Image by mailanmaik from Pixabay

Fedora 28: Updates for Czech, Catalan, Greek, and Lithuanian Users

Continuing my previous article I’d like to write about the more recent updates in date formats in glibc. These updates will be included in Fedora 28 final release. On March 29 a new version of glibc 2.27-8 has been released in f28 branch. Together with the unreleased version 2.27-7 it features the correct date formats in Czech, Catalan, Greek, and Lithuanian.

Unfortunately, these updates have not been included in the recently released Fedora 28 Beta ISO image so all Fedora 28 users must update their systems first.

Czech

Bugzilla link: https://sourceware.org/bugzilla/show_bug.cgi?id=22963.

These changes are the most controversial. While talking to my Czech friends I had various answers to my question whether a genitive form of a month name in a date is obligatory in Czech language or not. Is April 10 in Czech 10. dubna or 10. duben? Because of these doubts the changes for Czech language were not included in glibc 2.27 initial release (February 1). But since the Czech translator has added the genitive forms of the months names to glib2 (whose aim is to provide the same features for the systems which do not support genitive forms of months names) I decided that there is no reason to wait any more.

So, this is a short message for Czech users: if you can see a date formatted incorrectly in Czech language because a month name should be nominative rather than genitive, then you must change the date format specifier from "%B" to "%OB" in the translation of an application as soon as possible. I am sorry about the confusion but other inflected languages require a genitive case here. The "%OB" format specifier has been introduced in order to support the cases where a nominative form is required.

By the way: probably the same problem will be in Serbian and Slovak but so far no changes have been introduced in these languages. It would be good to make some decisions before glibc 2.28 is released which is planned on August 1 this year, and better not in the last minute – one month or more before would be recommended.

Catalan

Bugzilla link: https://sourceware.org/bugzilla/show_bug.cgi?id=22848.

We are in April which is a good time to discuss the Catalan language because April in Catalan is abril and the date April 10 is 10 d’abril. The next month will be May (Catalan: maig) and the date May 10 will be 10 de maig.

As I wrote in the previous article, this update had already landed in Fedora Rawhide but now it has been also included in Fedora 28 repository. However, this is not the only change. It turns out that in Catalan the de preposition (or d’ if the following noun begins with a vowel) obligatorily must be added before the abbreviated months names, so there is not only d’abril but also d’abr.

This causes some problem: the ls command line utility which displays file modification timestamp (ls -l) limits abbreviated months names to 5 letters. Let’s see how it looks in Catalan if we use the correct genitive case:

$ LANG=ca_ES.utf8 ls -l
total 0
-rw-rw-r--. 1 rl rl 0  1 de ge 00:00 20180101.test
-rw-rw-r--. 1 rl rl 0  2 de fe 00:00 20180202.test
-rw-rw-r--. 1 rl rl 0  3 de ma 00:00 20180303.test
-rw-rw-r--. 1 rl rl 0 14 d’abr  2018 20180414.test
-rw-rw-r--. 1 rl rl 0  5 de ma  2018 20180505.test
-rw-rw-r--. 1 rl rl 0  6 de ju  2018 20180606.test
-rw-rw-r--. 1 rl rl 0  7 de ju  2018 20180707.test
-rw-rw-r--. 1 rl rl 0  8 d’ag.  2018 20180808.test
-rw-rw-r--. 1 rl rl 0  9 de se  2018 20180909.test
-rw-rw-r--. 1 rl rl 0 10 d’oct  2018 20181010.test
-rw-rw-r--. 1 rl rl 0 11 de no  2018 20181111.test
-rw-rw-r--. 1 rl rl 0 12 de de  2018 20181212.test

March and May are both displayed as de ma and June and July as de ju. I have already filed the request for enhancement against the coreutils project and it has been added upstream – we are waiting for the coreutils 8.30 release which I suspect will be in a month. Will it make it to Fedora 28 before the final release?

Greek

Bugzilla link: https://sourceware.org/bugzilla/show_bug.cgi?id=22937.

These changes are not revolutionary but still interesting. Greek is an inflected language (same as Slavic languages and Latin) and the differences between the genitive and nominative cases are visible even in abbreviated forms of some months names. For example, the month May in Greek is Μάιος in the nominative case and the genitive case is Μαΐου; the abbreviated forms are Μάι and Μαΐ, respectively. From now this difference is correctly supported in Linux.

The change is also visible in an output of ls -l command:

$ LANG=el_GR.utf8 ls -l
σύνολο 0
-rw-rw-r--. 1 rl rl 0 Ιαν   1 00:00 20180101.test
-rw-rw-r--. 1 rl rl 0 Φεβ   2 00:00 20180202.test
-rw-rw-r--. 1 rl rl 0 Μαρ   3 00:00 20180303.test
-rw-rw-r--. 1 rl rl 0 Απρ  14  2018 20180414.test
-rw-rw-r--. 1 rl rl 0 Μαΐ   5  2018 20180505.test
-rw-rw-r--. 1 rl rl 0 Ιουν  6  2018 20180606.test
-rw-rw-r--. 1 rl rl 0 Ιουλ  7  2018 20180707.test
-rw-rw-r--. 1 rl rl 0 Αυγ   8  2018 20180808.test
-rw-rw-r--. 1 rl rl 0 Σεπ   9  2018 20180909.test
-rw-rw-r--. 1 rl rl 0 Οκτ  10  2018 20181010.test
-rw-rw-r--. 1 rl rl 0 Νοε  11  2018 20181111.test
-rw-rw-r--. 1 rl rl 0 Δεκ  12  2018 20181212.test

Lithuanian

Bugzilla link: https://sourceware.org/bugzilla/show_bug.cgi?id=22932.

These changes are minor. The Lithuanian translator had just asked to use in glibc the same abbreviated months names as he used in glib2 and which are also provided by CLDR – so for example the abbreviated name of April will be displayed as bal. rather than Bal now.

This change could be already visible in the ls -l output – unfortunately, for now only the numerical date formats are used:

$ LANG=lt_LT.utf8 ls -l
viso 0
-rw-rw-r--. 1 rl rl 0 2018-01-01 00:00 20180101.test
-rw-rw-r--. 1 rl rl 0 2018-02-02 00:00 20180202.test
-rw-rw-r--. 1 rl rl 0 2018-03-03 00:00 20180303.test
-rw-rw-r--. 1 rl rl 0 2018-04-14 20180414.test
-rw-rw-r--. 1 rl rl 0 2018-05-05 20180505.test
-rw-rw-r--. 1 rl rl 0 2018-06-06 20180606.test
-rw-rw-r--. 1 rl rl 0 2018-07-07 20180707.test
-rw-rw-r--. 1 rl rl 0 2018-08-08 20180808.test
-rw-rw-r--. 1 rl rl 0 2018-09-09 20180909.test
-rw-rw-r--. 1 rl rl 0 2018-10-10 20181010.test
-rw-rw-r--. 1 rl rl 0 2018-11-11 20181111.test
-rw-rw-r--. 1 rl rl 0 2018-12-12 20181212.test

Is this only because the Lithuanian translators did not like the abbreviated months names and they decided that the ls command line utility should display only numbers? If this was the reason then you can restore the text format now. I can’t speak Lithuanian but I would suggest this form:

$ LANG=lt_LT.utf8 ls -l --time-style +"%b %e"
viso 0
-rw-rw-r--. 1 rl rl 0 saus.  1 20180101.test
-rw-rw-r--. 1 rl rl 0 vas.   2 20180202.test
-rw-rw-r--. 1 rl rl 0 kov.   3 20180303.test
-rw-rw-r--. 1 rl rl 0 bal.  14 20180414.test
-rw-rw-r--. 1 rl rl 0 geg.   5 20180505.test
-rw-rw-r--. 1 rl rl 0 birž.  6 20180606.test
-rw-rw-r--. 1 rl rl 0 liep.  7 20180707.test
-rw-rw-r--. 1 rl rl 0 rugp.  8 20180808.test
-rw-rw-r--. 1 rl rl 0 rugs.  9 20180909.test
-rw-rw-r--. 1 rl rl 0 spal. 10 20181010.test
-rw-rw-r--. 1 rl rl 0 lapkr 11 20181111.test
-rw-rw-r--. 1 rl rl 0 gruod 12 20181212.test

For now the dots after lapkr and gruod do not fit but, as I wrote above while discussing the Catalan language, the problem has been already fixed upstream and sooner or later the update will land in Fedora.

Summary

After adding Catalan and Czech support now we have 9 languages which display the dates correctly using the required genitive case (with previously supported Belarusian, Croatian, Greek, Lithuanian, Polish, Russian, and Ukrainian). Belarusian and Russian are not the only which require the different genitive and nominative forms of abbreviated months names, the same is required in Catalan (because of the de or d’ preposition) and in Greek.

Same as previously, if you see in the screenshots in this article any errors in date formats which can be fixed by translators, like missing punctuation marks or incorrect day/month order then please contact the translators of the respective applications.

10 thoughts on “Fedora 28: Updates for Czech, Catalan, Greek, and Lithuanian Users”

  1. Pingback: Links 12/4/2018: Stable New Kernels, Neptune 5.1 | Techrights

  2. Hello!
    Thanks for your hard work!
    I translate software in Occitan, you may know that Occitan and Catalan are/were a single language. Thus we share the same troubles about dates and their formats.
    I’m updating the GLIB files but I have no clye where to submit them.

    1. Rafał Lużyński

      You are welcome, Quentí. If you mean GLib (GLib2) then translations should be submitted via Damned Lies: https://l10n.gnome.org/. But I can’t see the Occitan language team in this service so maybe you should start it: https://wiki.gnome.org/TranslationProject/StartingATeam/. On the other hand the translation already exists and is updated sometimes: https://gitlab.gnome.org/GNOME/glib/blob/master/po/oc.po. The link inside says the website of the translation team is: https://totenoc.eu/.

      However, if you think about things like the list of month names etc. the authoritative place to submit is CLDR: http://st.unicode.org/cldr-apps/v#/oc/Gregorian/ and it does not show the nominative/genitive month names for Occitan. Then glibc and other systems update their data from this database. I understand your need. Please at least start the process of update of CLDR and then I will also update the glibc locale database. For other projects I can’t help more, only redirect you to the proper places.

        1. Rafał Lużyński

          1. I don’t know how to register to CLDR but here you can file a ticket (without registration): https://unicode.org/cldr/trac/newticket
          2. These files look more like for glibc rather than for glib2. Is that what you want? Yes, they should match CLDR so at least filing a ticket and drawing some attention would be helpful. Regarding glibc, I am going to open a new Bugzilla report for those languages which have not been updated in the previous development cycle. They are about 10.
          3. Nominative: when the month is named by itself (“April”, “this is April”, Polish: “kwiecień”). Genitive: when you want to express dependency (“the 20th day of April”, “beginning of April”, “20 d’abrial”, Polish: “20 kwietnia”) – this does not exist in English hence difficult to explain. Depending on the grammar rules this does not have to be really nominative/genitive. BTW, your files look correct.

  3. Hello Rafal,
    I’ve just created the issue for the CLDR.
    If you have any information to help spread the correct form of Occitan dates, you are very welcome! What applies to Catalan, has hight chances to apply to Occitan too.

    1. Rafał Lużyński

      That’s great, Quentí.

      1. Please provide a link to your CLDR ticket.
      2. How to introduce the correct Occitan dates to glibc – see this link: https://sourceware.org/bugzilla/show_bug.cgi?id=23140. You may contribute your patch or once I have the sufficient info I will do it. You may also open a new bug report just for the Occitan language if you want it to be processed separately from other languages.
      3. How to introduce them to GLib which provides the same features for GNOME and GNOME derived software running on older Linux systems: please contact the Occitan GNOME translation team (link above) or join them or if you are a member already then just fix the translation for GLib. If whole translation team is unresponsive we may contact some superpower people who will fix the translations like they did here: https://gitlab.gnome.org/GNOME/glib/commit/7dc3c8f.
      4. Does Occitan language require “de”/”d’” before the abbreviated month names, same as Catalan: “de gen.”, “de febr.”, and so on?
      5. Is it correct that GLib provides “abril” and “agost” unabbreviated? What about their genitive forms, wouldn’t “d’abril” and “d’agost” be too long for the abbreviated context?

      1. 1. Here is my ticket for CLDR: https://unicode.org/cldr/trac/ticket/11115
        2. How can I help you for the patch?
        3. I am planing to get in touch with the Occitan GNOME team this week-end.
        4. That’s a good question…What was decided for Catalan? The abbreviations should be short by nature, I don’t see the point of adding this “de/d’”. I think we should do the same as Catalan for Occitan about it.
        5. I don’t know why they wrote the complete word, might have considered they were short enough to fit. Again, I think we shouldn’t have De/D’ in the abbreviated writing.

        Many thanks again!

        1. Rafał Lużyński

          Thanks a lot! I have added a patch for Occitan to the bug report, it is extracted from your repository, please review: https://sourceware.org/bugzilla/show_bug.cgi?id=23140. This is only part of your changes. The others seem to be unrelated with dates so I think they should go to another bug report although they can be committed together. Regarding “de”/”d’” in abbreviated forms: Catalan locale data have this (see this article) and you have also added them to your repo so I copied as well. We have time to discuss if you want to change your mind, though.

Comments are closed.