Wiktionary:Grease pit/2009/November

November 2009

Page Wiktionary:All Wikisaurus pages WT:WSI needs a technical fix

This page has some sort of script to display all the pages in Wikisaurus. However, since the number of Wikisaurus entries now exceeds the number that can be displayed on one page, the list runs out at G.

Anyone know how to get this page to list all the Wikisaurus entries ? --Richardb 22:46, 1 November 2009 (UTC)

Someone "merged" the code for Special:AllPages and Special:PrefixIndex a while ago, in the mistaken idea that they were improving things; they broke a number of things in the process, and now no-one seems to be interested in fixing any of them. You could go enter a bug asking that the two functions be separated again, and proper function restored, but don't expect anything to happen ... Robert Ullmann 15:31, 4 November 2009 (UTC)

Can anyone give me a pointer or link to "enter a bug" ?--Richardb 00:38, 14 November 2009 (UTC)

Try bugzilla: -- Prince Kassad 11:13, 14 November 2009 (UTC)

New RC patrolling concept

Hello, is there any needs to add the site in https://bugzilla.wikimedia.org/show_bug.cgi?id=21517? JackPotte 13:03, 15 November 2009 (UTC)

I don't think we trust people enough to let anyone patrol edits, particularly given how stringent we are over formatting. Conrad.Irwin 13:36, 15 November 2009 (UTC)

I'd prefer this system to flagged revisions, simply because we can see a ! directly into the RC list. Moreover, it's already used in fr.w, and it.wikt. JackPotte 19:31, 15 November 2009 (UTC)

We already have patrolling enabled, and a patroller group (which we don't use as yet, limiting it to sysops). We could set up process for granting patroller to non-sysops, but that was generally thought to be un-needed. Users that aren't sysop or patroller don't see the flags or the links; it is invisible, except that anyone (IIRC) can read the patrol log. Robert Ullmann 13:07, 13 December 2009 (UTC)

Couple of things

1. There's been a request that {{fr-noun}} have a p2 option (2nd plural) like businessman has the plural businessmen or businessmans.
2. {{cite-book}} and other citations templates should have a translation parameter, right? See Citations:seignor.
3. There was a third thing, but I can't think of it right now

Mglovesfun (talk) 10:18, 20 November 2009 (UTC)

Chinese Cantonese and Mandarin use the same alphabet

Chinese Cantonese and Mandarin use the same alphabet, aside from trends that have resulted of illiteracy. All Chinese references should be grouped under one group.

So do German and English, should we group those two under ==Germanic== ? --Ivan Štambuk 13:28, 20 November 2009 (UTC)

Indeed. Mglovesfun (talk) 12:13, 21 November 2009 (UTC)

Non sequitur. Cantonese and Mandarin is not written in an alphabet. They both use a single logosyllabary, the Han script. 124.214.131.55 15:44, 21 November 2009 (UTC)

subbing western angle brackets w CJK

It appears that our angle bracket articles, 〈 and 〉, use CJK brackets (U+3008/9) rather than Western ones (U+2329/A). This appears to be a technical problem, because I'm taken to the same page regardless of which I enter into the search window, and when I tried correcting them at Template:punctuation, it didn't even register as an edit.

Unicode deprecates U+2329/A for mathematical use because of their "canonical equivalence" to CJK punctuation U+3008/9. Could this be the source of the problem? There are non-math uses, such as in epigraphy, and SIL IPA fonts support U+2329/A but not the math bra-ket angle brackets, so this Unicode's note does not mean they are obsolete. kwami 02:45, 24 November 2009 (UTC)

I've run across this with the Greek semicolon (U+037E) being converted to the regular semicolon ';' (U+003B). Is this due to the Unicode standard or the mediawiki implementation? Is there a list somewhere of the ones that are automatically converted? This list should be public. It would be similar to Appendix:Unsupported titles. --Bequw → ¢ • τ 05:05, 25 November 2009 (UTC)

This is due to MediaWiki normalizing everything to NFC (see w:Unicode equivalence#Normalization). It is a "feature" designed to cope with the fact that unicode has multiple ways of representing the same character (i.e. á could be U+00E1 or U+0061 U+0301 — either an accented A, or an A and an accent). Sadly, some bright sparks decided that some characters are so similar that they should have the same canonical form, which really begs the question of why they exist in multiple places in the standard anyway (as far as unicode is concerned the characters are identical). This lose matching of characters behaviour was supposed to be defined by NFKC which is compatibility decomposition, in which ¹ and 1 are considered to be equivalent, as they represent pretty much the same thing, or as close as makes no odds. MediaWiki's policy of converting to NFC is based on Unicode recommended practice, and helps our pages conform to the other 99.98%1 of the internet that uses NFC, so this bug is definitely a Unicode one. Conrad.Irwin 21:49, 8 December 2009 (UTC)

Started Appendix:Unicode normalization. I remember Stephen(?) talking about other "issues" with normalization (maybe involving diacritics). If there are, please note them. --Bequw → ¢ • τ 22:43, 9 December 2009 (UTC)

The reason for the actions of the "bright sparks" was one of the essential design objectives of Unicode/UCS: That data can be round-trip converted from any of a very wide range of national and international standards into UCS and back, with the result being identical to the input. So if any one of those standards has two different code points for the "same" character, UCS provides two different code points. Then one of those two (or more) is considered to be the normalized form. In a Unicode/UCS "native" application, like the wikis, normalization is standard practice, and the other code points never appear in the (stored) data. At the same time, an application like email can convert a sent message in (e.g.) Big5 to UTF-8, and the receiver can convert it to a desired target code; if that code is Big5, the data is exactly as was sent. So it isn't a "bug", it's a feature ... it was done that way intentionally. Robert Ullmann 11:21, 14 December 2009 (UTC)

Template:fro-decl-noun

This and the corresponding Template:fro-decl-adj seem to have a bug. See caitif (as of today's date) as it swallows up the following section in the box, although it shouldn't. Help please! Mglovesfun (talk) 13:24, 25 November 2009 (UTC)

fixed now, I think. Conrad.Irwin 16:16, 25 November 2009 (UTC)

Category:American English English

Hands up, who broke that? Mglovesfun (talk) 17:27, 26 November 2009 (UTC)

{{welcomeip}}

Can this be made so that the wikipedia paragraph is optional? I know nothing about templates L☺g☺maniac chat? 17:42, 26 November 2009 (UTC)

How about this? wp=no eliminates the Wikipedia paragraph. --Yair rand 17:51, 26 November 2009 (UTC)

Thank you very much! L☺g☺maniac chat? 01:46, 27 November 2009 (UTC)

random entry language?

i'm trying to use the random entry API. it doesn't appear that it honours the site language. for example, if i do,

http://en.wiktionary.org/w/api.php?action=query&list=random&format=json

the word returned can be any language. why doesn't the query honour the site language (http://en.wiktionary...)? is there a parameter i can pass to indicate the desired language? i could not find one.

i also saw the "random entry by language" link on the left here. that's fine, but i need to access this programmtically, through a web service.

any help is appreciated.

No there is no way to do this on-site, which is why the off-site tool was created. What do you need it for? I can provide the lists used to create the indexes in most languages which you can then randomise yourself, if that'd work - otherwise you can try asking User:Hippietrail if his random language by entry supports multiple returns in one request. (It does actually honour the language, this site has "all words in all languages" with definitions in English, compare http://fr.wiktionary.org ) Conrad.Irwin 22:51, 28 November 2009 (UTC)