Wiktionary:Grease pit/2014/June

Is it possible to remove transliteration for full-width characters only for ja-usex, so that nothing appears in furigana (ruby)? (transliteration in other places is still required, don't remove them from Module:ja)

Ａ：わかった？ Ｂ：おう。

Ａ: Wakatta? Ｂ: Ō.

A: "You see?" B: "Got it."

Full-width characters Ａ, Ｂ are only used here for display, they look better, not planning to use them in entries, in case anybody is concerned. --Anatoli ^{(обсудить}/^вклад) 12:58, 31 May 2014 (UTC)

@Haplology, @Wyang, @Eirikr, @Kephir - do you think you can fix that? (Without removing from Module:ja)--Anatoli ^{(обсудить}/^вклад) 02:39, 3 June 2014 (UTC)

Like the effect now? Wyang (talk) 03:11, 3 June 2014 (UTC)

Brilliant! --Anatoli ^{(обсудить}/^вклад) 03:31, 3 June 2014 (UTC)

Except for the part with 80 module errors... Chuck Entz (talk) 04:10, 3 June 2014 (UTC)

That's the tracking error returned by Module:zh-han, not related to this. Wyang (talk) 04:20, 3 June 2014 (UTC)

The first couple I looked at (which have since magically disappeared) were Japanese. As for the "tracking error"... that makes about as much sense as a "tracking hand grenade". There has to be some way to track such things without blowing stuff up so you can spot the mushroom clouds. That way you could look in a tracking category to find out that you didn't allow for variant radical characters, like 纟 (sīzìpáng) or 糹／纟 (sīzìpáng) instead of 糸 (mì), instead of flooding the script error page so no one could spot the half dozen or so errors that you didn't do on purpose.

Of course, it's easier for me to point out what I think you're doing wrong than to provide a solution- let alone do it better (or even half as well) myself- I guess I'm just being cranky since it's past my bedtime... Chuck Entz (talk) 07:24, 3 June 2014 (UTC)

Oops, I see that's @Kc kennylau's mess. Never mind... Chuck Entz (talk) 07:36, 3 June 2014 (UTC)

Template:ko-hanja has been unable to show Korean hangeul characters as wikilinks by default (as in without added wikilink formatting brackets) for about almost a month now in the "hangeul=" parameter. This used to work before and doesn't now. Bumm13 (talk) 00:13, 3 June 2014 (UTC)

In a previous discussion I noted that our current solution for Lua-cising {{langrev}} was not efficient enough and was causing timeouts. User:Kephir suggested writing a script that generates a reverse lookup table, and then loading it with mw.loadData so that it would remain loaded for the duration of the whole page. That would make the first use relatively slow, as it would have to back-translate every single language in Module:languages and make a table of them. But hopefully that process would be fast enough, especially if it only ever has to be done once per page.

But I've been wondering about an intermediate solution. It's basically the same, except that instead of generating a single large reverse lookup table, we generate one for each of the data submodules of Module:languages. So for Module:languages/data2 there would be a corresponding Module:languages/data2/by name and so on. The assumption is that like with regular lookups, a lot of pages will not need to do a reverse lookup for every single data submodule, but only a few. So this might speed it up more still. Does anyone have feedback on this? —CodeCat 18:19, 7 June 2014 (UTC)

How can we maintain the correspondence between the two pages (without relying on a bot that someone may someday decide to stop running)? DTLHS (talk) 18:22, 7 June 2014 (UTC)

The reverse pages aren't actual tables. Rather they contain a script that runs when the module is first loaded (this is possible as far as I know), and generates the tables on the fly. —CodeCat 18:30, 7 June 2014 (UTC)

I created both Module:languages/by name (which generates one table for all languages in one go) and Module:languages/data2/by name, and tested them both. Generating a single table for all languages takes about 600-700 ms on average, while doing it for data2 alone took around 20-30 ms. So it looks like the time spent generating the tables roughly scales with the number of languages to process (as one might expect), and there is no huge difference between generating one table for all languages at once and generating 29 separate tables for each of our data modules. Presumably the latter is slightly slower because there may be a bit more overhead involved, but it doesn't seem too significant.

However, if the table is generated in one go, then any page using it will always take a 600 ms hit on load time. If we use individual tables for each submodule, then it would be less overall because pages would generally not need to reverse-lookup one language from each of the 29 possible tables. So it would be faster in the end. The downside is that we'd have to copy the table generation script 29 times, which is less than ideal. —CodeCat 19:03, 7 June 2014 (UTC)

Not really, a generic generation script could be put on one page, while the 29 reverse-lookup pages would have stubs like return require('Module:languages/genrevtable')('Module:languages/data3/x'). This is not a big issue anyway, since I doubt that the reverse-lookup generation code would ever have to be revised. — Keφr 08:44, 8 June 2014 (UTC)

I may be reading this wrong though, because the time spent in Lua includes the time spent loading Module:languages/data2 and all the other data modules themselves. Every page will always load at least one of the data modules for other purposes, but probably many more, particularly for pages that use many different codes (translation tables or descendant trees for example). So that time would need to be subtracted. —CodeCat 19:07, 7 June 2014 (UTC)

I tested it again and it appears that loading all of our language data modules alone (via Module:languages/alldata) takes about 200-300 ms. Subtracting that from the total time spent creating the reverse tables leaves about 300-500 ms. —CodeCat 19:12, 7 June 2014 (UTC)

And I found out while working on Module:Unicode data that looking up phantom tables themselves tends to be somewhat expensive (hence all the memoisation there). — Keφr 08:44, 8 June 2014 (UTC)

Given that the language data modules aren't changed that often, it's a shame there's no way avoid using all those resources to recreate the lookup table on the fly with every page view. I wish there were a way have updating of a module-based table triggered by saving changes to one of the data modules, or to have something analogous to doing a null edit on a template to make it reflect changes to the templates it transcludes. If there were a simple, automated way to update the lookup table, we could rely on the admins who edit the data modules to do so. Chuck Entz (talk) 20:33, 7 June 2014 (UTC)

I've now deployed the basic all-language table in Module:languages, and changed the three templates that used {{langrev}} to use it instead: {{ttbc}}, {{trreq}} and {{langcat}}. It looks like it works well, or at least well enough to be usable. I think we can delete {{langrev}} now? —CodeCat 14:29, 10 June 2014 (UTC)

After advertising the switch from the template to the Lua version for a little bit, yes. (@msh210 in particular as a user of the template.) The Lua langrev is inherently up-to-date (unlike the template), and can be used directly and even substed...yay! I've added some documentation on how to use it. - -sche (discuss) 15:50, 10 June 2014 (UTC)

Thanks for the ping. How do I use the module instead (the way I use langrev, viz with subst)?​—msh210℠ (talk) 15:58, 10 June 2014 (UTC)

If you know a language's code (for example, "en") and you want to find out its canonical name, you can use this:

• {{subst:#invoke:languages/templates|getByCode|en|getCanonicalName}} (returns "English")

If you know a language's canonical name (for example, "English") and you want to find out its code, use this:

• {{subst:#invoke:languages/templates|getByCanonicalName|English|getCode}} (returns "en")

- -sche (discuss) 16:04, 10 June 2014 (UTC)

It's actually just {{subst:#invoke:languages/templates|getByCanonicalName|English}}. I made it so that the templated version returns the code directly, rather than call functions on the language object. I did this because it should be possible to determine whether a canonical name exists or not (invocation returns empty string if not). I don't know if I should have done it this way though, there may be a better way that makes more sense. —CodeCat 16:18, 10 June 2014 (UTC)

Thank you both. Looks like it works (without |getCode; I didn't try it with).​—msh210℠ (talk) 05:02, 11 June 2014 (UTC)

The template {{langrev}} can probably be deleted either now or in a reasonably short time (or perhaps redirected), but the subtemplates definitely can not, as they are used directly by WT:EDIT in a manner that can not be replaced by the new method. --Yair rand (talk) 05:05, 11 June 2014 (UTC)

Why can they not be replaced? —CodeCat 10:03, 11 June 2014 (UTC)

Replaced with what? — Keφr 11:01, 11 June 2014 (UTC)

Comment: A couple of years ago, I copied the trans adder (WT:EDIT) over to de.Wikt and tried get it to work there. Differences in entry format — such as the presence on en.Wikt and absence from de.Wikt of glosses atop trans tables, and of separate trans tables for separate senses — proved to be the main hurdle; once I set up a test entry with a modified, en.Wikt-style, gloss-having trans table, the trans adder worked, though it didn't move as quickly as it does here. de.Wikt has never had a Template:langrev or any subtemplates thereof.
The subtemplates of langrev add some peripheral functionality to WT:EDIT, pertaining according to its documentation to "autocomplet language names", a feature which takes enough seconds to kick in that I always finish typing the language name before it does (in the rare instances I'm not just typing in the language code), except just now when I intentionally typed "Irish" very slowly into ] to confirm that an autocomplete function did actually exist. WT:EDIT still works (core-functionality-wise) without the langrev subtemplates.
That's not an argument for deleting them, it's just to point out that they're not critical (which is good, because they're out of sync with Module:languages).
Can we find a way that WT:EDIT can search through the data modules, so that it can use them to autocomplete language names? Or should we have a bot periodically sync the subtemplates with the data modules? - -sche (discuss) 17:14, 11 June 2014 (UTC)

Hi! I am Ready Steady Yeti, and if you do not know me yet, I have been here for a few months, am really liking this project, and am making English and Danish entries, which I cannot say are perfect but I try the best I can because I actually have a goal for the "completion" of this project which can be viewed on my user page.

I have quite a history with the Danish language. I've self-teaching myself the language, and have been for almost 2 years now. I have been liking language since actually about 4 years ago when I started learning Spanish in school (which I do not speak anymore, because I completely ditched Spanish for Danish). None of my family or acquaintances know Danish or are from Denmark. I live in a southern US state, but I cannot remember exactly why I originally decided to learn Danish; I believe I read an article on the internet that was in Danish and it piqued my interest, but I'm not exactly sure if that's how it started. I have many methods of learning the language. I have 2 Danish learning books, and several regular books that were printed in Danish, such as Volmer: portræt af en samfundsstøtte. I also always use google.dk, and everything I can do online I do in Danish. I also chat with Danish people online a lot in Danish, which helps me greatly to be able to write in the language. Danish is a beautiful language (just my opinion). This is why I continue to learn it.

Anyway, as far as speaking Danish, I can write and type in a fluency of I'd say da-3, since I know more than just simple communication, but also some advanced words. My grammar is often very bad, but I do know the vocabulary at an advanced level. My speaking, as in like talking in real life, is unknown. I have no idea how well my Danish pronunciation is because I don't really have anyone to comment, but I listen to Danish music, videos, and other media, so I can kind of guess at how they talk. As far as speaking in real life though, it's much harder to understand Danish people when they're actually speaking than when they write. So I don't know if da-3 is an appropriate tag for me, but I use it because I'm able to write at that fluency, and that's all that really matters here, because we never speak to each other on wikis.

I'm not saying I'm an expert. I still have a lot to learn in this field.

So anyway, yeah, as for the real discussion. I have been looking at the Danish Wiktionary for a while now, as I am trying to see if they have articles for a word I add or not (which they usually don't but occasionally they do, which is why I'm starting this discussion). Well the Danish Wiktionary has a massive load of problems that I will introduce to you now.

Well the first and foremost problem that I see that needs to be fixed more than anything else is the Danish Wiktionary itself and its entire layout. In its entry layout, it is nothing like it is here. It horribly represents pretty much all section headers using templates, such as { { -noun- | en } }, for the only purpose that I see of helping to categorize words, which we can do on our own here without those ridiculous templates. I mean don't you think that this layout is a bit too complicated for some people to understand, such as newcomers? I say, because I think very technically and I've been using the internet for a long time (since I was 2 years old to be exact), I think I know how to understand the source code. But this layout is just completely unnecessary not just because of newcomers not being able to understand it. It needs to be like our layout, and by the way instead of using the common { { head } } template they use silly things like { { pn } } to represent a noun, and I don't even know what the rest of their "heads" are. It is ridiculous. And they barely even use plural templates. I think that we should make their layout just like or very similar to our layout. They also have a huge lack of proper templates for lots of things, such as word forms, which is part of the reason why they have almost no entries at all for word forms, which I will get to later in this announcement. We should have the entry layout changed, first of all, having bots remove all the ridiculous messy layout that they have and replacing them with a layout that is not alienated to everyone besides the sysops. Secondly, we need new templates, such as the head template and word form templates for all languages we can, imported, and if necessary translated as well, straight from the English Wiktionary, and I somehow don't even think we need bots for this, but we could use bots if we wanted.

The second major problem that the Danish Wiktionary has (and trust me, many Wiktionaries have this problem actually), is that it has a huge lack of content. If you know what I mean by that. I will explain. The basic words that every Wiktionary needs are the basic words, at the very least, for their own language, which the Danish Wiktionary lacks a lot of those basic words. Like why is da:den missing? I'm sorry but that is just ridiculous. Besides that, as for the fact that Denmark's second language is English, it's English is lacking in the same way. Danish Wiktionary has certain specific words that the few users that they have have added. But what about the Malagasy Wiktionary? Despite the fact that Malagasy is a very rare and uncommonly known language, the Malagasy Wiktionary, thanks to User:Jagwar, is the second most successful Wiktionary there is (we are first, remember that?). As far as I know, he used bots to manipulatively import entries from other Wiktionaries and translate them, and is doing this very successfully, because this Wiktionary has an unbelievable amount of words from not only its own language but from various other languages as well. I think someone needs to do this same thing to the Danish Wiktionary, use bots to import entries straight from other Wiktionaries and have some method of translating them. The language that is specifically needed more than anything else are Danish entries. We could import some entries for Danish terms from other Wiktionaries, and we should also use online ordbogs (Danish dictionaries) to use to add Danish terms in the Danish Wiktionary. Second most important language that must be there is English. I've done some entries on my own in English, but it's a pain in the ass to have to do it all manually you know. So we should import all English entries straight from here including word forms, meaning we'd have to make the proper templates, once again. Thirdly, I'd say we should focus on importing all possible terms (and worm forms, you know the drill) from the following languages:

• Swedish
• Norwegian (Bokmål and Nynorsk)
• German
• Icelandic
• Faroese
• Greenlandic

I say these languages because they are languages for words that I'd imagine Danish speakers would commonly want to look up on a site like Wiktionary. Additionally, since Wiktionary is for all words in all languages, we should import more of these common languages as well:

• French
• Spanish
• Portuguese
• Italian
• Dutch
• Polish
• Japanese
• Chinese
• Arabic
• Hindi

And since ALL languages, after that we should import ALL possible languages, like rare ones like Malagasy and Slovenian and Romanian and Azerbaijani and all the other languages you can possibly think of.

I know this isn't exactly about the English Wiktionary (well it kind of is, since we would import from here). I honestly have no idea how this kind of stuff works. These are just my ideas. The point is, how do you think we can implement this for the Danish Wiktionary to have as many terms as possible. I really want to look up any term here and there be a "Dansk" link at the side. Thanks for the help. I appreciate it. Ready Steady Yeti (talk) 21:27, 8 June 2014 (UTC)

First of all, the number of entries is only one measure of success- the entries have to have content that is correct, usable and useful. While there's considerable debate right now over whether the type of definition-less entry that would result from a mass import is better than no entry at all, I don't think anyone would claim that such an entry begins to compare with an entry that's been prepared by a human being who knows the language. Although it might have more entries, I don't think the Malagasy Wikipedia is anywhere near to being in the same league with the German, French, or Russian Wiktionaries (to name just a few).

Read through the current debate in the Beer Parlor about the mass creation of Ukrainian entries, and you'll see that it's not an unmixed blessing, and not the kind of thing to do without a consensus. In all of your discussion of Danish Wiktionary, I see one thing lacking: any mention of what Danish wiktionarians think about the idea. It may not have as much content as we have, or as many editors, but there are people there who have put in considerable time and effort to get it to where it is now. They deserve to have a say in how- or whether- this should be done. If you want to accomplish anything, you need to get them to sign on.

You're young and energetic, and you have some big ideas. On the one hand, there are amazing things accomplished by people who were too inexperienced to know that they were impossible. On the other hand, fools rush in where angels fear to tread: we've been burned badly by many prolific editors who knew little about the languages they were editing, but plowed right ahead anyway- we're still repairing the damage, years after they were banned.

Besides, getting things done by consensus can be slow and sometimes maddening, but the alternative is alienating the people who do all the work and who are the life's blood of a wiki. Chuck Entz (talk) 22:55, 8 June 2014 (UTC)

I saw a module error caused by using the "lang=diq" in the {{wikipedia}} template, so I thought I would fix it by using the the code that we accept, zza. The only problem is that there was no conversion of our zza code to the diq code used by the Zazaki Wikipedia, so it ended up linking to a "not found" page at English Wikipedia.

I could probably trace down the place where such conversions are set, but I thought it better to leave it to those who deal with such things in case there was some factor I wasn't aware of. As it turns out, Zazaki Wikipedia has no page with that particular spelling anyway, but it would be nice for editors to be able to link to any pages that do exist. Thanks! Chuck Entz (talk) 21:55, 8 June 2014 (UTC)

This should do the trick. - -sche (discuss) 22:07, 8 June 2014 (UTC)

Thanks! I knew it would be fairly simple, but I didn't want to blunder in without knowing why things are set up the way they are. Chuck Entz (talk) 22:22, 8 June 2014 (UTC)

No problem. I've added some documentation to WT:LANGCODE in bid to increase institutional knowledge of the workings of Template:wikimedia language. - -sche (discuss) 22:48, 8 June 2014 (UTC)

If a language has automatic transliteration, {{transterm}} incorporates it:

• {{transterm|слово|lang=ru}} → =

But if a language does not have automatic transliteration, {{transterm}} does not allow a transliteration to be specified manually:

• {{transterm|מִלָּה|tr=mîllá|lang=he}} → =

(And if a language has automatic transliteration but someone desires to override it, e.g. in the case of что, that is also not possible.) Can someone modify the template so that it does allow transliterations to be set manually? - -sche (discuss) 21:17, 11 June 2014 (UTC)

It now accepts tr= like {{m}} and {{l}} do. —CodeCat 21:47, 11 June 2014 (UTC)

Thanks! - -sche (discuss) 21:51, 11 June 2014 (UTC)

While previewing this edit before making it, I got the following messages, in red (the first one is normal, it's the other two I find interesting):

Remember that this is only a preview. Your changes have not yet been saved! → Go to editing area

Warning: Template include size is too large. Some templates will not be included.

Warning: This page contains at least one template argument that has a too large expansion size. These arguments have been omitted.

The category is also in several automatically-added cleanup categories, viz. Category:Pages where template include size is exceeded, Category:Pages containing omitted template arguments and Category:Pages with script errors. (Category:Latin orthographic variant forms is also in these categories.) Anyone know what's causing this and how to fix it? PS, would it be better grammar to have the error message say "an expansion size which is too large"? - -sche (discuss) 23:02, 11 June 2014 (UTC)

It looks like it has something to do with cases where the label ("preverb") does not exist. That can probably be handled better, I'll have a look. —CodeCat 23:11, 11 June 2014 (UTC)

In any case, a short-term fix would be to either add these categories to {{poscatboiler}}, or to remove the template from them. —CodeCat 23:17, 11 June 2014 (UTC)

After creating the 𩙿 (U+2967F) article, I copied its "Japanese" (L2) section from the 飠 article as the Japanese information is the same. For some reason, the "kun=" (Japanese kun form) reading information isn't displaying for me (in two different browsers). I don't know if it has something to do with the character being in the CJK Unified Extension B range or if it's just some other miscellaneous bug. Bumm13 (talk) 06:51, 12 June 2014 (UTC)

Thanks to whoever fixed this! :) Bumm13 (talk) 12:52, 16 June 2014 (UTC)

For some reason, User:Conrad.Irwin/editor.js doesn't work for Cantonese, even though checking for Interwiki seems to be disabled (or was disabled). E.g. on elder_sister#Translation trying to add 姐姐 (ze4 ze1). Clicking "Preview translation" has no effect. "Loading..." shows indefinitely long. --Anatoli ^{(обсудить}/^вклад)

Checking for interwikis appears to be on. Cantonese has the Wiktionary prefix listed as "zh-yue", but https://zh-yue.wiktionary.org doesn't exist, which is probably what's causing the problem. --Yair rand (talk) 07:20, 12 June 2014 (UTC)

Removed the {wiktprefix:"zh-yue"}, seems to be working now. --Yair rand (talk) 07:22, 12 June 2014 (UTC)

Thank you! --Anatoli ^{(обсудить}/^вклад) 07:28, 12 June 2014 (UTC)

I tried to use the wizard provided to add a usex on авіаза́сіб (aviazásib). That's the result in this revision:

1. Template:Cyrl

   aviazásobom

   by air

I meant this:

1. авіаза́собом ― aviazásobom ― by air

Can this be fixed? Besides, some languages don't require transliteration, this should be optional. The wizard forced me to add transliteration. --Anatoli ^{(обсудить}/^вклад) 23:57, 12 June 2014 (UTC)

Transliterations are no longer required by the script. The template {{lang}} is now used instead of the deleted script templates. Do we no longer have any sort of standard for example sentences? ELE still says that both translations and transliterations are supposed to be on their own lines, below the sentence itself. --Yair rand (talk) 22:47, 15 June 2014 (UTC)

Thank you but the template should use the language codes and automatic transliteration. As for transliteration on their own lines, the rule must be out-of-date, also the rule that words must not be linked. Inline usexes are also used quite heavily.--Anatoli ^{(обсудить}/^вклад) 23:27, 15 June 2014 (UTC)

Can someone please either unprotect this page or change its text from "#redirect ]" to "#redirect ]"? Thanks. —Mr. Granger (talk • contribs) 14:05, 14 June 2014 (UTC)

 Done; thanks for noticing the issue. - -sche (discuss) 14:08, 14 June 2014 (UTC)

Since these are being cleaned up/de-sphagettified and Luafied, thanks to CodeCat, I had an idea to throw in with all the other changes:

Right now, many of the category templates display a row of wikilinked breadcrumbs representing the category tree the category belongs to. When one of those is a redlink, clicking on it goes to the usual empty edit window, at which point one has to enter the correct template with the correct parameters- with very ugly results if one gets so much as a single character wrong.

It occurred to me that the template that creates those wikilinks should have all the information necessary to preload any redlinks with the correct template with the correctly-spelled parameters already included. This would change creating the categories from an advanced test of typing skills and of knowledge on how to find resources on Wiktionary (quick- what's the code for Benue-Congo?) to a no-brainer.

Is this idea workable, and can we do it? I'm not talking about today, but whenever the rework gets around to that part of the interface. Thanks! Chuck Entz (talk) 19:02, 14 June 2014 (UTC)

I certainly think it's feasible, especially as we already have a lot of the code for it, it would only need some modification. I think it's a very good suggestion too. —CodeCat 19:05, 14 June 2014 (UTC)

When I search a latin verb, the inflection table that used to come up doesn't anymore. for example, for the verb divido, when I scroll down to Latin-Inflections, it says "Module error" in red text. When I click on the "Module error", the following error is shown: "Lua error in Module:la-verb at line 234: attempt to index field 'subtype' (a nil value) Backtrace: Module:la-verb:234: in function "get_regular_stems" Module:la-verb:298: in function "?" Module:la-verb:31: in function "chunk" mw.lua:478: ?" Can someone please look at this? Thanks. — This unsigned comment was added by Ws04 (talk • contribs).

The error may have been introduced in diff (the module is called by the inflection templates), I can't be sure. - -sche (discuss)

Wyang (talk • contribs) seems to have fixed the main problem. There are still 13 entries (for now- there may still be edits working through the queue) that didn't clear up after a null edit, and there are 5 conjugation templates that also are still showing module errors after null edits. Chuck Entz (talk) 03:38, 16 June 2014 (UTC)

It has been brought to my attention that the chapter field in the quote-book template puts its value into the displayed quote preceding the book's title. It would be a lot better if it were to appear after the title. This should be fairly easy to arrange, and I would be grateful if someone would arrange it. ReidAA (talk) 04:29, 16 June 2014 (UTC)
Note that the value of the chapter field should be shown after that of the section field. ReidAA (talk) 04:43, 16 June 2014 (UTC)

Now at ], where it belongs. Striking here.​—msh210℠ (talk) 07:21, 20 June 2014 (UTC)

Discussion moved from Template talk:term#Automatic transcription appears to override manual transcription?.

Compare the following attempts to refer to Moksha эльде:

• эльде (eľďe) (no transcription set; outputs elʹde)
• эльде (eľďe) (transcription set to äĺďä; still outputs elʹde)

Also, from a quick look around, automatic transcription is apparently language-specific? OK, someone needs to turn it off of Moksha anyway since the Cyrillic orthography fails to distinguish /ɛ/ and /e/. But this override thing is a clear bug at least. --Tropylium (talk) 13:52, 16 June 2014 (UTC)

This is definitely a bug that needs fixing. As for Moksha itself, I don't know much about it but distinguishing between two sounds is not always necessary in a transliteration. --WikiTiki89 14:11, 16 June 2014 (UTC)

Not a bug but a feature, Module:mdf-translit uses standard Moksha transliteration. Moksha doesn't need manual transliteration. Tropylium, if you wish to use a different standard, you need to explain why and what system you propose to use. Here's the current standard WT:MDF TR. --Anatoli ^{(обсудить}/^вклад) 14:30, 16 June 2014 (UTC)wiki

Do you think it's not a bug, it's a feature is idiomatic? — Keφr 14:58, 16 June 2014 (UTC)

I don't know but I meant it literally. --Anatoli ^{(обсудить}/^вклад) 22:01, 16 June 2014 (UTC)

What Anatoli said. If you want to change the Moksha transliteration rules, please start a discussion on the WT:MDF TR page, preferably linking to scholarly sources documenting your scheme (I am not saying your scheme is necessarily worse than the current one). --Vahag (talk) 14:36, 16 June 2014 (UTC)

Regardless of how perfectly correct the automatic transliterations are, there should always be a way to override them. Even if they should not be overridden in the main namespace, it may be useful in discussions. --WikiTiki89 20:13, 16 June 2014 (UTC)

I think it was decided some time ago that for certain languages the transliteration was "perfect" and should therefore override any manual ones, as they were occasionally incorrect. —CodeCat 20:23, 16 June 2014 (UTC)

Maybe you should re-read what I just said. --WikiTiki89 21:51, 16 June 2014 (UTC)

For discussions one could avoid using templates, which use automated transliteration. The automated transliteration could be turned off if a language can't use it but it's not proven that it's the case with Moksha. --Anatoli ^{(обсудить}/^вклад) 22:01, 16 June 2014 (UTC)

But that's a workaround. If a language's transliteration is perfect, then we could just remove all the erroneous uses of tr=, but we should still be able to override it when we want to. Templates should be as flexible as possible so that they can serve a wide variety of expected and unexpected uses. Making them intentionally restrictive is pointless and annoying. --WikiTiki89 22:17, 16 June 2014 (UTC)

But that's why there are languages, for which automatic transliteration overrides manual, since there are too many existing transliterations and users may continue adding non-standard translit. A bot can remove but someone may re-add. As I said, there were no reasons stated for the current transliteration to be overridden. If Moksha has exceptions, which merit manual transliterations, they should at least be mentioned. The original poster wasn't complaining about exceptions but our methods, if I understand correctly. --Anatoli ^{(обсудить}/^вклад) 22:27, 16 June 2014 (UTC)

Users could just as easily add other junk to a page as an incorrect transliteration. And I already said that the override should probably not be used in the main namespace, but who knows, a reason may come up even for that. Foresight is not 20/20. --WikiTiki89 22:59, 16 June 2014 (UTC)

I'm not sure you understood what I meant but anyway, let me clarify. We only have one current transliteration per language. You know that I am pro-phonetic transliteration for languages with exceptions. With Cyrillic languages it's only Russian and Mongolian, which are not 100% phonetic and its a dictionary style to use word stresses for Slavic languages (Macedonian word stress is about 90% predictable). If word stresses are inserted for all Slavic languages, then manual transliteration would only be required for exceptions but majority of languages might rely on 100% automatic transliteration. With Mongolian (Cyrillic) we have neither native speakers nor enough info to cater for exceptions, so Cyrillic Mongolian transliteration overrides manual (for the lack of a different solution or lack f information). We assume that Moksha has no exceptions, i.e. pronunciation may be derived from the script, if phonetic rules are known. Tropylium (the original psoter) didn't specify that "э" or "е" may have different readings in various words/positions, he/she was just not happy with the way both letters are automatically transliterated as "e" when palatalisation is not marked - like in Russian, consonants must be palatalised in Moksha in front of "е", so "д" in эльде must be . Note that many casual Russian contributors use "'" to mark palatalisation, like "d'ér'evo" for "де́рево"

Korean and Japanese kana now have fully phonetic transliteration, which is phonetic (almost automatic with Japanese, 100% automatic with Korean), Lao is almost 100% predictable but has rare Thai borrowings with traditional spelling. Transliterating Thai automatically would be much more challenging and manual override should be allowed. North Indic languages, apart from Sanskrit have exceptions, some predictable (dropping inherent "a"), some are not (letters without nuqta when it is implied), so 100% automatic transliteration won't work. --Anatoli ^{(обсудить}/^вклад) 00:50, 17 June 2014 (UTC)

Then you misunderstood what I said. Occasionally it might be useful to use a non-standard transliteration scheme, especially on discussion pages, for some languages even in etymology sections. There is no reason for the template to prevent you from doing so. --WikiTiki89 01:14, 17 June 2014 (UTC)

That transliteration, if it's different from our default (Wiktionary standard) should be specific, perhaps, tr2= or named transliteration, such as "yale=", with params such as sprs=y (suppress=yes) or something but I don't think it's necessary, if the discussion is not about the word but transliteration, then just type it or copy paste, for example "hangul" is deried from McCune–Reischauer "han'gŭl", you don't have to use "tr=" here, just use the literal "han'gŭl", using templatised 한글 (han'geul) will only show currently accepted standard transliteration for Korean, no need to force it to show "han'gŭl", unless the Korean module couldn't handle transliteration correctly. Why would you otherwise, use a transliteration, which is not approved or default? --Anatoli ^{(обсудить}/^вклад) 01:26, 17 June 2014 (UTC)

tr2= and yale= don't exist. If you want to make the templates more complicated by adding them then go ahead, but the better solution is to just allow overriding. Yes, there is always the workaround of not using the template, but the templates are more convenient (which is what they are for). --WikiTiki89 03:06, 17 June 2014 (UTC)

You can't help without complicating, if we allow overriding, then we get what we had before - thousands of non-standard transliterations, which was especially striking with Armenian, Georgian and many other languages. Vahagn asked for override for a reason. tr= calls automatic transliteration, if exists and there is only one method per language. In most cases, the transliteration used is the one that's required. "äĺďä" is not the current standard transliteration for Moksha "эльде". To use transliteration other than standard we should use other (new) templates, which could take tr= or rom= or don't use templates at all. For example, {{Zhuyin}} converts pinyin "lìshǐ" to Template:Zhuyin.

One could make specialised templates for any additional romanisation or conversion. --Anatoli ^{(обсудить}/^вклад) 03:24, 17 June 2014 (UTC)

• I agree with Wikitiki89. People are smarter than machines, and when machines get in their way, people are good at working around them. If someone is adding a word, and really feels that a certain transliteration should be used for that word, then I'd rather let them (provisionally) use that transliteration; and if we think they're wrong, then we should discuss with them and reach consensus. Otherwise, there are three possible outcomes: (1) we get our way, by mechanized bureaucracy and faceless bullying; (2) they get their way, by circumventing our template system, discarding all language-tagging, etc.; or (3) neither of us gets our way, because they discard their edit and don't add the word. (I'm sure some of you will be happy with outcomes #1 and #3, and will accept outcome #2 because you're sure that you can devise smarter machines that will patch that hole as well; but I'm hoping that most Wiktionarians still have the good sense to recognize that all three of these are bad outcomes.) —Ruakh_TALK 04:05, 17 June 2014 (UTC)

• Only enable manual translit for these languages when not in main namespace, and turn off the transliteration field for these languages in translation tables. Transliteration should be consistent when presented to readers. Wyang (talk) 04:24, 17 June 2014 (UTC)

While not totally disagreeing with Ruakh or Wikitiki89, I'd like to try to rephrase what I said before. The rationale for the override was that manual transliteration was redundant as a minimum, non-standard or wrong as a maximum. No transliteration can ever please everyone and you both know it. Now, with cases when we do really need to use a different transliteration - there are three reasons, in my opinion 1. because we don't like it - suggest to use a different system, 2. exceptions - languages with exceptions allow manual override, 3 - I accept the system but I want to use something else still - don't use templates or asks for a different conversion template. Does it sound fair? The first and the third reason are in conflict with each other, therefore there must be an additional parameter/template for this. @Vahagn Petrosyan could you show some examples of bad transliteration, I don't remember.

@Wyang. Yes, it sounds reasonable but someone may still want to use automatic transliteration in discussions, so again - multiple templates or parameters may be the way to go. E.g. if tr2= is specified, tr= is suppressed or something similar. --Anatoli ^{(обсудить}/^вклад) 04:34, 17 June 2014 (UTC)

There are two types of automatic transliteration: 1) when |tr= is empty, all languages with a specified translit_module in Module:languages/data2 etc. will be transliterated; 2) when |tr= is non-empty, languages specified in override_translit in Module:links will have the |tr= value overriden by the automatic transliteration. We could just further restrict the second condition, making it happen only if the page is in the main namespace. Wyang (talk) 04:39, 17 June 2014 (UTC)

I know but (I repeat) someone may still want to use automatic transliteration in discussions. One such application is, especially important, IMHO, is Wiktionary:Translation_requests but not only, automatic transliteration is helpful in many cases for people not familiar with native scripts, even outside the main namespace, e.g. if I want to say that Korean 반장 (banjang) (one of the senses) is related to Chinese 班長／班长 (bānzhǎng) - the automatic transliteration immediately shows phonetic resemblance. --Anatoli ^{(обсудить}/^вклад) 04:57, 17 June 2014 (UTC)

Making automatic transliteration not overridable still generates the automatic transliteration when there is no transliteration specified. The automatic transliteration is not displayed only when a transliteration is specified. Wyang (talk) 05:28, 17 June 2014 (UTC)

I know but for some languages it may be preferable to override. Even if I specify a non-standard transliteration "sri lankāva" for ශ්‍රී ලංකාව (śrī laṁkāwa), it will still show the correct "ś‍rī laṁkāva" because it has override_translit. --Anatoli ^{(обсудить}/^вклад) 05:37, 17 June 2014 (UTC)

I actually think it would be far better for Korean to have override_translit because translations may soon be loaded again with Yale, McCune–Reischauer, etc. --Anatoli ^{(обсудить}/^вклад) 05:40, 17 June 2014 (UTC)

• WikiTiki89, Ruakh, you do not get to complain until you clean up thousands of entries in Category:Terms with manual transliterations different from the automated ones/hy, Category:Terms with manual transliterations different from the automated ones/ka, etc. and then pledge to monitor the categories for the rest of your lives. I do not intend to waste my time so that some пижонs could have the option of playing with transliteration on their talk pages. --Vahag (talk) 08:09, 17 June 2014 (UTC)

• I would tend to agree with Ruakh that we should allow manual override of automatic transliterations, except that I can see two big, connected problems with doing that, which Vahag hints at. (1) Not all languages have automatic transliteration, so users may supply manual transliterations not because they want to override the automatic ones in a particular case, but because they don't realize a transliteration would be provided automatically, and they think they have to set one manually. To catch such instances, we could (as I think we already do) category cases where manual transliteration overrides automatic transliteration. But (2) if there are more than a few dozen or hundred instances where we've intentionally overridden automatic transliteration, it will probably be difficult to spot new additions to the category. For example, if there are 250 words in Language X where we want to override automatic transliteration, 250 words out of, say, 300 000 words in the language is such a small percentage that it doesn't stop it from being worthwhile to turn automatic transliteration on for that language, but 250 is a large enough number that spotting the newcomers when the number of entries in the category goes up to 252 will be nontrivial.
  To address the first issue, we could use a tr2= parameter as was suggested above, so that it's harder for people to override manual transliterations without realizing that's what they're doing. To address the second issue, I suppose we could maintain an appendix of some sort of 'approved' cases of transliteration-override, and have a bot work from that to find new cases of transliteration-override... is that feasible? Is that worthwhile? - -sche (discuss) 15:04, 17 June 2014 (UTC)

OP here again. OK, I follow the concern that allowing too much manual transcription leads to a lack of standardized transcription — though also the counterargument that entirely disallowing this leads to people looking for workarounds. What's even the point of having a tr= field, if it's not actually usable? The suggestion of a tr2= might be a decent compromise.

(Also FWIW, this all really ought to be explained in the relevant template documentation.)

Contra Anatoli upthread, I am however indeed mainly complaining about being unable to render phonemic distinctions that the native script does not capture! (Specifically for the purposes of an etymology section or appendix.) I have by this point one main question: who decided and where that "Moksha doesn't need manual transliteration" or that "all languages written in the Cyrillic alphabet, other than Russian and Mongolian, are 100% phonetic"?? This is nonsense that is contradicted for plenty of few minority languages of Russia. E.g. Tundra Nenets, whose orthography does not distinguish /i u/ from /iː uː/; Mansi, whose orthography does not distinguish /e/ from /ə/; Erzya, Udmurt, Komi and Mari, whose orthography uses iotated vowels for both /ʲV/ and /jV/; etc. These are basic facts that anyone familiar with the languages in question ought to know. I can dig up sources if necessary, but I'd like to see where this all came from, first. If it was decided by "some пижонs" who knows nothing about the languages, spending a day or two digging up sufficiently explicit sources really should not be necessary for overturning this.

Though I guess there's also a transliteration (orthographic) vs. transcription (phonetic) issue here. I wouldn't mind automatic transliteration at all, if there were a way to independently indicate phonetic transcription. --Tropylium (talk) 12:31, 18 June 2014 (UTC)

I didn't say that any language is 100% phonetic but the scripts were made with the pronunciation already in mind, even if they don't match exactly the native pronunciation. In Russian, there are traditional spellings and most minority languages just didn't have such a long Cyrillic spelling tradition to have a traditional spelling. It's true that there are big differences between the spellings and the pronunciation but that's with phonetic transcription in IPA, not transliteration, isn't it? Long vowels are still transliterated as short vowels. Do you suggest to render iotated and palatalised consonants with different symbols? If yes, then what transliteration system is that? If you only need this for a demo, it may not work. As I said, we don't have people to work with Mongolian Cyrillic, so Mongolian automatic transliteration overrides manual, regardless of non-phonetic spellings. Note that with Russian, which has expected sound changes - voicing, devoicing, vowel reduction, consonant assimilation, deassimilation, etc. - are not rendered with special transliteration (molokó, not malakó, zub, not zup, etc.), only unpredictable exceptions are transliterated manually and those exceptions are documented, categorised and looked after. --Anatoli ^{(обсудить}/^вклад) 13:06, 18 June 2014 (UTC)

• Late to the party but User:Gilgamesh~enwiktionary's mistaken/idiosyncratic treatment of ⟨Ξ⟩ as ⟨ks⟩ instead of ⟨x⟩ in violation of all standard systems is a reminder that, no, your systems are not perfect; this is a bug not a feature; and the template needs a manual override. — LlywelynII 21:58, 27 July 2015 (UTC)

The headers added by Template:fake== and Template:fake=== still used the old (sans serif) font, and so don't match actual L2 and L3 headers, which use serif font. - -sche (discuss) 04:47, 17 June 2014 (UTC)

They still use sans-serif for me. — Keφr 05:44, 17 June 2014 (UTC)

For me too, which is what -sche said. The "real" L2 and L3 headers use a serif font; the "fake" ones use a sans-serif. I can't say I find this particularly tragic, though. —Aɴɢʀ (talk) 08:43, 17 June 2014 (UTC)

No, the real headers use sans-serif. — Keφr 13:40, 17 June 2014 (UTC)

Ew. 2006 called; they want their skin back. —Aɴɢʀ (talk) 13:59, 17 June 2014 (UTC)

I prefer an ugly but stable skin to keeping getting annoyed with whatever UI changes the WMF feels like making. Which includes serif headers. — Keφr 17:20, 17 June 2014 (UTC)

The feedback tool disappears after you use it. Each section must be transcluded to its own talk page. Lysdexia (talk) 07:42, 17 June 2014 (UTC)

Yellow links don't work for me for some reason (ticked, save and hard-refreshed). Tried Firefox and IE. --Anatoli ^{(обсудить}/^вклад) 14:33, 17 June 2014 (UTC)

What generates yellow links? I get red, orange, maroon, two shades of blue, and black, but never yellow. DCDuring TALK 17:22, 17 June 2014 (UTC)

He probably means the orange links. What generates maroon links? I've never seen those. —Aɴɢʀ (talk) 20:37, 17 June 2014 (UTC)

He is probably referring to visited and unvisited links. --WikiTiki89 21:30, 17 June 2014 (UTC)

That was my thought, too, but then what generates the second shade of blue? Internal vs external links? Goodness, it's all so confusing! haha. - -sche (discuss) 21:54, 17 June 2014 (UTC)

brown leaf — Keφr 21:29, 17 June 2014 (UTC)

Orange, not yellow, sorry - WT:PREFS Color links to language sections that don't exist orange where they would otherwise be blue. --Anatoli ^{(обсудить}/^вклад)

A color that appears marooon on my laptop, but some less distinct color on my tower's screen, is what shows up for visited redlinks. And I forgot green. DCDuring TALK 22:04, 17 June 2014 (UTC)

There are quite a number of different colors that links can have.

No idea what's causing the problem. The script works for me on Firefox, so I suspect something might just be broken by your personal JS. --Yair rand (talk) 23:00, 17 June 2014 (UTC)

What are other prerequisites for the orange links to work? Yesterday I tested on Belarusian за́яц (zájac) (Russian entry existed but not Belarusian). If an entry exists but in a different language, it should be orange. Is that a correct assumption? Using a different word, since заяц has been added: абза́ц (abzác) the link appears blue for me, not orange and it's a different computer. --Anatoli ^{(обсудить}/^вклад) 23:42, 17 June 2014 (UTC)

The default font that is used for Wiktionary content has Cyrillic characters as well, but actual Cyrillic content is displayed in Helvetica instead. I don't think it looks very nice when Cyrillic is shown in a rather different font, it would look better if they matched. How could this be done? —CodeCat 20:22, 18 June 2014 (UTC)

I agree that Cyrillic should use the same font as the rest of Wiktionary if that font accommodates Cyrillic. While we're on the topic, the formatting imposed by {{m}} now puts Cyrillic in italics, e.g. {{m|ru|гит}} currently displays as гит (git), looking like sum but with the s in mirror writing. That's not too bad as far as I'm concerned (I know other people object to italic Cyrillic), but Early Cyrillic (Cyrs) is also displaying in italics, e.g. {{m|cu|гит}} currently displays as гит (git), which looks like crap. —Aɴɢʀ (talk) 20:35, 18 June 2014 (UTC)

I reinstated the italics just now. I don't know if many people will disagree with that, but I feel that if Cyrillic italics confuse people, then those are probably the people who are not native readers of the script. And those are exactly the people for which transliterations may be useful. Personally, I find Cyrillic italics very confusing too, but I still think we should have them for consistency with Latin script.

As for Old Cyrillic looking bad in italics, then I think we should look for better fonts, not remove the italics. It looks fine to me in any case. —CodeCat 20:39, 18 June 2014 (UTC)

FWIW, the previous discussion of Cyrillic and italics was Wiktionary:Beer parlour/2013/July#Scripts_and_italics. People seemed about evenly divided on the question of whether Cyrillic should be italicized or not. - -sche (discuss) 20:58, 18 June 2014 (UTC)

The issue here is (I think) that non-Slavic languages using Cyrillic script use certain characters that are missing in the default fonts. -- Liliana • 21:02, 18 June 2014 (UTC)

For such cases, we often fork off a more specific variety of the script, like for many Arabic script languages. Would that not work here? —CodeCat 21:04, 18 June 2014 (UTC)

That certainly applies to Old Church Slavonic, as the default fonts tend not to include characters like ꙑ, ꙗ, ѩ, and ѭ. As the English Wikipedia, our target audience is not native readers of Cyrillic, but learners of Cyrillic, who may have mastered the roman form of the letters but not the italic form; and even for those of use more familiar with the italic form, things like питати are confusing because it looks so much like numamu—and that's a word in a language that uses both scripts. You say that гит (git) looks fine to you; what does ѩзꙑкъ (językŭ) look like? On my computer, the Early Cyrillic forms are not true italics but merely obliques and look very bad and unnatural (as bad and unnatural as oblique kanji and katakana look). —Aɴɢʀ (talk) 21:55, 18 June 2014 (UTC)

Or maybe set it for Cyrl and then reset it for all Slavic languages? There should be much less Slavic than non-Slavic languages using Cyrillic script (ru, uk, be, bg, sr, mk - anything else?) -- Liliana • 09:44, 19 June 2014 (UTC)

Also Rusyn (rue). —Aɴɢʀ (talk) 10:22, 21 June 2014 (UTC)

Btw, it would be "sh" rather than "sr". Also, whether a language is Slavic or not is irrelevant. The real question is whether italics are used in that language. Here is an example of Mongolian in italics. --WikiTiki89 13:19, 21 June 2014 (UTC)

This is not about italics or not italics, it's whether the language needs special fonts. Slavic ones don't, but Mongolian for example has the letter Өө which is not present in contemporary fonts. -- Liliana • 13:27, 23 June 2014 (UTC)

Except that it is. All the letters in scripts using Cyrl are present in the fonts we use for Cyrl. --WikiTiki89 14:12, 23 June 2014 (UTC)

Yes, and now people are whining because Russian uses a totally different font from the default. This is what the whole topic is about! -- Liliana • 19:07, 24 June 2014 (UTC)

Russian does not use a different font from other (modern) Cyrillic languages. --WikiTiki89 20:21, 24 June 2014 (UTC)

It uses a different font from English though and that's the issue at hand! -- Liliana • 20:23, 24 June 2014 (UTC)

So then the question is whether font used for English supports the non-Slavic characters you are referring to. The font used for English on my computer is Arial, which does support those characters. The problem with Arial is that it does not use actual Italics, but just slants the letters, which I think is just poor style (this applies to English as well). --WikiTiki89 22:05, 24 June 2014 (UTC)

It also uses a different font than Old Cyrillic on my computer - Old Cyrillic appears in the same font as English. —CodeCat 21:38, 24 June 2014 (UTC)

Yes, I was referring to Cyrl, not Cyrs. On my computer, Cyrs shows up in a different font called "Bukyvede" that resembles old Cyrillic manuscripts, probably because I installed it, but I don't remember whether I did or whether it came with Windows 7. --WikiTiki89 22:05, 24 June 2014 (UTC)

Incidentally, at diff you also put Greek into italics, but the Greek alphabet does not distinguish between roman and italic letter forms like Latin and Cyrillic do. Greek fonts tend to be somewhat oblique, but Greek doesn't have italics in the usual sense (apart from careless desktop publishing, an environment where you can find anything in italics, even East/Southeast/South Asian characters, Arabic, Hebrew, etc., none of which have any business being italicized). Please at least restore the italic inhibition for Cyrs, Grek, and polytonic, because italics are completely inappropriate for those scripts. And please consider restoring it for Cyrl as well, since the last time this was discussed 11 months ago, there was no consensus to display Cyrillic in italics. —Aɴɢʀ (talk) 22:05, 18 June 2014 (UTC)

I support using the same font for Cyrillic and Latin. I also support Cyrl in italic (but not Cyrs) and I just enabled italic for Armn. If people can't read italic Cyrillic, let them look at the transliteration. A good Russian dictionary would use italics to mention a word, e.g. here. --Vahag (talk) 07:59, 19 June 2014 (UTC)

I just got a talk page message here about this. The screenshot they attached shows that the fonts being used for Old Cyrillic are very different from the ones I'm seeing. They complain that the font is now too small, whereas I see Cyrs in the same font as regular Wiktionary text, so to me the extra 25% size makes it look oddly large, while not having it in italics looks odd when other Cyrillic text is. I suppose this is a very old problem with web design. But if our goal is to make Wiktionary look right for everyone, we really need to do something about this. Everyone needs to see the same fonts at the same sizes. There's no way we will get everyone here to agree on font display if everyone sees something else in the first place. —CodeCat 12:43, 19 June 2014 (UTC)

My font is similar to the one in that image. I don't think it's necessarily too small, but being oblique definitely makes it very hard to read. —Aɴɢʀ (talk) 14:23, 19 June 2014 (UTC)

Italicised Cyrillic is quite common and standard and used for various applications, even it's not normally used for book or movie titles like e.g. italicised English. Not sure about non-Slavic languages but I can't imagine that Kazakh, Kyrgyz, Tajik or minority languages of Russia never use italics, since this is/was quite normal and common in the USSR and Russia and punctuations, styles and many other things were for minority languages languages were just copied from Russian. In the previous discussion on italicised Cyrillic I did admit that italicised Cyrillic is significantly different from normal and may cause problems to non-native readers of Cyrillic but I didn't say it's not used, just to be clear on the distinction. I haven't seen Old Church Slavonic or Old Russian written in italicised script, so it's OK to NOT allow italicised Cyrs. @Vahagn Petrosyan can Georgian also be italicised? --Anatoli ^{(обсудить}/^вклад) 03:34, 23 June 2014 (UTC)

I know very little about Georgian, Anatoli. FWIW, I looked at a couple of linguistics books in Georgian: they use bold to mention a term and never use italic. --Vahag (talk) 07:30, 23 June 2014 (UTC)

Angr wrote “our target audience is not native readers of Cyrillic, but learners of Cyrillic, who may have mastered the roman form of the letters but not the italic form....” Our audience does read English, but it is very broad and we cannot pigeonhole it so tightly. Some readers find any foreign scripts and even many Latin characters and diacritics completely inaccessible, so one could argue that everything should be romanized and neutered. On the other hand, some use many languages and scripts, and some at an advanced native level.

But we have chosen to render foreign text in its native form, and supplement accessibility with romanizations. So our native renderings should be as true to native form as possible, for integrity and consistency, and for didactic purposes. If we present a language in a stilted form, then we should add a warning to readers that they need to go elsewhere to see how it should really be used.

I don’t see the point in avoiding oblique Cyrillics because some letterforms are new to some people, because all the letterforms are new to many people anyway. —Michael Z. 2014-06-29 16:17 z

These principles are broken in the bizarro world of editing Russian of course, where editors transcribe the spoken word as if it were Chinese or something, and refuse to transliterate the written forms at all. —Michael Z. 2014-06-29 17:31 z

Anyone learning Cyrillic would find it beneficial anyway to learn Italic Cyrillic. --WikiTiki89 16:40, 29 June 2014 (UTC)

I created a vote: Wiktionary:Votes/2014-06/Allowing Cyrillic to be italicized. —CodeCat 17:54, 29 June 2014 (UTC)

As there has been no further discussion, I've started the vote now. —CodeCat 14:02, 12 July 2014 (UTC)

Lately I have been getting "Unresponsive script" messages regularly, 1-3 times a day.

The most recent message is got is:

Script: https://bits.wikimedia.org/en.wiktionary.org/load.php?debug=false&lang=en&modules=jquery%2Cmediawiki%2CSpinner%7Cjquery.triggerQueueCallback%2CloadingSpinner%2CmwEmbedUtil%7Cmw.MwEmbedSupport&only=scripts&skin=vector&version=20140617T182240Z:69

1. Am I the only one getting such messages?
2. Is this a sign of some kind of real problem?

It's a little annoying, but only requires that I click a button to stop the script, so, unless there is a real underlying problem, this does not require action. DCDuring TALK 22:29, 20 June 2014 (UTC)

I get an unresponsive browser every time I open RFV/RFD. But no message about scripts. Try with ?debug=true added to the URL. — Keφr 07:57, 21 June 2014 (UTC)

Never mind. Problem not limited to Wiktionary or even WMF. Maybe Firefox. DCDuring TALK 17:33, 25 June 2014 (UTC)

Category:Candidates for speedy deletion now contains several categories full of templates that need to be deleted. I have no idea how to delete them, though. There's just way too many. Is there a way to mass-delete pages? —CodeCat 16:46, 21 June 2014 (UTC)

Run a bot on your account (having granted yourself the bot flag beforehand). Otherwise no, apart from Special:Nuke, which seems not to be applicable here. — Keφr 18:08, 21 June 2014 (UTC)

Apostrophes in {{IPA}} are now showing up as superscript j's. See for instance iri#Esperanto and bona#Catalan. I assume this is related to the edits that were made to the template in the past couple of days. —Mr. Granger (talk • contribs) 23:29, 21 June 2014 (UTC)

I’ve removed the code that is causing this until we figure out what to do. It’s not really the code’s fault, because iri, bona, etc. should be using ˈ (IPA stress mark) and not ' (dumb apostrophe), which is XSAMPA for ʲ. — Ungoliant ^(falai) 23:41, 21 June 2014 (UTC)

I imagine we may have to leave the '→ʲ code out permanently, because the misuse of apostrophes in place of stress marks has been common at all stages of Wiktionary's history. Incidentally, I've tried to compile all possible pronunciation-section problems into one list here (linked to from WT:TODO), in case any can be fixed by bot: for most languages, it would seem that changing apostrophes to stress marks could be done by bot. (But for languages that have ejective consonants, apostrophes may be misused for the ejective mark.) - -sche (discuss) 00:50, 22 June 2014 (UTC)

And in Gaelic languages, it's not unknown to use apostrophe for a prime, which is a non-IPA but used in textbooks (such as Ó Siadhail's Learning Irish) way of marking a palatalised consonant. --Catsidhe ^{(verba, facta)} 01:20, 22 June 2014 (UTC)

We should add a tracking category until the uses are fixed. DTLHS (talk) 00:51, 22 June 2014 (UTC)

It's a bad idea to convert X-SAMPA to IPA in {{IPA}}, we should have a new template for this, otherwise it leads to unwanted behaviors like this.

@Angr Please read this discussion.

I had created a Lua version of the template at Module:User:ZxxZxxZ/IPA2, there's a list of all standard IPA characters there. There's also a list of obsolete and "non-standard" characters, entries which use them would be put in a tracking category, and entries using any other characters (i.e. invalid characters) would be put in a cleanup category. --Z 07:07, 22 June 2014 (UTC)

Well, but now there are other entries using the straight apostrophe in its XSAMPA meaning of ʲ, so if we just turn it off completely, those entries will get broken. Fortunately, there is another way of represent ʲ in XSAMPA, namely _j, so if we remember to always use that, then we can leave ' undefined. Maybe it could cause a module error, warning the user to correct to either " (for stress) or _j (for palatalization). If this is the only problem, then I don't think it's a bad idea at all to convert XSAMPA to IPA in {{IPA}}, especially considering the large number of entries that still use the "normal" g instead of the IPA ɡ. By allowing the existing {{IPA}} template to convert XSAMPA into IPA, we no longer have to track down and fix all of those. —Aɴɢʀ (talk) 13:52, 22 June 2014 (UTC)

The problem is in the overlap between them. I don't know if there are many cases where the same character can be interpreted in different ways depending on whether it's treated as IPA or X-SAMPA. But there are probably a few of such cases, which will cause problems if any entries mix the two (inadvertently in case of "g" or "'"). —CodeCat 13:55, 22 June 2014 (UTC)

Actually, there aren't any cases where the same character can be interpreted in different ways depending on whether it's treated as IPA or X-SAMPA. The straight apostrophe is not an IPA character; it has no meaning in IPA. The problem is that it looks enough like the IPA stress mark that some people have misused it that way. —Aɴɢʀ (talk) 14:32, 22 June 2014 (UTC)

Are there any other characters that are likely to be misused in a way that will cause conflict between IPA and X-SAMPA? —Mr. Granger (talk • contribs) 16:00, 22 June 2014 (UTC)

I can't think of any, but I'm always surprised by people's creativity when it comes to misusing the IPA. The bigger problem is people using Kirshenbaum instead of X-SAMPA, because there are a lot of characters that have one meaning in Kirshenbaum and a different one in X-SAMPA, including notably ', which is the stress mark in Kirshenbaum but the palatalization marker in X-SAMPA. —Aɴɢʀ (talk) 17:02, 22 June 2014 (UTC)

Looking at an X-SAMPA chart, maybe "!"? I imagine people might use it for "ǃ", but in X-SAMPA it seems to represent "↓". But fortunately alveolar clicks and ingressive airflow are both rare, so that shouldn't be a big problem. I guess there's also a risk of people using G for ɢ and R for ʀ, but that would be a pretty dramatic substitution. —Mr. Granger (talk • contribs) 17:20, 22 June 2014 (UTC)

@Mr. Granger: possibly `, which is misused the same way the apostrophe is. And does the comma stand for anything in X-SAMPA? People sometimes misuse it for IPA ˌ. - -sche (discuss) 20:37, 22 June 2014 (UTC)

This is just one of the reasons it's a bad idea to mix IPA and X-SAMPA. --WikiTiki89 17:40, 22 June 2014 (UTC)

Well, maybe, but it's an even worse idea not to. —Aɴɢʀ (talk) 17:49, 22 June 2014 (UTC)

Umm... why? --WikiTiki89 18:10, 22 June 2014 (UTC)

Because of all the problems people have inputting IPA. I think a lot of people just don't bother with pronunciation information because it's such a hassle to the input the characters. —Aɴɢʀ (talk) 19:29, 22 June 2014 (UTC)

But that can be solved without mixing them. --WikiTiki89 19:43, 22 June 2014 (UTC)

Why not use a different template? —Mr. Granger (talk • contribs) 17:53, 22 June 2014 (UTC)

That's a better idea. --WikiTiki89 18:10, 22 June 2014 (UTC)

You mean bring back {{X-SAMPA}}? —CodeCat 18:15, 22 June 2014 (UTC)

Yes, and have it automatically convert from X-SAMPA to IPA. —Mr. Granger (talk • contribs) 18:21, 22 June 2014 (UTC)

I think there is a ramification in all this that hasn't been considered yet. If our entries end up containing X-SAMPA in their code, do I need to learn X-SAMPA in order to edit their pronunciations? —CodeCat 18:26, 22 June 2014 (UTC)

I was thinking we could have a bot periodically convert {{X-SAMPA|...}} to {{IPA|...}}. --WikiTiki89 18:39, 22 June 2014 (UTC)

That's the advantage to having one template do both IPA and X-SAMPA: people can use either system depending on what they're more comfortable with. —Aɴɢʀ (talk) 19:36, 22 June 2014 (UTC)

You may have misunderstood what I meant. A bot converting {{X-SAMPA|/@"baUt/}} to {{IPA|/əˈbaʊt/}} would mean that IPA and X-SAMPA are kept separate and that in the end we are left with only IPA. --WikiTiki89 19:43, 22 June 2014 (UTC)

But until the bot makes its rounds, if CodeCat spots an error in a transcription that's generated by {{X-SAMPA}}, she has to learn X-SAMPA in order to correct it. If we have a single template that does both, she can make the correction directly in IPA. So if it says {{IPA|/@"baut/}} by mistake, someone with no knowledge of X-SAMPA can correct that to {{IPA|/@"baʊt/}} and it'll still come out right. —Aɴɢʀ (talk) 20:24, 22 June 2014 (UTC)

A template named {{IPA}} with non-IPA characters in it... that's just asking for trouble, IMO. I like Wikitiki's idea. And couldn't the {{X-SAMPA}} template display "IPA: /IPA symbols/"? That way, people who don't know X-SAMPA get the same benefit you speak of CodeCat getting from an {{IPA}} template that accepts X-SAMPA input and displays IPA, while at the same time, there's no confusion about whether the input is IPA or X-SAMPA. - -sche (discuss) 20:37, 22 June 2014 (UTC)

That's exactly what I had in mind {{X-SAMPA|/@"baUt/}} and {{IPA|/əˈbaʊt/}} would display the same thing (and maybe the former can even be substed). --WikiTiki89 21:40, 22 June 2014 (UTC)

We don't have to use separate templates: make the first parameter a strictly-enforced IPA input, but have a named X-Sampa parameter for those who don't want to bother with direct IPA input (heck, we can throw in a Kirshenbaum parameter while we're at it). The error message for using non-IPA characters in the IPA field should tell people there's an X-Sampa option. Chuck Entz (talk) 20:57, 22 June 2014 (UTC)

But it makes much more sense to use separate templates rather than different parameters in one template. It's an arbitrary difference on the technical side, but a semantic difference on the user side. --WikiTiki89 21:40, 22 June 2014 (UTC)

I'm not so sure of that. The average user just wants to produce IPA output- do they really need two different templates to do basically the same thing? Chuck Entz (talk) 22:14, 22 June 2014 (UTC)

From a technical standpoint: Currently, if {{IPA}} is used without any content in its first parameter, it (appropriately) categorises the use as an error; if we were going to change it so that people could leave the first parameter blank and set an X-SAMPA parameter, we would have to change that. I think it would be tidier to use a separate {{X-SAMPA}} template; a separate template would also be simple to track (via Whatlinkshere). From a user-friendliness standpoint: Is a noob more likely to figure out that {{IPA}} accepts an X-SAMPA parameter, or that an {{X-SAMPA}} template exists? My guess is that it's the second one, or that it's a tie. We could even add the X-SAMPA template to the edittools, potentially with a note like "if you don't know IPA" to indicate that providing IPA is preferred. - -sche (discuss) 00:10, 23 June 2014 (UTC)

Did you notice that ULS provides an X-SAMPA input method? Why not point people to that instead? — Keφr 19:51, 22 June 2014 (UTC)

What is ULS? —Aɴɢʀ (talk) 20:24, 22 June 2014 (UTC)

That thing we have been all throwing obscenities at a few months back. Gear icon next to the interwikis list. There is an "Input tools" tab, in which you can enable an X-SAMPA input method for inputting IPA. I never use it (I prefer to use the IME integrated with my operating system), but hey, it is there. — Keφr 21:02, 22 June 2014 (UTC)

This thing, to be specific, if anyone wants to read up about it. - -sche (discuss) 00:13, 23 June 2014 (UTC)

I'm willing to undo the changes to {{IPA}} (and {{rhymes}}) that made them invoke Module:IPA if someone is willing to either (1) run a bot on a regular basis that fixes invalid IPA characters like g (should be ɡ) and : (should be ː) and the affricate ligatures like ʦ, ʧ, ʤ (should be t͡s, t͡ʃ, d͡ʒ etc.), or (2) set up a maintenance category listing pages where invalid characters are used inside {{IPAchar}} (which will include both {{IPA}} and {{rhymes}}). —Aɴɢʀ (talk) 07:01, 28 June 2014 (UTC)

Okay, a Feedback comment brought it to my attention that my change to how {{IPA}} works totally broke {{IPA letters}} (whose existence I was previously unaware of) in a way that I don't know how to fix, so I've gone ahead and undone my changes. {{IPA}} now works the way it used to up until about 8 days ago. I still hope someone can find a way to solve the issue of invalid IPA characters inside {{IPAchar}}, though. —Aɴɢʀ (talk) 11:26, 28 June 2014 (UTC)

Done: Category:IPA pronunciations with invalid IPA characters, Category:IPA pronunciations with obsolete or nonstandard characters, also Category:IPA pronunciations with invalid representation marks --Z 14:47, 28 June 2014 (UTC)

Thanks! I have to say, though, there are several entries in Category:IPA pronunciations with invalid IPA characters in which I can't find any mistakes. —Aɴɢʀ (talk) 15:22, 28 June 2014 (UTC)

Anything with a space will get added- not sure if that's intentional. DTLHS (talk) 21:38, 28 June 2014 (UTC)

Spaces need to be allowed. —Aɴɢʀ (talk) 21:42, 28 June 2014 (UTC)

What's invalid about 곤밥? - -sche (discuss) 17:08, 29 June 2014 (UTC)

The IPA there isn't enclosed in either slashes or square brackets, so it's unclear whether it's a broad/phonemic or a narrow/phonetic transcription. —Aɴɢʀ (talk) 17:28, 29 June 2014 (UTC)

Ah. That seems to be a problem many {{ko-pron}} entries have. They're probably all (or mostly) supposed to be /broad/, but I suppose they should be fixed in some human-checked way, e.g. with AWB, just to be sure. - -sche (discuss) 17:46, 29 June 2014 (UTC)

On the entry for cén, basically everything has module-errored. The page is practically unreadable because of all these module errors. Velociraptor888 09:31, 22 June 2014 (UTC)

Fixed already. Just purge the page. User:CodeCat accidentally deleted Module:languages. — Keφr 10:28, 22 June 2014 (UTC)

Thank you! Velociraptor888 11:35, 22 June 2014 (UTC)

Yes, I'm sorry about that. I was using a bot to delete all the language code templates, and didn't notice that the module was also categorised in there, so it was deleted along with the rest... I restored it as soon as I saw what happened. —CodeCat 10:30, 22 June 2014 (UTC)

When I view lists of Macedonian words in the category section, they are not in the proper alphabetical order. The list appears to go "ѕ", "j", "љ", "ќ", "џ" and then "а", "б", etc. See here: ]. I think that this should be mended. Could someone take a look at it? Martin123xyz (talk) 20:16, 22 June 2014 (UTC)

I'm seeing this too, except for me it goes ', Ѕ, Ј, Љ, Ќ, Џ, Б, В, without an А in sight. - -sche (discuss) 20:45, 22 June 2014 (UTC)

Until custom collation orders for categories are implemented, we can't fix this. We're stuck with the Unicode collation order. —CodeCat 20:47, 22 June 2014 (UTC)

I seem to have been using the wrong apostrophe sign for the entries of Macedonian verbs starting with a syllabic "r". I used the apostrophe I have on my Latin keyboard outlay because I didn't know what button it was on the Cyrillic one and also because I though that there's no difference between a Latin and a Cyrillic apostrophe. However, there apparently is, and now all the entries of words starting with a syllabic "r" that I've made have the Latin type whereas all of the ones made by other users previously have the Cyrillic one. One reason why this is problematic is that the two apostrophes are alphabetized on different positions in lists. Could someone fix these entries for me? I don't know how to edit page titles (or delete pages). When I asked how, it was simply done for me. Martin123xyz (talk) 20:20, 22 June 2014 (UTC)

1. Move entries to the correct titles (click on More button, you will see this option) - this will also create a redirect from the original spelling. 2. Request deletion of the original spelling, if you don't think they should be kept. --Anatoli ^{(обсудить}/^вклад) 03:24, 23 June 2014 (UTC)

I have moved the entries to the correct titles with the "More" button. However, I now cannot access the original pages to request deletion. They no longer appear on the list of nouns in the category "Macedonian Nouns" where they used to. Martin123xyz (talk) 07:11, 23 June 2014 (UTC)

They are no longer nouns but redirects, you can still access them and edit. --Anatoli ^{(обсудить}/^вклад) 11:37, 25 June 2014 (UTC)

@Martin123xyz Please also fix the broken links at Special:WhatLinksHere/'рскавица and Special:WhatLinksHere/'рж. --WikiTiki89 11:58, 25 June 2014 (UTC)

I have fixed something - I don't know if it's all alright now, though.

There's no such thing as a distinct Cyrillic apostrophe in Unicode. The entries have been moved from a version using the "typewriter" apostrophe ' to a version using the "curly" apostrophe ’, but both characters are equally Latin and equally Cyrillic. Since our custom here is to always use the "typewriter" apostrophe (e.g. don't; m’), I think these Macedonian entries should be moved back to the version with ', but the redirects kept (just as don’t and m’ exist as redirects). —Aɴɢʀ (talk) 19:44, 26 June 2014 (UTC)

I was thinking the same thing, but there is a difference here though. English Wikipedia uses the typewriter apostrophe for contractions, while Macedonian Wikipedia uses only the curly apostrophe for initial syllabic r, but the typewriter apostrophe for transliterations of the English apostrophe for example. --WikiTiki89 21:19, 26 June 2014 (UTC)

Come to the category chat to learn more. --Lo Ximiendo (talk) 03:19, 23 June 2014 (UTC)

Fixed. Chuck Entz (talk) 04:19, 23 June 2014 (UTC)

Thanks. The sorting errors on Vietnamese entries have been bugging me too. --Anatoli ^{(обсудить}/^вклад) 04:23, 23 June 2014 (UTC)

(pinging: Keφr, Yair rand, CodeCat, Ruakh, Kenny, ZxxZxxZ from workgroup tech)

I just wrote a rough first draft of this module. It allows one to conveniently ping users interested in a particular topic. It works like this:

1. Users add themselves to Module:workgroup ping/data indicating their areas of interest.
2. When someone else wants to catch attention of a particular interest group, they use the {{wgping}} template. For example, {{wgping|sla}} will ping all the users interested in maintaining Slavic language entries.

I filled Module:workgroup ping/data with some sample data. Feel free to create new groups, add or remove yourself. This is just a proof-of-concept; I already have some ideas for changes (for example, requiring subst: for the template, to fix group members at the time of writing the post). And tell me what you think. — Keφr 21:01, 23 June 2014 (UTC)

Is there still a limit (set by the site's software) on how many users can be pinged at once, or has that been lifted? In an ArbCom case on WP some time ago, a user pinged all the Arbs at once, so none of the pings went through. - -sche (discuss) 21:09, 23 June 2014 (UTC)

I might remember something like that, but I would have to dig though MediaWiki source code to be sure. — Keφr 21:23, 23 June 2014 (UTC) (wrong URL anyway)

There is at least an upper limit of 100 users. — Keφr 21:27, 23 June 2014 (UTC)

Is there any way to tie this into the babel templates somehow? DTLHS (talk) 21:13, 23 June 2014 (UTC)

I think no. Well, you might try abusing transclusion of Special:Whatlinkshere and work from that, but some sparse testing suggests it would not work. — Keφr 21:23, 23 June 2014 (UTC)

That's what I thought. It just seems like a shame because we already have lists of users by language that will be somewhat duplicated. DTLHS (talk) 21:25, 23 June 2014 (UTC)

Plus that wouldn't separate knowledge of a language from interest in a language. --WikiTiki89 21:32, 23 June 2014 (UTC)

I have not set any fixed definition of a "work group" or an "interest group". I figure users could be pinged from WT:TRREQ as well. This module is whatever you make out of it. — Keφr 21:48, 23 June 2014 (UTC)

We were discussing the idea of automatic groups from Babel templates. --WikiTiki89 22:20, 23 June 2014 (UTC)

Turns out I was wrong about the Babel thing. While writing Module:tagged but not listed, I realised that I can use the very same "preprocess and unstrip <DynamicPageList>" trick to list users in Babel categories like Category:User en. The problem is, categories might become huge, while editors become inactive. I would at least need another category for inactive users to make this workable. (Or maybe parse Special:RecentChanges? That is doable too.) — Keφr 11:52, 24 November 2014 (UTC)

Turns out no after all. The unstripping trick does not work at template substitution time, and I doubt that it works while Echo parses diffs either. (Even if it works in the latter case, we could not maintain historicity: the list of pinged users would change over time, and no longer reflect which users were actually pinged when the post was made.) — Keφr 12:34, 24 November 2014 (UTC)

Amazing idea! — Ungoliant ^(falai) 21:14, 23 June 2014 (UTC)

I really want to undo this edit, but the anon has it right, the {{IPA}} template breaks when refs are inserted inside it. How can this be fixed? --WikiTiki89 02:54, 25 June 2014 (UTC)

I don't like the new IPA template. I cannot use <> or ~ inside the brackets now, as desired here, because things that should be interpreted as non-IPA are interpreted as IPA now. This new template should be moved to something like {{IPA-simp}}, leaving the original {{IPA}} functioning like before. Wyang (talk) 03:04, 25 June 2014 (UTC)

The subject of references inside {{IPA}} came up once before when a module or bot (I forget which) to check for the use of nonstandard symbols like " inside {{IPA}} was first being designed and I pointed out that things like {{IPA|/fu/<ref name="foo">Foobar</ref>|/fo/|lang=en}} existed. At the time, some other users argued that such things should be handled with {{IPAchar}}, like this. But using references within {{IPA}} was still possible at that time; if it's no longer possible, something must have changed. - -sche (discuss) 03:09, 25 June 2014 (UTC)

What changed is the support of X-SAMPA characters intermixed with IPA, which I think is a really stupid idea in the first place. --WikiTiki89 09:07, 25 June 2014 (UTC)

Then let's undo that change. The other discussion showed support for your two-template idea. - -sche (discuss) 17:31, 25 June 2014 (UTC)

Even if we undo support for X-SAMPA in the IPA template, we still shouldn't be putting ref tags inside the IPA template. It's just for displaying IPA. Put refs at the end, outside the template. —Aɴɢʀ (talk) 19:31, 26 June 2014 (UTC)

The mods in almost all the cases are no longer active in the Korean Wiktionary. And I want to bring the Russian verb conjugation templates here to the Korean Wiktionary. But the problem is that I don't know how. I wonder if anyone could help an ailing version of Wiktionary. Thank you. --KoreanQuoter (talk) 11:44, 25 June 2014 (UTC)

BTW, I wouldn't respond until Saturday. But thank you, again. --KoreanQuoter (talk) 11:54, 25 June 2014 (UTC)

Nobody? --KoreanQuoter (talk) 08:41, 28 June 2014 (UTC)

I am trying to add hærfest as a translation of autumn. When I type in the language code, the word, and the gender, and then click "Preview translation", the usual pop-up box comes up, but with an error: ERROR:TypeError: undefined is not a function. --WikiTiki89 13:03, 25 June 2014 (UTC)

It was the {{trreq|or}} that would follow the Old English translation. — Ungoliant ^(falai) 15:36, 25 June 2014 (UTC)

Oh. I presume that's a known issue, so I can close this discussion. --WikiTiki89 17:27, 25 June 2014 (UTC)

Hi,

In Slovene, we have the word Franche-Comté (like in English and in French). But when we want to create a link to the Slovene section for this word (even if it doesn't exist now) with the {{l}} template, the link will be a link to Franche-Comte. A such link is not correct, is it? — Automatik (talk) 22:51, 25 June 2014 (UTC)

That's because {{l}} strips the accents from words. I'm not sure how we would prevent it in this case. —CodeCat 22:54, 25 June 2014 (UTC)

(e/c) @CodeCat Perhaps we should provide an override for the accent removal? I imagine something like {{l|sl|Franche-Comté|target=Franche-Comté}}, which can maybe be shortened even to {{l|sl|target=Franche-Comté}}. --WikiTiki89 22:58, 25 June 2014 (UTC)

If that is how it should work, then we might as well just make it {{l|sl|Franche-Comté|Franche-Comté}}. That is, always disable the removal of diacritics if the alt parameter is present. This will break some of our current entries, but probably not too many and we can track them down first. On the other hand, it is kind of cumbersome to do it this way, but considering we've run into this problem only now, I don't think it's a serious issue. —CodeCat 23:07, 25 June 2014 (UTC)

If we have {{l|sl|Franche-Comté|Franche-Comté}} do that, it may not break many manual links, but it will break some logic in inflection table modules, where this feature is actually useful. --WikiTiki89 23:14, 25 June 2014 (UTC)

True. What's important though is that this change is compatible with a wide range of templates and their parameters. We could give all of them target= parameters, but there may be another way. Just like we have the character * in front of words to indicate that they are reconstructions, we could think of some other character that indicates that the link should receive no postprocessing. —CodeCat 23:22, 25 June 2014 (UTC)

That's also a good idea. What characters are not allowed in entry titles? --WikiTiki89 23:26, 25 June 2014 (UTC)

Alternatively, instead of using a single character, we could use a prefix string, like raw: for example. This is pretty much guaranteed to never occur at the start of a page title? The question then is what happens when this is combined with *, like *raw: and raw:*. Presumably the former should not process the name but still link to appendix pages, while the latter should link to an actual page beginning with * (like *nix). —CodeCat 23:36, 25 June 2014 (UTC)

That would work. I imagine raw:* being much more useful than *raw:, since we don't really do any postprocessing of reconstructed languages. --WikiTiki89 23:42, 25 June 2014 (UTC)

Not all appendix entries are reconstructed languages, though. We have quite a few Latin ones as well. —CodeCat 23:54, 25 June 2014 (UTC)

That's a good point. But they are still only reconstructed terms, which are likely to follow our notion of Latin orthography, which is not going to include words that are required to be written with a macron or anything like that. --WikiTiki89 00:02, 26 June 2014 (UTC)

How about just a colon? It would create the advantage of not having the syntax of vanilla markup and {{l}} diverge too much. (And it already works for "*".) — Keφr 05:52, 26 June 2014 (UTC)

I thought about that before, but thought it would be counterintuitive given the existing purpose of a plain colon. But now that you bring it up again, I realize that it makes perfect sense. In fact, I'd even say that we should just check for the presence of a colon anywhere, that way other namespaces can be linked to without screwing up the link. But then the question remains of whether we should add the language section fragment after a raw link; it would be convenient in most cases, but might not always be desired. --WikiTiki89 06:03, 26 June 2014 (UTC)

Such target-page-related issues (diacritics removal, linking to appendices e.g. for *nix) rarely occurs, we can simply introduce an on-off parameter for linking templates to disable the functionality of make_pagename in line 216 completely. Let's avoid manipulating the string as far as possible, introducing a new character like "*" or "raw:" in the string may complicate our codes too much, without much benefit. --Z 08:14, 26 June 2014 (UTC)

I've implemented the : prefix for raw links now: {{l|sl|Franche-Comté}} > Franche-Comté but {{l|sl|:Franche-Comté}} > Franche-Comté. Likewise, {{l|en|*nix}} > *nix but {{l|en|:*nix}} > *nix. —CodeCat 19:43, 29 June 2014 (UTC)

We should have checked for existing usages with initial ":" first. --Z 20:08, 29 June 2014 (UTC)

They would have worked pretty much the same anyway: *nix. See? — Keφr 20:22, 29 June 2014 (UTC)

And links to ":" don't work even in raw code: ]. So nothing should be broken. —CodeCat 20:24, 29 June 2014 (UTC)

I know, I was talking about interwiki links. --Z 21:03, 29 June 2014 (UTC)

Other Wiktionaries have genitive forms for nouns, such as the Swedish Wiktionary. Like "toad's", "frog's", "dogs'", "Ohio's", "Teletubby's", "Teletubbies'", "log's", etc. Please see sv:frog and sv:toad for an example. Why do we not have these forms? We have various word forms for other languages, including genitives and things with apostrophies, so the thing I don't get is why do we not have these? They are just as good of word forms as anything else. And other Wiktionaries are being a bit more clever than us at our own language. Please, let's get bots, or something, to add all these genitive forms that are missing from the English language. Let's put it in our policy now, there are 4 verb forms, nominative singular, nominative plural, genitive singular, and genitive plural. We are missing the genitives. Who's with me? Vote? Rædi Stædi Yæti {-skriv til mig-} 15:44, 26 June 2014 (UTC)

There has already been a vote where it was decided not to include these. The reason is that these are not true genitive forms. Rather, the suffix is a clitic and attaches to the whole phrase, rather than to the head of the phrase like in a real genitive case. —CodeCat 15:47, 26 June 2014 (UTC)

Why don't we tell this to the Swedish Wiktionary then, and tell them to stop adding the genitive case and confusing us with it? Rædi Stædi Yæti {-skriv til mig-} 15:50, 26 June 2014 (UTC)

Feel free to tell them. I'm not confused. But if you want to understand what I mean, look at The king of France's crown for example. The genitive marker follows France, but that doesn't mean that word is in the genitive. If it were, then the phrase would mean "The king of the crown of France" where it of course really means "The crown of the king of France". —CodeCat 15:52, 26 June 2014 (UTC)

Regardless of whether you call it a genitive or a possessive particle, the reason we don't include it is because it is easily predictable. --WikiTiki89 15:55, 26 June 2014 (UTC)

Does that mean that genitive forms are only useful in other language Wiktionaries, as someone who knows almost no English, but speaks Arabic or some random language like that, would have no idea what "dog's" means, despite the fact that they know what "dog" means? Whereas here, it is already obvious. Rædi Stædi Yæti {-skriv til mig-} 15:58, 26 June 2014 (UTC)

Each Wiktionary makes its own decisions. Editors here believe that even foreigners would have no problem figuring out what "dog's" means if they know what "dog" means. --WikiTiki89 16:09, 26 June 2014 (UTC)

The problem is actually more serious than that. It's that if you treat "dog's" as a form that is independent from "dog", then almost every word in the English language has a genitive form, even verbs, adverbs, prepositions and so on. Some examples: the dog I never saw's tail, the man who walked quickly's briefcase, the band I have an album of's latest single. —CodeCat 16:11, 26 June 2014 (UTC)

Well let's see. We have lots of Danish entries but we are lacking in many areas, I am saying once again. I just found out that the forms we use are not all of the forms that Danish uses in conjugation. Such as let's take antyde for an example. We only support the things for imperative, present tense, past tense, and past participle forms of antyde. However there are many other forms, and I will name a few as examples. I believe the following are called passives:

• antydes
• antydedes
• antydets
• antydendes

Also we are missing future tense:

• antydende

Where are these forms in our templates? And are there more that I have not yet mentioned? (I am not a native speaker)

Please, let's make a conjugation template or something for Danish verbs. If not, let's improve the Template:da-verb template. Rædi Stædi Yæti {-skriv til mig-} 04:49, 27 June 2014 (UTC)

I believe there are an active and passive tense, much like Swedish. Why don't we make a conjugation table? This would make things much more clear. Rædi Stædi Yæti {-skriv til mig-} 05:14, 27 June 2014 (UTC)

Fill in as you wish and put at Template:da-conj-whatever:

<div class="NavFrame">
<div class="NavHead">Conjugation of {{m|da||{{{1<includeonly>|{{PAGENAME}}</includeonly>}}}}}</div>
<div class="NavContent">
{| class="wikitable inflection-table"
<!-- your table goes here -->
 |}
</div></div><noinclude>{{documentation}}

— Keφr 09:01, 28 June 2014 (UTC)

Shouldn't we include emoticons in Wiktionary? I mean, I kind of was trying to look up a specific emoticon here but we didn't have it. Is this includable for the dictionary, emoticons? Because I don't see any here at all. Rædi Stædi Yæti {-skriv til mig-} 05:12, 27 June 2014 (UTC)

We have a few: Category:Emoticons by language. Someone of them need to be subpages of Unsupported titles because the first character of pagenames can’t be :. — Ungoliant ^(falai) 08:32, 27 June 2014 (UTC)

Not sure whether they are "words in a language", but neither are the mathematical symbols. What we shouldn't do at least is add the silly, complicated ones that only occur in lists: they should be held to the usual attestation standards (use, not mention). Equinox ◑ 13:38, 27 June 2014 (UTC)

Well what I'm thinking is we can make "Translingual" entries for these emoticons. Rædi Stædi Yæti {-skriv til mig-} 14:59, 27 June 2014 (UTC)

If you can attest them in multiple languages, then surely. (By the way, WT:BP seems better suited for this question.) — Keφr 15:04, 27 June 2014 (UTC)

I'm more than positive that we can find most emoticons in Google Books in a flash. The question is is this allowed, and how can we implement this? Rædi Stædi Yæti {-skriv til mig-} 15:07, 27 June 2014 (UTC)

-1. Yes; 0. Open a page, 1. Type in a definition, 2. Paste citations, 3. Save page, 4. Done. — Keφr 15:12, 27 June 2014 (UTC)

I tried to create eSports by typing the word into the Search box, and was taken straight to esports, which at the time was solely the Catalan entry. I then clicked on a redlink, and was taken successfully to page creation. I don't know if this is a bug or a feature. If it's a feature, I found nothing on it while quickly scanning the obvious page creation information.

I also double-checked: if you try to create (from the Search box) an oddly capitalized word that does not exist in all-small, you will be taken to page creation. Choor monster (talk) 13:04, 27 June 2014 (UTC)

It's a feature, and a relatively new one. The same is true for diacritics: if you enter a word with a diacritic in the search box, and if that word doesn't already exist as an entry, but another word with similar or no diacritics does exist as an entry, you'll be taken to the existing entry instead. You'll only be taken to the Search page if there are multiple entries with other diacritics, since in that case the search feature won't know which one you wanted. However, if you move your cursor down to the bottom of the suggested entries in the search box, where it says "containing..." and click there, you'll be taken to the Search page no matter what. —Aɴɢʀ (talk) 14:05, 27 June 2014 (UTC)

Perhaps in the case of a single valid entry, the entry should be displayed, but the typed-in version should be made available as a red-link below the top of the entry. Sort of like WP redirects. Choor monster (talk) 20:47, 27 June 2014 (UTC)

I've been thinking about making a conjugation template for Volapük verbs, most likely based on the one from the Portuguese Wiktionary. Any ideas and questions? --Lo Ximiendo (talk) 06:15, 28 June 2014 (UTC)

My bet for the conjugation table: {{vo-conj-verb}} followed by the verb's stem. --Lo Ximiendo (talk) 23:48, 28 June 2014 (UTC)

Hello. Since yesterday, every time I visit this website, I find out that I am logged out. I am telling it to remember my login and everything, but it just keeps on logging me out. This is really annoying, and also means I have to get a new temporary password each and every time, as for some reason, it just doesn't accept the current one. Is there a solution to this? Velociraptor888 08:57, 28 June 2014 (UTC)

Can a template be set up for Den Danske Ordbog, which can be used for reference purposes in Danish entries? I was thinking of something like “Grease pit/2014/June” in Den Danske Ordbog. Or is there already a template under a different name? Donnanz (talk) 14:36, 28 June 2014 (UTC)

No reason you can't do it yourself at {{R:Den Danske Ordbog}} (ie, Template:R:Den Danske Ordbog) or {{R:Danske Ordbog}}, using as models {{R:ODS online}} for categories and {{R:Webster 1913}} for including bibliographic info and for linking to the site (or to a Danish or English WP article) and to a specific page of any online edition, if that is possible. Linking to a specific page is not always straightforward. DCDuring TALK 22:54, 28 June 2014 (UTC)

I have tried setting it up, but I'm going wrong somewhere, or the link to the dictionary won't work. The link would need to use http://ordnet/ddo/ordbog?query?= . I also tried ODS online, but that leads you to the old dictionary (pre-1950). Donnanz (talk) 00:52, 29 June 2014 (UTC)

• I am basically monolingual and no template maven either, but I have made some reference templates work. Where can I find a copy of your efforts? DCDuring TALK 01:08, 29 June 2014 (UTC)

Here: https://en.wiktionary.orghttps://dictious.com/en/Den_Danske_Ordbog

The website isn't helping at the moment - it has decided to crash. http://ordnet.dk/ddo (I left out .dk before). Donnanz (talk) 01:21, 29 June 2014 (UTC)

I deleted it because templates need to start with "Template:" I created {{R:Den Danske Ordbog}} and gave it my best shot- but I'm no template maven either. The tricky part was getting it to output an = instead of interpreting it as part of the template syntax. You also got the query syntax a little muddled, but I saw a couple of good ones in some Google search results. Here's an example of use: {{R:Den Danske Ordbog|dansk}} gives “dansk” in Den Danske Ordbog

Chuck Entz (talk) 01:47, 29 June 2014 (UTC)

I realised my effort had to be deleted. I have just tried it for soveværelse, and it worked well, taking the letter æ into account. Thanks a million, Chuck Entz! Donnanz (talk) 07:16, 29 June 2014 (UTC)

• Funnily enough, dansk is one of those words hinted at by DCDuring - "Linking to a specific page is not always straightforward." There are separate entries for the noun and the adjective - the adjective appears at dansk2 - http://ordnet.dk/ddo/ordbog?select=dansk,2&query=dansk. I will have to find a way round that. Donnanz (talk) 08:06, 29 June 2014 (UTC)

When I go to create a new citations page, the edit window opens with the following preloaded:

 {{documentation subpage}}
 {{documentation needed}}<!-- Replace this with a short description of the purpose of the template, and how to use it. -->
 <includeonly>
 ]<!-- replace this category with the category of your choice -->
 </includeonly>

This is also preloaded when I go to create a template documentation subpage, but it's to be expected and desired there. It's the preloading of it onto citations pages that should be stopped... - -sche (discuss) 22:24, 28 June 2014 (UTC)

I could probably fix this, but I forgot which script this code is in. Can someone tell me? —CodeCat 13:11, 30 June 2014 (UTC)

MediaWiki:Gadget-DocTabs.js (formerly MediaWiki:Common.js), and I already fixed it. — Keφr 14:06, 30 June 2014 (UTC)

Thanks! - -sche (discuss) 14:08, 30 June 2014 (UTC)

@Kephir Is there a reason why this was moved to its own dedicated script? —CodeCat 14:40, 30 June 2014 (UTC)

(for some reason the ping did not work.) To decrease its impact on other scripts; an error in it will not prevent others from running. And I have been planning to split out scripts from MediaWiki:Common.js in general. I wanted to move most of MediaWiki:Common.js to the LegacyScripts gadget, and start working on replacements. Right now looking at it makes me think of The Gospel of the Flying Spaghetti Monster. — Keφr 16:26, 30 June 2014 (UTC)

Our current and longstanding practice is to wrap collapsible tables in a pair of "div" elements with the "NavFrame" class, which are then handled by JavaScript (MediaWiki:Common.js) accordingly. However, this arrangement is kind of cumbersome. It means that the outer div can never adjust its width according to the table that's inside it. Instead, we have to do things like trying to pick some sensible default width and put up with wrapping if the content is too wide. Or alternatively, some of our collapsible tables are always as wide as the page, even if we don't need that much space (it looks very bad on wide displays like 1920x1080). Of course this is not ideal. So I wonder if we can move away from this practice somehow, and instead make the table itself collapsible, without wrapping it in anything. Of course collapsible things that are not tables, like translations or derived terms lists, will not change. —CodeCat 13:23, 29 June 2014 (UTC)

Maybe display:table-cell and friends can provide a fix? —Michael Z. 2014-06-29 15:27 z

Some collapsing divs have text outside tables (i.e. see recolher). How would that be dealt with? — Ungoliant ^(falai) 17:15, 29 June 2014 (UTC)

We can put that in the table. I think it would look nice to put it in the table even if we don't get rid of the divs. It looks pretty bad right now. —CodeCat 17:19, 29 June 2014 (UTC)

We could use the <caption> tag (|+ below {| in wiki markup). But there is a problem; <caption> cannot be styled the way NavFrames can. Also, collapsing a table will also collapse its width to match the caption text (though that is also true of captionless tables). — Keφr 18:11, 29 June 2014 (UTC)

Isn't it possible for JavaScript to determine the width of the table before it is collapsed, and set the collapsed-state width accordingly? —CodeCat 18:22, 29 June 2014 (UTC)

No, it is. But not if the table is collapsed before the script runs (which is what we do with NavFrames now). — Keφr 18:29, 29 June 2014 (UTC)

The tables should start expanded though, so that the page's contents is accessible even if JS is turned off. —CodeCat 18:36, 29 June 2014 (UTC)

Right now I believe they start "expanded", but the JS collapses them before they are even rendered, so I don't the width would be determinable at that point. --WikiTiki89 18:42, 29 June 2014 (UTC)

Right. What happens is, the rendered HTML of every page has <html lang="en" dir="ltr" class="client-nojs">, and there's some early-running JS that changes the class from client-nojs to client-js. (That's not an enwikt-specific thing; it's handled for us by MediaWiki.) In MediaWiki:Common.css, we specify that .client-js .NavFrame .NavContent should not be displayed. It's not a perfect system — it means that if a browser manages to add the .client-js, but then runs into problems handling later JS, then the user will not have any way to expand the content — but it's not obvious how it can be improved. —Ruakh_TALK 19:00, 29 June 2014 (UTC)

Then I guess the question is, is it so bad that the table width changes when it's collapsed? I think the current alternative, with tables that don't resize at all, is worse. —CodeCat 14:41, 30 June 2014 (UTC)

Can someone edit Module:la-verb so that the imperative headers don't show for verbs that have no imperatives (such as this one)? Esszet (talk) 00:41, 30 June 2014 (UTC)