Page MenuHomePhabricator

Decide what to do about Manual of Style pseudo-namespaces conflicting with Mooré Wikipedia
Open, Needs TriagePublic

Description

See https://meta.wikimedia.org/wiki/Requests_for_new_languages/Wikipedia_Moor%C3%A9#Comments.

(LangCom has decided that this is not their problem at https://meta.wikimedia.org/wiki/Talk:Language_committee#Approval_Request_for_Moor%C3%A9_Wikipedia)

Possibilities include:

  • Accept that the English Wikipedia is going to get screwed. This seems seriously unwise and likely to incite a user revolt.
  • Create the wiki at some other name like "mos-x-deconflict.wikipedia.org" (using the BCP47 private subtag syntax is probably the least bad). This would work, probably without major changes, but doesn't feel right.
  • Create the wiki at mos.wikipedia.org and add hacks to the dumpInterwiki script in MediaWiki-extensions-WikimediaMaintenance so, either only for enwiki or for all wikis mos: doesn't become an interwiki. I've written a patch for this locally, and it's similar to what we already do for Swedish ("s:" isn't an interwiki on Swedish Wikipedia because they abbreviate "saint" as s:t), but that caused problems like T177913 and probably others.
  • Convert the MOS: enwiki pseudo-namespace to a real namespace (which overrides interwikis), and do something similar with WikimediaMaintenance to add an interwiki with some other name like "mos-x-deconflict"
  • Someone on enwiki suggested making interwiki prefixes case-sensitive and having "mos": point to the project while "MOS:" is a pseudo-namespace. This isn't how either concept currently works, and will probably break things elsewhere
  • Something else I haven't thought of?

Event Timeline

Pppery renamed this task from Decide what to do about English Wikipedia Manual of Style pseudo-namespace conflicting with Moore Wikipedia to Decide what to do about English Wikipedia Manual of Style pseudo-namespace conflicting with Mooré Wikipedia.Apr 25 2024, 11:07 PM
Pppery updated the task description. (Show Details)
Bugreporter subscribed.

Note: Chinese Wikipedia also has such pseudo-namespace.

Convert the MOS: enwiki pseudo-namespace to a real namespace (which overrides interwikis)

Regardless this should happen, fake namespaces just causes issues.

List of pages that will be affected:

Generated using Wikimedia Global Search on 2024-04-26 11:28

WikiPage title
dag.wikipediaMOS:CHANGEDNAME
dag.wikipediaMOS:PEOPLETITLES
dag.wikipediaMOS:JR
dag.wikipediaMOS:CITIZEN
dag.wikipediaMOS:LEADBIO
dag.wikipediaMOS:SURNAME
dag.wikipediaMOS:NICKNAME
dag.wikipediaMOS:SR
dag.wikipediaMOS:NATIONALITY
dag.wikipediaMOS:JOBTITLE
dag.wikipediaMOS:LASTNAME
dag.wikipediaMOS:NICK
dag.wikipediaMOS:BLPLEAD
dag.wikipediaMOS:MULTIPLENAMES
dag.wikipediaMOS:NICKNAMETHE
dag.wikipediaMOS:JUNIOR
dag.wikipediaMOS:MRS
dag.wikipediaMOS:OFFICE
dag.wikipediaMOS:OPENPARABIO
dag.wikipediaMOS:NAME
dag.wikipediaMOS:NE
dag.wikipediaMOS:SECTIONCAPS
dag.wikipediaMOS:SELFREF
dag.wikipediaMOS:TVSPLIT
dag.wikipediaMOS:JOB
dag.wikipediaMOS:MULTINAMES
dag.wikipediaMOS:REGNAL
dag.wikipediaMOS:ROLEBIO
dag.wikipediaMOS:QUOTENAME
dag.wikipediaMOS:FIRSTBIO
dag.wikipediaMOS:FAMILYNAME
dag.wikipediaMOS:CREDENTIAL
dag.wikipediaMOS:MR
dag.wikipediaMOS:BIRTHDATE
dag.wikipediaMOS:NICKBOLD
dag.wikipediaMOS:JRSR
dag.wikipediaMOS:NEE
dag.wikipediaMOS:NAMES
dag.wikipediaMOS:LAYOUT
dag.wikipediaMOS:CREDITS
dag.wikipediaMOS:ISMCAPS
dag.wikipediaMOS:LCITEMS
dag.wikipediaMOS:COMMONNAMECAPS
dag.wikipediaMOS:MS
dag.wikipediaMOS:PSEUDONYM
dag.wikipediaMOS:NAMEFMT
dag.wikipediaMOS:LEADREL
dag.wikipediaMOS:GIVENNAME
dag.wikipediaMOS:STRUCTURE
dag.wikipediaMOS:INITS
dag.wikipediaMOS:DOC
dag.wikipediaMOS:ALLCAPS
dag.wikipediaMOS:NICKCRUFT
dag.wikipediaMOS:BIRTHPLACE
dag.wikipediaMOS:REFERS
dag.wikipediaMOS:COMMONNAMES
dag.wikipediaMOS:BOLDTITLE
dag.wikipediaMOS:BIO
dag.wikipediaMOS:PATRONYMIC
dag.wikipediaMOS:MISS
dag.wikipediaMOS:COMPASS
dag.wikipediaMOS:SMALLCAPS
dag.wikipediaMOS:INSTITUTIONS
dag.wikipediaMOS:RACECAPS
dag.wikipediaMOS:CONTEXTBIO
dag.wikipediaMOS:BOLDLEAD
dag.wikipediaMOS:FULLNAME
dag.wikipediaMOS:LIST
dag.wikipediaMOS:INITIALS
dag.wikipediaMOS:HYPOCORISM
dag.wikipediaMOS:LEADNO
dag.wikipediaMOS:BADNICK
dag.wikipediaMOS:POLITICALUNITS
dag.wikipediaMOS:DOCTOR
dag.wikipediaMOS:BLP
dag.wikipediaMOS:BIRTHNAME
dag.wikipediaMOS:SAMESURNAME
dag.wikipediaMOS:ROYAL
dag.wikipediaMOS:CELESTIALBODIES
dag.wikipediaMOS:CAPS
dag.wikipediaMOS:FILMPLOT
dag.wikipediaMOS:FILMNOW
dag.wikipediaMOS:FILM
dag.wikipediaMOS:FILMGENRE
dag.wikipediaMOS:FILMCOUNTRY
dag.wikipediaMOS:MOVIEPLOTS
dag.wikipediaMOS:PLOTBLOAT
dag.wikipediaMOS:FILMPRODUCTION
dag.wikipediaMOS:FILMLEDE
dag.wikipediaMOS:FILMLEAD
dag.wikipediaMOS:FILMCAST
dag.wikipediaMOS:LEAD
dag.wikipediaMOS:LEADNOTUNIQUE
dag.wikipediaMOS:PROPER
dag.wikipediaMOS:THEINST
dag.wikipediaMOS:LEGALNAME
dag.wikipediaMOS:ETHNICITY
dag.wikipediaMOS:SPACEINITS
dag.wikipediaMOS:PHD
dag.wikipediaMOS:GOD
dag.wikipediaMOS:AT
dag.wikipediaMOS:COLLAPSE
dag.wikipediaMOS:NUM
dag.wikipediaMOS:NOSECTIONLINKS
dag.wikipediaMOS:TIES
dag.wikipediaMOS:HEAD
dag.wikipediaMOS:STYLERET
dag.wikipediaMOS:HON
dag.wikipediaMOS:LIGATURE
dag.wikipediaMOS:LIFE
dag.wikipediaMOS:MAIN
dag.wikipediaMOS:
dag.wikipediaMOS:OPEN
dag.wikipediaMOS:FIRST
dag.wikipediaMOS:US
dag.wikipediaMOS:1STOCC
dag.wikipediaMOS:NOTUSA
dag.wikipediaMOS:HONORIFIC
dag.wikipediaMOS:BLPCHRONO
dag.wikipediaMOS:HONOR
dag.wikipediaMOS:REVEREND
dag.wikipediaMOS:HONOUR
dag.wikipediaMOS:DAME
dag.wikipediaMOS:GENDERID
dag.wikipediaMOS:ASTRONOMICALBODIES
dag.wikipediaMOS:PEOPLANG
dag.wikipediaMOS:GENRECAPS
dag.wikipediaMOS:MILTERMS
dag.wikipediaMOS:HEADCAPS
dag.wikipediaMOS:GEOUNITS
dag.wikipediaMOS:SENTENCECAPS
dag.wikipediaMOS:LISTCAPS
dag.wikipediaMOS:MARINE
dag.wikipediaMOS:VAR
dag.wikipediaMOS:STYLEVAR
dag.wikipediaMOS:SO
dag.wikipediaMOS:PREFIX
dag.wikipediaMOS:SIR
dag.wikipediaMOS:LADY
dag.wikipediaMOS:GID
dag.wikipediaMOS:LORD
dag.wikipediaMOS:DEADNAME
dag.wikipediaMOS:CHRONOLOGICAL
dag.wikipediaMOS:BLPTENSE
dag.wikipediaMOS:SINGULARTHEY
dag.wikipediaMOS:LEADELEMENTS
dag.wikipediaMOS:NB
dag.wikipediaMOS:NEOPRONOUN
dag.wikipediaMOS:LEADCITE
dag.wikipediaMOS:INTRO
dag.wikipediaMOS:BEGIN
dag.wikipediaMOS:LEADSENTENCE
dag.wikipediaMOS:CITELEAD
dag.wikipediaMOS:LEADCLUTTER
dag.wikipediaMOS:USA
dag.wikipediaMOS:1STABBR
dag.wikipediaMOS:LEADORDER
dag.wikipediaMOS:SECTIONHEAD
dag.wikipediaMOS:SECTIONS
dag.wikipediaMOS:SECTIONSTYLE
dag.wikipediaMOS:CONSISTENT
dag.wikipediaMOS:COMMONALITY
dag.wikipediaMOS:ARTCON
dag.wikipediaMOS:BLANKLINE
dag.wikipediaMOS:POSTNOM
dag.wikipediaMOS:ENGVAR
dag.wikipediaMOS:HEADINGS
dag.wikipediaMOS:RETAIN
dag.wikipediaMOS:NOBACKREF
dag.wikipediaMOS:SIGNIFCAPS
dag.wikipediaMOS:PROPERNOUN
dag.wikipediaMOS:HYPHENCAPS
dag.wikipediaMOS:BIOCAPS
dag.wikipediaMOS:ANGLO-
dag.wikipediaMOS:EMPHCAPS
dag.wikipediaMOS:PROPERNAME
dag.wikipediaMOS:CAPSACRS
eo.wikinewsMOS: sukerkonsumadon ni forte reduktu
gur.wikipediaMOS:EMPHCAPS
gur.wikipediaMOS:TMCAPS
gur.wikipediaMOS:GOD
simple.wikipediaMOS:RETAIN
simple.wikipediaMOS:SCIENTIFIC
test.wikipediaMOS:L
en.wikipediaMOS:THEY
en.wikipediaMOS:THE
en.wikipediaMOS:PSEUDO
en.wikipediaMOS: OVERLINKING
en.wikipediaMOS:EL
en.wikipediaMOS:EM

Update: the list only includes non-redirect pages as redirects can not be searched via global search; See T363568: Global search can not find redirects by title.

Interesting. I see that all the MOS on the Dagbani Wikipedia were copied from the English Wikipedia and the text remained untranslated. I am keenly waiting for feedback regarding the English pages.

List of pages that will be affected:

Generated using Wikimedia Global Search on 2024-04-26 11:28

That list is not exhaustive, because there are a lot more in enwiki.

  • Create the wiki at some other name like "mos-x-deconflict.wikipedia.org" (using the BCP47 private subtag syntax is probably the least bad). This would work, probably without major changes, but doesn't feel right.

This is a no-go from my POV. We're trying to minimize the amount of irregularities on Wikimedia language codes vis-à-vis ISO codes, and this would break with that.

  • Create the wiki at mos.wikipedia.org and add hacks to the dumpInterwiki script in MediaWiki-extensions-WikimediaMaintenance so, either only for enwiki or for all wikis mos: doesn't become an interwiki. I've written a patch for this locally, and it's similar to what we already do for Swedish ("s:" isn't an interwiki on Swedish Wikipedia because they abbreviate "saint" as s:t), but that caused problems like T177913 and probably others.

Sounds doable, but how will interwiki links to the Mossi Wikipedia be added to enwiki then? AFAIK, "under the hood" Wikibase still inserts interwiki links with the [[old:syntax]], it's just that it happens out of sight.

  • Convert the MOS: enwiki pseudo-namespace to a real namespace (which overrides interwikis), and do something similar with WikimediaMaintenance to add an interwiki with some other name like "mos-x-deconflict"

Agree with @Peachey88 that this should be done regardless.

  • Someone on enwiki suggested making interwiki prefixes case-sensitive and having "mos": point to the project while "MOS:" is a pseudo-namespace. This isn't how either concept currently works, and will probably break things elsewhere

I've been thinking the same thing. Not to make all interwiki prefixes case-sensitive, but this one specifically. But I haven't looked at the relevant code yet to see how feasible that is.

Convert the MOS: enwiki pseudo-namespace to a real namespace (which overrides interwikis)

Regardless this should happen, fake namespaces just causes issues.

Note: Chinese Wikipedia has four fake namespaces for shortcut only: MOS, LTA (long-term abuse), NC (naming convention), NT (notability guideline). Though lta, nc and nt are not valid ISO language codes for any language.

Pppery renamed this task from Decide what to do about English Wikipedia Manual of Style pseudo-namespace conflicting with Mooré Wikipedia to Decide what to do about Manual of Style pseudo-namespaces conflicting with Mooré Wikipedia.Apr 26 2024, 3:11 PM

Someone on enwiki suggested making interwiki prefixes case-sensitive and having "mos": point to the project while "MOS:" is a pseudo-namespace. This isn't how either concept currently works, and will probably break things elsewhere

Note: since search box will try case-insensitive matches, users can type mos:var in lowercase and be redirected to the correct manual of style page. Such workflow will break.

Sounds doable, but how will interwiki links to the Mossi Wikipedia be added to enwiki then? AFAIK, "under the hood" Wikibase still inserts interwiki links with the [[old:syntax]], it's just that it happens out of sight.

Presumably the same way interwikis to Swedish Wikisource are currently handled, with a "mos-x-foo" interwiki existing and being used. I admittedly haven't looked into this locally, and don't feel like setting up Wikibase locally to do so.

  • Create the wiki at mos.wikipedia.org and add hacks to the dumpInterwiki script in MediaWiki-extensions-WikimediaMaintenance so, either only for enwiki or for all wikis mos: doesn't become an interwiki. I've written a patch for this locally, and it's similar to what we already do for Swedish ("s:" isn't an interwiki on Swedish Wikipedia because they abbreviate "saint" as s:t), but that caused problems like T177913 and probably others.

Sounds doable, but how will interwiki links to the Mossi Wikipedia be added to enwiki then? AFAIK, "under the hood" Wikibase still inserts interwiki links with the [[old:syntax]], it's just that it happens out of sight.

If "mos:" only doesn't work at enwiki, then "m:mos:" would still work there, correct? (Currently, for example, both links to "[[m:ms:]]" and searches for "m:ms:" [in the search bar at the top of every page] work to get you to the Malay Wikipedia. But that search doesn't work on the Special:Search page.) OTOH, this wouldn't work if the prefix is disabled everywhere.

Of course, adding local magical interwiki links of the kind provided remotely by Wikidata would not be possible, but would the ones provided by Wikidata still work?

"m:mos" would indeed. I have no idea how Wikidata interacts with this system, though.

Although it looks like Meta is currently treated by the interwiki system as a Wikipedia even though it isn't one, so if the decision is to de-interwiki "mos:" on all Wikipedias (as opposed to de-interwikiing it only on en, zh, and dag wikipedias) it will need more special-case code.

Although it looks like Meta is currently treated by the interwiki system as a Wikipedia even though it isn't one

Bummer. Is this a holdover from the days when Meta was considered the "metawikipedia"? Might it be easier to change that than to figure out some other workaround (assuming it is necessary)?

The easiest solution in my view is to let mos: become a normal interwiki prefix like all others. Yes, this breaks a couple thousand links on the English Wikipedia. Most of them (those not in old edit summaries) can be fixed by a few bot runs. For the people who type in shortcuts or follow links from old edit summaries, just temporarily (for a year or two) turn the target pages on the new Moore Wikipedia into soft redirects with information that this is a deprecated shortcut.

I do not think weird special cases in the interwiki system are a good idea. (And I say that as an enwiki editor who likes the MOS: shortcuts and has defended them at RFD for more than a decade).

Dzahn subscribed.

Removed Wiki-Setup because that is about creating new wikis and I think not really related.

Disagree - this is fundamentally about the creating new wikis process.

The easiest solution in my view is to let mos: become a normal interwiki prefix like all others. Yes, this breaks a couple thousand links on the English Wikipedia. Most of them (those not in old edit summaries) can be fixed by a few bot runs. For the people who type in shortcuts or follow links from old edit summaries, just temporarily (for a year or two) turn the target pages on the new Moore Wikipedia into soft redirects with information that this is a deprecated shortcut.

I do not think weird special cases in the interwiki system are a good idea. (And I say that as an enwiki editor who likes the MOS: shortcuts and has defended them at RFD for more than a decade).

I think you're vastly underestimating the scale of the problem:

Unfortunately I don't think breaking all those links on en.wikipedia is viable. Even if you could convince the community to start using longer WP: style shortcuts (good luck!) and update the documentation everywhere, you'll still break all the uses in past edit summaries. A quick query suggests that's around 42,000 uses of MOS: in June 2023 alone.