Hoi, In the last month, this list has gained many new subscribers; they are people who are members of a language support team. As you may understand, the Wikimedia Localisation team is not able to understand all the intricacies of the 280+ languages we support at the Wikimedia Foundation. It is for this reason that we need help and the best way is when people who are knowledgeable enough provide us with their knowledge.
Recently we have added support for grammatical gender and plural for JavaScript. As you may know, we already supported gender and plural in PHP for quite some time. As part of the process Santhosh made an inventory of the differences between the implementation in MediaWiki and in the standard for such things, the CLDR. I blogged about the result on the Wikimedia Foundation blog.
As I also reached out to the Wikipedias of the language that have an issue, we gained both more people who are interested to provide support for their language but as relevant, we learned that entering data in the CLDR is overly complicated. Even our developers, who are also linguists or language experts have a hard time entering the data for plural support.
Even when there is no information in the CLDR about things like plural support, we want to support this in MediaWiki. When the information we get about the rules for plural support are given to us with references either on the Internet or in publications, it is relatively straightforward to create a bug report to the CLDR. What we are looking for are people who can help us with getting the necessary information so that we can implement it in MediaWiki, have it tested by a language community and then post a bug report at the CLDR. If need be, our developers can help with the formulation of plural rules in the format of the CLDR
Another really interesting development was for the Arabic script. We learned about the Amiri font, a font that supports all the characters of the Arabic script and consequently is able to support not only Arabic but also languages like Urdu, Kashmiri etc. We wrote to the developer of the script and he was happy for us to use his font in a WebFonts implementation.
Not so funny is that we learned that the support of the Arabic script in MediaWiki is problematic. It turns out that we do not fully comply with the rules for merging parts of a character (including diacritics). This is not only an issue for MediaWiki but also for some of the modern browsers. We will be talking about how MediaWiki is to be fixed with Brion Vibber at the upcoming FOSDEM conference in Brussels and we have posted bugs for the browsers.
At some stage we will have fixed MediaWiki and at that time we will need people to test their language and see if everything looks as you expect it.
When your language has problems with fonts or input methods, we are quite happy to support your language as well. For WebFonts we need a freely licensed font that can be used on all the major operating systems and preferably on the versions that are still in use (several versions of IE are technically unable to support WebFonts at all). For input methods we need a mapping from the international keyboard to the keyboard as expected. This includes information on the rules of how characters are to be combined.
As a member of a language support team you are VERY welcome to come forward with requests for the support of fonts or input methods. We do expect your involvement with both the implementation and the testing. We are also very appreciative when you communicate this widely in your language community. As always we can test these things at translatewiki.net. This is where we run the latest MediaWiki software before it goes into production on any of the Wikimedia Foundation's projects. Thanks, Gerard
http://blog.wikimedia.org/2012/01/13/addressing-the-many/ http://ultimategerardm.blogspot.com/2012/01/supporting-font-for-arabic.html http://translatewiki.net/wiki/Language_support_team
Hello, Gerard, where are the current rules documented and where should the correct rules be gathered/reported? I've seen Nike's list here: https://translatewiki.net/wiki/Plural Are all explicit rules used in MediaWiki defined in that plural-gettext.txt, and should corections be reported on talk or somehow in the table or...? For instance, I read that Italian doesn't have custom rules anywhere, hence defaults to «plural rule "n != 1"», which is correct, so I don't have to do anything, right? Thanks, Nemo
Hoi, When the information for a language is correct, there is nothing to be done for PLURAL. As I understand it, you are right when you state that the plural-gettext.txt is where the information can be found. In essence it means for me that in addition to a potential difference with the CLDR there may be a difference with the implementation in gettext as well. As you may know, gettext is used inthe implementation of many freely license applications. Thanks, Gerard
On Sat, Jan 14, 2012 at 7:06 PM, Federico Leva (Nemo) nemowiki@gmail.comwrote:
Hello, Gerard, where are the current rules documented and where should the correct rules be gathered/reported? I've seen Nike's list here: https://translatewiki.net/wiki/Plural Are all explicit rules used in MediaWiki defined in that plural-gettext.txt, and should corections be reported on talk or somehow in the table or...? For instance, I read that Italian doesn't have custom rules anywhere, hence defaults to «plural rule "n != 1"», which is correct, so I don't have to do anything, right? Thanks, Nemo
Mediawiki-i18n mailing list Mediawiki-i18n@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/mediawiki-i18n
Gerard Meijssen, 17/01/2012 08:35:
When the information for a language is correct, there is nothing to be done for PLURAL. As I understand it, you are right when you state that the plural-gettext.txt is where the information can be found. In essence it means for me that in addition to a potential difference with the CLDR there may be a difference with the implementation in gettext as well. As you may know, gettext is used inthe implementation of many freely license applications.
Thank you. And what do we need to do when it isn't correct?
Nemo
Hoi, There are three places where you are heard; this mailing list, the [[Support]] page at translatewiki.net and, the IRC channel at irc://freenode/mediawiki-i18n .
The first thing is that we REALLY want to know when it is not good. The next part is to define what is good and find documentation that proves it. Finally we will create an entry in the reporting system of Unicode for the CLDR. NB No definition is not good either ... even when everything is standard because how do we know ? Thanks, Gerard
On Tue, Jan 17, 2012 at 9:27 AM, Federico Leva (Nemo) nemowiki@gmail.comwrote:
Gerard Meijssen, 17/01/2012 08:35:
When the information for a language is correct, there is nothing to be done for PLURAL. As I understand it, you are right when you state that the plural-gettext.txt is where the information can be found. In essence it means for me that in addition to a potential difference with the CLDR there may be a difference with the implementation in gettext as well. As you may know, gettext is used inthe implementation of many freely license applications.
Thank you. And what do we need to do when it isn't correct?
Nemo
Mediawiki-i18n mailing list Mediawiki-i18n@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/mediawiki-i18n
mediawiki-i18n@lists.wikimedia.org