - of which most don't have that many files or new files. This task would take probably no more than 20 edits per month after the initial run. I do not believe the speed limit is based on consensus mind you. Let's discuss bot speeds on Commons talk:Bots as its a more general question if this is worth tracking and enforcing.
-- とある白い猫 ちぃ? 02:36, 7 April 2013 (UTC)
Please notify me on my talk page if there are any developments here. I cannot watch a page forever. :/ -- とある白い猫 ちぃ? 22:56, 20 April 2013 (UTC)
  • Per the talk page I have adjusted the speed limit to 12 edits per minute (1 per 5s), but I'm not willing to go higher than that without very good evidence or directions from above (i.e. developers/techs of some kind). So please adjust this request to that rate. If the actual number of edits is as you expect, then this will take you only 2 minutes per month after the initial runs. --99of9 (talk) 12:48, 22 May 2013 (UTC)[reply]
  • I agree with Ö that linking to the nom would be very useful, especially since this allows others to verify the bot's edits. How is your progress in implementing this? If it's too hard for you, I suppose I'm ok with running the job as it is now, with a view to crowdsourcing the nom-linking. --99of9 (talk) 12:48, 22 May 2013 (UTC)[reply]
    I unfortunately haven't focused on it yet. I want to see how the bot preforms with this task as is to establish the workload. One possibility is to add the new images into a temporary category for crowd-sourcers to plow through. -- とある白い猫 ちぃ? 14:36, 23 May 2013 (UTC)

I propose to approve this bot once the speed limit is adjusted to 12 per minute. --99of9 (talk) 12:48, 22 May 2013 (UTC)[reply]

Exactly why do you want to force a speed limit? Speed limit for bots do not make any sense as I discussed on the policy page which didn't get much of a reply. I am just trying to understand the reasoning. I don't think the bot would make 12 edits per minute for this task (as typically you do not get 12 promotions even on en.wikipedia I think) but I am running multiple tasks and sometimes they may overlap. -- とある白い猫 ちぃ? 14:31, 23 May 2013 (UTC)
As a precaution to protect the server we need *some* limit (e.g. if supercomputer bot wanted to upload 1 million files per second, would you approve that??). I do not know what the ideal balance between bot operators and server protection is, but per that discussion I have doubled the speed limit and brought it into line with the absolute maximum on the meta policy page. You are going to have to ask someone above my paygrade if you want approval beyond the policy limits. --99of9 (talk) 14:49, 23 May 2013 (UTC)[reply]
I am not trying to be difficult. My question is more fundamental/philosophical, why do we care about edit speed at all when developers do not. When asked developers stated that such a protection isn't needed. Performance is their worry, not ours. There actually is an edit limit for bots too so it is not like a bot can upload a million files per second. Mind that it is impossible for bots to hit such edit speeds due to latency reasons alone (bot has to request a page, receive the page, parse the text on the page, send the modified content, request a confirmation (that it saved), receive the confirmation - even through API this takes time) unless they basically operate next to the servers or run multiple (unrelated) tasks at the same time through multiple computers (like my case). I just do not see the technical problem that prompts for us to even care about bot edit limits.
Mind that I am not complaining about my own bots edit speed. It will probably never hit either limits even if I don't bother with a timer due to latency. I am just saying the policy doesn't make sense.
-- とある白い猫 ちぃ? 17:32, 23 May 2013 (UTC)
If you can get an appropriately authorized developer to say that on record, then I'd be happy to see the meta policy changed. --99of9 (talk) 18:36, 23 May 2013 (UTC)[reply]
I am trying. Some people uttered en:Wikipedia:Don't worry about performance. -- とある白い猫 ちぃ? 20:11, 29 May 2013 (UTC)

Some people have pointed to WP:PERF, which is basically what I'm going to reiterate. Basically as long as a bot is well behaved and obeys maxlag, people shouldn't worry about setting artificial limits on bot rate for "performance" reasons. It's up to the developers and ops team to enforce these sorts of rules at the software level, which we do. That's not to say you should write a stupid bot that overloads the API just because a developer says don't worry about it--remember that we should all be good API citizens and keep in mind there's other people using the API too. Of course communities can impose their own limits if they still want (for social reasons, or to make cleanup easier if something goes wrong), but please don't do it in the name of performance. Note that this is all just generic commentary because I was asked to weigh in...I've got no opinions on this actual bot in question or the tasks it wants to perform. ^demon (talk) 20:28, 29 May 2013 (UTC)[reply]

I'd like to clarify (as above) I did not seek dev comment in relation to my bots edit limits (discussed above). I have no intention of trying to punish API as I am a good API citizen. :) -- とある白い猫 ちぃ? 05:28, 30 May 2013 (UTC)

Unless there are further objections, I propose that we approve this bot. We will still need discussions to decide how the speed policy issue should be handled (preferably at meta first). Following ^demon's comments, I can't see the simple addition of a handful of templates crashing the servers, so I'm ok with approving this particular bot, even at this speed. --99of9 (talk) 15:09, 30 May 2013 (UTC)[reply]

I concur. --Dschwen (talk) 15:17, 30 May 2013 (UTC)[reply]
I will voluntarily use a speed limit beyond whats approved here for the time being as the workload for this bot wouldn't be that high. As I said my reservations mainly was the existence of the general speed limit which I feel needed more discussion. That of course as you said should be discussed at meta. -- とある白い猫 ちぃ? 01:33, 3 June 2013 (UTC)

Approved. --Dschwen (talk) 04:02, 3 June 2013 (UTC)[reply]