Talk:Special language codes

Moravian

Hello,

how can be add to Wikipedia project its Moravian language mutation? Moravian language hasn't ISO code. Moravia is country in czech republic, but Czech wikipedia censor any information about Moravia. They don't accept Moravian nationality. 83.240.78.219 11:18, 2 March 2011 (UTC)

Currently, the language proposal policy requires an ISO 639 code before a new language Wikipedia can be created. So I guess the first step would be to get an ISO 639 code, which Moravian doesn't have currently. PiRSquared17 (talk) 22:44, 11 March 2019 (UTC)

Tarantino

The entry about Tarantino is possibly wrong:

ISO 639-3 lumps it with Italian, as with most varieties of northern Italy.

According to Wikipedia, Tarantino is a dialect of Sicilian (language code scn), not Italian (language code ita).
Best regards. Naudefj (talk) 06:55, 25 September 2013 (UTC)

According to linguistic maps, Tarantino is more closely related to the Neapolitan group of dialects than to Sicilian, although Apulian dialects to its south are in the Sicilian group. You can refer to the English Wikipedia article w:Tarantino dialect or to it:Dialetti italiani meridionali. PiRSquared17 (talk) 22:50, 11 March 2019 (UTC)

This is not contested that this is a Southern dialect, but yes the sentence used may be confusive (as if it was saying it was a Northern Italian dialect like Emilian, Friulan, Genovese, Ligurian, Piemontese or Venitian... that have their own ISO 639-3 codes! that sentence about northern dialects is in fact false for most important Northern dialects). Anyway I did not create that sentence, just fixed the translatability. I had noted that statement but waited for reactions of Italians. But it's a fact that Tarantino is still not separately encoded and many documents from this area are tagged with "it" (ISO 639-1) and then do not even use the ISO639-3 code [scn] of Sicilian as well (and also frequently merge into [it] the contents in Sardinian, or even the Corsican [co] variant in North Sardinia, derived from medieval Genovese. verdy_p (talk) 02:10, 12 March 2019 (UTC)

Per File:Italian Dialects.png (label #39 in Puglia) and w:Tarantino dialect, I think it makes more sense to talk about roa-tara as part of nap than as part of scn. But Ethnologue lists nap as Neapolitan/Calabrese-Lucanian, without reference to Barese/Tarantino (or Abruzzo, Molise, southern Lazio dialects for that matter). On the other hand, Ethnologue does say that "Pugliese" is included in scn, but I assume they mean Salentino and not the northern dialects of Puglia. PiRSquared17 (talk) 02:36, 12 March 2019 (UTC)

Note that this page was initially created (by someone else) to create a more complete reference for an old request (sent in 2015) to Unicode CLDR. This request has been denied and closed because the submitter wanted to add some of the non-standard Wikimedia (& Translatewiki.net) language codes to the CLDR project (which correctly honors only requests about standard codes): this is not the problem of CLDR but a pure problem of Wikimedia alone (which also infects other projects, including many incorrect tagging spread in their wikis, including the content of the Wikidata database). Note that wikimedia "interwikis codes" are not a problem (they are not restrited to just language codes and are only used to create links to Wikimedia domain names, which are not part of the language tagging: there's no such restriction to use only BCP 47 tags in DNS domain names, so this is also not a request to change the interwiki codes, or Wikimedia domains, or internal database names or aliases).

Here we are only concerned by what we need in CLDR data (for translation of labels or some comment properties), in interwiki.net (for translations of MediaWiki UI or some of its extensions or opensourced bot scripts), and in what we insert in lang="..." and xml:lang="..." attributes in HTML/XML/CSS (which must conform to BCP 47).

And this page currently incorrectly states that some codes are invalid when they are still perfectly correct for BCP 47 (even if some of them they have been retired in ISO 639, or recategorized from single languages to macrolanguages or deprecated for splitting them in separate languages, such as old codes [bh] for Bihari or [qu] for Quechua). As well we continue remaining concerned about the non-conforming use of pseudo-variant extensions (that should have used private use codes or should have been registered by a formal request to the IANA database, such as "formal" for German and "informal" for Dutch, because they only exist to satisfy user requests and solve conflicts in some Wikimedia projects). Even the [simple] code used by Wikimedia for interwiki or domain names is valid for such use, but invalid for use in lang=..." and xml:lang="..." attributes.

As well [zh-yue] remains perfectly valid in BCP 47 (even if it is now deprecated in favor of [yue] by a specific entry added to the IANA database. So there's no emergency to change these codes, unlike the very incorrect codes [als] and [nrm] used in Wikimedia!). verdy_p (talk) 18:36, 12 March 2019 (UTC)

How to request changes to language codes (not subdomains)?

I know subdomain changes are doomed to stall/fail, but is there any way to request a change to language codes in general? There is already a difference internally. For instance bat-smg.wikipedia.org has <html lang="sgs"> PiRSquared17 (talk) 08:22, 29 August 2020 (UTC)

You don't need to do that, this has been reported since long and requested multiple times and it is tracked in Phabricator: not ignored but there are complex things to do to change a site name (the change for the interwiki so that it works with the new code is already effective, and wikilinks should already work; however URLs cause problems (notably from external sites or wikis). as well ther are changes to do in Wikidata (need some bot to perform long and costly checks); changes also in some Wiktionnaries when they used the wrong code for their tagging (normally BCP 47 should already be used for classification, however this is not blocking the change of the wiki site name).

Finally there is a lengthy complex process to change the domain name (because domain names have very long cachability): note that domain names are not required by the DNS standard to be valid BCP 47 codes, they can be quite arbitrary (notably subdomain names which are always assigned privately, here by Wikimedia). So somain names are independant of BCP 47 language codes and also independant from interwiki codes (which is an internal setting of the Mediawiki parser), thelselsves also indepenadnt of the SQL database name storing the wiki (and under which the wiki content is exported: this SQL name contains not just the "wiki language code" but also an abbreviation of the site's project type, glued together; it could be arbitrary and would not change the exposed codes except for those looking for database dumps area; but note that even in this download area, the export is not made immediately there, but passes some process to compress it and move it to an appropriate public area, so it is already renamed and public exports are not a real problem). Other things that cause wiki renaming to be pending is that wikis in Wikimedia share a common infrastucture that require common configuration files and settings: you cannot change these without affecting as well the other wikis using the same shared settings. Such change have to be scheduled, as it also requires a shutdown and maintenance. So it must be announced to the community in advance, there must be enoung admins interested to follow the bug and prepare what could be wrong, some testbed wiki may be needed to be setup (it requires hardware resources, and human time).

"bat-smg" to "sgs" is not ignored, in fact it's one of those changes tracked since long (including "zh-classical" to "lzh" or "zh-min-nan" to "nan", or "roa-tara" to "scn-x-tara", or "nrm" to "nrf", the msot critical case as "nrm" is the only wikicode that conflicts with another language that can't be created). "bat-smg" does not conflict with anything in BCP47 and "sgs" has still not been used so there's no emergency. All you can do is to use the "sgs:" interwiki prefix (in wikilinks of wikipages) as it already works, and help fixing the various HTML pages (notably the lang="*" attribute in HTML tags) or some bad CSS or javascript that makes incorrect assumption: this does not require any admin msot of the time, and most work to do is there (if it can't be automated by accepted bots with strict rules for their page edits), or old tempaltes using bad assumptions or not recognizing "bat-smg" as an old alias for "sgs" or the reverse: this is what could break a wiki after it is renamed, even if the renaming will keep the former code as an alias for some long time (probably one year or more, as long as the usage of the old code does not decrease to become insignificant, notably by visits from externals or search engines): Wikimedia has a policy to preserve working URLs even if these are redirects, so it has to maintain the compatiblity (only abusing pages can be deleted). verdy_p (talk) 11:53, 29 August 2020 (UTC)

Yeah, but what I asked is how to change the language code used by MediaWiki, not the subdomain. For example, I think "arc" should be changed to "syc". There's already a request to change the subdomain which is stalled for the reasons you mentioned, but in the meantime they could at least change the MediaWiki language code so that the HTML lang attribute is syc, etc. How would I go about requesting that? PiRSquared17 (talk) 21:26, 29 August 2020 (UTC)

Given that you are on the talk page of Special language code, you can see that this topic is already covered and liks to a Phabricator issue tracker. This talk page is not followed. The only thing you can do is to post your comment on the Phabricator thread linked here. an hope that other persons will join your request to push up some priorities so that this request is not stalled for long without any progress. Take some contacts in your communities, ask them to support your request on Phabricator and submit their ideas, or explain the workarounds and the work done, and that constantly needs to be done repeatedly just to converge to some stable enough solution. Document as much as you can the workarounds you implement, possibly create your own tracker page in your favorite wiki to measure how much you are impacted and what can be done (with costly workarounds that need to be tracked) and these that are blocked because of that (including in terms of usability) and the impact it has on your language.

Wikimedia works with cooptation, the best arguments by a single person, even when they are fully explained and justified is not enough: Counters really matter! You need more people with you. I support your request but cannot act myself to change anything. verdy_p (talk) 23:15, 29 August 2020 (UTC)