Topic on Talk:Universal Language Selector/Compact Language Links/Archive 1

A lot of useful feedback from several people was sent here. Relevant tasks discussed are:

https://phabricator.wikimedia.org/T70071 - prioritze links to featured pages

https://phabricator.wikimedia.org/T70077 - Prioritise interwiki to language "X" if a lang="X" attribute is present in the page's HTML

https://phabricator.wikimedia.org/T133029 - Fine-grained geolocation for looking up common languages in ULS

https://phabricator.wikimedia.org/T136248 - zh (Chinese) doesn't appear as a suggested language if the country is CN (China)

https://phabricator.wikimedia.org/T138973 - Show languages that a wiki community added to the top of the interlanguage list in the Compact list

I'm English, but I regularly read Wikipedia in Spanish, Catalan, German, French, Italian and Dutch. I tried to make this tool "learn" that those were the languages that I was interested in, but it would only learn four at a time - when I clicked the link to go to the fifth language, it knocked the first one off the list leaving me with Bengali, Welsh, Scottish Gaelic and Scots, none of which I'm interested in. I've been looking at Barack Obama on the English Wikipedia. Any chance we could just select the languages of interest in our preferences somewhere?

+1

Hi,

Thanks for the comment.

May I ask in which country you were and which page were you reading when you saw Bengali, Welsh, Scottish Gaelic and Scots?

Your user page says that you live in Catalonia. If you are connecting from Spain, then you are supposed to see Spanish, Catalan, Galician, Basque, Asturian, Extremaduran and English according to the CLDR Territory-Language information. You are also supposed to see the language of your browser and the languages that you previously selected. Of course, you will only see languages in which the page you're reading is available.

If you are connecting to the web from the U.K., then it's actually likely that initially you'll see Bengali, Welsh, Scottish Gaelic and Scots, because all these languages are spoken there.

We made two changes this week, which should address your issues:

• We increased the number of remembered languages from 5 to 9.
• When we run out of suggested languages to show, we are going to try to show some common world languages instead of showing the languages first languages alphabetically. With this change, you will more likely see German, French and Italian than Welsh, Gaelic and Scots (outside of the U.K.).

If the software deployment process goes as planned, you will see these changes on Friday.

Thanks for your reply. I'm reading in the UK. I'm not sure I explained myself well. To clarify, I'm not averse to seeing Bengali, Scots, Scottish Gaelic and Welsh before having selected any languages previously (as you say, those are languages spoken in the UK). I am just commenting that after having clicked the seven languages I mentioned, I would expect those seven languages to appear whenever the article I'm looking at has a version in those languages and yet when I look at the list, only four of them are shown - the other three (the first three I selected) seem to have dropped off the list.

As you say, I imagine that the increase in "remembered" languages from 5-9 should help matters, particularly in my case, but I really don't see why I can't simply select my preferred languages in the preferences instead of this complicated algorithm?

Thanks.

In case I am connecting from China, am I expected to see lzh, nan, hak, yue, cdo, wuu, etc. there?

By the way, is it possible to show those links base on context? For instance if a page is under a subategory for Japan or include Japanese text or Japanese language template, I believe it should prioritize displaying link to Japanese Wikipedia

Yes, more or less. In China you should see zh, yue, wuu, hak, nan, gan, ii, ug, za, mn, bo, ko, kk, ky, en, ru, vi, uz, lzh. See the table: http://www.unicode.org/cldr/charts/29/supplemental/territory_language_information.html

If you are in China, I'd love your feedback on whether it works. (Some of these languages have very tiny Wikipedias, however.)

Adding by context will probably be done in the future. See these tasks for example:

• https://phabricator.wikimedia.org/T70077
• https://phabricator.wikimedia.org/T70071

Also, we need to make the location more fine-grained for countries that have a lot of languages. For example, I guess that yue should have a higher priority in the South of China, while ug, ky and mn should have higher priority in the North. We are not doing it yet, but we may do it in the future, see https://phabricator.wikimedia.org/T133029 . It's a problem also for India, Russia, U.S., and other countries.

1. Unfortunately I am in Hong Kong now so I can't help you test how it would work.
2. But when I try the function in Hong Kong by browsing a random Korean Wikipedia article page on my device with device language set to Japanese, those links I get are: ar, ca, da, de, en, es, ja, zh-yue. It does not include zh which is the most-used language version in Hong Kong.
3. I think those phabricator issues did not mention about the possibility of determining context by category?
4. "North of China" or "South of China" are too much. Most languages are only popular in a few provinces. W:Language Atlas of China have some maps about language distribution in China. For instance, yue are basically unheard of outside Guangdong and Guangxi, ug is essentially exclusive to XInjiang according to that map but I think Guansu/Ningxia/Sichuan/Qinghua might also have a few ug community. Only a small part of XInjiang use ky or kk. Most mn speakers are in Inner Mongolia, with some spill over to Liaoning/Hebei/Jilin/Heilongjiang and two isolated area in Qinghua and Xinjiang
5. Despite there're many people speaking Mongolian in China, but most of them don't know how to write their language in cryllic script and thus linking the existing Mongolian wikipedia is useless to even those who speak Mongolian in China. To address this, it have been suggested multiple times that a "Mongolian WIkipedia written in Mongolian Script" should be created. But it seems like its creation is blocked by the task https://phabricator.wikimedia.org/T11436.

You say "device". Is it a mobile phone or a computer?

Android phone. Was using chrome on android.

OK, this page is only about desktop, not Android.

On Android it is done differently, and I think that it doesn't use geolocation at all. Desktop and mobile should be more similar, but they aren't yet.

If you can test anything on the desktop with the "Compact Language Links" beta feature enabled, it will be very useful.

@Amire80 When I was testing on the android, I was using the "Desktop version site", so I think it should follow desktop's behvior or was I just imagining?

And I just tried to test the behavior on my windows notebook via a Chinese VPN, on Japanese Wikipedia main page it give me ar/az/bg/ca/en/ko/ru/vi/zh, and on zh-yue wikipedia main page it give me en/ko/ru/vi/zh/wuu/zh-min-nan/hak/gan

Thanks a lot for the feedback!!!

Thanks to this report I found an important bug: https://phabricator.wikimedia.org/T136248

by the way, I saw languages like Yiddish and Crimean Tatar on Wikimedia RU's "language with official status" list https://ru.wikimedia.org/wiki/%D0%97%D0%B0%D0%B3%D0%BB%D0%B0%D0%B2%D0%BD%D0%B0%D1%8F_%D1%81%D1%82%D1%80%D0%B0%D0%BD%D0%B8%D1%86%D0%B0 , but they are not on the CLDR list. Are they imcluded in wikimedia's list?

Heh.

Yiddish is barely needed—it's spoken by a very small number of people in Russia, and it doesn't have any official status even in the “Jewish Autonomous Region”. That region is Jewish only in name and there are very few actual Jews there. So if Yiddish is not shown in Russia, it's not a problem. I guess that it appears on the Wikimedia RU site for nostalgic or symbolic purposes :)

Crimean Tatar is not CLDR because it is spoken in Crimea and no international organization recognizes Crimea as a part of Russia (but that's probably not a political statement from CLDR—they probably just didn't bother very much). For practical reasons, I wonder whether people who surf the web from Crimea are identified by our geolocation as Ukraine or as Russia. I need to find some people who actually live there. Thanks again for pointing this out!

According to http://qz.com/243619/crimea-just-switched-over-to-the-russian-internet/ , most Crimean internet users are now connecting via Russian network

Thanks for the link. I should still test it with real users, but this probably already means that we'll have to make some clever adjustments. We'll have to make some anyway, not only for this language.

Had something just changed that make me see those ace/kbd/af/ak come back to the compact list again?

They weren't supposed to go yet! The deployment of this change is happening later today.

Hello, Is it possible to create 'languages of interest' button?

What do you mean by this?

For example: I want 'English' always at top of the list, but Deutsch is always above it.

With the current way the software works, initially you will see no more than 12 languages in the list, and if English is one of the preferred languages, it will always be in that list. It will not necessarily at the top of the list because the list is sorted alphabetically and Deutsch can be above it. But is it hard to find the language you need in a list of up to 12 languages?

My understanding is that, If a user want to learn Polish starting from today, then he should be able to add Polish to his own personalized compact link.

This can already be done by clicking on Polish in the selector panel.

you are right

Was activated today on Wikipedia. I do not like the fact that I don't see all languages and want to shut it down. However since I work inter-wiki I have to shut it down 200x on all different languages. I want to be able to see the same list on all languages. So firstly I would like to have it completely removed. And as a constructive criticism I would like to ask that you leave regular editors the option to pick their own languages.

שלום עליכם, אמיר:

(I love this rtl/ltr stuff. Not.) I'm not a big fan of this functionality, myself. It would probably be useful if there were a way for a user to set this permission globally.

That said, it seems to me that on Judeo-Spanish projects (e.g., ladwiki), es and he will just about always deserve to be present contextually. I suppose most people visiting ladwiki regularly would probably have at least he as one of their selected languages after they've used this for a bit, but I'd be happier if it were always a default at ladwiki.

Thanks for the comment. I added this task: https://phabricator.wikimedia.org/T138973

It should cover your request, although we'll have to discuss the implementation.

Thanks very much. I'll need to add that page on ladwiki, but as you say, we'll discuss the implementation when the time comes.