Wiktionary:Grease pit/2018/April

zhx-teo and zhx-tai need to be added to MediaWiki:Gadget-TranslationAdder-Data.js so that translations will be nested. It seems like it should go under lzh: c, in the form of "zhx-teo": c, and "zhx-tai": c,. —suzukaze (t・c) 07:59, 4 April 2018 (UTC)

 Done —Suzukaze-c◆◆ 03:06, 13 April 2018 (UTC)

@Rua, Daniel Carrero, Erutuon Sometime in the last year or so, someone (probably one of the three people just pinged, since they're the ones who've been hacking on this code the most) changed the behavior of {{affix|...|lang1=LANG}} to add the text "Borrowing from ..." at the beginning. This didn't use to be the case, and I suspect it was an accident; in any case it has caused some seriously ugly results in probably hundreds of pages that I created from before that time. See дистальный for one of many examples, where the text now reads

Perhaps from Borrowing from German distal + -ный (-nyj).

where it used to read

Perhaps from German distal + -ный (-nyj).

I sort of noticed this yesterday and tried to fix things by adding the notext= parameter to {{affix}}, {{confix}}, etc., but this isn't the right solution, as it would require adding |notext=1 to hundreds of pages. I think we should simply revert the changed behavior and not include the "Borrowing from" text, because:

1. This is compatible with the way that {{bor}} now works.
2. Hundreds of pages are broken with the text present.
3. The current code is completely broken if you try to use lang2=, lang3=, etc.: {{affix|ru|distal|-ный|-ation|lang1=de|lang3=en}} → German distal +‎ -ный (-nyj) +‎ English -ation.
4. If there are pages that depend on the new behavior, they won't be horribly broken, because the "Borrowing from" text will necessarily occur at the beginning of a sentence, and the line will still be coherent without it.

If no one complains, I'll go ahead and fix this tomorrow. (It might be possible to clean this up with a bot, by (a) making the proposed change, and (b) having a bot go through and add the text "Borrowed from " to all occurrences of *fix that have lang1= in them, occur at the beginning of a line, and have a period at the end of the line.) Benwing2 (talk) 22:40, 8 April 2018 (UTC)

@Rua, Daniel Carrero, Erutuon Pinging again ... sorry, I don't know if multi-paragraph pings work properly ... :( Benwing2 (talk) 22:41, 8 April 2018 (UTC)

There was a conversation about this a while back, which I can't find at the moment, but I seem to recall that Rua also wanted this to be fixed in the way you describe, but left the project before it could be implemented. —Mahāgaja (formerly Angr) · talk 12:56, 9 April 2018 (UTC)

@Mahagaja, Benwing2: Wiktionary:Grease pit/2018/January § Affix with lang1 --Per utramque cavernam (talk) 13:16, 10 April 2018 (UTC)

@Mahagaja, Per utramque cavernam It looks like there was some module editing with unintended consequences. The Jan 2018 discussion just mentioned referred to a May 30 2016 edit by Rua that supposedly added the "learned borrowing" text, but at the time, the flag in question to format_borrowed() had the meaning of "notext", and the change by Rua was correct. It looks like someone (not sure who yet) later changed the calling sequence of format_borrowed(), removing (I think) the nocap param, without properly fixing up the callers, causing the call to format_borrowed() to accidentally instead specify the "learned borrowing" flag and not specify the "notext" flag. Someone else tried to fix this by disabling the "learned borrowing" flag, which left the normal borrowing text in its place. I will fix this up properly.

BTW I was gone for about 6-7 months, and in the meantime, Rua seems to have (a) adopted a new name, and (b) left the project. Anyone know what happened to her? Benwing2 (talk) 17:49, 15 April 2018 (UTC)

@Benwing2: This was the last discussion Rua participated in. —Aryaman^{A (मुझसे बात करें • योगदान)} 17:52, 15 April 2018 (UTC)

@Aryamanorora: Thanks. Benwing2 (talk) 17:56, 15 April 2018 (UTC)

@AryamanA: Oops. Benwing2 (talk) 17:57, 15 April 2018 (UTC)

@Benwing2: Yes, that and Wiktionary:Votes/sy-2017-11/Desysopping CodeCat aka Rua (which is mentioned in the discussion). --Per utramque cavernam (talk) 18:15, 15 April 2018 (UTC)

@per utramque cavernam: Thanks. IMO this is too bad. Rua definitely caused friction by her behavior but did a lot of good work on the project. Benwing2 (talk) 18:19, 15 April 2018 (UTC)

BTW I made the change to remove the "Borrowing from" text from {{affix|...|langN=...}}, and reverted the previous additions I made of a |notext= flag. Benwing2 (talk) 18:20, 15 April 2018 (UTC)

@Benwing2: Yes, I'm saddened by her departure too. And thanks for fixing {{affix}} (by the way, I'm not ignoring your Russian pings, but I'm not knowledgeable enough to contribute). --Per utramque cavernam (talk) 20:14, 15 April 2018 (UTC)

@Per Utramque cavernam: No need to respond to the pings, I'm just including you in case you are interested; if you'd rather I not include you, just let me know. Benwing2 (talk) 20:25, 15 April 2018 (UTC)

@Per utramque cavernam: Damn I need to be more careful spelling usernames. Benwing2 (talk) 20:26, 15 April 2018 (UTC)

@Benwing2: Copy+paste? --WikiTiki89 14:40, 16 April 2018 (UTC)

@WikiTiki89: It doesn't always work. --Per utramque cavernam (talk) 17:01, 16 April 2018 (UTC)

@Per utramque cavernam: You have to copy and paste directly from the link in the wikitext of the user's signature. Work's 100% of the time. --WikiTiki89 18:03, 16 April 2018 (UTC)

Surely preferences used to have a switch to enable hidden categories to be listed at the bottom of a page. Has it gone or is my memory going. — Saltmarsh. 14:15, 9 April 2018 (UTC)

It must still be there, because I still have this feature enabled. --WikiTiki89 14:18, 9 April 2018 (UTC)

It's in the appearance tab under advanced options. DTLHS (talk) 15:28, 9 April 2018 (UTC)

Thank you so much I thought I had searched everywhere - but probably spent most time looking through the long lists. I'm now thoughly humbled :) — Saltmarsh. 15:33, 9 April 2018 (UTC)

New Special:AbuseFilter/79 does not work. Should catch e.g. this deleted entry اختباراختبار (my test), and in general anything that contains "==English==" but has an Arabic title (a common form of disruptive edit recently). Can anyone help me debug? Perhaps using Arabic right-to-left chars in a regex is not properly supported? Equinox ◑ 23:08, 9 April 2018 (UTC)

@Equinox: Looking at the documentation, the filter is matching ==English== against the page title (article_text) and the Arabic letters against the added_lines. It should be the other way around. (I wonder if like tries to match the whole string or just find a matching substring.) — Eru·tuon 01:32, 10 April 2018 (UTC)

I also wonder if like doesn't recognize multi-byte characters (the Arabic) in a character set (i.e., treats them as '', not as codepoints U+63A, U+638, ...). If so, the continuation bytes could match non-Arabic characters. Might be safer to use rlike. And you could expand the regex to the whole Arabic block (''), or copy the pattern from Module:scripts/data. — Eru·tuon 02:12, 10 April 2018 (UTC)

It works now, thanks to Eru magic. Thanks! I have also set it to block the edits because we get a lot of these and they are worthless. Equinox ◑ 02:59, 13 April 2018 (UTC)

If you are expected a classic drunk Equinox story, unfortunately there isn't one. I can't join the wikt IRC even after resetting the router, which suggests somebody has applied a huge sweeping IP ban that covers the entirety of British Telecom or something (I was never actually banned; I assume this is someone else's ban). Fixy fixy? Equinox ◑ 02:58, 13 April 2018 (UTC)

I had the same problem, apparently my IP was banned on the Japanese wiktionary (?), it went away at some point. Crom daba (talk) 04:37, 13 April 2018 (UTC)

Hmmm, if only we had an equivalent to IRC, but something where your IP couldn't be banned by mistake, and with lots more features... —Μετάknowledge^{discuss/deeds} 05:25, 13 April 2018 (UTC)

As I stated before, proprietary systems are tacky and often shut down without warning. Enjoy your sandbox until the adults take it away. Then we'll welcome you on IRC -- along with the past victims of MSN Messenger, AOL AIM, Google Buzz, Google Wave, Orkut, Flock browser -- I could go on. Equinox ◑ 05:48, 13 April 2018 (UTC)

I'm confused, is this now the classic drunk Equinox story? It seems to have all the nonsensical makings of one. --Victar (talk) 14:47, 13 April 2018 (UTC)

Does the web interface work? And does it prevent you from joining the channel or any freenode server? - TheDaveRoss 12:56, 13 April 2018 (UTC)

(For what it is worth, none of the channel bans seem like they would impact you, so it must be a network ban.) - TheDaveRoss 13:03, 13 April 2018 (UTC)

Two points:

• This is likely bleed-over from a long-term abuser on en.WP who often uses BT. I am working on finding and alleviating.
• I will happily host an instance of Rocket.Chat, which is a flossy pseudo-clone of Slack, but Equinox is correct - such multi-channel communications systems have not proven to last, unlike IRC.

- Amgine/ ^t·e 15:09, 13 April 2018 (UTC)

Advice from irc://chat.freenode.net/wikimedia-ops is to try again. When they have done BT range blocks they are for as short a time as possible, usually a few minutes. - Amgine/ ^t·e 15:14, 13 April 2018 (UTC)

Is there a tool or way of finding the intersection (and moreover, the disjunct group) of two categories? For example, category:English adjectives and category:English words suffixed with -ly (which in fact would go to category:English words suffixed with -ly (adjectival)!). Sobreira ►〓 (parlez) 09:23, 13 April 2018 (UTC)

See https://petscan.wmflabs.org/ explained at m:PetScan. It returns 526 pages in Category:English adjectives and Category:English words suffixed with -ly, not in Category:English words suffixed with -ly (adjectival). --Vriullop (talk) 10:01, 13 April 2018 (UTC)

Ordinary Cirrus search supports category intersections. DCDuring (talk) 12:00, 13 April 2018 (UTC)

Cirrus search is at least 0.6% better than petscan because 'incategory:"English words suffixed with -ly" incategory:"English adjectives"' generated 529 hits. DCDuring (talk) 12:05, 13 April 2018 (UTC)

Adding in search '-incategory:"English words suffixed with -ly (adjectival)"' it returns 526 pages. There are 3 pages in both "-ly" and "-ly (adjectival)" categories. --Vriullop (talk) 12:35, 13 April 2018 (UTC)

Could someone who's able to edit Module:languages/data2 please add French as an ancestor of Haitian Creole? It's a French-based creole, but right now its ancestors are listed as unknown. —Globins 22:13, 13 April 2018 (UTC)

We have thus far chosen to consider the main lexifiers of a creole as distinct from its ancestors, where any exist. Were French to evolve naturally, it would not produce a creole, so Haitian Creole is not a descendant language despite most of its lexicon being descended from French lexical items. —Μετάknowledge^{discuss/deeds} 23:37, 13 April 2018 (UTC)

If this is the policy for all creoles, then I don't think it makes sense for English to be listed as an ancestor of Tok Pisin. —Globins 00:04, 14 April 2018 (UTC)

I agree. It's not policy, though, just our (imperfectly applied) standard so far. More input would definitely be good; I seem to remember @Mahagaja having discussed this in the past. —Μετάknowledge^{discuss/deeds} 03:19, 14 April 2018 (UTC)

Not in any informed way, I haven't. I think at some point I said that {{der}} was the only appropriate template for creoles and pidgins because their words are neither really inherited nor really borrowed from their lexifier languages, but other people disagreed with me. —Mahāgaja (formerly Angr) · talk 07:08, 14 April 2018 (UTC)

Whatever we do, I think we should standardize it. Does anyone else have an opinion on this? —Globins 13:47, 14 April 2018 (UTC)

I think we should call English a creole of French as well. I jest, but I agree, I think Haitian Creole should really be descended from French. The alternative is to say all the French vocabulary is borrowed, which seems the worst of the two options. --Victar (talk) 17:44, 14 April 2018 (UTC)

Or we could do what Mahagaja supports and what we've been doing up to this point, namely use {{der}}. —Μετάknowledge^{discuss/deeds} 18:02, 14 April 2018 (UTC)

(edit conflict) No, the alternative is to not say whether it's borrowed or inherited, just where it comes from. There are no decisive arguments one way or the other, and even if there were, it's likely they wouldn't apply uniformly in all instances. I don't believe that we really understand what happens when a pidgin is created and when it becomes a creole- there are elements of inheritance and borrowing and influence that are really hard to separate. It's best not commit to any interpretation, at least for now. Chuck Entz (talk) 18:10, 14 April 2018 (UTC)

@Metaknowledge, Chuck Entz, I don't think making French the ancestor and the use of {{der}} or {{inh}} to be mutually exclusive. --Victar (talk) 20:41, 15 April 2018 (UTC)

@Victar As long as there's a good chance that whatever we decide could be wrong, I'd rather not decide. Chuck Entz (talk) 20:54, 15 April 2018 (UTC)

@Erutuon What do you think about something along the lines of a |nopos= parameter so people don't have to learn how to use {{topic cat}}. Chuck Entz (talk) 18:25, 14 April 2018 (UTC)

@Chuck Entz: I think it's a good idea, though maybe |topic= would be a clearer parameter name. I've added the parameter, and it seems to work. — Eru·tuon 19:30, 2 July 2018 (UTC)

What does the tag "-th" mean? There's no explanation at Special:Tags. Thanks. – Gormflaith (talk) 00:43, 15 April 2018 (UTC)

It's referring to the IP range characteristic of a certain Thai editor who often adds incorrect information. @Chuck Entz created it, I think. —Μετάknowledge^{discuss/deeds} 00:44, 15 April 2018 (UTC)

Yeah, I noticed he was really the only one doing anything about it. – Gormflaith (talk) 00:53, 15 April 2018 (UTC)

Special:AbuseFilter/75. Equinox ◑ 00:45, 15 April 2018 (UTC)

@Equinox: you forget that some of us are the mere "public" :-) ("You may not view details of this filter, because it is hidden from public view.") – Gormflaith (talk) 00:53, 15 April 2018 (UTC)

I think people do that to stop the bad editor from studying the filter and coming up with workarounds. BTW, it would be nice for Special:Tags to indicate which numbered filters apply a given tag. Equinox ◑ 01:10, 15 April 2018 (UTC)

Yep. The main problem isn't bad information (though they do that, too), but adding labels and templates where they don't belong. They seem to be so mesmerized by our template technology that they add them everywhere, even in languages they know nothing about, even where the templates produce garbage unless you know the right parameters (which this editor certainly doesn't). I've seen them add a dozen or more conjugation templates to Spanish entries that already had working ones.

There's no way for a non-Thai to tell them apart from competent IPs when they're working with Thai, but Thai IPs working in other languages are few and far between, so I have the abuse filters tag the IP range editing entries with no Thai content. In my experience their edits are maybe 50% unnecessary clutter and 30-40% varying degrees of wrong- so definitely worth checking- but they seem quite normal if you're just skimming Recent Changes. Chuck Entz (talk) 05:08, 15 April 2018 (UTC)

I have a list of about 1,700 dinosaur names from Wikipedia at User:BD2412/dinosaurs. Is there some technical shortcut we can use to extract information from Wikipedia to create entries for these? bd2412 T 18:03, 16 April 2018 (UTC)

I think that would be undesirable. Plus, the most interesting part is often the etymology, which Wikipedia has usually messed up to some degree. If you ping me in about three weeks, I can spend some hours creating a few hundred or so. —Μετάknowledge^{discuss/deeds} 05:10, 20 April 2018 (UTC)

I have put a reminder in my phone. I tried to say "Metaknowledge" and it came out "meadow knowledge", but close enough. bd2412 T 18:13, 22 April 2018 (UTC)

• @Metaknowledge Three weeks later... bd2412 T 01:49, 14 May 2018 (UTC)

  Thank you for the timely reminder. I have just lost a family member, so I don't have quite the amount of free time I was anticipating. I'll see what I can do. —Μετάknowledge^{discuss/deeds} 03:42, 14 May 2018 (UTC)

  • I'm so sorry to hear that! No worries at all - this can wait. bd2412 T 17:51, 14 May 2018 (UTC)

Does somebody know why this doesn't work? --Per utramque cavernam (talk) 10:31, 19 April 2018 (UTC)

The problem is that the label "terms derived from English" belongs to a different submodule (Module:category tree/derivcatboiler) and Module:category tree/poscatboiler assumes that you are specifying a label that belongs to it. I guess you have to give the name of the submodule and the language code for English instead, but I can't get it to work. — Eru·tuon 16:58, 19 April 2018 (UTC)

@Erutuon: At first I thought the module was trying to prevent me from creating things like CAT:English pseudo-anglicisms... But autocat works, so it's not even that. --Per utramque cavernam (talk) 13:23, 26 April 2018 (UTC)

This is probably a stupid question, but why do edit summaries for reverts look like this? Is there a way to fix it besides manually deleting the tags? And is it just my browser? – Gormflaith (talk) 17:41, 19 April 2018 (UTC)

Not a stupid question. I've been meaning to ask that myself. It didn't used to do this. I can see three possibilities:

1. The gadget has been changed to use a bad tag that it didn't use before
2. The system has changed so that a formerly acceptable (or at least functional) tag is no longer allowed.
3. Either the system or the gadget has changed so that the output of the gadget is no longer being treated as html.

The last option is suggested to me by the fact that I see unconverted markup in the Recent Changes splash screen that I get when I revert or delete. I suppose it might be due to some kind of interaction with another gadget or appearance preference, but I thought I went through and unclicked everything that might do that, with no effect. Chuck Entz (talk) 04:30, 20 April 2018 (UTC)

Reconstruction: namespace has been around for more than 2 years now, and I'm still finding cases where redirects from old Appendix namespace to new Reconstruction namespace are broken. Can anyone create a category or list of instances where (1) a page named Appendix:Proto-$0/$1 redirects to Reconstruction:Proto-$0/$2, and (2) Reconstruction:Proto-$0/$1 is a redlink? In other words, where the bit after the slash in the Appendix: page's name is not the same as the bit after the slash in the name of the Reconstruction: page that the Appendix: page redirects to. For example, Appendix:Proto-Brythonic/neμ redirects to Reconstruction:Proto-Brythonic/neβ̃ and neμ is not the same string as neβ̃. However, this specific example wouldn't need to be listed because Reconstruction:Proto-Brythonic/neμ is a blue link. I'm only interested in cases where changing "Appendix:" to "Reconstruction:" yields a red link. Thanks! —Mahāgaja (formerly Angr) · talk 21:36, 23 April 2018 (UTC)

@Mahagaja I didn't find any pages matching that criteria. DTLHS (talk) 05:08, 25 April 2018 (UTC)

There are a few cases at Special:BrokenRedirects where a redirect from the Appendix space to the Reconstruction space is broken, but the ones I spot-checked are because the Reconstruction page was deleted and no-one thought to fix the Appendix redirect or the mainspace entries linking to it. - -sche (discuss) 05:34, 25 April 2018 (UTC)

I guess I should clarify that I don't mean literal broken redirects in the sense of a redirect to a redlink. I mean an Appendix page for one spelling that redirects to a Reconstruction page for a different spelling, while the Reconstruction page for the same spelling is a redlink. For example, until I fixed it two days ago, Appendix:Proto-Malayo-Polynesian/daRaq was a redirect to Reconstruction:Proto-Malayo-Polynesian/daʀaq, but Reconstruction:Proto-Malayo-Polynesian/daRaq was a redlink. I wish I could believe that that example was the last one that needed fixing, but I strongly doubt it. —Mahāgaja (formerly Angr) · talk 07:03, 25 April 2018 (UTC)

@Mahagaja User:DTLHS/proto redirects Is this what you wanted? DTLHS (talk) 01:50, 26 April 2018 (UTC)

@DTLHS: Yes, that's exactly what I wanted. Thanks! —Mahāgaja (formerly Angr) · talk 06:31, 26 April 2018 (UTC)

See Category:Tundra Nenets lemmas.

1. Many of them have ң, which is not a letter in Nenets, in place of the correct ӈ. ңылна and ӈылна (ŋilna) are doubling each other.
2. Many have diacritical marks over vowels like macrons and breves in their title. Those marks are not used in writing, only in dictionaries to disambiguate pronunciation. Thus, they must be treated like Latin macrons, Serbo-Croatian accent marks and the like: Template:l should strip them.
3. They use all kinds of Unicode apostrophes.

This needs to be fixed. Guldrelokk (talk) 05:59, 25 April 2018 (UTC)

@Guldrelokk: Is the information at w:Tundra Nenets language#Orthography accurate? I deduce from that that the only "real" letters with diacritics are ⟨ё⟩ and ⟨й⟩, and moreover that ⟨ʼ⟩ (U+02BC MODIFIER LETTER APOSTROPHE) and ⟨ˮ⟩ (U+02EE MODIFIER LETTER DOUBLE APOSTROPHE) are real letters too, but that ⟨ʺ⟩ (U+02BA MODIFIER LETTER DOUBLE PRIME), which is also used in some of our entries, is wrong and should be replaced with the double apostrophe. I assume ңyʼʼ, with two single apostrophes, should have a double apostrophe as well. Otherwise all breves and macrons (except the one over ⟨й⟩) as well as the grave accent on ңэ̀со should be stripped, and all ⟨ң⟩ moved to ⟨ӈ⟩. Is that right? —Mahāgaja (formerly Angr) · talk 17:51, 25 April 2018 (UTC)

@Mahagaja: As far as I can tell, everything is right, and мэ̇сь (mėsʹ°) should be moved to мэсь (mesʹ°) too. And U+02BC and U+02EE do seem the best option for the apostrophes, so entries that use either U+00B4 (нум´ (num´), сял´ (syal´)), U+2019 (ня’ (nya’), та’ (ta’), тэ’ (te’), ху’’ла (xu’’la)) or U+02BA (ваʺӑв (waʺăw°), иʺ (yiʺ)) should be moved. Guldrelokk (talk) 20:42, 25 April 2018 (UTC)

@Guldrelokk: I think I'm done. I've enabled diacritic stripping so that links that contain diacritics will point to pages without them, and I've moved a bunch (hopefully all) of the pages at CAT:Tundra Nenets lemmas to the correct forms, using only U+02BC for the single apostrophe and U+02EE for the double apostrophe. I've also fixed all the incoming links I could find. We don't seem to have any Tundra Nenets non-lemma forms. —Mahāgaja (formerly Angr) · talk 23:27, 25 April 2018 (UTC)

@Mahagaja: Great, thank you. Apparently there are no more issues. Guldrelokk (talk) 23:38, 25 April 2018 (UTC)

The case of the velar nasal is not quite as straightforward as the apostrophes or diacritics (thanks for these). ⟨ң⟩ is definitely used: e.g. in Tereshchenko's Nenets-Russian dictionary from 1965, or in Salminen's reference description from Routledge's 1998 handbook on Uralic languages. To be sure I believe hooked ⟨ӈ⟩ is the proper prescribed glyph, and the descender version has only been used when the hook version is not available (the former is generally preferred in the more widely spoken Turkic languages of Russia, while the latter is generally preferred in the small indigenous languages), but this is usage regardless. We should have something at WT:AYRK about this at minimum. Maybe even soft redirects from spellings with ⟨ң⟩ when attested? --Tropylium (talk) 01:46, 26 April 2018 (UTC)

@Tropylium: I do think that the use of ң, especially in later Western publications, was owing to printing issues. As early as in 1961 «Самоучитель ненецкого языка» by А. В. Алмазова used ӈ; I cannot consult Tereshchenko’s 1965 dictionary, her 1955 dictionary still had нг and its 2005 re-issue has ӈ. What is important is that today school materials use ӈ, that is how alphabet is taught to the people; scholarly publications are hardly relevant compared to that. And printing issues do not exist for an online dictionary, which can easily advocate proper usage: for example, Romanian entries use ș and not ş. It may be noted at WT:AYRK that ң sometimes appears in printed materials, but it would be much better to do without redirects, as they would break spelling consistency if ever used in entries. Guldrelokk (talk) 03:59, 26 April 2018 (UTC)

I also consulted the Routledge handbook. The letter is printed there by putting a comma under н. This isn’t even a separate glyph and can represent ң and ӈ equally well. Guldrelokk (talk) 04:21, 26 April 2018 (UTC)

It has other issues too, for example, л is printed as ʌ, which starkly contrasts with the otherwise used font and is not how it’s normally printed, and instead of Э̇ it has З̇, which is utterly wrong. They clearly didn’t have enough Cyrillic glyphs. Guldrelokk (talk) 04:28, 26 April 2018 (UTC)

If several different sources use ⟨ң⟩ instead of ⟨ӈ⟩, then I'd say we probably should have entries for them (at least the ones that are actually attested in those sources) and call them {{nonstandard spelling of}}, because there's a chance that someone will see those spellings in those sources and then look for them here, and possibly even create entries using ⟨ң⟩ if they find we don't have them. —Mahāgaja (formerly Angr) · talk 06:40, 26 April 2018 (UTC)

I don't have any problem with generally sticking to ⟨ӈ⟩ for now, but I will keep an eye out for what may occur in actual attestations. --Tropylium (talk) 14:31, 27 April 2018 (UTC)

@Mahagaja, can you also make links ignore the same diacritics for Forest Nenets (syd-fne)? Guldrelokk (talk) 01:16, 26 April 2018 (UTC)

@Guldrelokk: Done. We don't have any Forest Nenets lemmas to move. —Mahāgaja (formerly Angr) · talk 06:40, 26 April 2018 (UTC)

Any plainlinked English words used in the taxonomic definitions within {{taxon}} (which appears on more than 14,100 mainspace pages) now show "orange" instead of blue if the word has an entry. Can the prior behavior within {{taxon}} be restored? Perhaps {{taxon}} does not conform to current standards or perhaps there are no specific standards that bear on this problem. I hope that the module changes that presumably caused this problem don't have any other adverse consequences. DCDuring (talk) 17:21, 25 April 2018 (UTC)

For an example of the problem see the sole definition line at ]. DCDuring (talk) 17:22, 25 April 2018 (UTC)

Caryophyllineae is missing a parameter, so the template thinks the 3rd parameter should be linked as Translingual. DTLHS (talk) 17:27, 25 April 2018 (UTC)

The {{quote}} templates don’t italicize the title and other automatically italicized parameters on mobile. See accidentally on purpose (on mobile) — first and third quotes use template, second does not. – Gormflaith (talk) 13:54, 26 April 2018 (UTC)

It looks like a different stylesheet is loaded for mobile than for desktop, and the "cite" class is defined differently in each. - TheDaveRoss 14:05, 26 April 2018 (UTC)

Yeah, the mobile stylesheet is not as fancy as the desktop one, since it needs to keep text legible but keep itself small, since it's loaded for all mobile users. I added bits to provide correct display of right-to-left languages and suppress italics for scripts that don't use them after a recent thread here in the GP, but I don't know whether italicizing titles is important enough to add. I don't see a class named "cite" in either stylesheet, btw... - -sche (discuss) 14:58, 30 April 2018 (UTC)

Can anyone explain why {{auto cat}} is unable to parse category names that start with the language name "East Central German"? For some reason they all had either a module error or an invalid label error showing {{topic cat}} as the template used. At first I thought it was due to the language name including another language, but "Alemannic German" and "Pennsylvania German" don't have this problem. The categories that start with the "gmw-ecg" language code also work fine I was able to work around this by using the original templates ({{poscatboiler}}, {{derivcatboiler}}, etc.), but most people haven't worked with all the catboiler templates like I have- and they shouldn't have to. Chuck Entz (talk) 09:15, 29 April 2018 (UTC)

I've noticed before that {{auto cat}} doesn't work with newly added codes. I don't know if it starts working with them after a while, or if it never works with codes added after a certain date, though. And I don't know why it should be the case or if it's fixable. But it's a problem I've noticed before with other new codes. —Mahāgaja (formerly Angr) · talk 14:29, 29 April 2018 (UTC)

You need to edit Module:languages/canonical names / Module:languages/code to canonical name to get auto cat to work. This and other inconsistencies can be found at Module:data consistency check. DTLHS (talk) 16:19, 29 April 2018 (UTC)

I never knew about those two modules. I've added East Central German to them both now. It's a pity that they're necessary, i.e. that the necessary information can't simply be extracted from Module:languages/datax. —Mahāgaja (formerly Angr) · talk 17:19, 29 April 2018 (UTC)

@Mahagaja: Well, they aren't absolutely necessary. Before, two modules transcluded all the language data modules and compiled the code-to-name and name-to-code mappings, but I figured it might save some memory and processing time to do it by hand. I don't know how much it actually saves. I should also make it easier to update them. — Eru·tuon 18:00, 29 April 2018 (UTC)

If {{auto cat}} fails without them, then they come pretty close to being absolutely necessary. —Mahāgaja (formerly Angr) · talk 18:02, 29 April 2018 (UTC)

Right, Module:languages/canonical names is necessary in that way. I meant that using the manual update method isn't necessary; that is, it could be done automatically, at the cost of much more memory usage on some pages. — Eru·tuon 18:22, 29 April 2018 (UTC)

Is it feasible to have whatever module backends {{autocat}} get the canonical name/code from Module:languages in the old way, while modules/templates like {{t}} that are likely to be used on pages that run into memory errors still use the newer/efficient way? Or could the module(s) fall back on the old way if they don't find the name/code they seek in Module:languages/canonical names, without that requiring more memory even in cases where they do find what they seek in Module:languages/canonical names? - -sche (discuss) 15:06, 30 April 2018 (UTC)

Well, {{t}} doesn't use Module:languages/canonical names. It doesn't need to convert canonical name to code.

I guess I should have looked at usage more closely. It looks like, in the mainspace, Module:languages/canonical names is almost exclusively used in entries for given names. There are probably not many given names that have high memory usage: such entries tend have lowercase Latin or Cyrillic, or Han-script, titles.

The only mainspace module that uses Module:languages/code to canonical name (see the search insource:/Module:languages\/code to canonical name/) may be Module:translit-redirect, which is used all over the place, including in high-memory entries (transclusions), to choose the correct transliteration module for languages that use more than one non-Latin script. So it may not be a good idea to switch Module:languages/code to canonical name to the higher-memory version. Even though all the language data modules are probably transcluded in high-memory entries, accessing them another time to compile a code-to-name mapping (which would happen only once per page) would add some memory. — Eru·tuon 18:39, 30 April 2018 (UTC)

A lot of Finnish non-lemma entries use Template:plural of to represent plural terms. However, it is ambiguous and we would want to replace all instances with Template:fi-form of with case=nominative and pl=plural. I have generated a list of all Finnish entries using that template here and am now looking for a way to replace these automatically. I suppose using a bot would be the best solution: any input on this or perhaps suggestions for an existing bot that could do this (or do I have to make my own)? SURJECTION _{·talk·contr·log·} 14:42, 30 April 2018 (UTC)

w:WP:AWB? —Suzukaze-c◆◆ 00:09, 2 May 2018 (UTC)

I'll have to refer to that if/when I'll need some other repetitive task done. Thank you for the tip. SURJECTION _{·talk·contr·log·} 06:06, 2 May 2018 (UTC)

@Surjection Running now: . Wyang (talk) 05:19, 2 May 2018 (UTC)

Seems to be doing a great job, many thanks to you. SURJECTION _{·talk·contr·log·} 06:06, 2 May 2018 (UTC)

For preventing weird edits like this. —Suzukaze-c◆◆ 04:27, 1 May 2018 (UTC)

While we are at it, can we add a button saying something like "Are we missing an entry? Let us know." to allow the user to add Chinese-character ones to WT:Requested entries (Chinese) or some other page? Wyang (talk) 05:10, 2 May 2018 (UTC)