this patch fixes country_name and lang_name for armenian hy_AM locale
Created attachment 26 [details] fix 2 FIXMES in hy_AM locale
I've requested comments from Pablo Saratxaga and Eugene Sevinian, which have worked on this locale earlier.
Created attachment 27 [details] revised version of the patch this patch fixes LC_NAME FIXME and uses Eastern spelling for lang_name since that's the spelling that is being used in Armenia
As I've been requested advice, here it is. I've been involved in the first version of the file (written from input of people in Armenia); but I don't speak Armenian myself; the proposed patch looks ok for me. (note however that the hy_AM file still doesn't work, due to the fact that localedef isn't able to handle the definition of a new script after the copy "iso14651_t1")
The email to Eugene Sevinian bounced with "Connection timed out: retry timeout exceeded", so I guess is it unlikely to get a comment from him at this time.
Created attachment 48 [details] final version of the patch this is the final version of the patch.hy_AM It updates the obsolete web site address and removes all the ordering information from hy_AM since it has to be moved to iso14651_t1 ( see comment #4) this patch has to be applied against libc/localedata/locales/hy_AM
Created attachment 49 [details] adds ordering information into to iso14651_t1 this patch simply adds all the ordering information that was removed from hy_AM to iso14651_t1 It has to be applied against libc/localedata/locales/iso14651_t1 file
Created attachment 79 [details] translated LC_NAME section and updated Pablo's email
Created attachment 231 [details] fixed country_car
The hy_AM patch look good to me. The collate patch to iso14651_t1 is a bit new to me. I'm not sure how the iso14651_t1 content are generated, and do not know if it is OK to append it with the proposed patch. There are some minor problems with the locale. Testing it with http://www.student.uit.no/~pere/linux/glibc/check-locale will at the moment show - Missing '% Charset: <charset>' info - Missing quotes around category standard refs - yesexpr and noexpr have '.*' It would be nice to have these problems fixed as well, but it isn't too important.
Created attachment 380 [details] fixes %Charset: and missing quotes this patch fixes: - Missing '% Charset: <charset>' info - Missing quotes around category standard refs
Comment on attachment 49 [details] adds ordering information into to iso14651_t1 this patch is actually not necessary
Created attachment 383 [details] fix yesexpr and noexpr this patch also fixes yesexpr and noexpr ".*" problem. the only warning that check-locale reports are warnings about wrong section orders. Do you want me to fix those warnings too ?
No, you do not have to change the order. That check is most important for new locales. It is better for existing locales to keep their order, to reduce the cvs diff.
Created attachment 481 [details] Wrong bug number, sorry, this patch is for BZ58.
iso14651_t1 will not be modified. Since the patch depends on this I'll not apply it.
Actually, this patch doesn't depend on iso14651_t1. There is only one patch from comment #13 (id=383) that should be applied to libc/localedata/locales/hy_AM after which the Armenian locale can be easily compiled by running: $ localedef -f UTF-8 -i hy_AM hy
If I would use the patch from comment #13 then the collation handling would be just as in iso14651_t1 which, according to comment #7, is not sufficient.
Hi Ulrich, If you look at patch from comment #7 you will notice that Armenian letters in Unicode are already ordered correctly. (Capitals go from <U0531>-<U0556> and small letters go from <U0561>-<U0586>). Since Unicode already has the correct ordering isn't the default behavior already correct, which would make patch from comment #7 unnecessary? If not, what do you suggest to get the Armenian locale working?
This one appears to still be under discussion.
I applied the last patch.
Hi Ulrich, I just looked at Ubuntu 6.06 which includes the Armenian locale and I noticed 2 things: 1. It seems that they have all the fixes from my current patch + some extra patches like translit_combining 2. They didn't remove the collation from hy_AM like I did, which is also nice. I think we should consider using Ubuntu's changes instead of using my patch. I am attaching Ubuntu's hy_AM file.
Created attachment 994 [details] Armenian locale from Ubuntu Dapper 6.06
Why? First you tell me the changes are correct, then that they are not. I'm this -><- close to rip the whole thing out and remove it forever. What is the problem with the source now in CVS? Explain is detail, and fast.
There is nothing wrong with the current CVS version. What I was trying to say is that after doing a diff between the current CVS version and Ubuntu version I found 2 differences, which are: 1. They added a translit section for Armenian -> English transliteration and I can submit that as a separate self-contained patch. 2. They didn't remove the collating symbols, which means that they fixed the localedef parser. It is _not_ necessary because unicode ordering is correct. I am goint to write an email to Pablo now and ask him why did he add the collating symbols in the first place. So this bug can be closed now, and I'll submit further improvements as a separate bug.