Wikipedia:Administrators' noticeboard: Difference between revisions

X2-nuke interim period

Wow, cool! Glad we made some progress, and just trying to nail down the next steps to keep things moving smoothly. To recap my understanding:

• we are now in the "inoculation period" with a fortnight-timer which expires 23:22, 6 May 2017 (UTC) an interim period where we figure out how to implement this.
• during this period, anyone may tag articles in the list at Wikipedia:Administrators' noticeboard/CXT/Pages to review with the proper tag to prevent nuking two weeks hence

A couple of questions:

• do we have to recruit someone to write a script to do the actual nuking?
• what form should the actual "vaccination" tag have? In the proposal above, I just kind of threw out that expression: {{bots|deny=X2-nukebot}} but I have no idea how we really need to tag the articles, and maybe that's a question for the script writer?
• will the bot also observe strikeout type as an indicator not to nuke? A possible issue is inconsistent usage among editors: for example, some editors have not used strikeout for articles they have reviewed and clearly wish to save (e.g. see #1601-1622)

As for me, I will continue to tag a couple hundred more articles with language-tags as I did previously in the 1301-1600 range, to make it easier for everyone to find articles translated from languages they are comfortable working with, and that they therefore might wish to tag. Mathglot (talk) 02:28, 23 April 2017 (UTC) Updated by Mathglot (talk) 09:02, 23 April 2017 (UTC)[reply]

• Let's make two lists, one of articles to delete and the other of articles to retain for the moment. I don't think that it will be necessary to formally request a bot. We have quite a few sysops who could clean them all out with or without scripted assistance.—S Marshall T/C 15:55, 23 April 2017 (UTC)[reply]

  I would implement it as a giant sortable wikitable - Something that looks like this:

Tazerdadog (talk) 17:17, 23 April 2017 (UTC)[reply]

Isn't the current list easier to deal with than creating a new table, or two new ones? Can we just go based on strikeout type, or add some unambiguous token like, nuke=yes in the content of the items in the enumerated list that need to be deleted? I'm just trying to think what would be the least work to set up, and easiest to mark for those interested in vaccinating articles.

If we decide to go with a table, I might be able to use a fancy regex to create a table from the current bullet list. Although I definitely see why a table is easier to view and interpret once it's set up, I'm not (yet) persuaded that there's an advantage to setting one up in the first place. For one thing, it's harder to edit a table than a bullet list, because of the risk of screwing up cells or rows. Mathglot (talk) 18:59, 23 April 2017 (UTC)[reply]

The real advantage of the table is the ability to sort by language. This way, if we have a volunteer who speaks (for example) only English and Spanish, they can just sort the table by language, and all of the Spanish articles will be shown together. It's harder to edit, but in my opinion, the ease of viewing and extracting the information far outweighs this.

I have created a list that removes all struck items at Wikipedia:Administrators' noticeboard/CXT/Pages to review/Tazerdadog cleanup list. I'm currently working on getting rid of the redlinks as well. Once that is done, we can move to a vaccination model on the articles that have not been cleaned up in the articles thus far. The vaccination can take virtually any form as long as everyone agrees on what it is - I'd recommend that we vaccinate at the central list/table rather than on the article however. Once the two weeks expire, it's trivial to extract the unvaccinated articles and poke a sysop for deletion. Tazerdadog (talk) 00:07, 24 April 2017 (UTC)[reply]

@Tazerdadog: This was posted over at Wikipedia talk:Administrators' noticeboard/CXT/Pages to review as well but wanted to mention it here. Timotheus Canens has created a language-sortable table in their sandbox at User:Timotheus Canens/sandbox that I think is similar to what you were thinking. Mz7 (talk) 04:06, 25 April 2017 (UTC)[reply]

And we may have to recreate the table, as I didn't notice it and have been continuing to mark language codes on the main list (and shall continue to do so, unless someone yells "Stop!"). Also, not sure how trivial it is: given a full set of instructions what to do, then, yes, it's trivial, but this is not formatted data (yet) and there are all sorts of questions a sysop might have, such as, what to do with ones marked "moved", or "redirected", and other situations I've come across while going through the list that don't spring to mind. We don't want to burden the sysop with an illy-defined task, so all of those situations should be spelled out before we ask them to take their time to do it, as if there are too many questions, they'll either give up, or they'll do whatever they feel like. Mathglot (talk) 06:20, 25 April 2017 (UTC)[reply]

@Timotheus Canens: Tazerdadog (talk) 05:55, 26 April 2017 (UTC)[reply]

And am still doing so on the main page, and so have at least six others since the message just above this one was written. Mathglot (talk) 11:53, 4 May 2017 (UTC)[reply]

X2 countdown and vaccinate indicators

Floating a proposal to get the clock started on the two weeks. Any user can write "Vaccinated" (or anything equivalent , as long as the meaning is understood) on the list on the same line as the Strike out any article they want to vaccinate. I can then go through and use regex to remove the vaccinated articles line-by-line from the delete list. I will then separate out the articles with no substantive commentary attached (anything beyond a language or a byte count is substantive) for an admin to delete or draftify. Any article which has been individually substantively discussed will be evaluated independently. If this is OK, we can start the clock. Tazerdadog (talk) 22:23, 6 May 2017 (UTC) Updated Tazerdadog (talk) 06:07, 7 May 2017 (UTC)[reply]

People are already using strikeout type as the "vaccinate" flag so no additional method is needed though I see nothing wrong with using both, if someone has already started with the the other method. Mathglot (talk) 22:51, 6 May 2017 (UTC)[reply]

Also, I have been placing substantive commentary on plenty of articles, with the intention of facilitating the work of the group as a whole, in order to aid people in deciding whether that article is worth their time to look at and evaluate. In my case at least, substantive commentary does not indicate a desire to save, and if you intend to use it that way in the general case, then you need to suggest another indication I can use as a "poison pill" indicator to ensure it is nuked despite the substantive commentary. Mathglot (talk) 23:31, 6 May 2017 (UTC)[reply]

@Mathglot: Strikeout works even better than my idea, as it is easier to write the regex for. I was figuring that substantive commentary at least deserved to be read before we nuked them, although unless a comment was actively positive on the article I would have sorted it as a delete. If you want every article you commented on to be deleted, I can use your signature as the poison pill. Otherwise, use what you want, just make sure it is clear what it is. Ideally, place it at the start of a line, so I don't have to think when writing the regex. Tazerdadog (talk) 06:07, 7 May 2017 (UTC)[reply]

@Tazerdadog: If you need a tester, feel free to shoot me a pattern; I'm a bit of a regex wonk myself, plus I have a nice test app for it. Can't use my sig as poison pill, cuz often my commentary is unsigned cuz I did them 20 or 50 at a time, with the edit summary carefully explaining what was done, but no sig on the individual line items. Beyond that, quite a few have commentary by multiple people, so even if I did comment (and even sign) others may have, too. The only clear way to do this, afaics, is to have an unequivocal keep (or nuke) indicator (or more than one is okay, if you want to OR them) but anything judg-y like "substantive commentary" seems risky to me. In the latter case, we should just get everyone to review all their edits they forgot to strike, and strike them now, or forever hold their peace. In my own case, no matter how positive my comment, or how long, if there's no strike on the article title, it's a "nuke". It occurs to me we should poll everyone and get positive buy-in from all concerned that they understand the indicator system, to make sure everyone knows "strike" equals "keep" and anything else is nuke (or whatever we decide). It won't do to have 2,000 articles nuked, and then the day after, "Oh, but I thought..." Know what I mean?Mathglot (talk) 06:50, 7 May 2017 (UTC)[reply]

@Mathglot: I think the solution is to draftify until everyone agrees that no mistakes have been made, then delete. I'm happy to do the grunt work of the manual checking of longer entries, and I don't think it is particularly risky to do so. However, the vast majority are short, and can and should be handed with a little regex script. We do need to make sure that the expectation of strikeout = delete instead of strikeout = resolved was clear to all parties. As for a deleteword, literally anything will do if it is unique and impossible to misinterpret. I would recommend "kill" as this deleteword, as it is clear what the meaning is, possible to write the regex for, and currently has only a couple of false hits in the page that can be worked around easily. Does this work for you?

The reasoning for checking longer entries is to try to catch entries like this:

|Battle_of_Urica -seems fine, at least not a translation issueElinruby (talk) 19:14, 3 November 2016 (UTC)[reply]

Tazerdadog (talk) 08:02, 7 May 2017 (UTC)[reply]

@

Draft namespace is debatable, but might be the right solution.) As far as regexes, I count 738 <s> tags, 732 </s> tags, 587 keepers, and 2785 nukers as of May 7 ver. 779254187. Mathglot (talk) 22:40, 7 May 2017 (UTC)[reply

]

@

WP:SOFTDELETE, i.e. if someone asks for a small number to be restored after they have been deleted so that they can work on them they can just ask any admin to do so. I think that's all that needs to be resolved for now, so I'm going to go ahead and start the two week countdown until someone yells at me to stop. Pinging some participants: @S Marshall:@Elinruby:@Yngvadottir: Tazerdadog (talk) 23:12, 7 May 2017 (UTC)[reply

]

For clarity, the process is: At the deadline, June 6, 2017 all struck articles listed at Wikipedia:Administrators' noticeboard/CXT/Pages to review will be retained, and all unstruck articles will be deleted. Articles with significant commentary attached will have the commentary read before the deletion, but the default is the struck/unstruck status unless the commentary indicates clearly the opposite result is better. The work "kill" may be added to unambiguously mark an article for deletion. On or after June 6th, the regex nerds will compile a list of articles to delete and retain. The delete list will be moved to draft space (or subpages of Wikipedia:Administrators' noticeboard/CXT/Pages to review), where it will be audited briefly just to make sure nobody made a systematic error, then deleted. Tazerdadog (talk) 23:12, 7 May 2017 (UTC)[reply]

Per #deadline it's June 6. Your clarifications on "draftify" and the process all sound good, otherwise.
P.S. Note that one article matches /kill/i but none matches /\bkill\b/i. Mathglot (talk) 23:28, 7 May 2017 (UTC)[reply]

Fixed, I was unaware of that discussion, thank you. Tazerdadog (talk) 23:34, 7 May 2017 (UTC)[reply]

@Mathglot and Tazerdog: so for purposes of making life easier I will strike what I think should be struck. At one point people were checking my work so I was rather tentative initially. I am following the regex discussion but haven't used it in a while so save me the trouble of looking this up -- did you conclude that "kill" would be useful, or not? Elinruby (talk) 00:58, 8 May 2017 (UTC)[reply]

@Elinruby: If the title is strikeout type, it will be kept; if it isn't, it won't. Placing "Kill" on an article has no effect at nuke time, but it does have a beneficial effect now:, i.e., it saves time for others. It lets others know that you have looked at this one and found it wanting, so they should save their breath and not even bother looking at it. For example: You marked #18 Stevia_cultivation_in_Paraguay "really, really bad". That was enough for me not to bother looking at it, so you saved me time, there. If you want to place "kill" on the non-deserving items you pass by, that will help everybody else. I may do the same. But in the end, on Nuke day, the "kill" markings won't have any effect. Make sense? Mathglot (talk) 01:20, 8 May 2017 (UTC)[reply]

@Mathglot: yeah it does, thanks. And indeed I seem to be the most inclusionist in the discussion so if I think it's more work than it's worth I doubt that anyone else in the discussion would disagree. Elinruby (talk) 01:31, 8 May 2017 (UTC)[reply]

Re-pinging@Tazerdadog: on Elinruby's behalf for confirmation. Due to the ping typo above, he may not have seen this, and it's really his call, not mine. Mathglot (talk) 01:35, 8 May 2017 (UTC)[reply]

Mathglot's interpretation above is basically correct. Please do not duplicate work you've already done just to add the kill flag, but please strike entities that could be ambiguous (I will manually evaluate your intention based on comments that you left, but the default is the struck/unstruck status unless you are clear in your comments otherwise). Please do use these flags from now on, or on any where your intention is unclear. Tazerdadog (talk) 02:15, 8 May 2017 (UTC)[reply]

Tazerdadog I'm looking at formation of the strikeout tags enclosing the linked titles, and found 43 anomalies that might trip up the nuke pattern. I'll probably starting fixing these tomorrow. Mathglot (talk) 09:44, 14 May 2017 (UTC)[reply]

assumption for User space items

@

That was my assumption as well, all entries outside of mainspace should be fine. Tazerdadog (talk) 01:39, 9 May 2017 (UTC)[reply]

rescuing clobbers by CXT

@

I'm concerned that there may be an unknown number of formerly good articles of long standing in the list that we don't want to delete, simply because they got clobbered by CXT at some point and thus ended up in the list, and time ran out before anybody got a chance to look at them. If I can get a list of potential clobbers in the next week, I will check them all out. (Am betting it's less than a couple hundred, total; but maybe S Marshall would help out, if it turns out to be more than that.) Shouldn't be too hard to create such a list:

# Print out names of Titles in CXT/PTR that may be clobbers of good, older articles. 
# (Doesn't handle the case where oldest version is CXT, followed by user edits to make it good,
#    followed by 2nd cxt later which clobbers the good version; but that's probably rare.)
#
For each item in WP:CXT/PTR list do:
  $line = text from next <ol> item in list
  If the bracketed article title near the beginning of $line is within s-tags, next loop
  Extract $title from the $line
  If $title is not in article space, next loop
  Read Rev History of $title into array @RevHist
  Get $oldest_es = edit summary string of oldest version (last index in @RevHist)
  If 'ContentTranslation' is a substring of $oldest_es, next loop
  Pop @RevHist: drop oldest summary from @RevHist so it now contains all versions except the oldest one
  If 'ContentTranslation' is a substring of @RevHist viewed as a single string, do:
    Print "$title possibly clobbered by CXT"
End For

Are you able to create a list like this, or do you know someone who could? Thanks, Mathglot (talk) 21:56, 18 May 2017 (UTC)[reply]

• Why not just ask the deleting administrators to check the translation is the first revision before they push the button?—S Marshall T/C 23:32, 18 May 2017 (UTC)[reply]

That would be a shitton of work for the deleting admin. Tazerdadog (talk) 00:57, 19 May 2017 (UTC)[reply]

3.6 metric shit tonnes, to be exact. ;-) And thanks for the ping, Taz. Mathglot (talk) 01:56, 19 May 2017 (UTC)[reply]

@Tazerdadog: I think I've maybe got your query: I see from Samtar's query that you use MySQL. If that's the case, then to do this, I think you can take Samtar's query 11275 exactly as it is, with one more WHERE clause, to exclude the oldest revision:

AND WHERE rev.date > @MIN_REV_DATE

where @MIN_REV_DATE is either separately selected and assigned to a variable [as there would be one min value per title, it would have to either be an array variable or more likely a 2-col temp table with title and MIN date, which could be joined to rev.] Edited by Mathglot (talk) 18:50, 19 May 2017 (UTC), or probably more efficiently, a subquery getting the oldest rev date for that page using standard "minimum value of a column" techniques. So the result will be a subset of Samtar's original query, limited to cases where ct_tag was equal to 'ContentTranslation' somewhere other than in the oldest revision for that page. (By the way, I don't have access to your file structure, so I have no idea if 'rev.date' really exists, but what I mean by that, is the TIMESTAMP of that particular revision, whatever the field is really called. Also, again depending on the file structure, you might need to use techqniques for groupwise minimum of a column to get the min rev date for each page.) Mathglot (talk) 03:09, 19 May 2017 (UTC)[reply]

@Mathglot: Unfortunately, I've never used MySQL before. I was hoping I could muddle through with some luck and googling, but I had no such luck. Tazerdadog (talk) 03:53, 19 May 2017 (UTC)[reply]

@Tazerdadog: And I could totally do it if I had the file structure but I don't; but my strong hunch is that this is very easy, and needs one additional "WHERE" plus another query (probably the groupwise MIN thing) to grab the min value to exclude in the new WHERE. OTOH, if you have access to Quarry, shoot me your query by email if you want, and I'll fix it up, and you can take that and try again, and with several back-and-forths I bet we can get it. Or if you've got zip, I can try a few establishing queries for you to try, and then we can try to build the real one depending on the results you get from those. (Or, we can just wait for someone else to do it, if they will; it really should only take minutes.) Mathglot (talk) 05:11, 19 May 2017 (UTC)[reply]

@Tazerdadog and S Marshall: I don't think this is getting enough attention, and your previous request appears to have stalled at V Pump. This is not good. We need to get this list. Is there someone you can lean on, or request help from, to kick-start this? Alternatively, if someone will give me access to Quarry, a MySQL account permitting SELECT and CREATE TEMPORARY TABLE (or even better, MEMORY table) and a pointer to the file structure descriptions, I can do this myself and create a list to protect these articles. Mathglot (talk) 06:32, 22 May 2017 (UTC)[reply]

*Bump* Mathglot (talk) 18:19, 27 May 2017 (UTC)[reply]

Thanks,

Cheers, Tazerdadog (talk) 23:52, 8 July 2017 (UTC)[reply]

Fixing ping @S Marshall: Tazerdadog (talk) 23:54, 8 July 2017 (UTC)[reply]

I would like to start removing invisible characters from pages. -- Magioladitis (talk) 14:14, 26 June 2017 (UTC)[reply]

Please provide a lot more details about this and why it is necessary. — xaosflux ^Talk 14:33, 26 June 2017 (UTC)[reply]

They can break links and citation templates, but I was sure that Yobot already did such a thing. Jo-Jo Eumerus (talk, contributions) 14:58, 26 June 2017 (UTC)[reply]

Jo-Jo Eumerus True but I was asked to request re-approval. -- Magioladitis (talk) 15:02, 26 June 2017 (UTC)[reply]

Xaosflux Per CheckWiki documentation "This could be a problem inside an article.", per AutoEd documentation "These characters are hard to remove by hand because they all "invisible", but they can cause problems and unnecessarily increase the page's size.". Inside URLs, images they can break filenames and urls, etc. -- Magioladitis (talk) 15:06, 26 June 2017 (UTC)[reply]

Oppose in general. No opposition for fixing things that are broken (the URLs, broken images, etc.), but I would likely oppose removing an invisible character in some other random space that isn't causing a problem (unless accompanied by another substantive fix). Edits removing invisible characters that aren't breaking anything puts this in

• Per a recent discussion, I would suggest we tread carefully due to a tban that is in place. Dennis Brown - 2¢ 15:32, 26 June 2017 (UTC)[reply]
  • @Dennis Brown: The topic ban specifically included an exemption to discuss whether COSMETICBOT applies to Magioladitis' own bots (so long as not excessive), so there's not much danger here. In any event, since he's seeking consensus, COSMETICBOT itself is not too relevant. It can be overridden by consensus. I'm just noting that I see no reason to make an exception here for cases where things aren't broken. ~ Rob₁₃^Talk 15:34, 26 June 2017 (UTC)[reply]
    • My note was really a note to all of us more than him, to be careful so he isn't trapped into not being able to make a request. Maybe I'm over cautious. I agree, fixing things that really aren't broken seems to be unnecessary load. Dennis Brown - 2¢ 15:37, 26 June 2017 (UTC)[reply]
  • BU Rob13 This is CHECKWIKI related in case you missed that. In fact, this is CHECKWIKI error 16. I already noted that this is CHECKWIKI related in my reply above. Thanks, Magioladitis (talk) 15:54, 26 June 2017 (UTC)[reply]
    • I am recused from handling BAG actions related to CHECKWIKI because I anticipated drama if I were to do so. This is not BAG related and I am not commenting in my capacity as a BAG member. ~ Rob₁₃^Talk 17:06, 26 June 2017 (UTC)[reply]
  • Dennis Brown I plan to fix these pages in addition to other fixes. -- Magioladitis (talk) 15:59, 26 June 2017 (UTC)[reply]
  • Dennis Brown There is no tban in place about this request. -- Magioladitis (talk) 16:00, 26 June 2017 (UTC)[reply]

• Is this a request to get approval to submit this as a BRFA, or just edit you want to make without a bot flag using your own account? — xaosflux ^Talk 16:17, 26 June 2017 (UTC)[reply]

Xaosflux My own account. I don't need to come here for my bot acccount. I can just fill out a BRFA asaik. -- Magioladitis (talk) 16:25, 26 June 2017 (UTC)[reply]

This is related to the ArbCom restriction requiring consensus to do purely cosmetic edits semi-automatically, for context. ~ Rob₁₃^Talk 17:07, 26 June 2017 (UTC)[reply]

• @Magioladitis: Thanks, do you have any estimate for how many of these there are and at what edit rate you plan to run? There seem to be recent concerns from other editors that you are flooding watchlists. — xaosflux ^Talk 18:13, 26 June 2017 (UTC)[reply]

• Xaosflux Probably 3,000-4,000 pages since there are also bots that remove some of these. Moroever, this was one by Yobot for 7 years or so. -- Magioladitis (talk) 18:18, 26 June 2017 (UTC)[reply]

  Thanks, is there a reason this wouldn't be appropriate work for a bot to (to avoid lots of watchlist hits)? — xaosflux ^Talk 18:40, 26 June 2017 (UTC)[reply]

  • Xaosflux Yes, it can be done by Yobot. Some cases need manual attention though when it comes to non-breaking spaces. I will agree this is is done by Yobot and leave less than 100 edits to be done manually. -- Magioladitis (talk) 19:13, 26 June 2017 (UTC)[reply]
    • So why do you want to make these thousands of edits by hand, triggering watchlist notifications, instead of doing this with a bot account? I understand the minority that will need special editing, is the problem that they can't be filtered out? — xaosflux ^Talk 23:04, 26 June 2017 (UTC)[reply]
      • Xaosflux I can do them by bot if I asked. -- Magioladitis (talk) 06:16, 27 June 2017 (UTC)[reply]
      • Xaosflux Wikipedia:Bots/Requests for approval/Yobot 55. -- Magioladitis (talk) 06:25, 27 June 2017 (UTC)[reply]
• Support - I see no harm in Magioladitis performing the requested edits. Beyond My Ken (talk) 17:12, 26 June 2017 (UTC)[reply]

• Given the latest, I've lost all faith that Magioladitis gets it. Beyond My Ken (talk) 19:59, 1 July 2017 (UTC)[reply]

• Beyond My Ken Per ArbCom I have to ask permission to do changes that do not change the visual outcome, I have no restictions to make any other changes. Connecting the two kinds of edits it's interesting.-- Magioladitis (talk) 20:10, 1 July 2017 (UTC)[reply]

• I'm not connecting the "two kinds of edits", I'm connecting the one kind of behavior on your part in both discussions. To answer the question you originally posted here before you changed it, "Why did [I] expect?" - I expected an editor of your experience and length of time here, an administrator no less, to understand what they're being told, and not to Wikilawyer every goddamned thing that comes down the pike. That complaints keep coming, and that you attempt to talk your way out of them is one of your big problems, and the reason I changed my !vote here. (And, BTW, please don't send me any more sarcastic "Thanks for your edit" notifications, they are not wanted or appreciated, and, again, are an instance of un-adminlike behavior on your part.) Beyond My Ken (talk) 20:16, 1 July 2017 (UTC)[reply]

• Beyond My Ken I appreciate any comment that will help us understand the problem and solve it. So, would you still be OK with a bot task Wikipedia:Bots/Requests for approval/Yobot 55? Is the problem the edit from my main account? Is the problem that the edit does not change the visual outcome? Would you be OK if another editor or bot peform these edits? If for example I ask someone else to do these edits, that would be OK? Thanks again and happy editing!-- Magioladitis (talk)

• That response is typical. You attempt to gaslight your way out of a discussion which is not about the technicalities of your request, but about your behavior, and your apparent inability to understand that your tether is getting thinner and thinner. You are damn near exhausting the patience of the general community, and from the recent discussions, you seem to have already run out of it in the community of your bot-running peers, the people who, one would think, would be behind you. These are the reasons I'm withdrawing my !vote (without, however, changing to "oppose", at least at this time). Do me a favor, please, and don't ping me again. Beyond My Ken (talk) 22:33, 1 July 2017 (UTC)[reply]

• Who cares about me? I want the task to be done. I don't care if this is going to be another use or a bot or software that will disallow this characters to be entered (that was my proposal to MW programmers). So support the task and find someone to do it. -- Magioladitis (talk) 07:26, 2 July 2017 (UTC)[reply]

• Tentative support these are often changes that won't make very much differences, and will often be de-facto
  b} 17:51, 26 June 2017 (UTC)[reply
  ]
  • The list Regex: \u200E|\uFEFF|\u200B|\u2028|\u202A|\u202C|\u202D|\u202E|\u00AD can be removed. -- Magioladitis (talk) 18:18, 26 June 2017 (UTC)[reply]
    • Would those leave explicitely declared characters untouched, like in
      b} 20:00, 26 June 2017 (UTC)[reply
      ]
      • Headbomb Yes. Anything given explicitelly (by visible text or by templates) won't be touched. -- Magioladitis (talk) 20:45, 26 June 2017 (UTC)[reply]
• Support. For one thing, they're not technically cosmetic, because when you're viewing the code, the end result is the same as what you started with. Moreover, we routinely have bots going around and doing this already at Commons (one of the more common ones is removing
  RTL markers from category texts), because as noted above, they can cause problems when editors don't realise that they're there. For example, run a search for "soft hyphen" at Wikipedia:Village pump (technical)/Archive 60. Or see the "Weird pipes display issue" section at Wikipedia:Village pump (technical)/Archive 102, which discusses the invisible Zero-width non-joiner character. And finally, the mere fact of Commons bots doing this long-term is a demonstration that a bot can do this; it's not a CONTEXTBOT situation. Nyttend (talk) 22:25, 26 June 2017 (UTC)[reply
  ]
  • @
    WP:COSMETICBOT. A cosmetics-only change is defined as one which does not affect the visually-rendered result of the page (and also doesn't affect accessibility issues). This is definitely a cosemtics-only change as defined by the policy. If the community wants to make an exception in this case, it can of course do that. ~ Rob₁₃^Talk 00:48, 27 June 2017 (UTC)[reply
    ]
    • This causes accessibility issues for everyone, as I explained in the rest of my sentences. Nyttend (talk) 05:06, 27 June 2017 (UTC)[reply]

• Without more detail on what is actually being proposed to be removed, and what the specific benefit is, I couldn't support. Hchc2009 (talk) 06:32, 27 June 2017 (UTC)[reply]

@

b} 09:33, 27 June 2017 (UTC)[reply

]

Is the argument here that invisible characters are already banned from the Wiki, and that this is simply a question of how they're removed, or that we're proposing to ban them? Or that the proposal is to remove some invisible character? Hchc2009 (talk) 16:41, 28 June 2017 (UTC)[reply]

Hchc2009 If possible we should replace all by the visible counter-parts. -- Magioladitis (talk) 19:09, 28 June 2017 (UTC)[reply]

• Oppose If that many need to be done it can be done via a bot and BRFA. Otherwise any support will be used by mag to condone running a bot on his main account to make what may/or may not be purely cosmetic edits. Notice the caveats from the above editors 'can cause' not 'will cause'. Without a clearly defined list and the problems each edit causes, given his history, I suspect the reason this is here and not at BRFA is that a run removing thousands of invisible characters that may/may not cause problems would not pass. Send it to BRFA and let them decide. Its what its for. Only in death does duty end (talk) 10:33, 27 June 2017 (UTC)[reply]

@

b} 11:00, 27 June 2017 (UTC)[reply

]

A BRFA (Wikipedia:Bots/Requests for approval/Yobot 55) has been opened for this now. As mentioned above there may be edge cases that need manual editing, but if most edits can be done via a structured bot job that alleviates my excessive watchlist hits concern above. — xaosflux ^Talk 11:08, 27 June 2017 (UTC)[reply]

Let me be blunter. BRFA is a venue specifically to vet large-scale edits that would be best suited to a BOT/Automated process. AN is a noticeboard that attracts a wide array of editors & admins who may not have knowledge of Mag's history, the contentiousness of edits like this, the 'fixes' mag is on a crusade to implement by any means possible. Any 'support' here is essentially (given Mag's unique interpretation of what is/is not allowed when making gen fixes) giving Mag carte blanche to make thousands of semi-automated (given the speed of his editing history, I heavily doubt the 'semi' there) edits of dubious usefulness from his main account. I would rather not open up the floodgates to someone who has multiple restrictions related to automated editing. So no, I oppose any attempt here to give him 'permission' to do something that would best be evaluated at BRFA. If he wants to run a bot, he can run a bot. If the condition/requirement of passing BRFA for the task is that he gain consensus here, then no, he has shown he has zero judgement in when to apply controversial edits like this, so it would still be oppose. Only in death does duty end (talk) 11:10, 27 June 2017 (UTC)[reply]

@Only in death: BRFA is never where we assess consensus for a task. That is always kicked to another venue. In any event, this is getting a bit confused. Magioladitis, could you clarify whether you're seeking consensus for semi-automated or automated edits here? Those are rather different, and may elicit different levels of support. I'd be more likely to support a flagged bot doing this than semi-automated, although I don't know if I'd support either. ~ Rob₁₃^Talk 14:10, 27 June 2017 (UTC)[reply]

Sigh, if you want to be picky about the wording. BRFA is where you decide if a bot-task can go ahead. Which is the appropriate venue for this. If the decision to approve the task is reliant on the requester showing consensus exists to make the changes in the bot-task, and this is considered a valid place to gain that consensus, the answer would still be no from me as the above request is too vaguely worded and boils down to 'I want to remove invisible characters that may or may not affect the articles in some manner' and has not provided sufficient detail to show that a)they are needed, b)the articles in his list have been sufficiently identified to contain invisible characters that cause an actual problem. Only in death does duty end (talk) 14:22, 27 June 2017 (UTC)[reply]

@

WP:ARCA, but the community has no direct control over venue on this. ~ Rob₁₃^Talk 14:47, 27 June 2017 (UTC)[reply

]

Well in that case, if the scope of this request is to make semi-automated edits of no demonstrated need on his main account that affect thousands of articles... the answer is still no. Only in death does duty end (talk) 15:29, 27 June 2017 (UTC)[reply]

(edit conflict) @BU Rob13: don't preclude that these could be done semi-automated, using a bot account - there is nothing wrong with that model in general and avoids flooding recent changes/watchlist. — xaosflux ^Talk 14:23, 27 June 2017 (UTC)[reply]

If the scope of a task was tailored to only fix invisible characters causing issues and was done semi-automated on a bot account, I would definitely support that. ~ Rob₁₃^Talk 14:47, 27 June 2017 (UTC)[reply]

Xaosflux I would like to do this in automated way but also use my account for some cases because I would like to check some edits to avoid any mistakes. For instance, AWB can't remove an indivisible character if this is the only edit done. I don't wish to reply to RU Rob13 because he already said he'll stay away from CHECKWIKI related tasks. -- Magioladitis (talk) 16:39, 27 June 2017 (UTC)[reply]

• Just for clarity (and I will not reply further on this topic), I recused from handling CHECKWIKI bot tasks as a BAG member while explicitly stating I may comment as a normal editor. I've never stated I would "stay away from CHECKWIKI related tasks". ~ Rob₁₃^Talk 23:28, 28 June 2017 (UTC)[reply]

• Support Above someone suggested that invisible characters are only important when they break url, links and the like. It's worth noting that they also break searching, to varying degrees. All the best: Rich Farmbrough, 18:09, 30 June 2017 (UTC).[reply]
  
• Oppose 3,000 – 4,000 pages? No, if this task is to be done, it should be done through a bot-flagged account. ​—DoRD (talk)​ 02:39, 2 July 2017 (UTC)[reply]
• Oppose per DoRD. Beyond My Ken (talk) 09:27, 4 July 2017 (UTC)[reply]

• Beyond My Ken Can you please confrim that the oppose is only on the "from a normal account" part? -- Magioladitis (talk) 16:14, 9 July 2017 (UTC)[reply]

• Conditional Support I think removing these invisible characters would be a positive thing, and nobody has shown that there's any potential to harm any pages. I do have concerns about using Mag's personal account, due to the Watchlist spam potential mentioned above. However, I think Yobot doing this task through BRFA at a reasonable rate (with Mag manually handling the few the need individual attention) is perfectly fine. I believe overriding COSMETICBOT is justified here. The Wordsmith^{Talk to me} 16:21, 5 July 2017 (UTC)[reply]
• Support task, oppose implementation: I take the view that hidden characters can cause problems in urls, editing and other accessibility problems. Therefore, fixing these is a task that is a) should be done and b) suitable for a bot. However, this proposal was suggesting that this would be done on Magio's user account. Editing 3000+ pages on a user account in an automated way will clog watchlists and is far more suited for a bot task. Done sensibly on a bot account and I have no opposition to the task. TheMagikCow (T) (C) 10:18, 6 July 2017 (UTC)[reply]

• TheMagikCow, User:The Wordsmith I can do this task from Yobot's account. -- Magioladitis (talk) 16:15, 9 July 2017 (UTC)[reply]

Magioladitis. We know this can be done from Yobot's account. So, are these (proposed) edits going to be done from Yobot's account (and only and exclusively so)? —Sladen (talk) 17:35, 9 July 2017 (UTC)[reply]

Sladen If the consensus is "bot account only" I am doing from there. If the consensus "you may also use your normal account", I'll use both ;) -- Magioladitis (talk) 17:43, 9 July 2017 (UTC)[reply]

• Support the task. @Magioladitis:, as far as I know there is no restriction/policy that says that a bot account can't be used for some occasional 'manual' edits to supplement a bot-task (as long as they fall (strictly) under the same approved BRFA and that there is consensus for the edits). Heck, for me, if you have the necessary approvals to perform a task, and you want to use the bot account to do 4000 repetitive edits completely manual, why not? --Dirk Beetstra ^{T C} 17:36, 9 July 2017 (UTC)[reply]

Running at 1 edit per second for an hour is not an acceptable edit rate for non-bot flagged accounts making this type of edit. — xaosflux ^Talk 15:11, 29 June 2017 (UTC)[reply]

Skr15081997 seems to understand articles belong in child categories, not parent categories but their actions run counter to that. This needs to stop. Chris Troutman (talk) 15:28, 29 June 2017 (UTC)[reply]

Support reversion - unnecessary over-categorisation. PamD 16:54, 29 June 2017 (UTC)[reply]