User talk:Conrad.Irwin/¹

I archive my talk page when it gets to ~75 topics by moving the first 50 to a new subpage. Please do not edit the archive pages, if you want to talk about something again - copy it back to my current talk page or just start a new topic there and link back.

• User_talk:Conrad.Irwin/1 Jan 07 → Feb 08
• User_talk:Conrad.Irwin/i Feb 08 → Apr 08
• User_talk:Conrad.Irwin/一 Apr 08 → Jun 08
• User_talk:Conrad.Irwin/α' Jun 08 → Jul 08
• User_talk:Conrad.Irwin/١ Jul 08 -> Nov 08
• User_talk:Conrad.Irwin/१ Nov 08 -> Mar 09
• User_talk:Conrad.Irwin/¹ Mar 09 -> Dec 09
• User_talk:Conrad.Irwin/₁ Jan 10 -> Mar 10

If they don't want us to use the templates, then delete them! WritersCramp 14:10, 14 March 2009 (UTC)Reply

I wish they would :D. Nah, the right hand ones are OK when you don't have an image, but if you have lots on the right-hand side the page looks wierd (to me anyway). It's not a big problem. Conrad.Irwin 16:17, 14 March 2009 (UTC)Reply

Sorry, I'm new to wiktionary, but under the entry for gamin, your bot added ænigma. This is not really an anagram, since the neither the æ character nor any analogue for it exist in the original word, except for the a, which is used at the end of ænigma. Just thought you might like to know. Sorry if I didn't post to your talk page right, I couldn't figure out what you were trying to say. 76.125.225.194 22:19, 17 February 2010 (UTC)Reply

I encountered the word in a book I was reading (), but I guess I should have checked the other results. Teh Rote 20:05, 14 March 2009 (UTC)Reply

Cool, it'd be worth adding that to the citations page so that we can find it again quickly. Conrad.Irwin 20:08, 14 March 2009 (UTC)Reply

Hey again. I was asking Atelaes back in january how to make a index and he mentioned a slow way of doing it and certain magical ways of creating one with the help of a great wizard called Conrad =). I have added alot of words since january and it would be fun to have an index, If you find the time to help me it would be awesome. -Edelstam 17:11, 25 March 2009 (UTC)Reply

Hey Conrad, been a while, thanks for updating the mapudungun index again. I can't get on the irc anymore because got this TOR program on my computer and I can't get it to work with some IRC servers. So I'll have to ask you here. I have been working some with trying to start up the mapudungun wikipedia and there they are working with some separate solutions for the different orthographies(what I mistakenly have called alphabets) within one wikipedia. I looked some at the serbian wikipedia where they use two different alphabets for example. I got some comments about the mapudungun language category here on english wiktionary, that it was confusing to have words from two different orthographies in one place. So I was thinking if we could possibly adopt some solutions the serbian language category has here. They have for example a index with the two different alphabets, look here. Since we only have words in two orthographies it would mean a similar solution to the one the serbian category uses. What do you think about this, is it alot of work? -Edelstam 20:14, 26 April 2009 (UTC)Reply

Hi Edelstam, the main problem is automatically telling the two apart. If there is some indication in the page text, or even a way of testing the page titles, that will tell me which alphabet which entry belongs to, then it shouldn't be that hard. I can obviously look for some letters that only appear in one of the alphabets, but I don't know enough about Mapudungun to know if I will always be able to tell the difference. Conrad.Irwin 20:19, 26 April 2009 (UTC)Reply

Yeah I was thinking about how to separate the words since they don't look different like the Serbian Cyrillic- and Latin alphabet words. Right now all words written in the raguileo orthography have this in the article tho: (using Raguileo Alphabet), look at cew for example. Some words are the same in both orthographies and in those cases they look like this word: awka. Words written in Unified alphabet look like this: chafod. Can you use that for the bot? -Edelstam 20:47, 26 April 2009 (UTC)Reply

Yes, that's enough. I'll try to change the indexing scrit at some point. Conrad.Irwin 08:19, 28 April 2009 (UTC)Reply

Hello. You are an admin here, could you please see what's wrong with this entry? Looks like it's caused by a recent change to template:attention which can be edited only by admins. 213.163.65.33 03:46, 28 March 2009 (UTC)Reply

Now fixed, thanks for noticing. Conrad.Irwin 10:00, 28 March 2009 (UTC)Reply

This diff raised a question for me: There's a direct link to en.wp, plus the wp template which links to a disambiguation page. What would be the appropriate layout/template use? I really don't know. - Amgine/^talk 16:45, 29 March 2009 (UTC)Reply

Would you be willing to look atthis convo? I wonder if you understand the wotd template better than anyone. Many thanks. -Atelaes λάλει ἐμοί 19:27, 29 March 2009 (UTC)Reply

Please don't remove the "Edit" link. I use that all the time. --EncycloPetey 21:46, 29 March 2009 (UTC)Reply

Sorry. I'd neglected to appreciate that it appears in other places. Do you use the link that appears on the Main Page, or exclusively where it appears elsewhere? Conrad.Irwin 22:30, 29 March 2009 (UTC)Reply

Hi Conrad,

Since you commented at Wiktionary:Beer parlour#Transwikis from other Wiktionaries., I wanted to make sure you were aware of the resultant vote, Wiktionary:Votes/2009-03/Transwikis from other Wiktionaries.

—Ruakh_TALK 13:41, 31 March 2009 (UTC)Reply

I see you got rid of the #switch. I imagine it was in place so that people don't use this template to redirect to alternative spellings, misspellings, and the like: see template talk:user redirect. Did you remove it on purpose or en passant?—msh210℠ 22:41, 31 March 2009 (UTC)Reply

Deliberately, it seems that, if we are to have pages like WB:WB, they should use it as well. Conrad.Irwin 22:55, 31 March 2009 (UTC)Reply

I see no reason to keep WB:WB, and doubt there are other mainspace pages that are appropriate soft redirects, but perhaps there are. Very well.—msh210℠ 23:20, 31 March 2009 (UTC)Reply

I would also like to delete them, but I think someone told me not to (may have been a year or so ago, so might "just do it" now :D) Conrad.Irwin 23:22, 31 March 2009 (UTC)Reply

Are you being careful to use only the "I" individual languages? And skip the ones that we use the -1 code for? Robert Ullmann 09:50, 6 April 2009 (UTC)Reply

So we are going to have thousands of unchecked templates? Are you checking everything you should? (You are not.) Likely to be more of a mess than a help. (Sorry.)

Would be good to stop until a list of issues is addressed.

What are you using as the source? Robert Ullmann 10:16, 6 April 2009 (UTC)Reply

I had not looked at the "I" flag - as I was using the "Language Names Index" from http://www.sil.org/iso639-3/download.asp - which I had assumed very likely to be correct (it is the definition after all) However there do seem to some discrepancies between what we have and what they say; mainly only spelling. I had thought about checking against Ethnologue (run by the same organisation) but it doesn't seem to have been updated fully with the 2009 changes. I could change to using the full code-set and ignore those 62. Conrad.Irwin 10:55, 6 April 2009 (UTC)Reply

However current consensus seems to be to include them. See {{aka}} {{ara}} {{aym}} {{aze}} {{bal}} {{bik}} (I got bored of checking...) all macrolanguages done well before I started. Conrad.Irwin 11:30, 6 April 2009 (UTC)Reply

• aab Alumu-Tesu: {{{l|]}}}<noinclude>]</noinclude>
• abe Western Abnaki: {{{l|]}}}<noinclude>]</noinclude>
• abk Abkhazian: {{{l|]}}}<noinclude>]</noinclude>
• abl Lampung Nyo: {{{l|]}}}<noinclude>]</noinclude>
• ace Achinese: {{{l|]}}}<noinclude>]</noinclude>
• ach Acoli: {{{l|]}}}<noinclude>]</noinclude>
• acm Mesopotamian Arabic: {{{l|]}}}<noinclude>]</noinclude>
• ady Adygei: {{{l|]}}}<noinclude>]</noinclude>
• aht Ahtena: {{{l|]}}}<noinclude>]</noinclude>
• aib Ainu (China): {{{l|]}}}<noinclude>]</noinclude>
• ain Ainu (Japan): {{{l|]}}}<noinclude>]</noinclude>
• ajg Aja (Benin): {{{l|]}}}<noinclude>]</noinclude>
• amu Guerrero Amuzgo: {{{l|]}}}<noinclude>]</noinclude>
• anc Ngas: {{{l|]}}}<noinclude>]</noinclude>
• ang Old English (ca. 450-1100): Old English<noinclude>]</noinclude>
• apj Jicarilla Apache: {{{l|]}}}<noinclude>]</noinclude>
• apk Kiowa Apache: {{{l|]}}}<noinclude>]</noinclude>
• apl Lipan Apache: {{{l|]}}}<noinclude>]</noinclude>
• apm Mescalero-Chiricahua Apache: {{{l|]}}}<noinclude>]</noinclude>
• arc Imperial Aramaic (700-300 BCE): {{{l|]}}}<noinclude>]</noinclude>
• arc Official Aramaic (700-300 BCE): {{{l|]}}}<noinclude>]</noinclude>
• are Western Arrarnta: {{{l|]}}}<noinclude>]</noinclude>
• arn Mapuche: {{{l|]}}}<noinclude>]</noinclude>
• ast Asturleonese: {{{l|]}}}<noinclude>]</noinclude>
• ast Bable: {{{l|]}}}<noinclude>]</noinclude>
• ast Leonese: {{{l|]}}}<noinclude>]</noinclude>
• auq Korur: {{{l|]}}}<noinclude>]</noinclude>
• auy Awiyaana: {{{l|]}}}<noinclude>]</noinclude>
• ava Avaric: {{{l|]}}}<noinclude>]</noinclude>
• awb Awa (Papua New Guinea): {{{l|]}}}<noinclude>]</noinclude>
• azb South Azerbaijani: {{{l|]}}}<noinclude>]</noinclude>
• aze Azerbaijani: {{{l|]}}}<noinclude>]</noinclude>
• bak Bashkir: Bashkir<includeonly>{{#if: {{{1|}}}|]|]}}</includeonly><noinclude>]</noinclude>
• bas Basa (Cameroon): {{{l|]}}}<noinclude>]</noinclude>
• bav Vengo: {{{l|]}}}<noinclude>]</noinclude>
• bcl Central Bicolano: {{{l|]}}}<noinclude>]</noinclude>
• bej Bedawiyet: {{{l|]}}}<noinclude>]</noinclude>
• bem Bemba (Zambia): {{{l|]}}}<noinclude>]</noinclude>
• bez Bena (Tanzania): {{{l|]}}}<noinclude>]</noinclude>
• bfe Tena: {{{l|]}}}<noinclude>]</noinclude>
• bgc Haryanvi: {{{l|]}}}<noinclude>]
• bgk Buxinhua: {{{l|]}}}<noinclude>]</noinclude>
• bin Edo: {{{l|]}}}<noinclude>]</noinclude>
• bjn Banjar: {{{l|]}}}<noinclude>]</noinclude>
• bla Siksika: {{{l|]}}}<noinclude>]</noinclude>
• bno Bantoanon: {{{l|]}}}<noinclude>]</noinclude>
• bnv Bonerif: {{{l|]}}}<noinclude>]</noinclude>
• bnv Edwas: {{{l|]}}}<noinclude>]</noinclude>
• bot Bongo: {{{l|]}}}<noinclude>

There is no "consensus" to include maccrolanguages: they were wrongly created and must be deleted; we want templates for precisely the set of languages we want to use as L2 headers. In some specific cases we do want to use a macrolanguage code where it is better to treat the I languages as dialects (or in some combination). This requires looking at individual cases, with some knowledge of the language(s); this is why we want to create codes only as needed.

I'm afraid you are creating a massive raft of problems to add to the (fairly reasonable) number we already have. IMHO, it would be better to delete all code templates for languages not referenced anywhere at this time. We do carefully allow new contributors to add L2 headers and translations lines w/o having to have code templates, which can the be sorted with care. Robert Ullmann 11:48, 6 April 2009 (UTC)Reply

I agree there are difficulties with language templates, but which problems am I actually causing? It is just as easy, if not easier, to correct a template that is found to be wrong than for someone to lookup the language code for a language they are trying to add; and there will be (far?) fewer mistakes to correct than there are templates to create. Conrad.Irwin 12:09, 6 April 2009 (UTC)Reply

It is "recursion with a fixed iteration limit". SemperBlotto 16:25, 6 April 2009 (UTC)Reply

I went for "to a fixed depth" as, in my mind anyway, iteration is orthogonal to recursion. Conrad.Irwin 16:44, 6 April 2009 (UTC)Reply

Hi Conrad -- it's just User:JesseW/not logged in again. Thanks for the welcome, anyway! 75.214.226.186 18:04, 9 April 2009 (UTC)Reply

Oops, I guess you're right. The full text of the response's Status-Line is exactly “HTTP/1.1 404 Not Found”, consisting of HTTP-Version, Status-Code and Reason-Phrase. I'll adjust the phrasing. —Michael Z. 2009-04-11 22:56 z

I checked http://dictionary.reference.com/browse/upbeat

Are there upbeats, downbeats and somethign else? Or are there just upbeats and downbeats? RJFJR 15:59, 14 April 2009 (UTC)Reply

Just up and down, w:Beat_(music)#Downbeat is more how I understand the words. Conrad.Irwin 16:03, 14 April 2009 (UTC)Reply

OK, thank you for warnig. I'll keep that in my mind =) best wishes, Sinek 16:56, 14 April 2009 (UTC)Reply

Hi Conrad,

I tried to add a translation using the translation-thing, and it didn't work out. I filled out the form, and then I didn't see any way to save. So, I figured I'd continue playing around, and I clicked one of the arrows, and suddenly I got a save-button, which I clicked. But then I understood that all it was doing was saving the re-balancing, and suddenly it occurred to me to click the preview-button. But it was too late: a message came back that saving was complete, and I lost the save button. And as far as I could tell, there was no longer any way for me to recover what I'd entered; I'd have to redo it.

Also, since the preview-button clears out the form, and the undo-button doesn't restore it, there doesn't seem to be a way to fix a typo, to add a qualifier, etc. after clicking the preview-button, except by re-entering everything. (Admittedly, that wouldn't be a big deal if I had a Hebrew keyboard, but I don't, so it takes me a bit of effort to enter Hebrew text. I found it disheartening when my translation just disappeared into the ether.)

—Ruakh_TALK 17:48, 14 April 2009 (UTC)Reply

Hmm, it is necessary to click "Preview" before save, whether this is desirable is questionable, but I think generally it is; as typos are harder to spot when they are not in-situ. Maybe I need to make it clearer that "Preview" is the first step, but how to do that I'm not sure. (We should really organise some training for using Wiktionary...). I debated about making the "Undo" button re-fill-out the form, but wasn't sure whether that would be very usual either, as if people want something undone the last thing they want is it to come back and get in their way. The main problem is neither of those though, it needs to hold onto previous edits while saving, which I think I can fix. Conrad.Irwin 18:08, 14 April 2009 (UTC)Reply

I agree requiring Preview before Save is desirable, for the reason you say. (Perhaps the usual method of saving a page should also require it!  :-P ) But I agree with Ruakh that undoing should prefill, for the reason he says.—msh210℠ 18:34, 14 April 2009 (UTC)Reply

The “Add” label on the fields implies that using the fields performs the action. Might be clearer if it said “New”, or was removed altogether, and the action word were restricted to the button which performs the action. “Preview” is a bit confusing, because although that button does incidentally change mode to previewing the first time it's clicked, the editor only sees it add a translation every time it's clicked. —Michael Z. 2009-04-14 20:51 z

I was wondering why you so severely pruned this entry in this edit when you formatted the definition line? I could possibly see removing the hypernym entry as being covered by the definition, but not the rest of the pruning, and in any case, pruning of that extent would seem to warrant a more substantive comment than "fmt". — Carolina wren discussió 14:41, 15 April 2009 (UTC)Reply

I made this change. It seems to have overwritten yours, I thought it was supposed to give warnings of conflict in such cases, but don't remember getting one. Sorry. Conrad.Irwin 16:59, 15 April 2009 (UTC)Reply

No problem, just odd given that my edit had been made several hours earlier. — Carolina wren discussió 18:01, 15 April 2009 (UTC)Reply

Thanks for your warm welcome. Your boxes are a very useful tool for adding translations! --Hooiwind 17:32, 15 April 2009 (UTC)Reply

I am thanking you as well, Conrad.Irwin! The script for translations makes things faster. --Icqgirl 11:16, 2 May 2009 (UTC)Reply

The new button for adding translations is not working right for me: it responds only with "Please use a language code" but I do use a language code! However, I gather from the above message by Hooiwind that the tool is working all right for him. What is wrong? —AugPi 18:09, 15 April 2009 (UTC)Reply

It appears that it has stopped working under Internet Explorer since I tested it. I am investigating now, and hope to have it fixed in the next few minutes. Conrad.Irwin 18:15, 15 April 2009 (UTC)Reply

Something is amiss. Currently, none of the Translations sections are collapsed for me, and there is no option to Show/Hide. The Related terms boxes are also affected. --EncycloPetey 20:17, 15 April 2009 (UTC)Reply

I can't see this in Firefox, Safari, Opera, IE6, IE7, Chrome or even Konqueror. What are you using? Did a hard re-fresh do anything about it? Conrad.Irwin 21:01, 15 April 2009 (UTC)Reply

I'm using Safari, but the problem has (mostly) gone away now. The problem is now intermittent, and goes away if I empty the cache, then refresh the page. However, I have to do this again when I go back to the page again later. I'm also seeing an extra bullet point for a non-existent translation at the end of translation lists. --EncycloPetey 21:17, 15 April 2009 (UTC)Reply

Sorry, caused by me trying to fix the IE issues too quickly. Now fixed. Conrad.Irwin 21:33, 15 April 2009 (UTC)Reply

Hi,

I like your assisted translator. Well done! I don't use for Chinese entries, though because it doesn't allow traditional/simplified distinction. The other thing, I use Chinese, as the title, not Mandarin, which is the Standard Chinese dialect, anyway. By default, the entry is in written Chinese, shared by all dialects with some exception. The Mandarin's pinyin is also standard for China. Dialects could be separated by *: notation (Cantonese, Minnanhua, etc.

I created two templates recently:

• {{zh-tsp|中國|中国|Zhōngguó}} produces: trad. 中國／中国 (Zhōngguó), simpl. 中国 (Zhōngguó) (pinyin: Zhōngguó)
• {{zh-zh-p|北京|Běijīng}} produces: Template:zh-zh-p

If this is possible, could you add the functionality for the Chinese translations, so that:

• Chinese: trad. 中國／中国 (Zhōngguó), simpl. 中国 (Zhōngguó) (pinyin: Zhōngguó) (this entry has different traditional and simplified characters)
• Chinese: Template:zh-zh-p (this entry has identical traditional and simplified characters)

Anatoli 03:57, 17 April 2009 (UTC)Reply

Not in the near future, sorry. I was thinking of adding a field to the "More" view that allowed you to enter the entire line, so you could type "Chinese: {{zh-zh-p|...}}", it's not ideal but I think it might be helpful for some people. Supporting nested translations, for Chinese and Greek, is also on the todo list, but I need to have a chat with Ullmann about which languages are nested and under what circumstances. Conrad.Irwin 09:29, 17 April 2009 (UTC)Reply

Thanks. Anatoli 09:35, 17 April 2009 (UTC)Reply

After a year, I think it's time to re-run Index:Latin. There are many, many new entries, and many previously indexed entries have been removed as errors (especially those beginning with J and K). --EncycloPetey 19:25, 18 April 2009 (UTC)Reply

I'll try to get round to it tomorrow, I'll also update the formatting to match the other indices generated by Conrad.Bot (unless you have a strong objection thereto). Conrad.Irwin 23:23, 18 April 2009 (UTC)Reply

No objections, no. I don't use them all that much myself (except for cleanup purposes), but I know other people do make use of them. --EncycloPetey 04:15, 19 April 2009 (UTC)Reply

WHAT THE FUCK were you doing?

You sure as hell didn't check the results. I don't even know how to make the server store pages with no revisions, but whatever broken, fucked-up thing you did managed.

I am cleaning them up, one by one by page ID. It is 5 AM, I've been chasing this problem for several hours, and it looks like half a fucking day of manual cleanup. Thanks a lot.

Why ever didn't you at least READ the resulting templates with something? You would have found a problem? Eh? Robert Ullmann 01:57, 20 April 2009 (UTC)Reply

I did manually unmangle every pageID I could find so far, but someone is going to have to re-check all this crap. Robert Ullmann 02:08, 20 April 2009 (UTC)Reply

I will apologize for all the invective above. (:-) but I ask one thing: can you please learn why it is not a good idea to just blast through things? Adding all these required (and requires) checking all the work. Did you write a program to read all the templates and check them all? Or just write them and hope? ... best regards, Robert

The pages you have deleted, along with six others, threw an error on save. I made the, incorrect it would seem, assumption that an error on page save was equivalent to not saving. As it wasn't important to be utterly complete, I just logged "Error" and carried on. I can't vouch for the fact that all the others are fine, but given that everything you deleted logged an error I am confident enough that the others are not in this state. There is also a query running now on toolserver to find all pages with no revisions, just in case this problem has gone unnoticed elsewhere. (The people in #mediawiki seemed reluctant to admit that doing this was possible, and so now give the "recommended fix" as deleting the page...). Thank you for your help and time. Conrad.Irwin 09:08, 20 April 2009 (UTC)Reply

SELECT page_title FROM page LEFT JOIN revision ON rev_page = page_id WHERE rev_id IS NULL;
Empty set (52 min 47.11 sec)

Hi! the dump has run (unassisted, and okay ;-), so I did get all of them (from this case). I'm very curious how this managed to happen (and not very surprised that "they" would think it "impossible", it ought to be ;-)? How exactly did you do the "save"? (which code, index or api.php? etc?) And what was the error? (as far as you know). I'd like to see if I can maybe make it happen, and we can fix the underlying bug. Now that I've had some good sleep. Robert Ullmann 13:11, 20 April 2009 (UTC)Reply

I (foolishly) didn't record the actual error, I'm using mwclient, page = wikt.Pages; text = page.edit(); try: page.save(newtext) except: print "Error: %s %s" % (code, name). I can't really see anything in common with the edits, though there's clearly two main patches of problems, so I wonder whether it was something like database lag and a buggy anti-overload tool getting in the way. Conrad.Irwin 13:27, 20 April 2009 (UTC)Reply

page.save(text=newtext) (presumably?) can I see the actual code (you can email it if you don't want it floating about ;-). I'd like to set up the exact case, and then look at how it would trace through api.php. (but sleeping now I think) Robert Ullmann 22:22, 20 April 2009 (UTC)Reply

User:Conrad.Bot/iso-upload.py is it I think, I made a few modifications for testing after you reported the problem, but I think I undid them. Conrad.Irwin 22:36, 20 April 2009 (UTC)Reply

Doesn't look like anything odd on that side. Presumably I would have to hunt something in/under api.php that would cause an incomplete DB commit. (and I'm not doing that at 2AM ;-) thanks, Robert Ullmann 22:57, 20 April 2009 (UTC)Reply

Hello. I noticed that you reverted my self-revert on this entry. I initially entered a reference for its etymon under a "Notes"-thread (see the history of the entry) exactly as is done in Wikipedia. Then, the bot moved the "Notes"-thread on the top of the wikicode and it created a cite-error ("<ref> tags exist, but no <references/> tag was found"). The truth is that I don't know how to properly make a citation on Wiktionary yet, so I reverted myself (my rationale was that if the reference doesn't show up, there's no point having it in the code). But since you reverted me, the cite-error is back. Do you have a better idea on how to undo it while at the same time make the entry conform with the formatting standards of Wiktionary concerning citations? --Omnipaedista 18:31, 21 April 2009 (UTC)Reply

If not, I will probably remove the citation. But, I would really like to know your opinion first. --Omnipaedista 07:02, 25 April 2009 (UTC)Reply

Hi, sorry for not replying, isn't Atelaes fix correct? I just reverted you because it looked like you wre removing random stuff, sorry. Conrad.Irwin 08:55, 25 April 2009 (UTC)Reply

OK, no problem at all. I hadn't noticed Atelaes' intervention. The entry's fine now. --Omnipaedista 18:55, 25 April 2009 (UTC)Reply

Hello. Do you have here Recent Changes by language? Maybe only showing when a French entry is modified? --Rising Sun 13:46, 25 April 2009 (UTC)Reply

User:Visviva used to maintain lists at http://www.pnuteachers.com/wiktionary/, but they seem to have been turned off for the last week or so (and I've not seen Visviva around recently either). Conrad.Irwin 13:56, 25 April 2009 (UTC)Reply

Hi. Great work on making those indexes. However, here are some points:

1. By sorting Russian е and ё together, here is what I meant: your bot should temporarily replace every ё in the dump by е, sort such list as if ё did not even exist, than add ё-s back to the sorted list. This is the convention used in Russian dictionaries. Thus, these 5 words will be sorted like this:

Елисей, ёлка, еловый, ель, Ельцин

instead of

Елисей, еловый, ель, Ельцин, ёлка

as it is now.

Sorry, my fault - I have two normalization routines (one for the letters used as headings and page titles, the other for sorting) and I only fixed one of them. (will fix now) Conrad.Irwin 00:42, 26 April 2009 (UTC)Reply

Fixed. Conrad.Irwin 01:11, 26 April 2009 (UTC)Reply

2. The sorting of Armenian according to Unicode corresponds to old Armenian conventions in two points. First, is it possible to treat ու as a single, separate letter, and add words starting with it to Index:Armenian/ու? Right now they go under Index:Armenian/ո. And, whenever it's found in the middle of the word, it should be treated as the 34th in order, coming after ց and before փ in Index:Armenian, and not as a combination of ո + ւ. Besides, the letter և is treated by Unicode as the 6th in order, when it should be 37th - after ք and before օ. If you can't make this changes, no problem, old convention indexes are better than empty indexes.

This should be possible, I'll have a go over the weekend. Conrad.Irwin 00:42, 26 April 2009 (UTC)Reply

Now done, I think, but I find it quite hard to read the script, so I might have got it horribly wrong. Conrad.Irwin 01:11, 26 April 2009 (UTC)Reply

3. What happens if the translations or entries, from which the indexes were generated, are moved, changed or deleted? Will your bot automatically update the indexes? --Vahagn Petrosyan 22:47, 25 April 2009 (UTC)Reply

Yes, I run the script every fortnight or so - it takes about an hour to download the xml dumps process them and re-upload the indices. Conrad.Irwin 00:42, 26 April 2009 (UTC)Reply

I noticed that you'd been cleaning out the /0 pages, and have reverted the bots overwrite of you. Next time it runs, it should see the fixes you have made to the pages and not re-create the mess. Conrad.Irwin 01:11, 26 April 2009 (UTC)Reply

4. Just wanted to let you know that in Opera there is a minor indention problem in all indexes. See this screenshot for example. --Vahagn Petrosyan 22:47, 25 April 2009 (UTC)Reply

There's always one :D, I'll see if I can work out what's causing this, and then see if I can fix it for opera without breaking everyone else. (I suspect it's something property of the headings) Conrad.Irwin 00:42, 26 April 2009 (UTC)Reply

Awesome, you got it all right! Two, teeny-tiny things. If I move the remnants belonging to Index:Armenian/ու from Index:Armenian/ո to Index:Armenian/ու (just 3 articles at the bottom linking to the letter itself), will the bot overwrite me next time? Also, if I put a text like "Transliteration: č" at the top of each index page, will the bot overwrite me next time? --Vahagn Petrosyan 01:37, 26 April 2009 (UTC)Reply

Yes, it will. I can fix those "ու", not sure what to do with transliteration - I suppose if you pointed me to a list and showed me how you wanted it to be formatted, I could get Conrad.Bot to add it (it would then be easy to add for Russian too) - but it would not be trivial to change the code not to overwrite everything. Conrad.Irwin 07:34, 26 April 2009 (UTC)Reply

Thinking about it, I could add a trans= parameter to the {{index/Armenian}} call, which should then allow you to format it however. I'm intending on stealing the top list from Appendix:Armenian script. Conrad.Irwin 07:48, 26 April 2009 (UTC)Reply

I just noticed that after you added Index:Old Armenian to bot's maintenance list (which is sorted according to Unicode) the sorting rules of Index:Armenian have defaulted to Unicode as well. Can you return the two rules for modern Armenian, please? Namely:

• Index:Armenian/և should be 37th - after ք and before օ (and not Unicode's 6th)
• Index:Armenian/ու should be considered an independent letter: 34 th - after ց and before փ, and words starting with it should be removed from Index:Armenian/ո. --Vahagn Petrosyan 04:43, 23 July 2009 (UTC)Reply

I see you fixed everyting. Thanks! --Vahagn Petrosyan 07:31, 8 August 2009 (UTC)Reply

I just subst'd the template, from merge and replaced the second person singular. - Francis Tyers 10:17, 26 April 2009 (UTC)Reply

Alright, thanks. I'll fix the template. Conrad.Irwin 10:55, 26 April 2009 (UTC)Reply

As you can see in this edit, the assisted translations feature does not recognize or warn about edits made between the time the page is opened and the time an edit is saved, resulting in a loss of content/edits. See the edit history around that edit for a clearer picture. --EncycloPetey 04:39, 28 April 2009 (UTC)Reply

I am working on the fix for this. Conrad.Irwin 08:15, 28 April 2009 (UTC)Reply

This word is the plural of szülő (parent) and has the "plural of" template in its definition line - it is still listed in the index. Not urgent, but when you have a chance, would you please take a look at it. And thanks again for regularly updating the index! --Panda10 22:53, 29 April 2009 (UTC)Reply

It is linked to as a translation of parents. Fixing this is likely to take some time, as I had assumed that translations go only to the lemma form. Conrad.Irwin 00:10, 30 April 2009 (UTC)Reply

What do you think about creating a template such as {noindex} that would be added to entries the editors don't want to include in the index? Your code would check for this template. --Panda10 02:33, 10 May 2009 (UTC)Reply

Hmm, it sounds like a reasonable idea - and a similar template for including irregular form-ofs in the index might be nice, but a better name than {{NOINDEX}} which is already used to stop search-engine indexing of pages is needed. Perhaps {{doIndex}} and {{dontIndex}}? Conrad.Irwin 13:44, 10 May 2009 (UTC)Reply

It sounds good. --Panda10 19:06, 10 May 2009 (UTC)Reply

Hi ConnelCirwin. I've been improving my metadata server.

I want to build the master data from all the master sources automatically and periodically.

So I'm reading the master list of wiktionaries and the master list of ISO language codes and I've found these Wiktionaries which are not ISO codes:

• tokipona: A constructed language with no ISO code
• roa_rup: Aromanian, so why not the standard code rup?
• mo: Moldovan/Moldavian, to be deprecated in favour of combining with Romanian ro
• zh_min_nan: Some long-running dispute about the correct language code I think
• als: Tosk Albanian, its a code on Ethnologue but not in ISO, need to know more
• simple: Simple English, not a separate language
• sh: Serbo-Croatian, deprecated by ISO

So how should we handle such cases? I haven't yet included the English Wiktionary's Template namespace titles which are 2 or 3 letters or which are members of the language code category. There may be further anomolies there. — hippietrail 08:11, 30 April 2009 (UTC)Reply

For Serbo-Croatian read Wiktionary:About Serbo-Croatian (and see the discussion site providing justification), we had already been discussing this in the BP some five or six weeks ago and this is the outcome. I found the link - the discussion is here, but to-day it was archived, so one can no longer partake thereof. The uſer hight Bogorm converſation 08:22, 30 April 2009 (UTC)Reply

I do not wan to be inquisitive, but wherefore „Hi Connel“ on Conrad's talk page? The uſer hight Bogorm converſation 08:24, 30 April 2009 (UTC)Reply

See Wiktionary:Language code extensions; most of these are stupidity by the WM "language committee" which seems to be terminally clueless. (they denied Jerrais a WP because it "had not code", then allowed Norman, and give it the code for Narom (nrm). fortunately no wikt yet) Specifically:

• tokipona: we ignore, as with tlh (Klingon)
• roa_rup: standard code didn't exist at the time, should get changed
• mo: thus was/is a standard code, we can sort it when mo.wikt is closed/merged
• zh_min_nan: was supposed to be temporary, invented by IETF/IANA, not us should become nan (also zh-yue, the same, but has no wikt)
• als: Tosk Albanian; but used for "Alsatian", which does need an extension code, this is committee stupidity. however, they also decided to put the dictionary inside the pedia (Wort: namespace), so we just ignore it ...
• simple: should have been en-simple, but harmless
• sh: sh.wikt is basically dead but not closed yet, sr, hr, bs etc wikts are alive and well; eventually the Serbian will all be identified as such (as well as what is Bosnian and Croation) and the extraordinarily offensive Serbian Nationalist "Serbo-Croatian" crap will be removed

We've done a reasonable job of sorting these so far. Robert Ullmann 08:30, 30 April 2009 (UTC)Reply

• Actually there is a context my question. I'm making a tool to serve metadata about languages, designed to be used by other tools such as cirwin's editor.js so that they know whether to include gender fields, use script templates, need transliterations, provide interwiki links, etc. So I want to know whether and how to support things we use here but the standards don't.
• As for "extraordinarily offensive crap", some places are lucky. ISO allows Tibet no region code and tow's the government line on calling Burmese "Myanmar". I'm sure there's plenty more to be enraged about but all of that is besides the point and off topic.
• So how should we handle old Cyrillic Romanian/Moldovan/Moldavian spellings? I guess just Cyrl is fine but ro-Cyrl or does MD belong in there somewhere? — hippietrail 08:54, 30 April 2009 (UTC)Reply

Firstly, I am still called Conrad :p. Secondly, I've taken a completely pragmatic approach to this; although the situation on-wikt is not ideal, I've just copied it - doing anything else really isn't useful for what editor.js does. For a more generic meta-data thingy, I suppose it would be silly to copy wiktionary; so how you handle that I'm not really sure - probably with multiple "override" sets that can be used in place of the true values when a different "context" is requested by the client. The main problem I am coming across currently is that our Template:zh returns "Mandarin" (the name of the Wiktionary) and not "Chinese" (the name of the language) - but I don't know enough about the issues to deal with it. Conrad.Irwin 09:24, 30 April 2009 (UTC)Reply

Oh, ignore me, someone has already fixed it. Conrad.Irwin 09:25, 30 April 2009 (UTC)Reply

So do you need data for the languages codes which we have language templates for including nonstandard templates and all the language codes which have Wiktionaries whether or not those codes are standard or whether or not we use those codes on enwikt? I suppose I should include fields for "in iso", "in wikimedia" and "wiktionary template".

Which means the only real problems are where one of these three uses a different code to one of the others, or when any two of them use the same code for different languages. And this does seem to be the case for a few where ISO uses one code and either wikimedia or enwikt or both use a different code. — hippietrail 09:52, 30 April 2009 (UTC)Reply

• (@Robert Ullmann) Serbian Nationalist "Serbo-Croatian" crap - the policy guideline is proposed by Ivan, who is an administrator of Croatian origin. I suggest you discuss this with him, but this policy (about Serbo-Croatian) is not going to be rescinded without a community consensus. Furthermore, it is completely futile to squander resources on three dialects elevated to the state of languages, when with the help of the SC header this is succinct and concise. And Bosnian and Serbian wiktionaries are no more alive than Serbo-Croatian, see the BP discussion for that, where evidence was presented. A veritable example for nationalist crap is the this article about the Montenegrin dialect. The uſer hight Bogorm converſation 09:11, 30 April 2009 (UTC)Reply

It seems discovery of valid Wiktionary language codes is not straightforward. We reserve all 2- and 3- letter all lowercase ASCII alphabetical template names as Language templates which resolve to the name of the language. This works for language codes of the form xx and xxx. So all can be discovered by searching the Template namespace.

But as far as I can see we don't reserve hyphenated language codes and we don't restrict membership of Category:Language templates either. If I search for all templates in the category beginning with two lowercase letters I get these results which are not in ISO 639-3:

• aoq: Ammonite -- no ISO 639-3 code. nothing in Ethnologue.
• ast-leo: Leonese -- ast is Asturian. Ethnologue says Leonese is a dialect of Asturian. code leo must be unofficial.
• bat-smg: Samogitian -- Ethnologue says its a dialect of Lithuanian which is lt or lit. ISO 639-3 has no code bat.
• be-x-old: Belarusian (Tarashkevitsa) -- nothing in Ethnologue.
• bh: Bihari -- has a Wiktionary, seems to have progressed from being a singular code to a collective code to being removed.
• cbk-zam: Zamboanga Chavacano -- cbk is Chavacano. Ethnologue says Zamboanga is a dialect. code zam must be unofficial
• el-it: Salentine Greek -- nothing on Ethnologue.
• eml-rom: Romagnolo -- Ethnologue says eml is for Emiliano-Romagnolo.
• fiu-vro: Võro (already discussed, should be moved to vro)
• fr-ca: Canadian French (nominated for deletion)
• fr-nng: Guernésiais -- Ethnologue counts it as a dialect of Fench. code nng must be unofficial
• fr-nnj: Jèrriais -- Ethnologue counts it as a dialect of Fench. code nng must be unofficial
• fr-nnx: Norman -- Ethnologue counts it as a dialect of Fench. code nng must be unofficial
• map-bms: Banyumasan. ISO counts it as a dialect of Javanese jv. code map doesn't exist in ISO 639-3
• mo: Moldavian -- has a Wiktionary. now covered by Romanian ro
• mol: Moldavian -- see above
• nah: Nahuatl -- has a Wiktionary. was an ISO 639-2 code but is not an ISO 639-3 code. There are many codes for Nahuatl languages/dialects
• nap-cal: Calabrese -- Ethnologue says nap is for Napoletano-Calabrese.
• nds-nl: Dutch Low Saxon --
• no-rik: Norwegian Riksmål -- historic predecessor of Bokmål but seems to have no language code of its own.
• roa-rup: Aromanian (has a Wiktionary, already discussed)
• sfk: Safwa -- seems to be an erroneous dupe of sbk.
• simple: Simple English (has a Wiktionary, already discussed)
• sr-mon: Montenegrin -- code mon must be unofficial
• suh: Suba -- Ethnologue gives suh for Suba but ISO 639-3 gives sxb. Ethnologue does not list sxb at all.
• szk: Sizaki -- Ethnologue says Sizaki has ISO 639-3 code szk but ISO 639-3 has no such code.
• tokipona: Toki Pona (has a Wiktionary, already discussed)
• twf-pic: Picuris. Ethnologue says Picuris is a dialect of Tiwa. code pic must be unofficial.
• wwg: Woiwurrung -- seems to be an Australian language without a code. wwg is not in Ethnologue or ISO 639-3, nor is the spelling Woiwurrung anywhere in Ethnologue, perhaps there is another spelling?
• zh-classical: Old Chinese
• zh-cn: Simplified Chinese
• zh-min-nan: Min Nan -- has a Wiktionary
• zh-tw: Traditional Chinese
• zh-yue: Cantonese
• zkm: Maikoti -- Ethnologue says this code has been retired and code kjl for Western Parbate must be used instead.

I'll try to look into each of these a little further. — hippietrail 09:42, 30 April 2009 (UTC)Reply

Could you check my request here? 50 Xylophone Players talk 07:56, 2 May 2009 (UTC)Reply

I just added support for the pl2 and dim2 parameters in {{nl-noun}}, was this what you wanted - I wasn't certain from the context. Conrad.Irwin 13:23, 2 May 2009 (UTC)Reply

Yeah, but for some reason I can't seem to get an alt diminutive to apppear in my sandbox. 50 Xylophone Players talk 11:02, 4 May 2009 (UTC)Reply

nvm, I figured it out. 50 Xylophone Players talk 11:03, 4 May 2009 (UTC)Reply

"When you approve someone, don't forget to go to Special:UserRights and mark them as autopatroller." Oh, is it enough for one person to support the vote? I was regarding it as something like an admin vote, where a decision was made after several people had contributed viewpoints. Equinox ◑ 23:35, 2 May 2009 (UTC)Reply

Do you think they are proper nouns? I do. Equinox ◑ 13:46, 8 May 2009 (UTC)Reply

In that they refer to one specific entity I suppose so. Conrad.Irwin 14:00, 8 May 2009 (UTC)Reply

Would you please take a look at this template? There is an if statement that should put infinitive forms in Category:Hungarian infinitive forms and main infinitives in Category:Hungarian infinitives, but it puts everything in the latter category. I use the {{hu-ni}} switch to determine the category, but it doesn't work. Thank you. --Panda10 12:39, 9 May 2009 (UTC)Reply

I replaced if with ifeq, it seems to work now. Thanks. --Panda10 13:22, 9 May 2009 (UTC)Reply

I can see only examples of plural for the second meaning. Perhaps it's worth noting in the entry. --Derbeth ^talk 10:41, 10 May 2009 (UTC)Reply

or just having a lend?

I can't see what is wrong with putting a link to a transcription of a text of a war diary in with the citation on "Cornstalk"? "Tommy Cornstalk" is certainly not a dictionary.

As for your RfV on the slang meaning of "bosky", if you had an olive complexion (like me) and been around factories, farms and football training areas in Australia in the 50s, 60s, and 70s (like me) you would have heard the word used to refer to you generically if they didn't know your name. "Hey, bosky" would be used to attract your attention just as if you had fair or blond hair you would be called "snowy" or "snow" or red hair "bluey" or "blue". It is is likely that most of these usages have not found their way into even a slang dictionary. The usage I have got for the word comes from a novel. I have to track it down on my bookshelves.

Please leave me alone.

Is there a guide on how to use the citations page somewhere?Albatross2147 02:18, 11 May 2009 (UTC)Reply

Yes, there is one: WT:CITE. The uſer hight Bogorm converſation 07:17, 11 May 2009 (UTC)Reply

I was not stalking you in particular, just looking at Special:RecentChanges in order to find contributions that needed improvement (see meta:Help:Patrolled edit for more information). I had a brief look on http://books.google.com, and http://groups.google.com and could find no written evidence to support what you were saying at bosky (which is needed by WT:CFI). You are right there is nothing wrong with the citation you added, but there is also nothing useful about it; which is why I linked you to the Help page and did not remove it from the entry. Conrad.Irwin 08:36, 11 May 2009 (UTC)Reply

I understand but Googlebooks is far from exhaustive. The usage of "bosky" I put in is almost certainly archaic and may only have been used in a region ie. inland NSW and Queensland. I am going to try to find the book I saw the usage in. Why isn't a link to a book's online transcript useful? Albatross2147 03:25, 14 May 2009 (UTC)Reply

I read the guide and am not much wiser I fear. I shall have a play around. Albatross2147 03:28, 14 May 2009 (UTC)Reply

Hi, sorry, I'm not great at explaining things. Essentially the quotations you added were "mentions" of the word, they did not show it being used, they showed it being discussed. We generally only accept the former for WT:CFI (though whether it is documented well, I don't know); while there is no harm in adding quotes that show the word being discussed, they don't show the reader (of Wiktionary) how the word is used. Conrad.Irwin 08:51, 14 May 2009 (UTC)Reply

Dear Conrad, the Spanish Index generated by your bot code contains quite some verb forms, which have not been detected as such. These are mainly the words ending in "ríais" and "rían". As all Spanish words with these endings are verb forms you could maybe consider to detect them and exclude them from the index, which should be IMHO easy to implement into the code. Matthias Buchmeier 09:26, 13 May 2009 (UTC)Reply

Sure, I'll try to up date it before the next run. (Seems to me that I just need to exclude definitions containing "(first|second)-person singular of" in addition to all the current rules that only match if the definition contains "form of"?) Conrad.Irwin 09:30, 13 May 2009 (UTC)Reply

Hello! We've really been enjoying your script over at se:wikt. I have a small request, however. Do you think you could make it so that it shows green links for template:irrnoun? Thank you! Tygrrr 17:15, 13 May 2009 (UTC)Reply

I've done that. It creates the forms using the {{simple:noun}} template, as that is what I saw done elsewhere. Conrad.Irwin 19:16, 13 May 2009 (UTC)Reply

Conrad, I've noticed that the masculine plural is not properly accelerated for this template. The feminine singular and plural both work fine, but the masculine plural is generated as a feminine entry for some reason. I can't figure out why (in part because the coding is a bit beyond me). --EncycloPetey 03:11, 14 May 2009 (UTC)Reply

Now fixed, though the masculine singular doesn't seem to accelerate for me. Conrad.Irwin 08:46, 14 May 2009 (UTC)Reply

It shouldn't need to, since the masculine singular is the lemma form. --EncycloPetey 04:36, 15 May 2009 (UTC)Reply

Would you mind having a {{look}} at this discussion and seeing what you think?—msh210℠ 21:31, 3 June 2009 (UTC)Reply

Hi there, it's w:User:Hangfromthefloor from #mediawiki on irc.freenode.net from Friday evening. I'm extremely grateful for your help, and even though your solution didn't work perfectly in the end, I appreciate all the effort you made trying to make it work. So, thanks once again! Hangfromthefloor 02:30, 6 June 2009 (UTC)Reply

I've added more than 1200 words in the past few months (which more than doubles the number of raw entries we had (not counting "Translation section entries" in the Index). It's probably time to re-run this. Reminder: alphabetization is the same as for Spanish. --EncycloPetey 16:26, 8 June 2009 (UTC)Reply

I can do this when I get back to my computer. It was last run on the 27th of May, and I had intended to do it on Saturday, but forgot. Conrad.Irwin 16:36, 8 June 2009 (UTC)Reply

On an unrelated issue: Do you know why the server is acting so sluggish today? I'm trying to add Galician verb forms via bot, and all I get is "Pausing 5 seconds due to database server lag." over and over. ...and I can't recall the (UNIX) code to stop a running process. --EncycloPetey 16:57, 8 June 2009 (UTC)Reply

Not sure, maybe they are doing the work to get the new search engine installed on Wiktionary too (well, I can dream :D). Conrad.Irwin 17:10, 8 June 2009 (UTC)Reply

(Oh, and CTRL+C should halt it, if not use CTRL+Z to send it to the background and then ps ax and kill to get rid of it). Conrad.Irwin 17:11, 8 June 2009 (UTC)Reply

Thanks. Are you ready (and willing) to branch out into other Romance languages? This would be relatively easy, since they all use Latin-based alphabets. If so, I can work up details (if you need them) for alphabetizing Romanian, Catalan, Portuguese, and French. All of these are major languages that are well-represented on Wktionary, but which have old Index pages in dire need of updating. Starting an Index for Asturian, Neapolitan, Sicilian, and Occitan could also be useful, although each of those would be less likely to need updating, as people seldom contribute in those languages. --EncycloPetey 19:05, 9 June 2009 (UTC)Reply

Yes, certainly, most of what has been holding me off has been that some of the indexes for these languages already contain lists of red-links, but I can always do the same as for Index:Russian and move them somewhere (i.e. Wiktionary:Requested Entries:<Language>/<letter>). Conrad.Irwin 11:27, 10 June 2009 (UTC)Reply

Catalan: Vowels may have an acute accent (only É é, Í í, Ó ó, Ú ú), a grave accent (only À à, È è, Ò ò), or a diaresis (only ï and ü, and I don't think this occurs over capital letters). I'm uncertain, but I believe indexing order is in the sequence: unmarked, acute, grave, diaresis. This sequence applies for ordering in the event of a "tie" (otherwise identical spelling), but the marked vowels are considered equivalent with the unmarked vowels otherwise, so words beginning with A and Á are intermingled. The only consonant oddities I know of are the c-cedilla (Ç ç), which comes right after C/c, and the ela geminada, which is used to distinguish two "L"s from a double "LL" (as in anul·lar). I do not know whether the double "LL" (without ela geminada) is indexed as a separate letter, or not. User:Carolina wren should be able to confirm or amend this information. --EncycloPetey 15:19, 10 June 2009 (UTC)Reply

This comes from the IEC's grammar:

First sort criteria is indeed with all accents (cedilla and the middle dot of the ela geminada in addition to the diaeresis, acute, and grave) ignored. Hyphens are also ignored unless what follows the hyphen is captalised, in which case it is treated as two separate words.

Second sort criteria is the position of an accent. The later the accent occurs in the word, the earlier it sorts.

Third sort criteria is which accent. Grave comes last, but I believe the relative order of diaresis is a moot point, as I don't believe there are any words in Catalan where a diaresis contrasts with acute or grave. In any case the IEC grammar merely asserts that grave comes after acute.

Examples

1. forca < força < forcall
2. cella < cel·la < cellut
3. suplica < suplicà < súplica (no accents < grave on 7th letter < acute on 2nd letter)
4. sénia < sènia ( acute on same letter < grave on same letter )
5. beneit < beneït
6. mallar < mà-llarg < mànec (sorted as mallar < mallarg < manec )

Finally there are three sanctioned methods of sorting multi-word phrases, with apostrophes ignored in all three.

1. Word by word

   mà < mà a mà < mà d’obra < mà d’urpa < mà de morter < mà invisible < mà morta < maça

2. Word by word, ignoring prepositions and articles

   mà < mà invisible < mà a mà < mà morta < mà de morter < mà d’obra < mà d’urpa < maça

3. All smashed together

   mà < mà a mà < maça < mà de morter < mà d’obra < mà d’urpa < mà invisible < mà morta

Other than the second method would be a lot of effort to get a bot to do, no comment as to which would be "best". — Carolina wren discussió 20:38, 10 June 2009 (UTC)Reply

French: Ther are a number of accented or modified characters, but none of them are considered separate letters. They are all indexed along with the same chaacter unmarked. For consonants, there is the c-cedilla (Ç ç). Vowels may have an acute accent (É é only), a grave accent (À à, È è, Où, où), a circumflex ( â, Ê ê, Î î, Ô ô, Û û), or a diaresis (Ë ë, Ü ü). I believe the order I've given is that used for breaking spelling "ties", but I do not know French well enough to be certain. User:Lmaltier could probably tell you. There is also an uncommon digraph Œ /œ, and I'm not certain but I believe it is indexed as if it were Oe / oe. --EncycloPetey 17:08, 10 June 2009 (UTC)Reply

For French, I think that EncycloPetey is right. But you should add ÿ (used in a few proper nouns) and ï (used in many words, e.g. naïf). Also: ä (used in länder), ñ (used in cañon). Other foreign words with diacritics are sometimes used with their original spelling, but this is not a general use. Lmaltier 18:06, 10 June 2009 (UTC)Reply

I'm not certain, but I believe that French also considers where the accent is to be more important than which accent when it comes to sorting. — Carolina wren discussió 20:38, 10 June 2009 (UTC)Reply

There needs to be provision for the sorting of expressions starting with preposition + article, such as (deprecated template usage) à l'abandon and variable preposition: should stuff starting in (deprecated template usage) aux ((deprecated template usage) aux abris) and (deprecated template usage) du go after (deprecated template usage) auvergnat and (deprecated template usage) dryade of after (deprecated template usage) à and (deprecated template usage) de? Where should (deprecated template usage) d'abord be sorted?. I've been whiling away time with a mock up and am struggling with this. Circeus 17:57, 23 June 2009 (UTC)Reply

What to do with compound vs. uncompounds? (IMHO the ideal sensible option is to list them next to each others no matter how we deal with hyphenation in general) cf. abri sous roche vs. abri-sous-roche (the later is the one listed in dictonaries because of the French typograpic definition of an entry, but appears much, much rarer than its hyphenated counterpart), antiaérien vs. anti-aérien... Circeus 18:05, 23 June 2009 (UTC)Reply

Portuguese: Vowels may have an acute accent (Á á, É é, Í í, Ó ó, Ú ú), a grave accent (À à, Ù ù), a circumflex ( â, Ê ê, Ô ô, although  seems very rare), a tilde (à ã, Õ õ), or a diaresis (Ü ü, which seems to occur only in the combination Qü or qü). Indexing sequence seems to be in the order: unmarked, acute, grave, circumflex, tilde, diaresis (although I'm making a bit of a guess on the last three). This sequence applies for ordering in the event of a "tie", but the marked vowels are considered equivalent with the unmarked vowels otherwise, so (for example) words beginning with A and Á are intermingled. The only odd consonant character is: Ç / ç, which is considered equivalent to C /c for indexing (but follows in the event of otherwise identical spellings). Portuguese also has the following digraphs: ch, lh, nh, rr, but I do not know the current conventions for indexing them. Traditionally, they were considered separate letters. You might inquire of User:Daniel. to see whether this is still the case. --EncycloPetey 15:19, 10 June 2009 (UTC)Reply

Various ortographic reforms ocurred in Portuguese language through the last hundred years, the last one being in effect since January 2009. Therefore, a relatively high number of different rules produced words to be included here in Wiktionary, including many that would probably be defined as {{obsolete spelling of}}.

• The Portuguese alphabet is composed of the same 26 letters of the English alphabet and in the same order, from A to Z. This includes the letters K, W and Y, that were once considered not part of the alphabet and are still avoided in favor of substitutions with the same sounds. For example, in Portuguese karate is caratê and Kenya is Quênia.
• There are the following letters with diacritics: á, à, â, ã, ç, é, ê, í, ó, ô, õ, ú (and their uppercase counterparts: Á, À, Â, Ã, Ç, É, Ê, Í, Ó, Ô, Õ, Ú). Some uses of the diacritics were eventually abolished, such as the distinction between tôrre and torre, platéia and plateia, among thousands of other words. All of these letters with diacritics are still very common in lowercase, but some are never used as first letter in any word, so their uppercase usage is restricted.
• The diaresis (Ü, ü) is now obsolete, but was also very common; it mainly ocurred in the combinations güe, güi, qüe and qüi to determine whether the vowel u would be pronounced. It could also be used to transform diphtongs into hiatuses: saüdade, uïvo, aïpo, etc.
• Another obsolete use of diacritics is the placing of a grave in certain vowels (À, à, È, è, Ì, ì, Ò, ò, Ù, ù) to change their sounds: pregar and prègar to distinguish between the translations to preach and to nail. Today, the grave has other use: it occurs only in the letter a, to mark certain contractions.
• There are the following digraphs: bd, bt, cc, cç, ch, ct, gd, gu, lh, mn, nh, pc, pç, pt, qu, rr, sc, sç, ss, tm, xc and xs.
• Additional digraphs, uses of diacritics and uses of K, W and Y may appear in words directly borrowed from other languages, such as in watt, kanji, walkie-talkie and the names Shakespeare, Loth and Müller.
• Digraphs and diacritics are not considered separate letters and do not interfere with the alphabetical order. For example, here are some words correctly sorted: cachorro, chaminé, coração, eleger, ética, excelente.
• In the event of a tie, the order or diacritics commonly used in dictionaries, including diaresis, is: unmarked, cedilla, acute, grave, circumflex, diaresis, tilde. Another example: agua, água, sanguinário, sangüinário. --Daniel. 05:47, 11 June 2009 (UTC)Reply

Romanian: It looks as though there's already a good guide across the top of Index:Romanian. There are a few letters (k, q, w, and y) that are non-native, but do occur in some borrowed foreign words. There is a "concave accent" that can occur over an "a" (as Ă / ă), a circumflex that can occur over "a" or "i" (as Â, â, Î, î), and a "cedilla" (actually a floating symbol not attached) that can occur under "s" or "t" (as Ş, ş, Ţ, ţ). Each of these modified symbols (vowels and consonants) is treated as a separate letter, and is indexed immediately following the same spelling unmarked. All of these except  may occur at the beginning of a word and should have separate index pages. Acute accents over vowels are possible in writing, but are only used in rare circumstances to differentiate between homographs. Of course, check with User:Opiaterein before starting, as he may have additional comments. --EncycloPetey 15:19, 10 June 2009 (UTC)Reply

Life has suddenly become unexpectedly busy, I will try and implement these over the course of the next fortnight, though I had been hoping to have them done yesterday. Conrad.Irwin 23:36, 12 June 2009 (UTC)Reply

Can I ask what the status on this currently is? Circeus 14:12, 14 July 2009 (UTC)Reply

Catalan and Romanian are being generated (and I've yet to receive any feedback about the inaccuracies in either). French gets exported, and translations for it found, but I have not yet written the sorter, I can do that tomorrow I think. Havent yet looked at Portuguese, again over the weekend is my aim. Conrad.Irwin 01:47, 1 August 2009 (UTC)Reply

Catalan is being indexed pretty much as expected as far as I can tell. However, would it be possible for the sorting bot to override its own rules and use the sort key if an inflection line template uses it? That would allow phrases such as a la romana to sort under romana, a la instead. (Sort key was added to the entry since the last indexing, so if this is already being done, just ignore me.) All the català inflection line templates that have a sort override use sort as the parameter, the same as {infl}. The suggested improvement would likely apply to the other languages as well. — Carolina wren discussió 01:30, 7 September 2009 (UTC)Reply

I could add support for it, for languages like English I just remove all the "the"s and "a"s from the front anyway - should I do that for "a" and "la" for Catalan as well as looking for the sort parameter? Conrad.Irwin 07:15, 7 September 2009 (UTC)Reply

No, don't try autostripping. Catalan is fond of using combinations of prepositions where English would use only one. We have one yellow link in the index "a dins" that I believe should stay in its current location, and there might be more worthy additions. Nor can "la" or the other articles be stripped automagically, as unless you know whether the following word is a noun or a verb, there's no way to tell whether a Catalan article is instead a Catalan pronoun. Which makes sense since they have the same etymological origin. It would be as if instead of "The boy hit the baseball with the bat," English used *"Him boy hit it baseball with it bat." — Carolina wren discussió 07:57, 7 September 2009 (UTC)Reply

I think this counts as a bug. --EncycloPetey 15:59, 17 June 2009 (UTC)Reply

Definitely. Should now be fixed (but it's hard to say for certain). Conrad.Irwin 02:39, 19 June 2009 (UTC)Reply

Not sure if this was discussed before. Could the code you've written for assisted translations be used to add requested entries without editing the page? It would be one entry field for the entire page and the code would place it in correct alphabetical order. E.g. on Wiktionary:Requested entries:Hungarian but it can be for any language. --Panda10 00:10, 19 June 2009 (UTC)Reply

Yes-ish, the code has been designed that such changes would require the "minimal" amount of Javascript. The -ish because Javascript cannot sort properly (though we can get it most of the way there), and the "minimal" in quotes, because such a change requires making an edit to the Wikitext and the HTML in parallel (which is, although not hard, not trivial to get right). I don't have any immediate plans to add this feature, but if you want to have a go, feel free (it should be very similar to the way that translations are currently added). Conrad.Irwin 00:15, 19 June 2009 (UTC)Reply

I am hesitant. I don't have a good understanding of how to do this and coding would take away all my time from adding/correcting Hungarian entries. Maybe there is someone else who is more knowledgeable and would be insterested in coding this project for the English requests. This should be much simpler than adding translations. It is really just adding a word (no gender specifications or other grammatical things) and if sorting is not perfect, there is no harm. --Panda10 02:04, 19 June 2009 (UTC)Reply

Hi cirwin. I had a hunch that paid off. There is an incompatibility between "paperview" and one of my JavaScript extensions. parjer.js does an alert "This doesn't look like a Wiktionary page. No can do I'm afraid." I boldly edited parser.js to add "Conrad Irwin / parser.js\n" at the begninning of all alert messages so they won't be so hard to find in the future. Feel free to change it however you think is best though.

Oh there are also two Google scripts included randomly from somewhere too but I think these were a read herring and not the things which were clashing with paperview. — hippietrail 02:33, 21 June 2009 (UTC)Reply

(excuse me, because my knowledge of English is not good)
Hello, I'm David, Admin of Georgian Wiktionary. I've seen English Translation template. It is very easy template for all users. I want this template in its Georgian Wiktionary. I have translated and have something done, but that is not as good as in English. Can you tell me how they made this template? Or is this a secret :) thank you from the outset. Dato deutschland 06:49, 22 June 2009 (UTC)Reply

Hi, I'm not sure what the problem is, but if it is because here they are all closed, and ka:მომხმარებელი:Malafaya they are open, then you need to change the javascript in ka:MediaWiki:Common.js. I am not sure otherwise. Conrad.Irwin 07:22, 22 June 2009 (UTC)Reply

Unsere Vorlage arbeitet. ist nur eine Problem. Englische template hat zusätzliche Funktion (Add translation, sehr leicht). But, georgian template has not this function. Entschuldigen sie bitte mir, wenn ich habe schlecht erklären. Dato deutschland 12:49, 22 June 2009 (UTC)Reply

You can copy User:Conrad.Irwin/editor.js to your wiktionary, but you will need to change it to make it work there. I am not able to make these changes for you, sorry. Conrad.Irwin 20:53, 22 June 2009 (UTC)Reply

Hi Conrad,

Just FYI, I modified the race-condition–handling code in MediaWiki:langcode2name.js and User:Conrad.Irwin/iwiki.js. In the latter case, the issue was that JavaScript is finicky: ! document.getElementById (for example) is true when document lacks a getElementById property, but ! langcode2name is an error if langcode2name is undefined. I don't know why this should be, but I Googled and I tested, and I'm pretty confident of it. In the former case, the issue is that the for(var prop in obj) construct iterates over the names of the properties, not their values, so in this case callback is the string 'iwiki' rather than the function add_prominent_interwikis.

(I don't know if similar race-condition–handling code is being used elsewhere on Wiktionary; if so, I imagine we need to change all of it.)

—Ruakh_TALK 03:13, 24 June 2009 (UTC)Reply

It's not used elsewhere, thanks for fixing it. Conrad.Irwin 07:44, 24 June 2009 (UTC)Reply

Hi, I figured you're busy lately, but this should be relatively easy. Can you make your bot populate Index:Old Armenian? No tweaks to sorting should be made whatsoever: just the standard Unicode one. --Vahagn Petrosyan 15:14, 1 July 2009 (UTC)Reply

Likewise Index:Hebrew, if you've got a chance. :-)   —Ruakh_TALK 17:24, 1 July 2009 (UTC)Reply

I've done Index:Old Armenian on one page, let me know what needs improvement. What do you want done with the Latin letters for Hebrew? Conrad.Irwin 23:08, 4 July 2009 (UTC)Reply

Excellent! Is there a way to index also the words in nested translations like in mouse? --Vahagn Petrosyan 00:35, 5 July 2009 (UTC)Reply

Thanks a bunch! Don't worry about the Latin letters — now that I know about them, and they're conveniently indexed, I can go through and take care of them. :-)   —Ruakh_TALK 17:57, 6 July 2009 (UTC)Reply

By the way, the Hebrew index is not-clickable in Opera. I mean the browser does not recognize the links except for the last letter ת. And it was that way even before automatic indexing. --Vahagn Petrosyan 23:30, 6 July 2009 (UTC)Reply

The code that you use to parse the dump, is it public, by any chance? I ask because a lot of the Hebrew errors are technically parsing errors rather than content error; but I'd rather fix our wiki-syntax to match what your parser will recognize, since I'm sure your parser isn't going to be the only one with this sort of problem. Thanks in advance! —Ruakh_TALK 03:48, 27 July 2009 (UTC)Reply

Umm, yes it is, but I'd rather fix the code than the entries - it's a rather huge mess of awk and python and perl held together with a bash script. What is the main problem? That way I can at least point you to the right file. You may get some idea of how it works by reading User:Conrad.Bot and looking in bash/create_indices.sh. Conrad.Irwin 07:39, 27 July 2009 (UTC)Reply

I think I came across two issues, both having to do with translations tables. One is lack of support for preposed named parameters; for example, {{t|sc=Hebr|he|מלכה|tr=milká}} was interpreted as referring to he, when in fact it refers to מלכה. The other is lack of support for {{he-translation}}; I don't even know what the issue is there, but sometimes it gets interpreted as a single space (]), sometimes as two (]). —Ruakh_TALK 15:08, 27 July 2009 (UTC)Reply

    5 0
    2 a
    1 d
    1 f
   22 h
    1 l
    4 m
    2 n
    3 p
    1 q
    2 s
    1 t
    2 y
    2 z
  698 א
  370 ב
  224 ג
  183 ד
  370 ה
   59 ו
   90 ז
  289 ח
  109 ט
  203 י
  206 כ
  605 ל
    1 ם
  762 מ
    1 ן
  246 נ
  254 ס
  265 ע
  259 פ
  154 צ
  291 ק
  198 ר
  390 ש
  240 ת
 6516 total

See . Same weird problem as before. --EncycloPetey 18:45, 4 July 2009 (UTC)Reply

Ok, I'll have another look then, (and secretly hope it's a caching problem and that they were still using an old version) Conrad.Irwin 23:05, 4 July 2009 (UTC)Reply

Help:Interacting with humans is a real step forward in explaining how Wiktionary works to newcomers. It was sorely needed, and hopefully it'll prove its worth in the future. Best, Knepflerle 02:48, 5 July 2009 (UTC)Reply

Hi, can I ask a favour of you? EncycloPetey created this template and asked me to comment, which I did, and he improved it so it's just about perfect now, with only one thing left he doesn't know how to do: Many masculine nouns in Scottish Gaelic have the same genitive singular and nominative plural. EP tweaked the template so that when one writes eg

• {{gd-noun|g=m|gen=balaich|pl=balaich}}

it appears as

• balach m (genitive and plural balaich)

(When the form is the same it affects eg the genitive plural or vocative, so it's helpful if it's pointed out.) Question is, using accelerated editing the "form-of" page has a definition line for the genitive only - is it possible to make it create both the genitive and the plural lines as at balaich, so that the line for the plural wouldn't have to be added manually? Another example of such word, where I haven't created the inflected forms' entry yet, is oileanach. Thanks in advance whether it is technically possible or not. --Duncan 16:19, 6 July 2009 (UTC)Reply

Yes, it's possible. I will try and get it to work for you. Conrad.Irwin 20:11, 6 July 2009 (UTC)Reply

I think it now works, if you clear your cache (ctrl+shift+F5) Conrad.Irwin 20:51, 6 July 2009 (UTC)Reply

For some reason, when I try it I get a "%23" at the start of the second definition line instead of a hash (#). --EncycloPetey 23:41, 6 July 2009 (UTC)Reply

Argh! it seems highly browser-dependent, which is why for the other hashes I use #, unfortunately, that seems to introduce an extra space. Meh, I'm not sure I can fix it trivially - a fix needs to be made to w:User:Lupin/AutoEdit.js - or, perhaps better, to stop using that entirely. Conrad.Irwin 00:08, 7 July 2009 (UTC)Reply

When I deleted "w:User:Lupin/AutoEdit.js" from User:Duncan MacCall/monobook.js the accelerated editing stopped working at all, but never mind - highlighting "%23" and overtyping it with "#" is still much better than copypasting the line and overtyping "genitive" with "plural". Another thing occurred to me in the meantime, though - my bad, I should have noticed it at once: the accelerated editing creates the inflected entries like "three apostrophes - inflected form - three apostrophes", thus only making them bold - but I think those too, not just the lemma, should show the gender of the noun - at least that's how I've been creating them until now, though I don't know about any policy requiring it, and it seems to me from eg the Spanish puños that this can be achieved - could you make the template work similarly to the Spanish one? Apologies and thanks in advance again, --Duncan 15:26, 7 July 2009 (UTC)Reply

Forget it, it doesn't matter any longer. Sorry for having wasted your time. --Duncan 17:23, 9 July 2009 (UTC)Reply

Though I can not detect any harm caused, I am getting a message that the parser has failed after every edit of an entry. DCDuring TALK 14:07, 7 July 2009 (UTC)Reply

Hi Cirwin. Last night I actually got around to using newNode() for the first time rather than just reading over code that uses it. Before I actually thought it read through one big object to create a tree of nodes. Now I realize that of course it does just create one node at a time but that calls can be easily nested.

Which all boils down to... I love it. Put it in Common.js ASAP (-: — hippietrail 02:13, 11 July 2009 (UTC)Reply

'd itbe myDell inspiron5160 isalredy2slow4Dragon/modern software?--史凡/ʂɚ˨˩fan˩˥/shi3fan2 (歡迎光臨/Welcome! 請也用/Please also use skype: sven0921為我/since I suffer RSI!) 07:07, 12 July 2009 (UTC)Reply

The piece of software I linked to is not speech recognition and requires very few resources to run (it has the added bonus of being much more accurate than speech recognition). Conrad.Irwin 10:31, 12 July 2009 (UTC)Reply

You commented in the BP. I assume you also know of the current vote (but am writing just in case not).​—msh210℠ 22:36, 13 July 2009 (UTC)Reply

As you must know, our search engine doesn't find terms like "stiff drink" when they are marked up per WT:ELE as "stiff drink". I also note that en.wikt gets a very high percentage of its hits from search engines (per Alexa). It would not be an evil use of meta-tags (or whatever they are called) to incorporate certain collocations which the search engines would not find due solely to wikimarkup. I took a look at Free Dictionary source for their "stiff" page. They span "stiff" "drink", so no markup interferes with the collocation. Another way of doing it would be the "Collocations" header that MZajac has been advocating, assuming that the search engine looks that deeply, especially "beneath" our show/hide templates. Any thoughts? DCDuring TALK 15:14, 14 July 2009 (UTC)Reply

Right now it's finding stiff, with the excerpt text “Adjective: A stiff drink; a stiff dose; a stiff breeze. Translations: of an object, rigid, hard to bend, inflexible. Finnish. fi | jäykkä ...” (I can't tell whether it is compiling this excerpt from the current entry text or has indexed an older version.)

If the search engine can't find phrases broken up by formatting, then it should (must) be fixed (regardless of any other workarounds we adopt in the meantime). Is there a bug report we can all vote for? If not, then let's file one. —Michael Z. 2009-07-14 15:30 z

Me too. Maybe something has been changed. Maybe there is a problem only in more specific circumstances. Maybe I erred. I'll have to pay closer attention to exactly what situations lead to unsatisfactory search results, if any. DCDuring TALK 16:34, 14 July 2009 (UTC)Reply

I like the collocations idea, except that I fear that such a section will get unwieldy fast (especially on certain pages). Allowing boldfaced common collocations (instead of merely boldfaced words) in example sentences is another idea, though not one I'm all that fond of. See {{Keyword}} for yet another idea.​—msh210℠ 15:46, 14 July 2009 (UTC)Reply

The template addresses the issue, of course, but at the cost of a lot of labor and, if done by bots, say, using COCA collocations, a lot of space. Using lists of the top COCA collocations to populate the keyword template if the text is not in the body might be seen as cheating. It would have to be explicitly sold to the search engine folks.

Perhaps we could look at COCA to find the top N collocations, insert the list in a namespace like "Talk", "Citations", or, dare I say it, "Collocations", and add usage examples manually for some of the redlinked ones based on that data, using a template {{usex}} to facilitate formatting and metatagging at the same time. DCDuring TALK 16:34, 14 July 2009 (UTC)Reply

Actually, I'm not sure that {{Keyword}} addresses the issue: search engines use keywords, perhaps, but I don't know whether Wiktionary's internal one does. Another idea is to not boldface anything in citations in the citations namespace (but to continue boldfacing in citations in the entry), and to by default search also in the citations namespace. Actually, I like this idea: it doesn't require "cheating" or new infrastructure. I don't know, though, whether we can change the default search namespaces. That should be easy to find out, though.​—msh210℠ 16:43, 14 July 2009 (UTC)Reply

I was going to stop complaining about our search engine until I have a better feel for when or how it fails from a user perspective. I would love to read up to 10 pages of more-or-less English on how the Mediawiki search software works. Any very easy low-resource-cost solutions, no matter how partial, are very desirable. DCDuring TALK 18:16, 14 July 2009 (UTC)Reply

You may be able to find what you are after at http://www.mediawiki.orghttps://dictious.com/en/User:Rainman/search_internals. I have recently filed a bug about the search engine, which currently doesn't index template expansion either (https://bugzilla.wikimedia.org/show_bug.cgi?id=18861), and been told that that is unlikely to be fixed in the near future - though feel free to file another one. When I tested it http://en.wiktionary.orghttps://dictious.com/en/Special:Search/"stiff drink" found it in stiff, but http://www.google.co.uk/?q="stiff+drink"+site:en.wiktionary.org didn't find it. It would certainly be worth including these as plain text, but I'd have thought google would get things right... Conrad.Irwin 23:26, 14 July 2009 (UTC)Reply

I think the issue with the built-in search may have been fixed. The issue you mention with Google just has to do with indexing delay; its cached version is from Jul 11, 2009 06:37:37 GMT, which was before DCDuring added the "stiff drink" sense and example. I'm sure it'll be fine once they re-index the page. —Ruakh_TALK 23:34, 14 July 2009 (UTC)Reply

Thanks for the link. Looks just about right for me. I'll be paying more exact attention to any search problems that I experience so I can be more specific. As Ruakh says, the problem with wikiformatting preventing MW internal search from working seems resolved. Whether google search has a problem with wikiformat is the next question. The last question regards what could be called "collocation stuffing". How can we productively insert common collocations what don't meet CFI into web pages? We could:

1. make sure that all headwords deleted by reason of being SoP appear in one or more relevant entries.
2. have a kind of a checklist to make sure that N of the most common collocations for a given headword appear in the entry. Or
3. could just stuff the top M collocations into metatags and be done with it.

I'm sure there are better ideas, too. And many drawbacks and implementation problems, not to mention questions as to desirability.

Per Alexa, we do seem to be getting a lot of our traffic from search engines already, much more than from WP. And we pass many users on to sister projects. Both of those facts should make us seem valuable to WM, I hope. Maybe we could become important enough to get more of their attention to our more distinctive needs. I get the impression that we are sucking hind tit most of the time. DCDuring TALK 00:24, 15 July 2009 (UTC)Reply

• I have three questions:
  1. How can I find out more about stop words as applied for en.wikt search?
  2. Do we have any ability to add or subtract from the stop word list specifically for en.wikt?
  3. Pending some kind of more definitive resolution of search engine handling of collocations, do you know whether, 1., hard redirects or, 2., soft redirects show up at all on google searches? At WT:RFD#check it out we have good content including usage examples that could go to a sense of check out or to check it out. There are also 2 hard redirects. Some collocations are now usage examples also. From the point of view of our internal search I think I understand the differences. But I don't know what works best for Google or other search engines, including OneLook.com, which has a fairly big list of collocations available for a secondary search. DCDuring TALK 03:20, 27 July 2009 (UTC)Reply

I've corrected myself at WT:BP#Automatic_transliteration.​—msh210℠ 17:01, 20 July 2009 (UTC)Reply

Hi Conrad ^_^ I have some questions that need your wizardry: Would it be possible to pre-process Wiktionary wikicode when a user visits a particular page, do some magic on it, send that magicaly processed wikitext back to MediaWiki and display it only then to the user?

E.g. suppose I want it so to split the Serbo-Croatian L2 section as three B/C/S section, taking into account scripts and the context labels, and display such split text to the user that would select to have split display (all three of the sections, or just one of them).

Except for a bit longer opening of pages, what other drawbacks could it have?

Suppose if we used subpages for non-English language sections, e.g. {{PAGENAME}}/hr, {{PAGENAME}}/bs, {{PAGENAME}}/sr, {{PAGENAME}}/sh that would be transcluded on the main page this would all be pretty much trivial.. --Ivan Štambuk 20:33, 20 July 2009 (UTC)Reply

It would be possible, a better idea would be to just find the section and duplicate it in javascript which would be considerably faster, if slightly harder to program correctly first time. (And which would also ensure that all the information for the three languages was in the page of the unified language). It might get slightly confusing, as we already have too many buttons and distractions, but providing some thought is given to the interface I don't see any show-stopping difficulty for this proposal. If you were to want to render wikitext and give the result to the user directly, User_talk:Conrad.Irwin/Api.js describes the library that I am currently merging into WT:EDIT, which would allow you to give the user rendered text trivially. JsMwApi().page(wgPageName).parse(newText, function (parsed) { document.getElementById('bodyContent').innerHTML = newText }) (Code like this is the basis for the preview that assisted translations provides). Conrad.Irwin 21:36, 20 July 2009 (UTC)Reply

That looks awesome! It's just that I have programmed javascript something like 2 times in my entire life (both times were cookie stealers for some dumb Croatian blogging service ^_^), so it might take me some time to simply get started (the SC->B/C/S conversion algorithm would would be some 50 lines of code max.). Would you be kind enough to point me to what exactly I must set up in my monobook.css to get started? I would be esp. thankful if you could provide me with the 3-LOC skeleton of a js function that appends e.g. "Hello world" to the page's wikitext. From there I can figure out the string processing on my own.. --Ivan Štambuk 21:54, 20 July 2009 (UTC)Reply

If you want hello world at the bottom of every Wiktionary page, try adding this to Special:MyPage/monobook.js (and clear your cache (ctrl+shift+F5) a lot):

var divToAdd = newNode('div');
divToAdd.innerHTML="<b>Hello</b> World!";
document.getElementById('bodyContent').appendChild(divToAdd);

There are both simpler (but buggy) and (almost pointlessly) more complicated ways to do the same thing, but that should work without breaking the current page. If you haven't already, I'd recommend getting Firefox with Firebug as then you can easily try out new things and it gives you easier to follow error messages. Conrad.Irwin 23:40, 20 July 2009 (UTC)Reply

Could you please take a look? Dutch diminutive plural forms are being created and categorised as plurals. This is wrong they should be using the following format: {{diminutive of|xxx|plural=1|lang=nl}} and I guess just like the singular forms it should be noted that they are neuter gender. 50 Xylophone Players talk 12:05, 28 July 2009 (UTC)Reply

Hi again, I see that you have fixed it, thanks. :) 50 Xylophone Players talk 13:02, 30 July 2009 (UTC)Reply

No problem. Conrad.Irwin 00:33, 31 July 2009 (UTC)Reply

Looks good so far. My only gripe would be that IMHO, a space should be treated as such, i.e. au contraire should precede aubaine, and a hyphen is probably best treated as going after a space (i.e. au-delà would be between au violon and aubaine, but given dictionaries appear to differ, I'll leave this to your discretion). AFAIK there is no standard treatment for this issue because French dictionary are very keen on the typographical word concept:a hyphenated word will have an entry even if the unhyphenated form is actually standard (cf. abri sous roche), and multiword expressions are systematically relegated as subentrie (in fact, in the Robert, they are not even considered subentries at all!).

Nonetheless, I feel strongly that putting a word break there to group such entries together is far more intuitive, and would suggest the same policy be applied to English: common sense leads one to look for at odds before Atacama rather than after atmosphere! Such a treatment also makes it possible to e.g. take at a glance a list of compounds starting with a preposition.

None of the reference book i personally own discuss alphabetical ordering or collation. If you want, I can look into and get back to you at another time? in any case, I remind much of the opinion that a space should be a separator. I was a bit confused by your comment re:trimming off various forms of the articles, so if you could show specific case so I understand your meaning.

Still, the index looks AWESOME and is great help in spotting weird stuff. Circeus 18:39, 1 August 2009 (UTC)Reply

A few extra ideas:

1. Since you're already pumping the translation sections (which allows me to fix all sort of small things), is it possible to also skim sections like, synonyms and derivatives for red entries? (at first I had "alternative forms" in these, but these entries are automatically discarded for being constituted only of a "form of" template...)
2. Since you're obviously checking whether a French section exists (or the index couldn't selectively link only when they exist), could you apply a formatting to entries where the page exist, but not the section (cf. Abdoulla before I created the section).

Circeus 19:49, 1 August 2009 (UTC)Reply

Yes, I also was unsure of space handling, not sure what made me choose the current method, but I can easily change that; as long as the collation looks right, I'm not too fussed about making it perfect. The forms of the article that I strip off are so, for example, "arithmétique", "d'arithmétique" and "de l'arithmétique" all appear next to each other - again if that's incorrect behaviour, it is easy to fix. I wasn't sure about whether au violin should be under "a" or "v".

I can try adding a scan for synonyms and derivatives - I tried once with Hungarian and found the results unpleasing, but maybe it will be better for a language with so many more entries, I'll have a go. It should be possible to colour the link if it doesn't have a local language link (I hadn't thought of that before, and might apply it to all languages). Conrad.Irwin 22:18, 1 August 2009 (UTC)Reply

• I would recommend not dropping "de" and "d'" at the beginning of word: it would be exceptional for them to be determiner rather than article preposition in that particular position within the lemma
• Regarding derivatives and synonyms, I meant it for locating additional missing entries (red and "orange" links), so IMHO the symbols can be dropped without problem. (I think it's safe to assume there will be less issues than there are with incorrectly entered translations) Circeus 01:50, 2 August 2009 (UTC)Reply

Hello there, I just would like to inform you that the Wiktionary logo 2009 refresh voting should start now, but so far we haven't come up with any rules yet. Since you were very active in the discussion, would you like to help out again? Thanks in advance. Wyvernoid 03:14, 8 August 2009 (UTC)Reply

I'm afraid I'm going to be mainly afk until the 30th, and was hoping that someone else would be able to finish things off. Conrad.Irwin 07:19, 8 August 2009 (UTC)Reply

Thanks for your advice. I got 'told off' on Wikipedia for not using that template so I tried to bring it across to Wiktionary but it depended on another template which in turn depended on another and so on. It seems over engineered and messy.

John Cross 09:13, 8 August 2009 (UTC)Reply

+pref-setin'>ta!!:D

• canitbemade2jump automaticaly2's'-section[or 'sop'directly?

*i'dlik2find entrys misin'fe dutchtranslations>how?--史凡>voice-MSN/skypeme!RSI>typin=hard! 16:59, 8 August 2009 (UTC)Reply

If you mean, could it be made to jump to the correct section of Wiktionary:Glossary, then the answer is yes, if you add some markup around each item in the list as follows:

* <span name="sop" id="sop">Sum of parts</span>

Then you could like to Wiktionary:Glossary#sop. Conrad.Irwin 19:46, 9 August 2009 (UTC)Reply

• er-"span/id"=?--史凡>voice-MSN/skypeme!RSI>typin=hard! 04:33, 10 August 2009 (UTC)Reply

I don't understand. Conrad.Irwin 21:19, 10 August 2009 (UTC)Reply

Do we know how search engines handle hard redirects vs soft redirects vs only-in vs full entries? How could I find out?

From a user point of view hard redirects afford some advantages IMO because the user needs one click less to get to meaningful content. It is also slightly quicker to add redirects than "form of" entries. But I suspect that they may not show up in search engine search pages. DCDuring TALK 18:13, 10 August 2009 (UTC)Reply

I don't know, I suppose by doing some searching you could work it out - the only one I tried "to kill the fatted calf" was not indexed. One of the advantages of the big search engines is that they do some of the form-normalisation for us, so that searches are more likely to end up at the right place without the need for the redirection - I suppose we have to remember that most people won't type into the URL bar, they'll use the search boxes or a search engine. Conrad.Irwin 21:13, 10 August 2009 (UTC)Reply

OK, experimentation it is: Google, Yahoo, Bing(?), OneLook, Answers.com.

It is my understanding that Lucerne search includes personal pronouns (he, his, etc) as well as "one" and (I think) "one's" as stopwords. It does not seem to include "someone", "someone's", "somebody", and "somebody's". I'm not sure that I'm thinking about this right, but if they were also on the stopword list, wouldn't that sometimes improve folks ability to find lemmas containing those words, eg, when they typed in other words (usually stopwords like "you", "your", "yours") ? DCDuring TALK 18:13, 10 August 2009 (UTC)Reply

It would increase the number of pages that matched the search, because it would not be searching for "someone's", but it would also bring up more false-positives. On Wiktionary, I think you are right, and having "someone" as a stop-word might be beneficial given that we have pages with "one" - I can't think of any phrases involving somebody that have only one other word (off the top of my head), but even those with only two words might be a bit of a problem. Conrad.Irwin 21:18, 10 August 2009 (UTC)Reply

Hello, Conrad. If I remmeber correctly, it was introduced by you, so there is some sort of feedback - when I add Bulgarian translations for nouns, the fields for gender and transliteration appear, but when I try to add Old Norse translations and I fill up the code non in the first field and click on the second field, both fields (check box and edit box respectively) for gender and transliteration disappear, whereas the first one (the check box for gender) should remain available, since the Old Norse language has three genders. Is this flaw easily reparable? I shall be adding Old Norse nouns manually in the meantime. The uſer hight Bogorm converſation 12:03, 13 August 2009 (UTC)Reply

Same problem with Gothic (got) too. The uſer hight Bogorm converſation 12:09, 13 August 2009 (UTC)Reply

Hi, the gender buttons are still there, just hidden under the "More..." link. I will look into the genders issue though, it should show all genders for any language that User:Hippietrail hasn't told it about. Conrad.Irwin 23:35, 13 August 2009 (UTC)Reply

They are ok now, but shotly after I had added the translation, AutoFormat unlinked Old Norse (as it should be, since this is a widely known language). Could you set up this language as unlinked? The uſer hight Bogorm converſation 18:55, 25 August 2009 (UTC)Reply

All it does is {{subst:non}}, so fixing Template:non should do it. Conrad.Irwin 22:30, 25 August 2009 (UTC)Reply

"i Sven, long time no chat. We'll have to catch up soon when I'm not so snowed under with work. :( Just to let you know, the "standard" formatting for Eng->Chi entries nowadays is:

• Chinese:
  • Mandarin: 叩頭／叩头 (kòutóu), 叩头 (kòutóu)

Please don't put them under "Mandarin"; it makes it harder to find, as most people look under Chinese when looking for translations. Cheers! Tooironic 22:18, 19 August 2009 (UTC)

   * inotisdn'agre,BUTedit-auto nodo,nmyhands notake the c/p ea time..:/"--史凡>voice-MSN/skypeme!RSI>typin=hard! 22:49, 19 August 2009 (UTC)Reply

When I asked people about this, the discussion got a little side-tracked and I have no patience with the politics, see Wiktionary:Beer_parlour_archive/2009/April#Nested_translations_etc. - I have implemented some forms of nesting, but they are not enabled, I will have to look again at it if I ever get some time. Conrad.Irwin 00:11, 20 August 2009 (UTC)Reply

Ariel Glenn overstates my expertise, we needed some table of equivalents so I consulted geography and library sites. Have only just seen you note and will look it over, over the next couple of days. —Saltmarsh^{απάντηση} 09:18, 27 August 2009 (UTC)Reply

Conrad - The transliteration table I set up at Wiktionary:About Greek/Transliteration created little interest at the time that I did it - looking back at it now there seem to be one or two possible "errors" which I should address (perhaps). Looking through the discrepancies:

1. It seems to sort out the ευ => ef/ev dichotomy alright.
2. el:προϊόν => proión != proion should IMO => proïón (see table)
3. el:αϊτός => aitós != aïtós     IMO aïtós
4. el:χειμώνας => cheimónas != khjimonas   yes cheimónas
5. el:σούστα => soýsta != sústa     IMO soústa is better (similarly with νούμερο)
6. el:ὀξύμωρον => ὀxýmoron != oxymōron   the breathing (ὀ) is AncientG, I don't think ModG has a noun (but a phrase οξύμωρο σχήμα)

Is there anywhere I could experiment in a Sandbox? Although I am short of time at present! —Saltmarsh^{απάντηση} 15:35, 27 August 2009 (UTC)Reply

I have contacted User talk:Doremítzwr and User talk:Rodasmith about this since both raised points about the previous table of equivalent values. User:Flyax has also entered discussion at Wiktionary talk:About Greek/Transliteration-new. I am away until 22 Sept and will pick the matter up then - when hopefully come consensus may be evident. —Saltmarsh^{απάντηση} 14:13, 4 September 2009 (UTC)Reply

I have finished this - bar tidying up the rules file - I think controversial points have been ironed out. Is there a time scale for implementation? —Saltmarsh^{απάντηση} 16:56, 23 September 2009 (UTC)Reply

Thanks a lot! I was hoping that it would already be done. We are waiting on the CTO of Wikimedia reviewing the extension as and when he gets round to it. Once it has been reviewed, it will then need another wait (hopefully much shorter) before it gets actually installed. https://bugzilla.wikimedia.org/show_bug.cgi?id=20246 is where my request to them is. Conrad.Irwin 22:27, 23 September 2009 (UTC)Reply

I haven't thought about this for a while - and my ear is often someway from the ground - is anything happening? —Saltmarsh^{απάντηση} 10:45, 1 June 2010 (UTC)Reply

Irritatingly not. I will try and pester them again at some point. Conrad.Irwin 16:51, 1 June 2010 (UTC)Reply

Hi Conrad, lately when I am adding two translations in the same trans table, it adds two Hungarian lines and I have to edit it to remove the duplicate line. Before, I believe, it just added the second translation with a comma. Thanks. --Panda10 14:24, 30 August 2009 (UTC)Reply

It definitely should just add them with a comma, I am currently investigating a similar problem that Guest63h brought to my attention, so maybe I'll find the problem. Conrad.Irwin 14:28, 30 August 2009 (UTC)Reply

Should be fixed if you clear your cache (ctrl+shift+F5). Seems to have been tied to using {{Xyzy}} in {{t}}, but I can't fathom why that should have changed anything. Conrad.Irwin 15:05, 30 August 2009 (UTC)Reply

Thanks. --Panda10 19:52, 30 August 2009 (UTC)Reply

Hi Conrad, how hard would it be to change {{hu-noun}} to display the plural of multi-word entries separately and not as one word? For example, the plural of baseball sapka (baseball cap) is baseball sapkák, but I'd like to display baseball sapkák. I think it would make more sense to treat the plural separately. Thanks. --Panda10 22:48, 10 September 2009 (UTC)Reply

About that hard, I hope. (as with other similar templates, when you link them manually, you must also bold them manually - this then allows for short notes to be included in the inflection line). Conrad.Irwin 23:00, 10 September 2009 (UTC)Reply

It works great! Thank you. --Panda10 01:32, 11 September 2009 (UTC)Reply

Hi Conrad. Could you explain the situation in re the purportedly “discarded” audio-recording button discussion mentioned herein please? It’s best to explain in that section of the Beer Parlour, since it’ll be of general interest. Thanks. † ﴾^(u):Raifʻhār ^(t):Doremítzwr﴿ 13:11, 12 September 2009 (UTC)Reply

Hi, you asked me yesterday on the IRC about Sven . . . I've been thinking about it and came up with a longer answer for ya.  :) My keyboard, I guess, just has a bad relationship with that channel or something because I have to type reeeeeeallly slowly there to be legible but here it works. (silly thing)
I think the problems are communication and hypersensitivity. Communication, because it is really hard to read his writing and so sometimes we can't tell what the problem is. Hypersensitivity, because he likes to write in all caps so it often sounds like he's shouting and so that further aggravates other people, and then that further aggravates him. I don't think it's your fault. However, I do sympathize with him ... It's hard being neutral. Very hard.
Anyway, I've now taken up translating for him in discussion pages. I wish there was another solution. :P L☺g☺maniac chat? 15:16, 12 September 2009 (UTC)Reply

• My solution is just to ignore him. SemperBlotto 15:21, 12 September 2009 (UTC)Reply

I wish. I don't think he'd appreciate it though. *sigh* L☺g☺maniac chat? 15:31, 12 September 2009 (UTC)Reply

I hope he appreciates your efforts. I do. I have to adopt SB's solution when my patience and kindness are on holiday together, as they often are. DCDuring TALK 15:39, 12 September 2009 (UTC)Reply

Thank goodness I have mine on a tight leash, otherwise they'd be on holiday by now... L☺g☺maniac chat? 15:48, 12 September 2009 (UTC) Reply

Thanks all. Conrad.Irwin 19:36, 12 September 2009 (UTC)Reply

I guess basically a bot would just have to look for the ==Mandarin== and ===Hanzi=== sections, then check if {{zh-hanzi| specifies sim= or tra=. If all that comes out positive, then the sim=/tra= should be copied to the {{yue-hanzi| under ==Cantonese== ===Hanzi===. If not... then nothin. :) — opiaterein — 21:03, 13 September 2009 (UTC)Reply

You might wish to see questions at Wiktionary talk:Votes/bt-2009-09/User:Conrad.Bot to do anagrams. Thanks!​—msh210℠ 23:49, 14 September 2009 (UTC)Reply

I'm sure I put a note somewhere about not using vote talk pages :p, thanks for the heads up. Conrad.Irwin 23:50, 14 September 2009 (UTC)Reply

I've noticed this happen several times now. Nadando 18:00, 15 September 2009 (UTC)Reply

Urm, ok. If I said "that shouldn't happen" it wouldn't help much. I think I need to re-think the timing issues with the balancers, maybe this is another aspect of them. Conrad.Irwin 20:43, 15 September 2009 (UTC)Reply

Now fixed. Conrad.Irwin 00:34, 21 September 2009 (UTC)Reply

Speaking as the 2009 UK French Scrabble Champion, you're right about French. Ignore case, diacritics, treat æ and œ as -ae- and -oe- and you've got it. Mglovesfun (talk) 21:41, 15 September 2009 (UTC)Reply

alright :). Conrad.Irwin 21:42, 15 September 2009 (UTC)Reply

I just came across cœurs, which is listed as an anagram of crus, which it isn't. That was this month. Could you fix it? PierreAbbat 22:59, 20 November 2009 (UTC)Reply

Yes, I've since patched the bot so it won't make more of these, but haven't gone back to fix them yet. Thanks for the reminder. (It now treats œ as "oe", is this correct?) Conrad.Irwin 23:02, 20 November 2009 (UTC)Reply

The A index page contains only 14 entries, it should be more than 875. Can you take a look whan you have a chance? Thank you. --Panda10 18:25, 20 September 2009 (UTC)Reply

Hi, sorry, they have broken the framework that used to upload these entries. I will try and get them working for next weekend. Conrad.Irwin 19:04, 20 September 2009 (UTC)Reply

Sure, no problem. Take your time. Thanks again. --Panda10 19:06, 20 September 2009 (UTC)Reply

Hi Conrad

After getting tired of updating stats @nl.wikt I decided to use the magic word PAGESINCATEGORY instead and so we now have a stats page that updates itself see. The functions may be a little hard on the server but I think it is functional enough to warrant that.

Greetings Jcwf 19:40, 20 September 2009 (UTC)Reply

Cool idea, that might be more useful than the current guesswork anyway. Conrad.Irwin 19:43, 20 September 2009 (UTC)Reply

Here are some tricks used on WMF (though they're mostly for skins/messages, but they're better than "served by" time): forceprofile=true and forcetrace=true. View the page source, a big comment at the end holds the data. Note the data can change depending on which apache you hit and how busy it is.

Zocky has this thing, but it requires core changes (some before/after hooks to the brace substitution) which make it a bit inefficient on WMF. You can see an example of it here. Anonymous editing is enabled so you could probably test your template there in a sandbox and action=profile it after. Splarka 06:06, 21 September 2009 (UTC)Reply

I may need to learn javascript.

So, the word lookup script appears to have become a very popular idea. The fr.wn contributors are all over it, and are thinking it needs to be added to the site and maybe to wikisource as well... and they've found a few bugs they'd like me to fix.

• trailing punctuation is included in the url. Clicking on cabinet, would retrieve cabinet,.
• preceding articles, such as l'article, are included in the url; in this case l'article.
• The most confusing bug is the quotes bug: in French « and ». Apparently this causes all clicks in a multi-line quote to be assigned to words in the first line of the quote, with terms vertically aligned to the left of the « being assigned to that character. This behaviour can be seen at this article.

Assuming you'd rather not become the godfather of this monster, where would you suggest I go online to learn how to code javascript? - Amgine/^talk 17:28, 25 September 2009 (UTC)Reply

Hi, the header ===Anagrams=== has to be in the page text, not in the template. Otherwise can't be found, used, ignored by whatever is parsing wikitext. And other issues ... (like edit sections etc)

(they all seem sort of pointless anyway to me? Why not get generate lists rather than add stuff to every entry?) I hadn't even noticed the proposal for this, too little time and too much going on! ;-) Robert Ullmann 13:02, 29 September 2009 (UTC)Reply

4nl--史凡>voice-MSN/skypeme!RSI>typin=hard! 01:09, 3 October 2009 (UTC)Reply

Fixed. Conrad.Irwin 09:01, 3 October 2009 (UTC)Reply

didnt notis,sory:/--ta4helpn!--史凡>voice-MSN/skypeme!RSI>typin=hard! 09:38, 3 October 2009 (UTC)Reply

Hiya. Now that you're bot-adding anagrams, do you think it'd be possible to add Hebrew to the code? Discussion at About:Hebrew yielded that the following rules should be followed:

• Ignore spaces and punctuation, noting that punctuation may include U+05BE, U+05F3, and U+05F4 in addition to the usual suspects.
• Any page with U+0591 through 05BD, or with 05BF through 05C7, or with 05F0 through 05F2, or with FB1D through FB28, or with FB2A through FB4F, which contains a Hebrew section, is a bad pagetitle (for Hebrew at least), and should not list anagrams, nor be listed as one (and if possible should be tagged for Hebrew attention).
• The letters U+05DA and 05DB are identical for purposes of anagrams. 05DD and 05DE are identical. 05DF and 05E0 are identical. 05E3 and 05E4 are identical. 05E5 and 05E6 are identical.
• Ignore redirects (don't list them as anagrams).

Thanks much.​—msh210℠ 16:00, 8 October 2009 (UTC)Reply

Alright. The bot is currently on hold until I reimplement it to not use the template system (initially through concerns by Ullmann and Hippietrail, but having seen people make a mess of editing the templates I now agree). I will try and get it running at some point, but I can't promise when. Conrad.Irwin 16:41, 8 October 2009 (UTC)Reply

I'm positively giddy about the bot's running once more, and am posting this to re-ask you to add Hebrew to your code if and when you have time and inclination. Thanks.​—msh210℠ 18:05, 16 November 2009 (UTC)Reply

I had entirely forgotten about this request, sure! though, they will have to wait until the second run (probably starting immediately after the first has (evenutally) finished. Conrad.Irwin 18:07, 16 November 2009 (UTC)Reply

Yeah, I figured you might have forgotten, considering all you're doing. Thanks much!​—msh210℠ 18:27, 16 November 2009 (UTC)Reply

BTW, not a big deal, but in alphagrams, for the pairs U+05D, U+05D, U+05(DF|E0), U+05E, and U+05E, I think the greater codepoint (i.e., the usual form of the letter, rather than the "final" form used at the end of a word) should be preferred. Thanks in advance! —Ruakh_TALK 22:52, 12 December 2009 (UTC)Reply

Alright. I'll try to arrange thatConrad.Irwin 23:30, 12 December 2009 (UTC)Reply

Would it be hard to get Index:Translingual going? Hopefully not since there's no language-specific rules for sorting. Thanks.--Bequw → ¢ • τ 14:43, 9 October 2009 (UTC)Reply

Hopefully not, though I'll have a think about the best way to divide it into sections when I try. Conrad.Irwin 15:01, 9 October 2009 (UTC)Reply

care to weigh in on its RFDO?--Bequw → ¢ • τ 01:53, 14 October 2009 (UTC)Reply

Hello, I request to have Index:Polish created, for an index of Polish words in Wiktionary. Is this easy? --Volants 12:36, 23 October 2009 (UTC)Reply

Sure, could you let me know how `exactly` the sorting works? Looking at w:Polish orthography there are several digraphs that are not included in the alphabet, are they simply sorted as though they were individual letters? Should c and ć be treated as different letters, or are they just forms of the same letter (same for other such characters). I'll try to get round to this at the weekend - I need to re-run all the others too. Conrad.Irwin 12:39, 23 October 2009 (UTC)Reply

"c" and "ć" are different letters. Digraphs that are not included in the alphabet. Here is the exact order: Image:Polish-alphabet.png. Maro 20:17, 23 October 2009 (UTC)Reply

Ok, let me know of any problems. Conrad.Irwin 13:53, 25 October 2009 (UTC)Reply

Thanks for the list. It may come in handy, I found many translations that need to be corrected. BTW, something is wrong on Index:Polish/ł, there are ła łą ła łą ła sections... But I think we don't even need separate section for each, where there are only hundred-two hundred words for a letter. Or even less. Maro 17:54, 25 October 2009 (UTC)Reply

There is a new green link in the Hungarian declension tables, for the plural accusative. When I create the form-of entry using that, it does not put the entry in any category. I had to replace the template in the definition line with hu-inflection of. Was there a change in the Hungarian templates? --Panda10 20:41, 23 October 2009 (UTC)Reply

I added some plural accusative rules to work with Esperanto, so they must have conflicted. I can remove them again as I ended up doing something totally different (but I assumed there'd be no side-effects). Conrad.Irwin 21:33, 23 October 2009 (UTC)Reply

Thanks. I cleared the cache and they are now gone. --Panda10 22:28, 23 October 2009 (UTC)Reply

Hey again Conrad. :) I was wondering if you could add code for the following three things, if it's possible, to my monobook.js whenever you have the time (as I am unable to do any kind of "programming" and stuff like this)

1. code to make edits for creating Hungarian noun form entries automatically minor, like how in creation.js URLs you have "&preloadminor=true"
2. same as above but to make the summary always be "n" or something (n standing for new that is)
3. finally, (and I realise this might be a bit much to ask) creation.js-style links for the form-ofs.

Thanks, I hope you can help. 50 Xylophone Players talk 16:34, 24 October 2009 (UTC)Reply

Maybe, how should the script tell you are creating a hungarian noun form?. Or would this be the same as just having the Hungarian forms all accelerated? Conrad.Irwin 11:36, 25 October 2009 (UTC)Reply

Well, I guess when I click on a link for one in a declension table. 50 Xylophone Players talk 11:56, 25 October 2009 (UTC)Reply

...for updating the Hungarian index. :) --Panda10 14:12, 25 October 2009 (UTC)Reply

I wanted to bring RHS ToCs up in the BP to see if it can get some widespread usage. Do you know of any outstanding systematic issues dealing with right-hand side ToCs? Additionally, I wanted to bring up the idea of scrollable, fixed-height ToCs ({{scrollable-toc}}, {{scrollable-tocright}}). Do you know if there's any problems with those? Thanks. --Bequw → ¢ • τ 00:03, 27 October 2009 (UTC)Reply

I would support moving it to the right (still). I am not a big fan of the scrollable idea - when the entry is big enough that it scrolls, it is big enough that you need to use the ToC to get to useful places. Conrad.Irwin 17:52, 27 October 2009 (UTC)Reply

With many users using the plain RHS ToCs, would it be useful to have additional editing guidelines to deal with situations where RHS elements get pushed out of their language section by big TOCs? Maybe we could say that, on entries that experience this problem, that these RHS elements should be "inlined. Images could go in galleries and navigation templates (both to related entries (eg {{ordinalbox}}) and to sister projects) could get put into See also sections. Is there anything that needs to be RHS? --Bequw → ¢ • τ 21:02, 27 October 2009 (UTC)Reply

As far as I see it, images are the only problem - floating navigation boxes are a bit icky. What to do with is complicated, I suspect a gallery at the bottom of the entry is called for. Conrad.Irwin 21:23, 27 October 2009 (UTC)Reply

Templates like {{top2}} specify table widths of 100%. As this can mess with RHS ToCs and other elements, is there anything that we can do? Can we put div tags in the top and bottom templates? --Bequw → ¢ • τ 16:43, 6 November 2009 (UTC)Reply

Like {{trans-top}} these could be wrapped inside a <div style="width:auto;margin:0px;overflow:auto;"> - but that would require updating everything that can work with {{bottom}} so that the </div> would be matched. (There's probably some way to bully CSS into treating the table like this directly, but I don't know what it is - Ullmann might). Conrad.Irwin 17:04, 6 November 2009 (UTC)Reply

Correcting now. - Amgine/^talk 17:40, 31 October 2009 (UTC)Reply

Hi there. Would you be willing to convert this to template form for us over at the Simple English Wiktionary:

<big>''']'''</big> <small>verb</small> #If you publish a book, an article, a song, etc. you make it available for other people to buy, read, listen to, etc. #: ''The study was '''published''' in the British Medial Journal.'' #: ''She '''publishes''' a monthly magazine.'' #: ''The government '''published''' the results on the Internet.'' <small>]</small>

We want to use it as a word of the week template where the arguement would be like {{wotw|<insert word here>}} and it would automatically set up a blank form where the word shows up and then has blank template like stuff where we can fill in the stuff that is needed like this: <nowiki> <big>''']'''</big> <small><insert part of speech here></small> <insert definition here> <small>]</small>

or something of this matter. Since I suck at templates, and you seem to be pretty good at them, would you be able to make sure a template for us? Thanks, Razorflame 19:20, 1 November 2009 (UTC)Reply

Is it a bug or a feature that club_butler requires underscores for spaces? For instance:

<hippietrail>	.? Bronx cheer
<club_butler>	Couldn't get any definitions for Bronx cheer.

<hippietrail>	.? Bronx_cheer
<club_butler>	Bronx_cheer — noun: 1. A razzing noise made with the lips and tongue; a raspberry

— hippietrail 08:16, 3 November 2009 (UTC)Reply

Probably a bug, its parser is a strange beastie (not written by me, you can get it at http://inamidst.com/phenny/), but I can probably fix it. Conrad.Irwin 08:45, 3 November 2009 (UTC)Reply

All i was trying to do is replace # Simple past and past participle of '''to ]'''. WITH # {{past of|]}}... Not sure how other changes made into it... TestPilot^{talk to me!} 02:16, 4 November 2009 (UTC)Reply

Ahh - figured that out. http://en.wiktionary.orghttps://en.wiktionary.org/w/index.php?title=attested&action=historysubmit&diff=7646333&oldid=941516 Edited old version - sorry, mine bad. TestPilot^{talk to me!} 02:34, 4 November 2009 (UTC)Reply

diff.​—msh210℠ 22:02, 4 November 2009 (UTC)Reply

Such "anagrams" are deliberately excluded in the current model - they should be linked to using {{also}}. Our definition of anagrams implies it is a re-arrangement of letters which I take to preclude the trivial case when the letters are in the same order (if the word "arrangement" was used it would be more ambiguous). Conrad.Irwin 23:03, 4 November 2009 (UTC)Reply

Thanks for the offer to automatically use your bot program on the Dutch Index. Dutch in general does alphabetize it's words the same way that English does. Sometimes Dutch dictionaries treat the digraph ij as its own letter, but that method is not followed by wikipedia, so it's just the regular twenty-six letter English/Latin alphabet, with the same order. Thanks, Mitchell Powell 03:00, 6 November 2009 (UTC)Reply

Hello!

I would like to know how could I get something like your personalized edittols working at pt.wikibooks, but I don't know where to start to look at... =/

Could you give me some direction?

Have a nice weekend! =D Helder 20:03, 6 November 2009 (UTC)Reply

You should be able to add importScriptURI('http://en.wiktionary.orghttps://en.wiktionary.org/w/index.php?title=User:Conrad.Irwin/edittools.js&action=raw&ctype=text/javascript'); and then create User:Heldergeovane/edittools. (Just copy mine for now, then when you know it's working, you can change it). Conrad.Irwin 22:01, 6 November 2009 (UTC)Reply

After some adjustments, it seems to work, but I needed to add the name we use for the main div of the edittools.

Do you mind of adding it to your js too? This way other users could just import it (as you suggested above), instead of make a copy =). The name "specialchars" (without the prefix "editpage-") is used at Wikimedia Commons (from where the edittols of pt.wb was copyed).

Thank you for your help... Now I can personalize my edittools with a lot of latex functions I use there =D Helder 13:40, 7 November 2009 (UTC)Reply

I've been testing and actually I haven't found a way of make them compatible. So, I just added a new div above our default edittols, with the id used by your script (and we can hide the default using a css rule with the id "specialchars"). Helder 17:13, 7 November 2009 (UTC)Reply

Or you could copy the script and edit it to suit your wiktionary, it has not been updated in a very long time and I'm unlikely to make further changes to it now. Conrad.Irwin 00:09, 8 November 2009 (UTC)Reply

Hi Conrad

I updated the index for Dutch by importing the Open Taal index, which gives a pretty exhaustive list of our language in proper official orthography. The lists are simply alphabetical and very lengthy. Is there any way to make them more accessible the way you have done that for the English list? If there is I would be much obliged if that could also be done #nl.wikti because we have the same problem there.

Jcwf 22:15, 6 November 2009 (UTC)Reply

Yes, I can run it through the last stage of the Conrad.Bot stuff - would you like me to insert all the Dutch that is present on Wiktionary at the same time? Conrad.Irwin 22:56, 6 November 2009 (UTC)Reply

The other thing you might want, though I won't do it unless you do, is to merge upper-and-lower case lists. Conrad.Irwin 22:59, 6 November 2009 (UTC)Reply

What en.wikti wants to do is up to the community here, but at nl.wikt we would like to keep the lists intact as they are, because as they are, they have received the official seal of approval of the Taalunie, the integovernmental body that regulates our spelling.

They represent a calibration set in a sense. So we do not want to alter them, unless the OpenTaal/Taalunie people come with an update. The Dutch present here (and at nl) contains entries with various orthographies, some of them historically obsolete ones (our spelling has changed numerous times), some of them downright miss-spellings. We certainly do not want to pollute the list with those. Jcwf 23:21, 6 November 2009 (UTC)Reply

These are now uploading, let me know of any mistakes you spot. Conrad.Irwin 00:00, 7 November 2009 (UTC)Reply

Thanks Conrad! `Jcwf 03:26, 7 November 2009 (UTC)Reply

Hi!!

I noticed this comment at MediaWiki:JavascriptHeadings.js:

// TODO \n+ doesn't seem to work here
res = spans.firstChild.nodeValue.match(/\/\*\s*<\/pre>+(==+)(.*?)(==+)+<pre>\s*\*\//,"gi");

I don't know what is the problem you had, but did you try to use "gim" instead of "gi"? o.0 Helder 13:09, 7 November 2009 (UTC)Reply

To be honest, I can't remember by now either - I seem to recall Hippietrail getting annoyed when I added a multiline flag to the original, but it works well enough for now. (It's his script) Conrad.Irwin 18:22, 7 November 2009 (UTC)Reply

I wasn't sure of which bots were approved for consensus search & replace but I thought yours was a good bet. Per WT:RFDO#Dialect etymology templates would you replace: I've orphaned the other with AWB already. This way we can see who's still using these old templates, and how long to hold onto them. If yours isn't well suited to this, is someone else's better? Thanks. --Bequw → ¢ • τ 18:21, 14 November 2009 (UTC)Reply

Yes, I can do this - I'm not sure whether I'll be able to run it at the same time as importing anagrams (which I'm hoping to restart now, and will take several days to complete); if I can't I'll let you know - then maybe User:Opiaterein would be another possibility. Conrad.Irwin 18:27, 14 November 2009 (UTC)Reply

Should now be done - let me know if there's more stuff to do. Conrad.Irwin 20:28, 14 November 2009 (UTC)Reply

Thanks! --Bequw → ¢ • τ 00:15, 15 November 2009 (UTC)Reply

Hi. I noticed that Conrad.Bot updates only the main page of Index for Old Armenian, Hilgaynon, Mapudungun and Polish. A bug? Or was it deliberate? Maro 20:30, 14 November 2009 (UTC)Reply

BTW, would it be possible to create indices for Kashubian, Lower Sorbian and Upper Sorbian? Maro 20:30, 14 November 2009 (UTC)Reply

Polish seems to be a bug, the other three are single-page indices (because they are so small). These indexes should be possible, but should I tree à etc. as an "A" with an accent (as French does), or as a seperate letter (as Polish does). And should "Ch" etc. be filed under "C" in the "h" section, or as "Ch" - a single letter? Conrad.Irwin 20:36, 14 November 2009 (UTC)Reply

Sorting order:

Kashubian: A, Ą, Ã, B, C, D, E, É, Ë, F, G, H, I, J, K, L, Ł, M, N, Ń, O, Ò, Ó, Ô, P, R, S, T, U, Ù, W, Y, Z, Ż

Lower Sorbian: a, b, c, č, ć, d, e, ě, f, g, h, ch, i, j, k, ł, l, m, n, ń, o, p, r, ŕ, s, š, t, u, w, y, z, ž, ź

Upper Sorbian: a, b, c, č, ć, d, dź, e, ě, f, g, h, ch, i, j, k, ł, l, m, n, ń, o, ó, p, q, r, ř, s, š, t, u, v, w, x, y, z, ž Maro 22:28, 14 November 2009 (UTC)Reply

These were done a while back, let me know of any problems. Conrad.Irwin 14:10, 16 November 2009 (UTC)Reply

Hi Conrad,

I notice that לחיים (l'khayím) is orange-ified as "partlynew" in the Hebrew index, apparently because it's given as a different part of speech (phrase) from the English word it's listed as a translation of (cheers, interjection). Is that something that can be addressed somehow, or do we basically have to choose between changing ]'s POS header, or accepting that it will be orange in the index?

Thanks in advance,
—Ruakh_TALK 21:04, 14 November 2009 (UTC)Reply

The actual problem is that it doesn't understand "{{non gloss definition}}" and, for templates it doesn't know, it uses the heuristic that if the definition is completely contained within one template then it is a form-of (and thus should not be in the index). I can fix this in the same way it treats {{given name}} and {{SI unit}} - are there any others you can think of? (While it would be nice to remove the partlynew on form-ofs that are translations, this is not trivial with the current architecture). Conrad.Irwin 21:35, 14 November 2009 (UTC)Reply

Oh, interesting. The problem is, not only need not a template-only def be a form-of, but also a form-of need not be a template-only def. (Some editors precede form-ofs with glosses.) I don't have a better suggestion, though, so until we come up with one, thanks for exempting {{non-gloss definition}}. :-)   —Ruakh_TALK 04:41, 15 November 2009 (UTC)Reply

Also, the bot seems to get confused by the way the Aramaic editors format their translations. For example, the first translations table at ] has this for Aramaic:

• Aramaic:

  Syriac: ܠܚܡܐ (lakhmā, lakhmo) m

  Hebrew: לחמא (lakhmā, lakhmo) m

; that is, it has two copies of the translation, one in the Syriac script, and one in the square script. The bot seems to mistake the latter for a Hebrew translation, so it adds לחמא to the Hebrew index (in orange).

Can it be changed either to understand this format, or, failing that, to discount lines that start with *:?

Thanks in advance! —Ruakh_TALK 21:22, 14 November 2009 (UTC)Reply

I can make it ignore Hebrew things that start with *: - this will also be a problem for WT:EDIT as it may find the nested Hebrew and try to add a Hebrew translation to it (if no *Hebrew exists yet) Conrad.Irwin 21:30, 14 November 2009 (UTC)Reply

Thanks! —Ruakh_TALK 04:36, 15 November 2009 (UTC)Reply

Hi Conrad,

At my admin vote, you expressed some concerns. Is there anything you’d like to elaborate on or that you’d rather I do differently, or was it mostly “added tools don’t seem useful”? Thanks for your comments!

—Nils von Barth (nbarth) (talk) 07:48, 15 November 2009 (UTC)Reply

I was just thinking back to events in 2008 - it was perhaps unfair to bring them up again after so long, but as I said, I haven't really had much to say to you since then! Congratulatuions for passing. Conrad.Irwin 08:55, 15 November 2009 (UTC)Reply

No worries – fair enough, and we did get off on rather the wrong foot (I had a few WT teething pains – “What, changing ELE isn’t ‘being bold’?”). Hope we’ve not had much to say since b/c I’ve learned how to not make messes (and just quietly worked), and look forward to working with you in future – thanks for the welcome!

—Nils von Barth (nbarth) (talk) 09:48, 15 November 2009 (UTC)Reply

Aren't IPA tags supposed to be preceded by a *, or is that only necessary when there's more than one thing under ===Pronunciation===? --Yair rand 02:20, 17 November 2009 (UTC)Reply

probably, I got rather frustrated trying to make it work. Conrad.Irwin 02:23, 17 November 2009 (UTC)Reply

It is supposed to have the * even if only one. Do note that AF will fix this automatically, not a big deal. Robert Ullmann 11:25, 18 November 2009 (UTC)Reply

Look at this. Has somehow lost the newline following the hrule, and the Lithuanian section header is treated as text. I modified AF to fix this case. Robert Ullmann 13:18, 17 November 2009 (UTC)Reply

Hrm, ok. Likely to be beacuse the ---- is in the anagrams section, but not part of it - should be fixable. Conrad.Irwin 13:26, 17 November 2009 (UTC)Reply

AF will fix these correctly: here (;-) Robert Ullmann 11:29, 18 November 2009 (UTC)Reply

I've now rebooted the bot with the right code - so it shouldn't happen again. Thanks. Conrad.Irwin 15:49, 18 November 2009 (UTC)Reply

Hello Conrad. Since you have created very nice index pages, would you update Index:Icelandic too? We have a quite good Icelandic section here, but the index is tiny --Volants 17:14, 17 November 2009 (UTC)Reply

Assuming that you want it sorted according to the list at w:Icelandic alphabet? Conrad.Irwin 17:28, 17 November 2009 (UTC)Reply

Let me know of any flaws you find. (it will be re-generated completely, with translation back-references like the others, next time I run all languages - probably this weekend). Conrad.Irwin 18:39, 17 November 2009 (UTC)Reply

See the edit summary.​—msh210℠ 21:07, 17 November 2009 (UTC)Reply

It certainly looks like it :( - I'm not entirely certain how it could have happened, but I'll try to fix it when I do. Thanks for letting me know. Conrad.Irwin 22:15, 17 November 2009 (UTC)Reply

same​—msh210℠ 20:55, 18 November 2009 (UTC)Reply

is this one of yours? see this edit ... Robert Ullmann 11:35, 18 November 2009 (UTC)Reply

Indeed - yet another: "that can't happen".... Conrad.Irwin 15:51, 18 November 2009 (UTC)Reply

Hi there. We need to talk about someone getting Darkicebot to be able to make form-of entries for nouns, adjectives, and verbs (Esperanto), that includes these characters (anywhere in the word):ĉ, ĝ, ĥ, ĵ, ŝ, ŭ. I don't know how to do it because the program that I use to enter the stems for the bot code currently does not allow these special characters to be added, and I don't know how to make them otherwise. Any ideas? Razorflame 23:28, 18 November 2009 (UTC)Reply

It should just be a case of changing template = """ to template = u""" and calling .encode('utf-8') on anything before you print it or write it. Let me know if you need a more helpful hand, I'm likely to be back on IRC in quarter of an hour. Conrad.Irwin 23:31, 18 November 2009 (UTC)Reply

Can you get back on IRC please? I want to talk to you more in depth about what you just told me on my talk page. Razorflame 01:05, 19 November 2009 (UTC)Reply

Well, I try to write a Catalan translation with automatical tools (showed in the same page) and sometime I discobert changes not succes, then I repeat the same operation and then appear translations duplicated. At the end I try do repair all that. I supouse it don't refresh enough rapidly. Sorry if I do anything wrong. --Maltrobat 17:43, 19 November 2009 (UTC)Reply

Sometimes that happens... next time, try to refresh the page before adding the word again. :) — opiaterein — 00:21, 20 November 2009 (UTC)Reply

Hey there. Darkicebot is now flagged as a bot and is currently running. Go check it out :). Cheers, Razorflame 00:16, 20 November 2009 (UTC)Reply

I'm wondering if it would be possible, using wiki syntax, to create an 'invisible' L2 header that redirects the TOC to another header? It would need to result in ^====\n in the article source. - Amgine/^talk 19:51, 22 November 2009 (UTC)Reply

You couldn't insert invisible headings into the TOC, but you could do something like:

<!--
 ==Heading==
 -->

and then fix the TOC with javascript. However this would break many many scripts both belonging to me and other people, that make the assumption that there won't be much inside comments. Conrad.Irwin 20:23, 22 November 2009 (UTC)Reply

In what way breaking? That is, the header is primarily to create an L2 entry in the TOC, and to store machine-readable instructions to find the actual data elsewhere for database reusers. I'm clearly being dense if this would somehow be information your scripts need? (I'm also not suggesting implementing this; I'm just looking for some compromise which does no harm.) - Amgine/^talk 20:38, 22 November 2009 (UTC)Reply

Take things like WT:EDIT, which need a corrsepondance between what can be seen in both places; things like the anagrams bot (and others) make the assumption that a language section runs from ==Language== to ---- or ==Language== whichever comes first, with yours there would be random open and close comment characters in unrelated sections (I can imagine AutoFormat reshuffling the sections on a page and commenting them out, and it certainly wouldn't make it easier to parse - certainly the way I do it, which is to first divide the page into language sections and assume that they are independant). Thinking about it more logically, you could wrap the entire section in <div style="display:none"> to get them into the TOC - but then the TOC links would only work in some browsers, and it has the same problem with putting random bits of markup in random places. Conrad.Irwin 23:42, 22 November 2009 (UTC)Reply

Please restore that page. I will continue to run my bot, but I will only run it for entries that are listed there. Cheers, Razorflame 03:18, 23 November 2009 (UTC)Reply

Thank you very much for your help and feedback on this. I understand that a dictionary is very different from an encyclopedia, in some ways much more formal, and fully understand that I don't understand the conventions I should follow - so thanks. It is a good word though. It came up in a Wikipedia article I was working on, needed a definition (after I figured out what it meant), and definitions do not belong in Wikipedia. Aymatth2 04:22, 23 November 2009 (UTC)Reply

Hi Conrad. Can your bot change all entries in Category:French noun forms into Category:French plurals. I think there must've been a bug somewhere beforehand. AFAICT all page in the aforementioned category are in fact plurals. --Rising Sun 11:32, 23 November 2009 (UTC)Reply

No don't, yet. See Wiktionary:Beer_parlour#Plurals_vs._nouns_forms. Plus some of them in that category aren't plurals. Mglovesfun (talk) 11:35, 23 November 2009 (UTC)Reply

Thanks! --EncycloPetey 18:26, 23 November 2009 (UTC)Reply

No worries, it was a surprisingly interesting word! (and not in any other dictionary I have access too, woohoo). Conrad.Irwin 18:28, 23 November 2009 (UTC)Reply

Looks like a good WOTD for December... --EncycloPetey 18:41, 23 November 2009 (UTC)Reply

Hello Conrad.Irwin, did you know that the French Wiktionary want this tool? I asked the import and I translated it but the position of the backup window is not correct for Frenchies. Is it possible to position it in the lower right corner? Also, they use the templates "trad-", "trad", "trad+" and "tradø" instead of templates "t-", "t", "t+" and "tø". Is it possible to change the script? Can you help me please? It would greatly help the French Wiktionary. Thank you, --Sniff 22:57, 23 November 2009 (UTC)Reply

Hi, yes (see on the page there). I don't have the time to maintain this tool on more than one wiki, but I am very happy to help if you have specific issues (and if I suddenly find I have a burst of spare wiki time - once the anagram bot is fixed assuming there are no more bug reports - I can hopefully help you with the first port). I will try and write out some detailed documentation for it this evening, though I have a few more hours of real work to do first. Conrad.Irwin 23:16, 23 November 2009 (UTC)Reply

User:Conrad.Irwin/editor_docs may be a good starting point (albeit long and boring). Conrad.Irwin 03:10, 24 November 2009 (UTC)Reply

I corrected the position, I changed the templates t for trad but with the templates used to creating tables ( (, - and ) ), it make an error. I don't know how I can resolve the error, my knowledge of Javascript does not allow me to correct that. Can you help me please ? Thank you in advance, --Sniff 22:26, 25 November 2009 (UTC)Reply

Hi, you should probably delete all the code inside TranslationLabeller{, as you don't seem to use glosses on fr.wikt. The English version of editor.js uses primarily the textual gloss to detect which table it is editing (this further helps prevent errors if a translation table in the HTML is not explicitly mentioned in the source code e.g. color). As you don't have this system on fr.wiktionary, you will probably need to redefine util.getTranslationTable in order to parse the french wiktionary's code to find the position of the wikitext for a translation table. The actual error is caused because within regular expressions you need to escape the (, so /\{\{( should be /\{\{\(. Thanks and good luck! Conrad.Irwin 00:06, 26 November 2009 (UTC)Reply

here --Vahagn Petrosyan 19:06, 24 November 2009 (UTC)Reply

Thanks for pointing it out. Conrad.Irwin 20:15, 24 November 2009 (UTC)Reply

Could you update the bot to do Icelandic anagrams? You'd just have to make sure it treats a/á, e/é, i/í, o/ó/ö, u/ú, y/ý, ae/æ, d/ð as separate letters. – Krun 16:16, 25 November 2009 (UTC)Reply

That should be possible. Conrad.Irwin 16:18, 25 November 2009 (UTC)Reply

Hi! It would be very nice if your bot could be updated to do Italian anagrams. I will be easier than French, Italian has just à è ì ò ù at the end. Thanks!! Pharamp 17:16, 25 November 2009 (UTC)Reply

By "at the end" what do you mean? I assume that (for the purposes of anagrams at least) á == a, etc.? Conrad.Irwin 20:14, 25 November 2009 (UTC)Reply

I think so (but actually à like verità). Mglovesfun (talk) 20:15, 25 November 2009 (UTC)Reply

I notice that a lot of Italian verb forms are anagrams of other parts - presumably this is useless and should be excluded? Conrad.Irwin 20:31, 25 November 2009 (UTC)Reply

• abbandonare#Italian: abbandonerà
• abbandonavo#Italian: abbondavano
• abbandonerà#Italian: abbandonare
• abbandoni#Italian: abbinando
• abbandono#Italian: abbonando, abbondano
• abbandonò#Italian: abbonando, abbondano

I meant like Mglovesfun said, "`" accent on the last syllable like lì or verità etc. which don't affect the pronunciation of the single letter (pronunciation of the French é is different than French è). For this I think that doing Italian anagrams can be quite easy: à will be a etc. (not separated like in Icelandic). For the verbs, that's true it is quite useless, but it will be funny to mark it, and this is also a French anagrams "problem" XD. Pharamp 11:55, 26 November 2009 (UTC)Reply

I didn't notice the abbandonare <-> abbandonerà forms in French, but it's not a huge problem so this can be done. Conrad.Irwin 12:05, 26 November 2009 (UTC)Reply

I don't think there is really the same problem in French. rejouais = jouerais comes to mind, but they're not from the same lemma (rejouer vs. jouer). Mglovesfun (talk) 12:08, 26 November 2009 (UTC)Reply

Yes, I agree, but also abbandonavo#Italian <-> abbondavano, abbandoni#Italian <-> abbinando, abbandono#Italian <-> abbonando, abbondano don't come from the same lemma, so I think that the bot should not exclude verb forms. Nothing comes to my mind in French like abbandonare <-> abbandonerà, but I think there's the possibility of a few cases. Pharamp 12:29, 26 November 2009 (UTC)Reply

Well, I was just going to exclude the case when they were verb forms of the same verb, but if you'l prefer them included that's fine too. Conrad.Irwin 16:18, 26 November 2009 (UTC)Reply

Hello, I noticed that your bot produces anagrams and alphagrams, I was wondering, since it does this, why doesn't it produce a reversed alphagram as well? (Z first, through to A)

76.66.194.154 12:24, 28 November 2009 (UTC)Reply

Is there much need for such a thing? I can imagine that if someone wanted to find anagrams they might search for the letters in alphabetical order. I can't think of any reason listing the inverse order would be useful, and besides it can be trivially discovered by just reversing the alphagram. Conrad.Irwin 12:45, 28 November 2009 (UTC)Reply

You can just read the alphagram backwards, right? Mglovesfun (talk) 12:46, 28 November 2009 (UTC)Reply

Reading it backwards won't come up in the searchbox. I'd say it has about as much use as the alphagram. 70.29.209.121 11:49, 4 December 2009 (UTC)Reply

Why would you search alphagrams? We're not a Scrabble site you know. Mglovesfun (talk) 11:59, 4 December 2009 (UTC)Reply

Why do alphagrams exist then? If it's not for Scrabble, then why are they there? 76.66.192.35 11:23, 9 December 2009 (UTC)Reply

""I can imagine that if someone wanted to find anagrams they might search for the letters in alphabetical order. I can't think of any reason listing the inverse order would be useful, and besides it can be trivially discovered by just reversing the alphagram. Conrad.Irwin 12:45, 28 November 2009 (UTC)"" Conrad.Irwin 14:54, 9 December 2009 (UTC)Reply

Hi, I was just looking through the XML dumps for Irish/index, and noticed that it might not be picking up the {{ga-verb}} tags (i.e., the verb bain by itself was missing. If you tweak your bot commands, you may want to search for that as well.

Thanks. Reidca 22:51, 3 December 2009 (UTC)Reply

Hi, the bot doesn't look at the inflection line, just at the definition lines and the headings; also, bain does seem to be on Index:Irish/b (in the third row and third column, with five asterisks). Have I misunderstood you? Conrad.Irwin 23:41, 3 December 2009 (UTC)Reply

So, there are a few digraphs that are always counted as individual letters and no letters are treated as equivalents, so no fancy bullshit like that :) Here's the alphabet:

A B C Ç D Dh E Ë F G Gj H I J K L Ll M N Nj O P Q R Rr S Sh T Th U V X Xh Y Z Zh

— opiaterein — 01:54, 4 December 2009 (UTC)Reply

Hi Conrad,