To this issue will be attached file with new locale settings for Bulgarian. This is new work intended non-maintained file.
Created attachment 4714 [details] new Bulgarian locale I would like to propose existing file (bg_BG) to be removed first from repository and after this new work to be added to repository.
You have to get the previous maintainer to agree with that.
It would be great to have at least some general summary of changes compared to the original locale. It seems it would be good idea to remove the generic comments about formats of various attributes and keep only locale-specific comments. The same goes for comments containing plaintext versions of attribute values. Your weekday abbreviations are wrong - if you specify Monday in week directive, the first day in the list must be Monday.
If I understand ISO/IEC TR14652:2002 properly if week start with Monday the lists with weekday names should start with Monday, but for GNU libc the lists must start with Sunday.
Hmm, you are right that strftime() does not honor week[2], I will have to fix that sometime *sigh*. In that case, it would probably be better to keep week[2]==19971130 and set first_weekday to 2, to have everything set up consistently. Anyway, trying to compile the locale throws a huge bunch of warnings.
About warnings no idea as my GNU glibc is ld version (2.7) and on other locales result is same as example: bg_BG:27: non-symbolic character value should not be used but line 27 is: title "Bulgarian locale for Bulgaria" About: LC_NAME: field `name_gen' not defined No salutation to all persons in Bulgarian. So what is correct : do not define or to define to empty string with comment "No salutation ..." ? LC_IDENTIFICATION: field `abbreviation' not defined What to set to this field BG-<VERSION> ? I don't know . It is defined only for uk_UA (ULU-2.1.12) and and en_ZA("Translate.org.za).
*** Bug 9865 has been marked as a duplicate of this bug. ***
Again, why should the existing locale be replaced? You have to get the author of the existing locale to agree.
Created attachment 5713 [details] version 3.0.3 with license update Updated license following recent discussion on libc-locales mailing list with subject "Many locale definitions need license change"
(In reply to comment #8) > Again, why should the existing locale be replaced? You have to get the author > of the existing locale to agree. Please see post http://sourceware.org/bugzilla/show_bug.cgi?id=9865#c3 . About attachments to this issue - it is new work and I would like to maintain in the future.
It seems to me I forgot to post what's new . The new version follow up http://zver.fsa-bg.org/pipermail/dict/2010-March/004652.html - a long discussion about some issues with Bulgarian locale . For public discussion version 3.0.1 was posted here http://zver.fsa-bg.org/pipermail/dict/2010-March/004743.html : A) brief description of updates. а) new author, copyright and contacts b) LC_IDENTIFICATION - new contacts - category compatible with IEC_TR_14652/2002 c) LC_CTYPE - no change d) LC_COLLATE - removed all extra collating sequences. Version 3.0.2 add "Capital letters go before small" on request. d) LC_MONETARY - no grouping, no thousand separator - currency with dot at end е) LC_NUMERIC - no grouping, no thousand separator . БДС ISO 31-0:1994 allow groping but is is not common f) LC_TIME - rewritten. see below for details g) LC_MESSAGES - no change h) LC_PAPER - no change i) LC_NAME - names with abbreviated profession, i.e. %S instead %s - abbreviation correspondind for mr. is with correct representation. j) LC_ADDRESS - country_name and lang_name in Bulgarian - postal address accordind to the norm of "Bulgarian Post" k) LC_TELEPHONE - without brace for domestic and international calls l) LC_MEASUREMENT - no change B) date and time description (LC_TIME) - description of era format - colon instead comma as time separator (preferred format discussed in the first link above) - clock time is in 24h format. Current version is ambiguous as describe 12h format without corresponding abbreviation for am/pm. New time output (for 1h 2m and 3s and for 13h 2m и 3s. for the flag %r): -------------------------------------------------------------------------------- 01:02:03 and 13:02:03 -------------------------------------------------------------------------------- Current is: -------------------------------------------------------------------------------- 1,02,03 and 1,02,03 -------------------------------------------------------------------------------- - rearranged days and month is date format New date output: -------------------------------------------------------------------------------- 2010-04-01 13:02:03: четвъртък 1 април 2010 13:02:03 EEST +%c чт 1 апр 2010 13:02:03 EEST +%x (date) 1.04.2010 +%X (time) 13:02:03 +%y (s-year) 10 +%Y (year) 2010 +%Ec (+era) чт 1 апр 2010 сл.н.е. 13:02:03 EEST +%Ex (date+era) 1.04.2010 сл.н.е. +%EX (time+era) 13:02:03 +%Ey (s-year+era) 2010 +%EY (year+era) 2010 сл.н.е. -------------------------------------------------------------------------------- Current is: -------------------------------------------------------------------------------- 2010-04-01 13:02:03: чт апр 1 13:02:03 EEST 2010 +%c 1.04.2010 (чт) 13,02,03 EEST +%x (date) 1.04.2010 +%X (time) 13,02,03 +%y (s-year) 10 +%Y (year) 2010 +%Ec (+era) 1.04.2010 (чт) 13,02,03 EEST +%Ex (date+era) 1.04.2010 +%EX (time+era) 13,02,03 +%Ey (s-year+era) 10 +%EY (year+era) 2010 --------------------------------------------------------------------------------
(In reply to comment #11) > +%EY (year+era) 2010 сл.н.е. And there is already a problem. You're completely overdoing it. Eras are pretty much only for Japan and nothing else. There are drastic changes. I understand you cannot get the previous maintainer to comment. But I need assurance. Did you announce the availability of the locale, have people test it, and report back? I have nothing else to get by.
Created attachment 5718 [details] C program to test locale The post for version 3.0.1(see link above) contain some tests. This is first test.
Created attachment 5719 [details] simple shell script to test locale second test from above post
Created attachment 5720 [details] perl script to test cyrillic sort I'm not perl expert and script will warn "Wide character in print at ./test-sort.pl line 4." but the last line is expected result.
Created attachment 5721 [details] shell script to test sort of punctuation+latin+cyrillic characters
(In reply to comment #12) > (In reply to comment #11) > > +%EY (year+era) 2010 сл.н.е. > > And there is already a problem. You're completely overdoing it. Eras are > pretty much only for Japan and nothing else. Why ? But AIX according to the internet pages define christian era (search for ibm aix localdef lctime). The definition look like: era "+:0:0000/01/01:+*:AD:%o %N";\ "+:1:-0001/12/31:-*:BC:%o %N" era_year "" era_d_fmt "" Many Bulgarian prefer neutral definition for New(Common) Era the file contain neutral abbreviations.
Created attachment 6048 [details] version 3.0.4 day in month and month without leading zero or space
*** Bug 13237 has been marked as a duplicate of this bug. ***
*** Bug 13949 has been marked as a duplicate of this bug. ***
Created attachment 6329 [details] update 'Bulgarian translators team' email "Bulgarian translators team" use new list address. Post to dict@fsa-bg.org(old one) are redirected to dict@ludost.net (new one).
I think Ulrich's questions are valid - these are very significant changes and you should get endorsement by the Bulgarian user community first.
I get consensus from bulgarian community . Since mail list was changed (moved to new host) , link to arhive http://zver.fsa-bg.org/pipermail/dict/2010-March/004652.html is not valid. The discussions related to bg locale are still available on list-mirrors. Quick review show that content of non existing "http://zver.fsa-bg.org/pipermail/dict/2010-March/004652.html" could be found on "http://bulgarian-translators-of-free-software.1048928.n5.nabble.com/3-0-1-tt4356742.html#none" (13 posts) - version 3.0.1. Version is based on discussion from : "http://bulgarian-translators-of-free-software.1048928.n5.nabble.com/-tt4356508.html" (63 posts) and some earlier posts.
Created attachment 6768 [details] restore previous copyrights
Created attachment 7265 [details] version 3.0.7 uploaded v3.0.7 with correct specification of category ( also include useless upstream fix for lang_lib as by definition if is not set is used value of lang_term )
there's a ton going on here, and bg_BG has been updated since. let's distill the changes down. we really don't want to see fully new locale files with tons of extraneous noise (like specs in the comments) added. some of these updates we'll just have to go with for now until we can get the details out of CLDR. * LC_COLLATE ... i'll just go with this one. * LC_MONETARY -mon_thousands_sep "<U00A0>" +mon_thousands_sep "" -mon_grouping 3;3 +mon_grouping -1 -grouping 3;3 +grouping -1 * LC_TIME -d_t_fmt "%x (%a) %X %Z" +d_t_fmt "%a %-d %b %Y %T %Z" -d_fmt "%e.%m.%Y" +d_fmt "%-d.%-m.%Y" -t_fmt "%k:%M:%S" +t_fmt "%T" +era "+:0:0/1/1:+*:сл.н.е.:%Ey %EC";/ + "+:0:-1/12/31:-*:пр.н.е.:%Ey %EC" +date_fmt "%A %-d %B %Y %T %Z" +era_d_t_fmt "%a %-d %b %EY %T %Z" +era_d_fmt "%-d.%-m.%EY" +cal_direction 2 These LC_TIME settings are wrong for sure: -week 7;19971130;4 +week 7;19971201;1 -first_weekday 2 +first_weekday 1 +first_workday 1 * LC_NAME -name_fmt "%s%t%g%t %m%t%f" +name_fmt "%S%t%g%t %m%t%f" -name_mr "г-дин" +name_mr "г-н" * LC_ADDRESS -postal_fmt "%f%N%a%N%d%N%b%N%sN%h, %e, %r%N%z %T%N%c%N" +postal_fmt "%a%N%f%N%d%N%b%N%s %h, %e, %r%N%z %T%N%c%N" * LC_TELEPHONE -tel_int_fmt "(+%c %a) %l" +tel_int_fmt "+%c %a %l" -tel_dom_fmt "(0%a) %l" +tel_dom_fmt "0%a %l"