@ArthurReutenauer I probably read the documentation to find out how to use it and discovered it didn't mention Welsh. I definitely looked later than 2009. I'd never even heard of polyglossia then ;).
@cfr Thanks for checking, I have no idea how the ‘ñ’ crept up there (probably a copy-paste from Babel). I take it Babel has similar problems? Did you report it to anyone then?
@cfr Well, again, it’s a flaw of the package, but to be fair the whole idea was that it was going to be experimental and benefit from people experimenting with it. Which has had limited success.
@cfr Well, here’s you chance to fix things, at least in Polyglossia. Do you have any pointer to an accurate and complete date specification? Even in Welsh, I think I can figure it out.
@ArthurReutenauer The mistake for dates is in everything TeX related because everyone copied the babel file and assumed it was right. (Why not?) So datetime had the same problem. It is also wrong in datetime2.
@cfr That’s just sad :-( And I can testify to the fact that François introduced errors when copying from Babel (mostly because of the conversion to Unicode), so it doesn’t get better ;-) But I’ve got useful reports over time.
@cfr I was going to make an update to Polyglossia in a few days (for other things, most notably a new languages), so it could get fixed very soon.
@ArthurReutenauer Here's an English version: bangor.ac.uk/cymorthcymraeg/dyddiadau.php.en. But do NOT take the months from the bottom table. The o is not used in formal date specifications: it is like saying 'of January' - you don't write the 'of'. But it causes a soft mutation, so the names you need for the spec are not there. Don't change the month names - those don't need changing. Does that make sense?
@ArthurReutenauer They do not have the mutation. And they shouldn't. If you say 'o...' then you want a mutation. But you don't write that. So we don't want the mutation in the .ldf. These are correctly unmutated in the current .ldf in Babel and Polyglossia. (And datetime.)
@ArthurReutenauer That is, the site is correct to mutate the names in the version they give but they are telling you how to say the date - not how to write it formally.
@ArthurReutenauer I think some other parts of the .ldf are less than ideal, but the rest is, at least, recognisably Welsh. It is the ordinals in dates which is really not good.
@cfr Yes, I see T mutates into D and M into F, that makes sense given what I know of Breton.
@cfr If you mean the captions, that’s an issue for every language - you wouldn’t guess how many discussions I’ve had about the translation of references as opposed to bibliography, or that of glossary (mostly I’ve witnessed the discussion of course, but there have been many - some of them before I even took over Polyglossia).
@ArthurReutenauer Does Breton have different kinds of mutations as Welsh does? Interesting that the soft one is similar. But it makes sense - Breton is recognisable when somebody says it, although it looks completely unfamiliar to me written.
@ArthurReutenauer Yes. I just remember my Welsh tutor marked stuff in red which Babel had auto-generated. (Polyglossia's file looks identical to Babel's.)
@cfr It may be only the spelling, I can’t remember exactly. But the voiceless > voiced mutation is very recognisable. And I seem to remember that Irish uses an ‘h’ to mark that mutation, i. e., “mh”, pronounced ... something like [f], I guess (and “bh” probably like [v]).
@cfr I’m sorry about that :-( But now we’re fixing that, I can also add a formal option to print the date with the “o”.
@ArthurReutenauer Oh, yes. We do the 'h's, too. Actually two different mutations involve additions of 'h's and there are some where you add an 'h' before a vowel which is not one of the three mutation types but somehow similar.
@cfr You’ve got all the details of Breton mutations on en.wikipedia.org/wiki/Breton_mutations (obviously) if you want to compare. The article says that Breton, like Cornish, never developed the “true” nasal mutation that Welsh has.
@cfr And it does seem like the mistakes in the .ldf files are coming from Breton, that does indeed have a ‘ñ’.
@ArthurReutenauer You'd need mutated month names for that option, though, rather than the standard ones. Don't apologise - I'm glad somebody is fixing at least one Welsh file for TeX!
@cfr Well, I’ve got them in the link you gave me. And you’re welcome! I’ll try to propagate the changes to Babel – I’ve got a plan.
@cfr When I’m done we can check if Gaelic, Irish and Breton need any improvement; and then tackle Manx and however many standards of Cornish there are these days ;-)
@cfr And then move on to extinct Celtic languages such as Gallic and the languages of Northern Italy. They had their own writing system – one that is imperfectly encoded in Unicode, as I understand.
@ArthurReutenauer They have more kinds than we do. Although they seem to affect fewer letters. Interestingly, that looks almost recognisable where the letters exist in Welsh or where I can guess an analogue. (No idea about z for example.) Similar functions, too. Although I notice that Breton has an indefinite article which is surely an extravagance ;).
@ArthurReutenauer We? ;) I only do Welsh.... Breton looks familiar. Gaelic of either variety is fairly alien. Don't know about Cornish. With Gaelic, I get the general impression that the sentence structure is somewhat similar but the words could mean almost anything ;).
@cfr Not at the moment, no. Obviously there is a submerged part of the iceberg, which is hyphenation patterns. For all I know, they have been compiled by Welsh language experts under the supervision of Yannis Haralambous (himself a TeX expert), but they may always contain mistakes that are usually only spotted when the word they affect happens to be hyphenated right at that point, hence not often.
@ArthurReutenauer I take back what I said about them having four kinds of mutation. There's only three, I think. We have something like the 'Mixed' one, too i.e. where one kind of mutation applies for some letters and another for other letters. (Usually where one kind isn't applicable to all the letters.)
@cfr I really don’t know that much about Breton; I only picked up a textbook when I moved to Brittany, which I soon put down once I realised that really absolutely no one spoke Breton (even in Brest, where I lived for one year).
@cfr Well, do tell me (or better, email tex-hyphen@tug.org) if you see anything.
@ArthurReutenauer I assume that the headtoname is in the sense of sending something to a person e.g. to Mike Roberts as opposed to e.g. to the dentist?
@ArthurReutenauer Yes. But I shouldn't have said 'as opposed to'. They are both people. I meant as opposed to e.g. to the dental surgery. I.e. place. The words for 'to' are different. 'At' is only for people.
@cfr OK, thanks. I’m afraid there’s no good fix for that. The way letter.cls, and presumably other letter classes, are written, they just put stuff at specific locations on the page (according to usual conventions, etc.); there’s nothing that prevents people to write the name of an institution instead of a person.
