Jump to content

Wikipedia:Bots/Requests for approval/Femto Bot 7

Add links
From Wikipedia, the free encyclopedia
< Wikipedia:Bots | Requests for approval
This is an old revision of this page, as edited by Rich Farmbrough (talk | contribs) at 18:40, 25 February 2016 (→Femto Bot 7). The present address (URL) is a permanent link to this revision, which may differ significantly from the current revision.

Femto Bot 7

Operator: Rich Farmbrough (talk · contribs · SUL · edit count · logs · page moves · block log · rights log · ANI search)

Time filed: 21:05, Sunday, January 17, 2016 (UTC)

Automatic, Supervised, or Manual: Automatic

Programming language(s): perl

Source code available: no

Function overview: Creates/updates lists of articles by location that require images.

Links to relevant discussions (where appropriate): Several year old request from User:WereSpielChequers.

Edit period(s): Will probably do a full scan monthly, might remove "fixed" items from the list daily.

Estimated number of pages affected: Initially one 185 pages for UK articles. May well add more, as there seems to be a demand for this type of report.

Exclusion compliant (Yes/No): n/a (will only be editing it's "own" pages).

Already has a bot flag (Yes/No): Bot flag need re-enabling.

Function details: Produces a report of pages in a given geographical area which require images. Details may vary by region. For example the UK page will have OSGB grid squares, which make it easier to find Geograph images.

UK example layout

Discussion

Where is the report produced at? How specific are the regions, e.g. how many report pages do you expect to create?

Approved for trial (UK report). Please provide a link to the relevant contributions and/or diffs when the trial is complete. Sounds like a straight-forward list-generating task that doesn't involve direct article edits. We can probably look at your sample report to see if there's any issues, though I can't foresee anything. —  HELLKNOWZ  ▎TALK 22:40, 17 January 2016 (UTC)[reply]

I've yet to see how big the (UK) report will be, but I expect a few thousand entries, which would go on one page - probably in my user space while I am trialling. Later it may go in Wikipedia space, probably under the "Database reports" hierarchy.
All the best: Rich Farmbrough, 21:59, 18 January 2016 (UTC).[reply]
We could also do with a list for Ireland, and ideally split out lists for Wales and Scotland. The finer the geography the better the chance of getting wikiprojects involved. At the other end of the scale, if there are other parts of the world with as many images on Commons it would be good to get lists (the UK is 0.1% of the Earth's landmass and about 10% of Wikimedia Commons, so there can't be many places currently as well covered on commons. But we might have photographers interested in lists of articles without images in an area they are or will be in. ϢereSpielChequers 13:41, 23 January 2016 (UTC)[reply]
Any updates, Rich? — Earwig talk 06:40, 31 January 2016 (UTC)[reply]
Yes I have, for example, this which is fairly raw still. All the best: Rich Farmbrough, 16:24, 1 February 2016 (UTC).[reply]
And here is the type of page generated for Aberdeen. All the best: Rich Farmbrough, 23:57, 18 February 2016 (UTC).[reply]
@Earwig: @WereSpielChequers: All the best: Rich Farmbrough, 23:59, 18 February 2016 (UTC).[reply]
How often is this going to edit, and how many pages per day? Ideally I'd like some assurance of hard maximums for now, given prior concerns from the community. --slakr\ talk / 04:34, 19 February 2016 (UTC)[reply]
I envision once per month, and possibly updates on demand, it will only be editing pages created by me to hold the results. As to the number of pages it may be one or two (British Isles or UK and Ireland), one per country/territory (Ireland, Scotland, Wales, NI, IOM, Jersey Guernsey) or one per primary subdivision for the UK and Ireland.
All the best: Rich Farmbrough, 11:52, 19 February 2016 (UTC).[reply]
Really useful reports, thanks Rich. As for updating the more frequently the better please, daily would be better than monthly. It is much easier to teach newbies to add images using this list if they don't have to be told to remove entries they have added images for. ϢereSpielChequers 13:27, 19 February 2016 (UTC)[reply]
It's quite a nice tool for spotting errors too. I have recovered an academy from the sea, and moved an arboretum to its correct country already. All the best: Rich Farmbrough, 13:52, 19 February 2016 (UTC).[reply]

Trial complete. (@WereSpielChequers:) User:Rich Farmbrough/temp138. Only surprise is that there are some empty subdivisions. All the best: Rich Farmbrough, 23:15, 21 February 2016 (UTC).[reply]

@Hellknowz:, @The Earwig:... any comments? All the best: Rich Farmbrough, 04:08, 25 February 2016 (UTC).[reply]
@Rich Farmbrough: That was wayyy more than "one page." In fact, you've made thousands of edits—and the bot is still running and has made, in some cases three edits to a single page in 24 hours. --slakr\ talk / 05:01, 25 February 2016 (UTC)[reply]
...and whatever's going on here and here is clearly broken. --slakr\ talk / 05:04, 25 February 2016 (UTC)[reply]
...blocked for now. Normally I wouldn't be as itchy with the trigger finger on this—and userspace edits are normally fine—but:
  1. There are bugs.
  2. The bot continued editing after the it was marked "trial complete."
  3. Mismatch between number of edits bot was claimed to be making per day and number of pages ("once per month" and "it may be one or two").
  4. Prior arbcom issues and sanctions related to automated edits.
Anyone is free to unblock at their discretion, but I don't personally feel comfortable with allowing this bot to constantly make edits over hundreds or thousands of pages, multiple times throughout the day, without some sort of clear boundary. The crossing of boundaries even within this BRFA is, for me, cause for concern—especially considering this is but one of hundreds of countries that this could happen on with this rather open-ended task description.
--slakr\ talk / 05:36, 25 February 2016 (UTC)[reply]
It's a shame that you used the block button rather than using your words.
Your "thousands of edits to hundreds of pages" are 666 edits to 185 pages.
Since I am running in user space there is no need for me to stop developing the bot merely because the trial is complete. BAG's imprimatur is only needed if and when the reports are moved to Wikipedia: space.
Your reference to "clearly broken" behaviour is quite accurate, however that diagnostic, identified a difficulty in following redirects, a feature I have now successfully added.
Pretty clearly until the bot was run there was no way to know the size of the data set. It is too large for a single page. Moreover the user requirements expressed by User:WereSpielChequers are for a more fine-grained solution. Thirdly the finer granularity enables smaller deltas.
As for bugs there are currently three known expressions, two of which are probably caused by Femto Bot not having the bot flag, and hence being limited to the number of items returnable from an API call, and one which is of unknown cause, but probably due to an API call failing to serve a page. I was hoping to resolve the third today.
This task has already been held up for nearly four years by bureaucracy, it is not good to hold it up further. The task as it stands meets with the criteria of being harmless and useful.
All the best: Rich Farmbrough, 14:24, 25 February 2016 (UTC).[reply]
14:24, 25 February 2016 (UTC)
Oh yes and the spec is clear "might remove "fixed" items from the list daily". All the best: Rich Farmbrough, 14:42, 25 February 2016 (UTC).[reply]
I think the main issue is that you specified "one page", the trial was for "UK report" (singular) without daily fixes and then you confirmed "one page", but we have 183 pages and many more edits. You did not communicate this, you simply went ahead with it. I gave you a quick trial to let you produce a report so we can focus on the merits of the task and not the operator. I would have approved the task as clearly useful for singular per-country pages with reasonable edit intervals. With ArbCom sanctions, I would have expected you to be triple-careful and I hoped you'd show us such diligence. But you have significantly miscommunicated and over-edited the trial. If such a trial is indication of future past approval, I cannot approve or endorse it. —  HELLKNOWZ  ▎TALK 15:25, 25 February 2016 (UTC)[reply]
In response to Slakr's query I said

As to the number of pages it may be one or two (British Isles or UK and Ireland), one per country/territory (Ireland, Scotland, Wales, NI, IOM, Jersey Guernsey) or one per primary subdivision for the UK and Ireland.

I don't think this could have been clearer. The time to have made objections would have been in response to that statement.
All the best: Rich Farmbrough, 18:03, 25 February 2016 (UTC).[reply]
Retrieved from "https://en.wikipedia.org/w/index.php?title=Wikipedia:Bots/Requests_for_approval/Femto_Bot_7&oldid=706862839"