Wiktionary:Grease pit/2017/September
Different Arabic and Persian fonts
editWhy do some Persian words appear on different pages (linked with a 'See also') from their Arabic etymons spelt exactly the same way? I don't know of any need for glyphic variation between the two languages, apart from new Persian letters like /p/ and /g/, and some problem with alif maqsura. Examples are كتف kitf/ketf (shoulder) and دليل dalîl (proof); but they appear on the same page (as I would expect for all such) with دار dâr (house) and دفتر daftar (notebook). (If I have a choice, the Persian font is much better looking, like proper print.) --Hiztegilari (talk) 10:55, 1 September 2017 (UTC)
- The first example features ك vs. ک. The second example features ي vs. ی. —suzukaze (t・c) 10:59, 1 September 2017 (UTC)
- In cases where the two glyphs are identical, though, such as كتف vs. کتف and دليل vs. دلیل, I wonder if a hard redirect would be a better solution. We can still use separate pages for words in which ك/ک and ي/ی appear differently. —Aɴɢʀ (talk) 13:19, 1 September 2017 (UTC)
- If we did that, the pagename would show codepoints that are never used for Persian, which is a problem from my perspective (people searching for Persian will probably use a Persian input method, which produces the right codepoints). With updating of
{{also}}
, being on separate pages poses no real issue in terms of users navigating to words. —Μετάknowledgediscuss/deeds 16:47, 1 September 2017 (UTC)- The correct solution to this problem would be to redesign Unicode from scratch. Until we can do that, we'll have to settle for having them on separate pages. --WikiTiki89 18:10, 1 September 2017 (UTC)
- If we did that, the pagename would show codepoints that are never used for Persian, which is a problem from my perspective (people searching for Persian will probably use a Persian input method, which produces the right codepoints). With updating of
- In cases where the two glyphs are identical, though, such as كتف vs. کتف and دليل vs. دلیل, I wonder if a hard redirect would be a better solution. We can still use separate pages for words in which ك/ک and ي/ی appear differently. —Aɴɢʀ (talk) 13:19, 1 September 2017 (UTC)
Diacritic stripping function
editIs there a diacritic stripping function built into Lua? If not, anyone know of a streamlined method? --Victar (talk) 15:26, 1 September 2017 (UTC)
- If there were, I wouldn't trust it. Module:links does something like that, depending on the language. Ask at the Grease pit for more information. Chuck Entz (talk) 15:48, 1 September 2017 (UTC)
- Thanks, I'll check it out. Whoops, I meant to post this there. Will move. --Victar (talk) 16:05, 1 September 2017 (UTC)
- What language, what diacritics do you specifically want to remove. DTLHS (talk) 16:43, 1 September 2017 (UTC)
- @DTLHS: I'm working on a declension table module for PII, so I'm just removing acute accents. --Victar (talk) 16:53, 1 September 2017 (UTC)
- If you have a list it can be added to Module:languages/datax. Or if you don't want it to apply to every PII entry you could just put a hardcoded table in your module. DTLHS (talk) 17:06, 1 September 2017 (UTC)
- @DTLHS: I need to strip acute accents from the declension suffixes if an accent exists on the stem. Module:User:Victar/iir-decl-noun --Victar (talk) 17:12, 1 September 2017 (UTC)
- Module:ine-common has a variety of functions for manipulating accents that might be useful. —Rua (mew) 17:32, 1 September 2017 (UTC)
- Perfect, thanks @CodeCat. --Victar (talk) 17:35, 1 September 2017 (UTC)
- Module:ine-common has a variety of functions for manipulating accents that might be useful. —Rua (mew) 17:32, 1 September 2017 (UTC)
- @DTLHS: I need to strip acute accents from the declension suffixes if an accent exists on the stem. Module:User:Victar/iir-decl-noun --Victar (talk) 17:12, 1 September 2017 (UTC)
- If you have a list it can be added to Module:languages/datax. Or if you don't want it to apply to every PII entry you could just put a hardcoded table in your module. DTLHS (talk) 17:06, 1 September 2017 (UTC)
- @DTLHS: I'm working on a declension table module for PII, so I'm just removing acute accents. --Victar (talk) 16:53, 1 September 2017 (UTC)
WM attention to "the two-stage page loading problem"
editHere is the beginnings of a discussion thread on Wikimedia-l on a matter directly relevant to us, based on my recollection of discussions:
Michael Peel email@mikepeel.net via lists.wikimedia.org 9:09 PM (4 hours ago)
to Wikimedia This is possibly the most annoying feature of the Wikimedia projects at the moment. You access a page. Then you start reading or editing it. And then suddenly the page jumps when a fundraising banner / central notice / gadget / beta feature loads. So you have to start reading the page again, or you have to find where you were editing again, or you have to undo the change you just made since you made it in the wrong part of the page.
I understand that this isn't intentional. Presumably there is a phabricator ticket about this. But how can we fix this - does this need more developer time, is this an external problem that we need someone else to fix, or is this a WONTFIX?
--
James Heilman jmh649@gmail.com via lists.wikimedia.org 11:25 PM (2 hours ago)
to Wikimedia I just put forwards a proposal to fix part of this issue Mike :-)
https://en.wikipedia.org/wiki/Wikipedia_talk:Twinkle#Button_load_issues
The TW button is easy to fix at least I am told, once I get consensus. Amir fixed one of the buttons earlier today.
- Does this also relate to the timing problem that leads to the all-too-frequent failure of certain JS-implmented features (show-hides most annoyingly) to load? DCDuring (talk) 05:37, 2 September 2017 (UTC)
- Should someone here who is technically knowledgeable follow this and push our view of the problem. DCDuring (talk) 05:40, 2 September 2017 (UTC)
- Hooray! The main problems in general editing (for me at least) are the entire tab bar (or the space directly below it), causing the edit box to jump downward, and the late-loading Citations tab on that bar, causing some tabs to jump rightward. Equinox ◑ 09:37, 2 September 2017 (UTC)
No more automatic link for plural creation?
editWhat happened to the automatic link that would load if you tried to create an English plural form? Now I just get a blank page. — SGconlaw (talk) 17:49, 2 September 2017 (UTC)
- You need to reenable it in your preferences. DTLHS (talk) 17:53, 2 September 2017 (UTC)
- Thanks. How did it get turned off? — SGconlaw (talk) 18:34, 2 September 2017 (UTC)
- By moving and/or renaming the file, which, if I remember correctly, was for good reasons. It would have been a good idea, though, to let people know it would happen. But then, hindsight is 20/20. Chuck Entz (talk) 18:41, 2 September 2017 (UTC)
- Thanks. How did it get turned off? — SGconlaw (talk) 18:34, 2 September 2017 (UTC)
- Now that we have the hindsight, I hope future changes of this kind will be seamless. Equinox ◑ 18:48, 2 September 2017 (UTC)
- Inconveniences likes this are good. It makes you reconsider whether you actually need the gadget. I might as well make this a feature - turning random gadgets off at random intervals. Dixtosa (talk) 19:28, 2 September 2017 (UTC)
- My immediate assumption when the form creation gadget stopped working was that it had broken, not that I needed to re-enable it. If there were some way to distinguish between those situations, the idea might work. (Or maybe you're joking.) — Eru·tuon 19:52, 2 September 2017 (UTC)
- A notice at "News for editors" would have been welcome. — SGconlaw (talk) 20:07, 2 September 2017 (UTC)
- My immediate assumption when the form creation gadget stopped working was that it had broken, not that I needed to re-enable it. If there were some way to distinguish between those situations, the idea might work. (Or maybe you're joking.) — Eru·tuon 19:52, 2 September 2017 (UTC)
Wildcard Contributions Searches
editThe Wiktionary:Per-browser preferences have stopped working for me. Given the sad state of my system, it's not worth the trouble of troubleshooting. There was only one thing it it I ever used: there was a thing that popped up in Special:Contributions pages that allowed me to do wildcard searches on IP. It wasn't exactly a masterpiece of user-interface design, but I was able to look for IP users who were changing their IPs, but staying within their ISP's local allocation. Would it be possible to have a regular gadget or a special page that would allow the same thing?
Basically it would take "77.37.156.*", or, even better, "77.37.156.23/24" as input and give a combined list of edits by every IP within that range. This capability is especially helpful when contemplating a range block, because it helps refine the range needed and makes it possible to check if anyone has made legitimate edits from within it. Thanks! — This unsigned comment was added by Chuck Entz (talk • contribs).
- The code behind the gadget was deleted. Pinging @TheDaveRoss. Dixtosa (talk) 07:15, 3 September 2017 (UTC)
- I had not realized that this was being used by anyone, or that it was added to the per-browser preferences. Restored. - TheDaveRoss 13:54, 3 September 2017 (UTC)
- @TheDaveRoss Well, now the user interface works like it used to, but it doesn't do anything- there's no list of edits. Chuck Entz (talk) 14:03, 4 September 2017 (UTC)
- @Chuck Entz, I have copied the latest source of the gadget from wp. Is it working now? Dixtosa (talk) 15:53, 4 September 2017 (UTC)
- Yes. Thank you! Chuck Entz (talk) 18:06, 4 September 2017 (UTC)
- @Dixtosa Did you move it to somewhere global so I can delete it again from my userspace? - TheDaveRoss 14:03, 6 September 2017 (UTC)
- @TheDaveRoss, now you can delete your page. Dixtosa (talk) 16:55, 6 September 2017 (UTC)
- @Dixtosa Did you move it to somewhere global so I can delete it again from my userspace? - TheDaveRoss 14:03, 6 September 2017 (UTC)
- Yes. Thank you! Chuck Entz (talk) 18:06, 4 September 2017 (UTC)
- @Chuck Entz, I have copied the latest source of the gadget from wp. Is it working now? Dixtosa (talk) 15:53, 4 September 2017 (UTC)
- @TheDaveRoss Well, now the user interface works like it used to, but it doesn't do anything- there's no list of edits. Chuck Entz (talk) 14:03, 4 September 2017 (UTC)
- I had not realized that this was being used by anyone, or that it was added to the per-browser preferences. Restored. - TheDaveRoss 13:54, 3 September 2017 (UTC)
Testcases fail but are identical?
editModule:ha-headword/testcases- as far as I can tell the expected and actual cells for "aeionū̀ (m)" and "aeionā̀ (f)" are identical, but still failing- why? — This unsigned comment was added by DTLHS (talk • contribs).
- It might be because the testcases use the combined Unicode character [A WITH ACUTE] while the module itself outputs the letter + the combining diacritics [LATIN LETTER A + COMBINING ACUTE]. Use
mw.ustring.toNFD()
to break A WITH ACUTE into pieces andmw.ustring.toNFC()
to do the reverse. —suzukaze (t・c) 23:42, 2 September 2017 (UTC)- Thanks, I didn't realize it was that easy. DTLHS (talk) 23:49, 2 September 2017 (UTC)
Mobile App Word Game
editHi,
I am looking for some advice on changing the downloaded version of the Wiktionary so that the size and formatting of the data is in a workable format for my purposes. I am investigating how to rearrange the data into the following columns (preferably with some type of delimiter):
COL1 (The word)|COL2 (The meaning)
Is this at all possible?
Regards,
John.
- Are you interested in a particular language? What about words that have multiple meanings? Are you interested in both lemmas and nonlemmas (cat vs cats)? DTLHS (talk) 16:41, 6 September 2017 (UTC)
inc-mgd
edit
This language (already added) needs MOD:Brah-translit as a translit module. —Aryaman (मुझसे बात करो) 20:05, 6 September 2017 (UTC)
- @Aryamanarora: Done. — Eru·tuon 21:51, 6 September 2017 (UTC)
When clicked, the arrow turns around as if to expand a list, but no list appears. (Using latest Chrome.) Equinox ◑ 21:16, 6 September 2017 (UTC)
- Fixed. DTLHS (talk) 21:17, 6 September 2017 (UTC)
Redlinks from reconstruction pages
editI was looking over Category:Proto-Indo-Iranian redlinks and I noticed that the redlinks from reconstruction pages, like those for PIE, aren't included in the list. Is that by design? Is there some way I can generate this list? --Victar (talk) 13:52, 7 September 2017 (UTC)
- I don't know the reason, but
{{redlink category}}
only categorizes in mainspace pages. That could be changed using{{#switch:}}
. — Eru·tuon 18:38, 7 September 2017 (UTC)- @Erutuon: Yeah, I don't see why not. Do we need a vote to add reconstruction pages? --Victar (talk) 02:42, 8 September 2017 (UTC)
- @Angr, CodeCat, JohnC5, Metaknowledge what are your thoughts on this? --Victar (talk) 17:20, 9 September 2017 (UTC)
- I don't have strong feelings one way or the other. I guess it makes sense to include links from the Reconstruction namespace as well, and probably Appendix namespace too. —Aɴɢʀ (talk) 17:24, 9 September 2017 (UTC)
- I'm fine adding the Reconstruction namespace, though I don't know why the Appendix should be added. Do we put any lemmata in the Appendix now? —JohnC5 21:56, 10 September 2017 (UTC)
- Okay, no objections so far and I can't think of a reason why not, so redlinks will now be tracked in the Reconstruction namespace. — Eru·tuon 22:39, 10 September 2017 (UTC)
- Thanks, @Erutuon! --Victar (talk) 21:40, 12 September 2017 (UTC)
- @JohnC5: We have lots of lemmas in Appendix namespace: Swadesh lists, lists of names, various other lists (e.g. Appendix:Burmese units of measure). —Aɴɢʀ (talk) 22:18, 12 September 2017 (UTC)
- @Angr: Yep, I wasn't paying attention. Thanks! —JohnC5 22:46, 12 September 2017 (UTC)
- I think it is useful to track the words in wordlists and stuff, so I've added the Appendix namespace. — Eru·tuon 23:20, 12 September 2017 (UTC)
- Hmm, @Erutuon, it doesn't seem to be working right, i.e Proto-Indo-European *kʷékʷlos is coming up in Category:Proto-Indo-Iranian redlinks. --Victar (talk) 05:41, 14 September 2017 (UTC)
- @Victar: Oh... good catch. That's because the page *čakrám doesn't exist. I'll have to make Module:redlink category check the pagename Reconstruction:Proto-Indo-Iranian/čakrám instead. — Eru·tuon 05:51, 14 September 2017 (UTC)
- Done (diff). — Eru·tuon 20:22, 14 September 2017 (UTC)
- @Erutuon: Gangbusters! --Victar (talk) 20:47, 14 September 2017 (UTC)
- @JohnC5: We have lots of lemmas in Appendix namespace: Swadesh lists, lists of names, various other lists (e.g. Appendix:Burmese units of measure). —Aɴɢʀ (talk) 22:18, 12 September 2017 (UTC)
- Thanks, @Erutuon! --Victar (talk) 21:40, 12 September 2017 (UTC)
what happened to the nocap parameter for the bor template?
editI get an error message when I use this now? What gives?! Word dewd544 (talk) 14:21, 7 September 2017 (UTC)
- See this conversation, essentially the inclusion of the "borrow" text is getting phased out. - TheDaveRoss 15:25, 7 September 2017 (UTC)
- In other words, if you want to write
{{bor|...|nocap=1}}
, write{{bor|...|notext=1}}
instead and write in "borrowing from" by hand. —Aɴɢʀ (talk) 12:20, 8 September 2017 (UTC)- Oy vey, what's with all the changes? I see. Well that's fine actually, as long as the ones that used "nocap" have been automatically converted to having "Borrowed from" as text before it. I guess this does make sense since it gives us more flexibility in using the template as part of an etymology, without having to use the "notext" parameter. Thanks. Word dewd544 (talk) 15:55, 8 September 2017 (UTC)
- I (and probably others) have been doing some of those conversions, hopefully we are well along the way to having everything cleaned up. - TheDaveRoss 17:58, 8 September 2017 (UTC)
- For the record, this change in
{{bor}}
was voted and approved here: Wiktionary:Votes/2017-06/borrowing, borrowed. --Daniel Carrero (talk) 18:02, 8 September 2017 (UTC)- Wait a sec, just to make things clear: so now it isn't even good practice to just start an etymology with the borrowed template without a notext, even if it's at the beginning of an etymology, because that will be phased out soon too? So always use notext=1 and write out "Borrowed or borrowing from..."? Word dewd544 (talk) 22:51, 22 September 2017 (UTC)
- Isn't it obvious? DCDuring (talk) 23:47, 22 September 2017 (UTC)
- DCD, you may think you're being funny, but it's really just low-quality trolling. Stop it. —Μετάknowledgediscuss/deeds 23:49, 22 September 2017 (UTC)
- My severe ADD prevented me from reading the aforementioned discussion and vote on the matter. Word dewd544 (talk) 01:00, 25 September 2017 (UTC)
- Isn't it obvious? DCDuring (talk) 23:47, 22 September 2017 (UTC)
- Wait a sec, just to make things clear: so now it isn't even good practice to just start an etymology with the borrowed template without a notext, even if it's at the beginning of an etymology, because that will be phased out soon too? So always use notext=1 and write out "Borrowed or borrowing from..."? Word dewd544 (talk) 22:51, 22 September 2017 (UTC)
- For the record, this change in
- I (and probably others) have been doing some of those conversions, hopefully we are well along the way to having everything cleaned up. - TheDaveRoss 17:58, 8 September 2017 (UTC)
- Oy vey, what's with all the changes? I see. Well that's fine actually, as long as the ones that used "nocap" have been automatically converted to having "Borrowed from" as text before it. I guess this does make sense since it gives us more flexibility in using the template as part of an etymology, without having to use the "notext" parameter. Thanks. Word dewd544 (talk) 15:55, 8 September 2017 (UTC)
- In other words, if you want to write
Lua memory errors related to Wikidata use in Module:senseid
editThere are quite a few pages in CAT:E with Lua memory errors. I think they are due to recent changes by @CodeCat in Module:senseid. Now, if a sense id formatted with {{senseid}}
is a Wikidata id (QN
), then the module does a bunch of stuff with the it: checking if the id is for a planet, continent, country, language, taxon, emotion, etc. Some part of this process is taking enough memory to cause quite a few pages to run out of memory.
I can see two obvious options: disable the Wikidata stuff in Module:senseid entirely, or create a list of the pages on which the function should not run (those currently running out of memory). — Eru·tuon 19:13, 8 September 2017 (UTC)
- I thought there was an agreement not to include anything which uses Wikidata in the main namespace without approval. Was use of senseid with Wikidata discussed? - TheDaveRoss 19:24, 8 September 2017 (UTC)
- The rationale was that since it currently only adds tracking categories it didn't need a discussion. DTLHS (talk) 19:28, 8 September 2017 (UTC)
(edit conflict) Hm, I might be wrong. I don't see {{senseid}}
in fish, one of the pages that recently ran over the memory cap, for example. Maybe it's something else. But I don't feel like going through and checking each page. — Eru·tuon 19:27, 8 September 2017 (UTC)
- Toggling the wikidata code on and off didn't seem to have much of an effect on memory- the template uses about 1 MB for the first invocation either way. — This unsigned comment was added by DTLHS (talk • contribs) at 19:28, 8 September 2017 (UTC).
- I was wondering this myself as the errors started appearing, but they seem to be unrelated to the Wikidata stuff. —Rua (mew) 19:35, 8 September 2017 (UTC)
- If I remove the translation table from language, usage drops from 50 MB to 10 MB. Was anything changed in Module:translations or the templates it depends on? Maybe a change in the Lua implementation itself? —Rua (mew) 19:44, 8 September 2017 (UTC)
- I've been looking at recent changes in the Module namespace. It's hard to answer your question, because there are a lot of modules involved in an actual use of Module:translations: language and script data modules, transliteration modules, other modules that are loaded by another module when that module is loaded. — Eru·tuon 20:02, 8 September 2017 (UTC)
- I still don't really understand why there are these extra restrictions on the use of Lua. —Rua (mew) 20:25, 8 September 2017 (UTC)
- What extra restrictions are you referring to? — Eru·tuon 20:59, 8 September 2017 (UTC)
- The memory limits based on how many Lua-fied templates you use on a page. A page shouldn't need more memory just because it has more Lua on it, because each template runs in its own sandbox. Different invocations don't need to share memory, so once one is finished processing, the memory should be freed. The only memory that should be kept across invocations is
loadData
stuff. If one invocation of{{t}}
uses X amount of memory, then each invocation of{{t}}
uses its own separate X amount of memory. The fact that memory usage accumulates when more invocations are added, indicates to me that there is something wrong with how Scribunto handles memory. —Rua (mew) 21:11, 8 September 2017 (UTC)- I'm tempted to agree, but I wish I knew just what all the accumulated memory was: previously loaded modules, module output, loaded data modules? It seems that previously loaded modules at least aren't shared between template invocations, as can be seen when I dump a representation of the
package.loaded
Lua variable on a page that uses other templates (see the current revision of Module:sandbox). But maybe there is other inaccessible memory usage in the Scribunto extension that counts towards the limit. — Eru·tuon 21:55, 8 September 2017 (UTC)- A possibility is that the software processes all Lua invocations in parallel threads, but using a shared memory pool. Since each template and module expansion is completely independent of any others, it's very parallelisable. On the other hand, the more parallel processing is done, the less memory each thread has for itself. If this is indeed how it's done, they could be more smart about it. For example, rather than just bailing out with an "out of memory" error, they could just throttle/stall some threads until some memory is freed up again once other threads finish, then resume. —Rua (mew) 22:39, 8 September 2017 (UTC)
- To me, it looks like the Lua invocations are processed in order, so the invocation at which memory runs out moves higher up the page when there is more memory used. I also recall an odd case in which one template (I think it was
{{zh-der}}
) ran out of memory, but the ones after it didn't. So it looked like the one template went over the limit, but then its memory was freed up and the smaller-memory templates after it were able to run. — Eru·tuon 19:37, 9 September 2017 (UTC)- But then if each individual template doesn't run over the limit, why do we run over the limit when we have many of them? That shouldn't happen. The page would be just fine if only it didn't try to use more memory than was available. —Rua (mew) 20:09, 9 September 2017 (UTC)
- Clearly there must be something adding more memory with each new use of Lua. But I don't know what or how. — Eru·tuon 20:16, 9 September 2017 (UTC)
- You'd almost think they didn't want us to use it, and would prefer us to start using slow and clumsy template logic again... —Rua (mew) 20:50, 9 September 2017 (UTC)
- All of the source code is fully viewable by anyone if you think you can improve it. DTLHS (talk) 20:51, 9 September 2017 (UTC)
- You'd almost think they didn't want us to use it, and would prefer us to start using slow and clumsy template logic again... —Rua (mew) 20:50, 9 September 2017 (UTC)
- Clearly there must be something adding more memory with each new use of Lua. But I don't know what or how. — Eru·tuon 20:16, 9 September 2017 (UTC)
- But then if each individual template doesn't run over the limit, why do we run over the limit when we have many of them? That shouldn't happen. The page would be just fine if only it didn't try to use more memory than was available. —Rua (mew) 20:09, 9 September 2017 (UTC)
- To me, it looks like the Lua invocations are processed in order, so the invocation at which memory runs out moves higher up the page when there is more memory used. I also recall an odd case in which one template (I think it was
- A possibility is that the software processes all Lua invocations in parallel threads, but using a shared memory pool. Since each template and module expansion is completely independent of any others, it's very parallelisable. On the other hand, the more parallel processing is done, the less memory each thread has for itself. If this is indeed how it's done, they could be more smart about it. For example, rather than just bailing out with an "out of memory" error, they could just throttle/stall some threads until some memory is freed up again once other threads finish, then resume. —Rua (mew) 22:39, 8 September 2017 (UTC)
- I'm tempted to agree, but I wish I knew just what all the accumulated memory was: previously loaded modules, module output, loaded data modules? It seems that previously loaded modules at least aren't shared between template invocations, as can be seen when I dump a representation of the
- The memory limits based on how many Lua-fied templates you use on a page. A page shouldn't need more memory just because it has more Lua on it, because each template runs in its own sandbox. Different invocations don't need to share memory, so once one is finished processing, the memory should be freed. The only memory that should be kept across invocations is
- What extra restrictions are you referring to? — Eru·tuon 20:59, 8 September 2017 (UTC)
- I still don't really understand why there are these extra restrictions on the use of Lua. —Rua (mew) 20:25, 8 September 2017 (UTC)
- I've been looking at recent changes in the Module namespace. It's hard to answer your question, because there are a lot of modules involved in an actual use of Module:translations: language and script data modules, transliteration modules, other modules that are loaded by another module when that module is loaded. — Eru·tuon 20:02, 8 September 2017 (UTC)
Swahili classes >2
editI can make pages for first, second, & third person conjugations of verbs, but what about the ma class, ki-vi class, & so on? Anjuna (talk) 05:33, 9 September 2017 (UTC)
- @Science Bird: You can create these pages as well, although creating any of these inflected forms is not nearly as useful as creating lemma forms with definitions, because the inflected forms can always be created later by robots. —Μετάknowledgediscuss/deeds 11:13, 9 September 2017 (UTC)
- Robots? Thank you. Is there a way to get them moving? Anjuna (talk) 13:44, 9 September 2017 (UTC)
- We call them bots, and it's simply a matter of asking here for someone with a bot to do a particular task for you. In some cases, it may be better to wait until it's more clear what will need to be done and until there's enough potential work waiting to make a bot run worth it- but I have no idea if this is one of those cases. Chuck Entz (talk) 13:54, 9 September 2017 (UTC)
- Robots? Thank you. Is there a way to get them moving? Anjuna (talk) 13:44, 9 September 2017 (UTC)
Distinguishing affixes that are identical in form
editI want to have separate categories for prefixes that all take the form ma- but are distinct in use. I feel like this problem has been solved before somewhere, but I don't know where and how. @CodeCat, do you (or does anyone else) remember? —Μετάknowledgediscuss/deeds 02:04, 11 September 2017 (UTC)
- @Metaknowledge: Category:English_words_suffixed_with_-er#mw-subcategories? —suzukaze (t・c) 02:06, 11 September 2017 (UTC)
- Thank you, the id1= (etc.) parameter does the job. —Μετάknowledgediscuss/deeds 18:56, 11 September 2017 (UTC)
Recentchanges tag filter inverse
editDoes anyone know if it's possible to filter recent changes for edits that are not labelled with a specific tag? DTLHS (talk) 04:19, 12 September 2017 (UTC)
- No it is not possible.5.178.149.114 05:42, 12 September 2017 (UTC)
Template:quote categorizes the entry as having a usage example
editFor example, פרזלא is in the category Category:Aramaic terms with usage examples when it shouldn't be, because it only has a quotation. --WikiTiki89 20:07, 12 September 2017 (UTC)
- Fixed. DTLHS (talk) 20:10, 12 September 2017 (UTC)
Visual Editor
editThe visual editor goes against all our style guidelines (see diff, for example). If this cannot be fixed, we should disable the visual editor entirely. --WikiTiki89 15:54, 13 September 2017 (UTC)
Abuse filter required
editWe're getting lots of English entries that are just the generated rfdef template, with no attempt to fill it in. (Part of speech varies.) Can we block these? Also, a lot of the entry titles are in Arabic script. Can we block any attempt to create an English entry with Arabic characters in the title? Equinox ◑ 22:52, 13 September 2017 (UTC)
- Yes please - these are getting annoying. SemperBlotto (talk) 11:01, 19 September 2017 (UTC)
- @Equinox: I am happy to try and create an edit filter, do you have a couple of diffs I could use as a basis/test? - TheDaveRoss 11:40, 19 September 2017 (UTC)
- This Arabic-script one was just created: معلومات_عن_العادة_السرية. Equinox ◑ 13:23, 19 September 2017 (UTC)
- Another: فتيات من ورقلة. Equinox ◑ 14:08, 19 September 2017 (UTC)
- Another, created as sign-language rather than English: پروتز. Equinox ◑ 17:22, 19 September 2017 (UTC)
- This last one is a bit different: Persian rather than Arabic, and a word that I think we actually ought to have (I reckon it means "prosthesis") rather than an entry that should never exist. —Μετάknowledgediscuss/deeds 22:24, 19 September 2017 (UTC)
- Another, created as sign-language rather than English: پروتز. Equinox ◑ 17:22, 19 September 2017 (UTC)
- I'm a little concerned about how to distinguish between the normal case with the new entry creator, where everything is preloaded with the empty templates and then edited by the contributor before saving, and the bad cases, where it's saved without being edited- they both start out identical. Chuck Entz (talk) 16:19, 19 September 2017 (UTC)
- I created Special:AbuseFilter/72 which is currently only tagging such entries (rfdef). If the user clicks a template and edits before saving to remove the
{{rfdef}}
it is not tagged, however if they create the entry with the{{rfdef}}
and intend to edit is further in subsequent edits we are never going to be able to capture that. - TheDaveRoss 11:42, 22 September 2017 (UTC)- I created some entries with
{{rfdef}}
only yesterday, for example ruibi. So we should probably choose some better way to recognise these edits. —Rua (mew) 11:52, 22 September 2017 (UTC)- It only flags new entries by anonymous users, so your creations would not be tagged. - TheDaveRoss 12:31, 22 September 2017 (UTC)
- I created some entries with
- I created Special:AbuseFilter/72 which is currently only tagging such entries (rfdef). If the user clicks a template and edits before saving to remove the
- They are still getting through. SemperBlotto (talk) 19:50, 23 September 2017 (UTC)
- I'm a little concerned about how to distinguish between the normal case with the new entry creator, where everything is preloaded with the empty templates and then edited by the contributor before saving, and the bad cases, where it's saved without being edited- they both start out identical. Chuck Entz (talk) 16:19, 19 September 2017 (UTC)
Improvements coming soon to Recent Changes
editHello
Sorry to use English. Please help translate to your language! Thank you.
In short: starting on 26 September, New Filters for Edit Review (now in Beta) will become standard on Recent Changes. They provide an array of new tools and an improved interface. If you prefer the current page you will be able to opt out. Learn more about the New Filters.
What is this feature again?
This feature improves Special:RecentChanges and Special:RecentChangesLinked (and soon, Special:Watchlist – see below).
Based on a new design, it adds new features that ease vandalism tracking and support of newcomers:
- Filtering - filter recent changes with easy-to-use and powerful filters combinations, including filtering by namespace or tagged edits.
- Highlighting - add a colored background to the different changes you are monitoring. It helps quick identification of changes that matter to you.
- Bookmarking to keep your favorite configurations of filters ready to be used.
- Quality and Intent Filters - those filters use ORES predictions. They identify real vandalism or good faith intent contributions that need help. They are not available on all wikis.
You can know more about this project by visiting the quick tour help page.
Concerning RecentChanges
Starting on 26 September, New Filters for Edit Review will become standard on Recent Changes. We have decided to do this release because of a long and successful Beta test phase, positive feedback from various users and positive user testing.
Some features will remain as Beta features and will be added later. Learn more about those different features.
If your community has specific concerns about this deployment or internal discussion, it can request to have the deployment to their wikis delayed to October 1, if they have sensible, consistent with the project, actionable, realistic feedback to oppose (at the development team's appreciation).
You will also be able to opt-out this change in your preferences.
Concerning Watchlists
Starting on September 19, the Beta feature will have a new option. Watchlists will have all filters available now on the Beta Recent Changes improvements.
If you have already activated the Beta feature "⧼eri-rcfilters-beta-label⧽", you have no action to take. If you haven't activated the Beta feature "⧼eri-rcfilters-beta-label⧽" and you want to try the filters on Watchlists, please go to your Beta preferences on September 19.
How to be ready
Please share this announcement!
Do you use Gadgets that change things on your RecentChanges or Watchlist pages, or have you customized them with scripts or CSS? You may have to make some changes to your configuration. Despite the fact that we have tried to take most cases into consideration, some configurations may break. The Beta phase is a great opportunity to have a look at local scripts and gadgets: some of them may be replaced by native features from the Beta feature.
Please ping me if you have questions.
On behalf of the Global Collaboration team, Trizek (WMF) 15:27, 14 September 2017 (UTC)
- @Trizek (WMF): I guess the Watchlist filter beta is out now. Here is my main serious issue with it: it makes the Watchlist take a good amount of extra time to load, which is a severe issue for those who refresh their Watchlist frequently. I would hope there would be a solution that allows the Watchlist to load in the normal amount of time when no filters are active. In fact I'm disabling this feature, as the Watchlist is now unusable for me. --WikiTiki89 16:41, 19 September 2017 (UTC)
This syntax worked fine the other day. Equinox ◑ 22:17, 14 September 2017 (UTC)
- @Erutuon DTLHS (talk) 22:22, 14 September 2017 (UTC)
- @DTLHS, Equinox: Thanks for the ping. Yes, it was me. I believe it's fixed now. — Eru·tuon 22:44, 14 September 2017 (UTC)
Can't reply to LiquidThreads posts
edit@IvanScrooge98 I use LiquidThreads on my talk page. When I click reply to answer someone's post, a new window with a spinny thing appears as usual. However, the spinny thing is supposed to disappear and an edit window should appear instead. Now, the spinny thing stays forever. This makes it impossible for me to post anything in LiquidThreads. Can this be fixed, please? —Rua (mew) 13:41, 17 September 2017 (UTC)
- @CodeCat: mmh, no clue. When I do, the “spinny thing” does not appear in a new window but right below the thread, giving me the same problem though. If I open the link in a new window, a perfectly normal editor appears, allowing me to reply. [ˌiˑvã̠n̪ˑˈs̪kr̺ud͡ʒʔˌn̺ovã̠n̪ˑˈt̪ɔ̟t̪ːo] (parla con me) 13:46, 17 September 2017 (UTC)
- Ah, that works for me. Thank you for suggesting that. Still, the problem stands. —Rua (mew) 14:36, 17 September 2017 (UTC)
Category:English terms spelled with 8 is currently manually added to this page. Module:headword is supposed to automatically categorise terms by their characters, making this manual category superfluous. However, {{en-noun}}
isn't adding this category automatically. Is something missing in our data? The same applies to Chinese on 69. —Rua (mew) 14:35, 17 September 2017 (UTC)
Something wrong with Tabbed languages
editSome pages aren't displaying Tabbed languages correctly. At bod, the ==Volapük== line is at the very bottom of the page and all the Volapük info is inside the Swedish tab. At faen, it's the ==Norwegian Bokmål== that's at the very bottom of the page and all the Norwegian Bokmål info is inside the Bislama tab. But other pages with multiple languages are displaying everything correctly. —Aɴɢʀ (talk) 09:14, 19 September 2017 (UTC)
- No ideas? It still isn't fixed. It's apparently related to the special characters in the language names, because changing them to ==Volapuk== and ==Norwegian Bokmaal== solves the problem, though obviously that's not a desirable solution. —Aɴɢʀ (talk) 10:54, 20 September 2017 (UTC)
- Do you know where the code for tabbed languages is? --WikiTiki89 14:39, 20 September 2017 (UTC)
- Never mind, found it. I'll take a look. --WikiTiki89 14:40, 20 September 2017 (UTC)
- Do you know where the code for tabbed languages is? --WikiTiki89 14:39, 20 September 2017 (UTC)
So here's what I found: A heading with no special characters looks like this in the HTML:
<h2>
<span class="mw-headline" id="Bislama">Bislama</span>
<span class="mw-editsection">...</span>
</h2>
While a heading with special characters looks like this:
<h2>
<span id="Norwegian_Bokmål"></span>
<span class="mw-headline" id="Norwegian_Bokm.C3.A5l">Norwegian Bokmål</span>
<span class="mw-editsection">...</span>
</h2>
I'm guessing the line <span id="Norwegian_Bokmål"></span>
is a new addition to support section links directly with special characters in addition to the old format. It doesn't look like the tabbed languages code knows how to handle that. --WikiTiki89 14:53, 20 September 2017 (UTC)
- Is it fixed? --WikiTiki89 15:16, 20 September 2017 (UTC)
- Seems to be. Thanks for your help! —Aɴɢʀ (talk) 15:29, 20 September 2017 (UTC)
Inflection templates categorizing in non-mainspace entries
editAppendix:Snowclones/don't X me, Citations:elephant juice for example. Can this be disabled? DTLHS (talk) 18:51, 21 September 2017 (UTC)
- It should be possible, but what are the cases in which categorization needs to be disabled? English headwords in Appendix and Citations namespaces, or any non-appendix-language headword in those namespaces? — Eru·tuon 00:58, 22 September 2017 (UTC)
- Well, appendix languages should only categorize in appendices, reconstructed languages only in the reconstructed namespace, everything else only in the main namespace. Would that work? DTLHS (talk) 05:24, 22 September 2017 (UTC)
- It's a little more complicated.
{{citation}}
uses{{cln}}
, which is handled by theformat_categories
function in Module:utilities, and Latin has some Reconstructed entries (*genuclum), but it's not a reconstructed language. So sometimes the function needs to add categories in the Reconstruction namespace for regular languages, and in the Citations namespace. It might be okay to have the function only add categories to an Appendix page if the language is appendix-constructed, though. I doubt someone would use the function to categorize other types of Appendix pages, because it assumes that the title is in the language in question. For the headword in a citations page, there probably has to be a specific exception. — Eru·tuon 05:45, 22 September 2017 (UTC)- I'm still in favour of deleting appendix-constructed languages altogether. —Rua (mew) 12:16, 22 September 2017 (UTC)
- Could there be "cat" and/or "nocat" parameters to allow for missing exceptions and reduce the need for excessive complication of template/Lua. DCDuring (talk) 12:20, 22 September 2017 (UTC)
- It's a little more complicated.
- Well, appendix languages should only categorize in appendices, reconstructed languages only in the reconstructed namespace, everything else only in the main namespace. Would that work? DTLHS (talk) 05:24, 22 September 2017 (UTC)
Extra IPA characters in box below edit area
editWould somebody be able to add ŋ̊ and ɲ̊ to the list of IPA characters in the clickable box below edit area (sorry, I don't know the technical name)? These are both used transcribing Icelandic words in IPA, and I am currently going through all the Icelandic nouns to add missing transcriptions so this would be very useful. Thanks, BigDom 10:45, 22 September 2017 (UTC)
- A better solution in the long term would be to have buttons that add combining diacritics separately. Then you could make any combination you want. Wikipedia does it this way. —Rua (mew) 12:15, 22 September 2017 (UTC)
- Yes, that would be even better. BigDom 13:46, 22 September 2017 (UTC)
- The list of characters is called CharInsert, and it's located at MediaWiki:Edittools, while the script that makes it actually work is at MediaWiki:Edit.js. — Eru·tuon 17:40, 22 September 2017 (UTC)
The entry dotard has been locked. Please add its Persian translation as follows:
- language code: fa
- translation: خرفت
- transliteration: xereft
Thank you 4nn1l2 (talk) 12:02, 22 September 2017 (UTC)
- Added and changed protection level. DTLHS (talk) 17:28, 22 September 2017 (UTC)
@Donnanz The Dutch tab is completely missing the categories. Instead they're placed under Latvian for some reason. —Rua (mew) 23:06, 22 September 2017 (UTC)
- @CodeCat: Did you ping the right person? I have never edited the page. DonnanZ (talk) 23:16, 22 September 2017 (UTC)
- <:ref name="LEV"/> is entered in the Latvian entry, but I have no idea why it comes before Dutch in the category listings and is separated from the rest of the Latvian categories. DonnanZ (talk) 23:53, 22 September 2017 (UTC)
- Me and names, bleh. It was the other person with a D. I still can't remember it. —Rua (mew) 11:07, 23 September 2017 (UTC)
- @Dixtosa I think. DTLHS (talk) 17:14, 23 September 2017 (UTC)
- Yes, thank you, yet another person with a D. —Rua (mew) 18:17, 23 September 2017 (UTC)
- @Dixtosa I think. DTLHS (talk) 17:14, 23 September 2017 (UTC)
- Me and names, bleh. It was the other person with a D. I still can't remember it. —Rua (mew) 11:07, 23 September 2017 (UTC)
Can someone change the ancestor for awa
(Awadhi) to pka
? And add MOD:hi-translit as the translit module for it. —Aryaman (मुझसे बात करो) 20:55, 24 September 2017 (UTC)
- @Aryamanarora: Done (diff). — Eru·tuon 05:21, 26 September 2017 (UTC)
@Wyang, Suzukaze-c: This page is exceeding the limit of expensive parser function calls (500). I suspect this is because of all the Chinese templates that retrieve pinyin from other pages. There was a module error in the Korean headword because Module:headword was trying to find out if the page Wiktionary:Korean transliteration existed, but I fixed that by stopping the function call. (So now the Korean headword doesn't link to the transliteration policy page.)
Errors may pop up again if more Chinese usage examples or compounds with entries are added, or when more compounds get their own entries from which pinyin can be retrieved. A temporary solution would be to remove usage examples or compounds.
Perhaps a more permanent solution would be having a whole host of data modules for pinyin, or a database somewhere else that somehow acquires and stores the pinyin from our entries automatically, allowing it to be retrieved by the module less expensively. — Eru·tuon 08:24, 28 September 2017 (UTC)
- Reduced the load a bit by removing some of the secondary compounds. Having a database-like structure would definitely be ideal. Wyang (talk) 09:05, 28 September 2017 (UTC)
Allowing Template:attention to ping people, is it feasible?
editIs it possible to modify {{attention}}
in some way so that, if you include a user's name in the text preceded by an @, then it pings the user in question to come look at the page? —Rua (mew) 12:05, 28 September 2017 (UTC)
- Pings have to be signed. Chuck Entz (talk) 15:06, 28 September 2017 (UTC)
- I would say it would be preferable anyway to ping them from the talk page instead. --WikiTiki89 18:03, 28 September 2017 (UTC)
News for editors link -- remove/fix?
editThe link at the top of the page, I mean. It makes the whole page sort of skip downwards when it loads, such that if you click before that, you probably misclick. Can it be removed (it doesn't really seem as important as its placement would indicate), or maybe have a space reserved for it or something Sorry if there's some obvious fix to it.__Gamren (talk) 15:51, 28 September 2017 (UTC)
- Never mind, it was just this thing: MediaWiki:Sitenotice.__Gamren (talk) 13:07, 14 October 2017 (UTC)
- Thank you! —Rua (mew) 15:24, 14 October 2017 (UTC)
Clueless Mobile Edits
editThere have been quite a lot of junk edits lately, and I suspect it has something to do with the capability of searching Wiktionary being added somewhere in the mobile interface without adequate explanation. That means people think that they're searching the web when they're actually searching Wiktionary, and that they end up editing or creating entries without realizing that's what they're doing.
What do you think about adding text to the edit screen saying something like:
- You are editing a page at Wiktionary, an online dictionary, and your content will be added to the dictionary if you proceed. If that's not what you meant to do, please click "Cancel".
I'm hoping this will reduce the volume of random garbage that keeps getting added to entries for no apparent reason, though not everyone doing this reads English, and some won't care. Chuck Entz (talk) 23:50, 28 September 2017 (UTC)
- I feel certain such people wouldn't read it, and don't read generally. Just more clutter. Equinox ◑ 09:15, 30 September 2017 (UTC)
Watchlist
editWhat the blazes has happened to the my watchlist (I'm not aware of changing anything):
- it's too "busy"
- I can't see the old "hide" options
- the current votes box is in the middle of the screen and squished "wanted"/"utilities" against the left margin
- I could go on! — Saltmarsh. 06:46, 29 September 2017 (UTC)
TextExtracts
editI came across butterfly and noticed that there is stuff before the page starts, that ends up in our search results and other forms of TextExtracts (like the sister search results on the Search page). It might be wise to add the "noexcerpt" class to templates like {{Commons}}
, or alternatively add noprint to the configuration of classes that are ignored when generating the short textextracts. TheDJ (talk) 14:09, 29 September 2017 (UTC)
Importing a large list of words into multiple appendices
editI've got a dump of all Russian terms from the Russian Wiktionary. I have removed all capitalised terms, terms with "-" and ".", so it's a list of all solid words. There are altogether 238,646 words. I'd like create some appendices, so that missing terms (red-linked) could gradually be filled. What is the maximum acceptable size of an appendix? I was getting errors about the size of the allowed files on 50,000 words, then reduced to 10,000. Now I am getting Wikimedia errors. Loading files like this will take for ever. Is there a quicker way? --Anatoli T. (обсудить/вклад) 06:30, 30 September 2017 (UTC)
- 10,000 words should be possible on a single page, assuming you are using plain links. What error are you getting? DTLHS (talk) 06:33, 30 September 2017 (UTC)
- I got it now, thanks, after two retries. This is the first portion of the dump: User:Atitarev/Ru-wiki-solid-Russian-terms-dump/1, which is only 1/24th of the total. --Anatoli T. (обсудить/вклад) 06:36, 30 September 2017 (UTC)
- @Atitarev I think I have a script which can remove the blue-links from those lists if you would like, or would you prefer to audit them and remove them manually? - TheDaveRoss 12:28, 2 October 2017 (UTC)
- @TheDaveRoss: Thanks, the dumps are now at [[User:Atitarev/Ru-wiki-solid-Russian-terms-dump]] - 24 pages. It's too much to audit manually. I did already a basic cleanup, as much as I could, even if it still contains a lot of non-lemmas. It would be great if you could remove the blue links. Ideally, only those that have the Russian L2 header only. I will also need to split them by the first letter alphabetically rather than by number but I'll do it myself if you don't. Each page currently has exactly 10,000 terms, except for the last one - @User:Atitarev/Ru-wiki-solid-Russian-terms-dump/24. Note that words starting with "ё" are in the last page but they should follow "е".--Anatoli T. (обсудить/вклад) 12:45, 2 October 2017 (UTC)
- @TheDaveRoss: Hello, please advise if you're still interested in helping and, which part you can do. --Anatoli T. (обсудить/вклад) 01:52, 9 October 2017 (UTC)
- I will try and get to this this week. - TheDaveRoss 13:44, 10 October 2017 (UTC)
- @TheDaveRoss: Hello, please advise if you're still interested in helping and, which part you can do. --Anatoli T. (обсудить/вклад) 01:52, 9 October 2017 (UTC)
- @TheDaveRoss: Thanks, the dumps are now at [[User:Atitarev/Ru-wiki-solid-Russian-terms-dump]] - 24 pages. It's too much to audit manually. I did already a basic cleanup, as much as I could, even if it still contains a lot of non-lemmas. It would be great if you could remove the blue links. Ideally, only those that have the Russian L2 header only. I will also need to split them by the first letter alphabetically rather than by number but I'll do it myself if you don't. Each page currently has exactly 10,000 terms, except for the last one - @User:Atitarev/Ru-wiki-solid-Russian-terms-dump/24. Note that words starting with "ё" are in the last page but they should follow "е".--Anatoli T. (обсудить/вклад) 12:45, 2 October 2017 (UTC)
- @Atitarev I think I have a script which can remove the blue-links from those lists if you would like, or would you prefer to audit them and remove them manually? - TheDaveRoss 12:28, 2 October 2017 (UTC)
- I got it now, thanks, after two retries. This is the first portion of the dump: User:Atitarev/Ru-wiki-solid-Russian-terms-dump/1, which is only 1/24th of the total. --Anatoli T. (обсудить/вклад) 06:36, 30 September 2017 (UTC)