Talk:Universal Language Selector

Jump to navigation Jump to search

About this board

Ynhockey (talkcontribs)

It appears that the extension uses a certain sans-serif font for Hebrew that is very bad at displaying nikud. Some fonts that are good at displaying nikud are: David, Narkisim, SBL Hebrew. Most others, especially sans-serif fonts, are terrible at it.

Can this issue be resolved? There is no problem with displaying Hebrew on any computer since Windows XP, so maybe there is no need for a web font at all. Or if there is, can it be a font that displays properly with nikud?

Reply to "Hebrew font problem"

Info request from event login

7
Mahitgar (talkcontribs)

Requesting kind Attn: ULS team Amir Aharoni, Niklas Laxstrom, Pau Giner, Santhosh Thottingal & Runa Bhattacharjee

Hi, Seasons Greetings

Ref: Bug 51349 and Extension:UniversalLanguageSelector/EventLogging queries

End of this july at mr-wikipedia we made certain changes to support ULS users with some additional help. On certain counts we found some positive impact. To have better (indepth) understandanding we request you to provide figures again for अक्षरांतरण and मराठी लिपी options of Marathi language ULS for month of August 2014. Or figures prefrably from this week day (or next week day) and again after 30 days.


(But we do not want figures of week ends that is not of saturday or sunday. Preferable any week day from Tuesday to Friday and Indian 'day' time is covered in the same.)

Runab WMF (talkcontribs)

Hello, Thanks for bringing this up. Could you let us know the specific type of data you are looking for? The queries may need to be modified based on the requirements.

Mahitgar (talkcontribs)

Hi,

Data I am looking for mr.wikipedia.org i.e. Marathi language wikipedia.

No. ULS option for Marathi What it is called in Marathi at ULS Data needed 1 Data needed २ Data needed 3 Data needed 4 Data needed 5
1 mr-transliteration अक्षरांतरण How many total users enabled mr-transliteration on 28th July 2014 or July 2014 How many total users enabled mr-transliteration on 1st August 2014How many total users enabled mr-transliteration on 31st August 2014 or for the month of August 2014 How many total users enabled mr-transliteration on 1st Jan 2015 or for the month of Jan 2015 How many total users enabled mr-transliteration on 28th Feb 2015 or for the month of Feb 2015
2 mr-inscript मराठी लिपी How many total users enabled mr-inscript 28th July 2014 or July 2014 How many total users enabled mr-inscript on 1st August 2014How many total users enabled mr-inscript on 31st August 2014 or for the month of August 2014 How many total users enabled mr-inscript on 1st Jan 2015 or for the month of Jan 2015 How many total users enabled mr-inscript on 28th Feb 2015 or for the month of Feb 2015

Why do we need above data (Purpose) ?

  • 1) Why we are expecting Data for the month of August and Jan Feb ?
Visitor hits usually are highest for month of August and Jan Feb for Marathi language Wikipedia both as per wiki stats and also as per google trends stats
  • 2) At the end of July 2014 we received an audio-visual clip support for ULS Marathi, using the clip we updated help pages and are running a special campaign for month of August 2014.
So far this campaign has shown excellent results on two fronts percentage of Roman script trial and errors has reduced substantially. Secondly hit ratio shows visit to related help pages per day increased from 1 or 2 to strait away 1000 per day. (We do not know exact figure of total hits for Marathi wikipedia per day but it is roughly estimated in between 3000 to 8000 per day. and 96 to 98 % of them are not likely to know how to do Marathi Typing online/Marathi Wikipedia)
  • 3) With this August campaign we have seen stunning increase by 1000 hits a day to Marathi Typing help pages unfortunately we have not seen proportionate increase in newcomer and anon edits so far. There is a need to find out what is the actual percentage of the people enabling ULS feature for Marathi language options. With the ULS eventLogin data provided by you we will get to know whether just people are visiting help pages but practically are not enabling ULS or they are enabling ULS but may be some thing else keeping them away from using the feature, So accordingly we can design question/survey to understand user behaviour better and improve upon our support and help pages.
  • 4) In any case we have done preparation for sept 2014 for one more campaign for editing mr-wikipedia. We hope sept 2014 campaign will either give more edits or some insight in to the issue why there had not been proportionate increase in edits but for proper comparison certainly we need eventLogin stats as requested above.

Presently we can see the number of hits to a wiki page on our own same way feature for ULS eventLogin stats visibility on our own will be useful to us. For an example data after september 2014 campaign also will be useful to us same way many time november also gets good traffic to mr wiki. Currently we are dependant on ULS team for stats so we are limiting ourselves to August 2014 and Jan -Feb 2015.

Looking forward to support on August 2014 stats at earliest . Thank you, warm regards and best wishes to ULS team

Runab WMF (talkcontribs)

Thanks for the details. Very helpful. I am passing the request onto Amir Aharoni for help with the event-logging data.

Mahitgar (talkcontribs)

Hi, This is a kind reminder. Undersigned is waiting for the requested support with crossed fingures.

Thanks and best wishes

Santhosh.thottingal (talkcontribs)
Mahitgar (talkcontribs)

Hi Santhosh, Thanks for your kind support, its very nice of you. These figures are very helpfull. While August 2014 (campaign period) figures signal kind of positive impact can be achieved for detail analysis I will be waiting for some stats figures from other place i.e. http://stats.wikimedia.org/EN/TablesWikipediaMR.htm for how many exact new users got added and there contribution etc.

In Agust 2014 campaign we tried to reach out with ULS inro video clip to aprox 1 lack Marathi people but minimum 30 thousand certainly visited one or the other help page. There are roughly 30 lack active Marathi internet users (6 lack regular or more than average active); roughly 2 to 4 % already know some system of unicode Marathi Typing with this campaign we could reach out to additional 1 percent (of 30 lack) .. So conservative estimate 1 in 30 persons would know Marathi Typing that means for a novice to seek a known person's help currently a freinds freind atleast would know a Marathi Typing method. Our (mr-wiki) target needs to be 1 in 5 i.e. 6 lack people should know ULS it self so contribution to mr-wiki improves. (Hopefully mr-wiki would be able to conduct a improved campaign for feb 2015 season)

Figures provided by you indicate in August 2014 ULS -mr options were activated 4680 times where as activation figure total for July 2014 was 2968 so net increase in activating ULS-mr options was 1712 (around 57% increase over a normal month figure) . 1712 additional activations devided by 30 days will be around 57 per day. If we consider average every user activates ULS 3 times in a day then we can safely assume around 19 new users every day activated ULS-mr since net trial edits are not substantially increased that means immidiate repeating of ULS on further days by same person may not be substantial with conservative estimates around 600 new users might have got aquinted with ULS-mr. ( 600 new users trying on 30,000 visitors to help pages comes to 2% and matches with over all Marathi peoples response rate of 1 in 50 persons usually responds to Marathi Wikipedia campaigns online or of line)

There can be several reasons for absence of net increase in edits out of that one could be what to write syndrome so this september we have already started a campaign. Second is new users may not be aquinted in advance with inscript and learning time requirement curve for inscript is longer. Phonetic and inscript 2 options are not much supported yet still hits are there on them. People may be choosing inscript since its nomenclature in ULS -mr translates to marathi script obiviously people may be activating it and not able to get conversant faster enough. Certainly this aspect needs to be investigated further with more surveys locally. And soon we will be doing that. And try improve support so we get better conversion to edits.

BTW new account creation using Devanagari scripts is increased 4 folds since this August and also with lot of grammer mistakes is a separate issue before us to handel and one more related issue already discussed at hi-wiki community and for that likely to come up with a feature request rfc in couple of weeks.

Thanks any way and warm regards

Reply to "Info request from event login"
Billinghurst (talkcontribs)

Is there a simple list of the available fonts, with links to examples, that can be utilised across the WMF wikis? About the only place that I can find anything that resembles a list is at the git repository, and such a list is hidden away, and not particular user friendly. Is such a list possible, and if it more relates to WMF only, should that then be placed at meta, or is it a subpage from the main namespace page, if more generic? Thanks.

Santhosh.thottingal (talkcontribs)

https://www.mediawiki.org/wiki/Universal_Language_Selector/WebFonts#Selection_of_fonts has a list of languages and default fonts. But there are more fonts than this that are not mapped to any language because they are special fonts to be applied outside the user preferences. Example: Unifraktur is supposed to be applied to a cetain section of wikisource content irrespective of users preference by specifying the font-family style explicitly. And these fonts are not meant for Mediawiki interface fonts. You are right, a list would be nice so that people can choose them and apply wherever relevant. A bit manual effort based on the git repo is needed. Is this something you can help us? Otherwise I will ask my team to consider doing it. thanks!

Billinghurst (talkcontribs)

It was I who raised the WS-related question, and the non-working of the explicit application of the webfonts. Once that issue is resolved, then I can look to assist resolving that component. So the questions that come with that are

  1. Can all the fonts be explicitly user-derived with the expressed methodology?
  2. Do all the fonts fall under this extension?
  3. So would a help page be best here (as a subpage presumably), or at one of the other WMF wikis if there a separate means to add fonts individually per wiki, or for WMF.
  4. Is that directory of the repo the only place that contains the fonts, or non-language specific fonts? If not, where are the others located?
Reply to "List of available fonts?"
Bill william compton (talkcontribs)

Devanagari numerals (१, २,...९) are default when you select Hindi irrespective of different options (BolNagri, phonetic, inscript, etc). In reality, Indo-Arabic numerals (0, 1, 2,...9) are used in Hindi newspapers, magazines, text books, government publications, etc. It really get on nerves if you want to write something in Hindi on Wikipedia because you have to switch over settings so many times. So it would be really helpful if developers could provide options for selecting numeral system also. Regards.

Mahitgar (talkcontribs)

Well it is for Hindi Wikipedians to discuss but with simple Control+M you can toggle (shift) between the scripts Roman and Devnagari. I do not think you need to switch over settings so many times. I tested १1२2३3४4 for Hindi लिप्यंतरण option just now. with ease I did १Control+M1२Control+M2३Control+M3४Control+M4 १1२2३3४4 it did not take many seconds. In any case you use shift button for some letters simillerly you are using Control+M instead so it does not increase many strokes. Seondly if you have an entire different keyboard for roman numeral+ hindi alphabates to type १1२2३3४4 this you will end up switching key boards. So rather using Control+M seems more logical to me still it is for majority of hindi wikipedians to decide and request the changes. Still I suggest a revisit to your request

Reply to "Numeral system"

Add more "Common languages" manually?

2
Gleki.arxokuna (talkcontribs)

Is there a way to always show some languages in "Common languages" block? My wiki has most content in a certain number of languages and some of them are never shown in ULS unless you find them via search.

Reply to "Add more "Common languages" manually?"

Custom fonts now more common than Flash in the web

2
Nemo bis (talkcontribs)
Reply to "Custom fonts now more common than Flash in the web"
Nemo bis (talkcontribs)

Why are the language names sometimes so small? 7.3px according to Firefox inspector on hu.wikt, and "magyar" looks even smaller. Only at the second attempt I looked more carefully and spotted the one familiar word ("italiano") hidden in the dialog, finding out that I wasn't in the wrong place after all.

Reply to "Small language names"

Edit diff that had made ULS non-functional on mr -wiki

3
Mahitgar (talkcontribs)

Hi,

Probably this edit difference had made ULS icons were not apearing to enable and ULS became non functional in important edit fields until the edits to MediaWiki namespace were reverted today.

Since edit making admin was not aware of technical aspects about enabling new gadgets, suitable technical guidance , if any will be welcome.

Siebrand (talkcontribs)

Gadgets provide such powerful functionality they can break all MediaWiki JavaScript, including UniversalLanguageSelector's. The best advice we can give is to enable the gadgets one by one, to find the one that breaks pre-existing JavaScript functionality. After finding the culprit, that one can be debugged, and remain disabled.

Reply to "Edit diff that had made ULS non-functional on mr -wiki"

Is there some expression that evaluates to the current language?

2
Larrydberg (talkcontribs)

E.g., something like {{CURRENTLANGUAGE}} that gives me the code for the current interface language?

Background, in case there's another solution: My wiki is multilingual. I use categories heavily, for example Category:Dances. The text at the top of that page should be displayed in the current interface language. So on that page I put {{int:CategoryDancesHeader}} and create pages MediaWiki:CategoryDancesHeader/XX as appropriate. This works fine.

For Hebrew I also must set the overall right-to-left orientation of the page. On a normal page I do that by putting {{PAGELANGUAGE:he}} at the top. But that doesn't work in this case. It looks like setting PAGELANGUAGE on a page doesn't effect pages into which that page is transcluded.

A solution is to set PAGELANGUAGE at the top of Category:Dances itself with something like {{PAGELANGUAGE:{{CURRENTLANGUAGE}}}}. How can I get my hands on the current language in order to do this?

I'd also think this functionality is useful enough to be provided directly, e.g., let {{SETPAGELANGUAGECURRENT}} to set the page language to the current language.

Nemo bis (talkcontribs)

No, there isn't. The request to add one has been rejected/delayed many times because it would allegedly break Wikimedia projects' caching system, see bugzilla:2085. Wikis are using a hack which achieves the same effect anyway (so allegedly breaking thier cache, but nobody cares), that is a custom message Lang called with int magic word, which will return the language code of the user's interface. I can't tell if it's the appropriate solution for you, but you can do the same or locally hack your MediaWiki installation as e.g. translatewiki.net does (I think some patches were published on the bug or elsewhere).

Reply to "Is there some expression that evaluates to the current language?"

Design considerations regarding usage of Autonym font in ULS

2
Patrick87 (talkcontribs)

The Autonym font is currently developed to be able to display all glyphs of all language Autonyms with the purpose to be used by ULS for language Autonyms

The question is: Does this really make sense? Do we really need a font that supports all glyphs? Wouldn't it make much more sense to design a font that only contained glyphs potentially missing in the system font?

E.g. Latin glyphs should be available in the system font on all systems if I'm not mistaken. Why do we add those to the Autonym font? Currently this causes already many bugreports because of a lack in font quality. Even if this is resolved the glyphs used in Autonym font will differ from the system Font of most OSes (e.g. Latin charakters contained in Arial on Windows look considerably different to the latin charakters of Autonym font). Therefore we're creating an inconsistent look of texts in the UI on purpose. Is this really what we want people to see in prominent spots when visiting Wikipedia?

Verdy p (talkcontribs)

No. This font is clerly NOT designed to support all glyphs needed to render any supported language. For example it MUST NOT be used to display translations. Its purpose is to display only the native language names, exactly those returned by {{#language:code}}, and only those texts! Its purpose is to allow correct visual identification of languages.

In addition it must be able to render any language code, i.e. lowercase letters a-z, digits 0-9 and the minus-hyphen '-' (needed for language variants) It may render the uppercase letters, allowed as equivalents for language-codes in BCP47 for use in the lang="" attribute, or any language code supported by Mediawiki on Wikimedia sites, though Wikimedia sites should only use lowercase letters for these codes). For this reason it will include the full printable ASCII range (U+0020..U+007E, i.e. 95 glyphs only); it should probably not include punctuations of ISO88-59-1 or windows-1252 outside ASCII.

The number of needed glyphs is then capped, and is not very large (limited to the list of supported languages).

But if we remove the ASCII characters then the ".autonym" CSS class should contain "font-family: Autonym, sans-serif". But some languages will render inconstantly when some characters are found only in the "Autonym" font, others being in the browser's default fonts for "sans-serif". Their glyph metrics won't match together.

So if we ever remove ASCII charcters from the "Autonym" font, then we MUST remove also all Latin letters, so that they will be all fond in the browser's default fonts for "sans-serif". Caveat: the metrics of Latin letters (in "sans-serif") may no longer match with metrics of other characters needed for other scripts (this will be a problem in case of dual scripts in one language name). So let's keep ASCII completely in that font, with other those Latin letters needed for supported language names.

My opinion is that "sans-serif" is present in the CSS class only for the case where the browser does not load the webfont from ULS, but it should NEVER be used if the Autonym webfont is loaded: the webfont must be complete for all supported languages, even if some of them may be rendered corectly only with the browser's default "sans-serif" font list.

It is absolutely not critical that the Autonym font looks different for displaying languages from the various fonts used for displaying the rest of pages or the UI, as long as it remains in a coherent sans-serif stroke style (for Latin, Cyrllic, Greek), or the default stroke styles most commonly used to display other scripts.

But I agree that all the supported glyphs in the Autonym font should have correct metrics, and good hinting for all resolutions (including small sizes >= 8px, as low as possible). Note that this font should be available in bold and italic for some UI cases (bold needed when the language name is selected in the current page, in a language bar.

Navigation language bars at top or bottom of pages (linking to translated version of the current page) should be deprecated as much as possible : the ULS should be used instead for selecting the page to render, written directly in the language selected for the UI. Pages can autodetect this language (including for their templates), using "autotranslation" based on {{int:lang}} (for example with LangSwitch on Commons or Meta, or with the translate extension).

Reply to "Design considerations regarding usage of Autonym font in ULS"