1. 07 Jul, 2019 6 commits
  2. 24 Apr, 2019 3 commits
    • Mathieu Bridon's avatar
      ci: Remove obsolete configurations · 03e1887b
      Mathieu Bridon authored
      We do not use Travis any more now that we have the Gitlab CI.
      
      Additionally, we won't be needing the Docker build hook as we are moving
      our images to the Gitlab registry.
      03e1887b
    • Mathieu Bridon's avatar
      ci: Publish libcangjie images to the Gitlab registry · 832d7742
      Mathieu Bridon authored
      These images will be used by the pycangjie CI, which builds against
      libcangjie.
      
      This requires changing a bit the Dockerfile, as it needs to pull the
      build-essential image from the Gitlab registry, not from the Docker Hub.
      832d7742
    • Mathieu Bridon's avatar
      ci: Build and test libcangjie · 2b37c7f6
      Mathieu Bridon authored
      This uses the new build-essential images published to the Freedesktop
      Gitlab registry:
      
      build-essential!7
      
      The CI builds and tests libcangjie on multiple operating systems in
      those images, exactly like Travis used to do it.
      2b37c7f6
  3. 16 Feb, 2019 1 commit
  4. 14 Feb, 2019 1 commit
  5. 13 Feb, 2019 4 commits
  6. 02 Jun, 2018 1 commit
  7. 20 May, 2018 2 commits
    • Mathieu Bridon's avatar
      tests: Fix literal strings · 111d6d9a
      Mathieu Bridon authored
      This was done a while ago, to make MSVC happier as it doesn't like
      non-ascii strings in C code.
      
      (Or at least, that was the case at the time. This was almost 5 years ago
      after all…)
      
          commit 273d280e
          Author: Linquize <linquize@yahoo.com.hk>
          Date:   Thu Oct 17 21:20:36 2013 +0800
      
              Do not write Chinese literal strings directly
      
              MSVC does not encoding Chinese literal strings as UTF-8
      
      However, for some reason the backslashes got escaped in this file (and
      only in this file), and we missed it on code review.
      111d6d9a
    • Mathieu Bridon's avatar
      docs: Remove trailing whitespaces · 0c373a11
      Mathieu Bridon authored
      0c373a11
  8. 09 May, 2018 7 commits
    • Mathieu Bridon's avatar
      data: Remove the section sign · f47875aa
      Mathieu Bridon authored
      This never should have been a symbol, it is a punctuation mark.
      
      However, it seems to be entirely unused in both Chinese and Japanese, so
      let's just remove it.
      f47875aa
    • Mathieu Bridon's avatar
      data: Remove the question mark · 0a563d64
      Mathieu Bridon authored
      Chinese and Japanese traditionally don't use the question mark. Instead
      they normally use the fullwidth question mark: '?'
      
      Some people might still want to use the question mark, but in that case:
      
      * they get it in ibus-cangjie when they enable the "Half-Width" option
        and type the '?' key;
      * they can always switch to the English keyboard layout;
      
      As a result, keeping this in doesn't seem necessary, much like we don't
      have codes to input the full stop or colon characters.
      0a563d64
    • Mathieu Bridon's avatar
      data: Remove some abbreviations · 9d0fc7a8
      Mathieu Bridon authored
      Those don't seem to be used in Chinese or Japanese.
      9d0fc7a8
    • Mathieu Bridon's avatar
      data: Remove maths symbols · d27962dd
      Mathieu Bridon authored
      There are much nicer ways to type maths, which academics are most likely
      already using. (e.g LaTeX, LibreOffice Maths, …)
      
      I'm keeping some of them though, that have documented non-maths usage in
      Japanese for example.
      d27962dd
    • Mathieu Bridon's avatar
      data: Correctly mark some punctuation characters · 756f8501
      Mathieu Bridon authored
      The two dashes (en and em) are sometimes used as punctuation (e.g to
      indicate ranges) and as such must be marked appropriately.
      
      As for ※, this is the Japanese reference mark (aka "rice symbol"), used
      similarly to the asterisk in Western languages.
      756f8501
    • Mathieu Bridon's avatar
      data: Better classify the Bopomofo tonal marks · 6258db00
      Mathieu Bridon authored
      Those aren't random symbols, they are the characters used to mark tones
      in Bopomofo/Zhuyin.
      6258db00
    • Mathieu Bridon's avatar
      data: Remove inaccessible characters · 4d8a41fa
      Mathieu Bridon authored
      About 80 characters in our table have no code associated, whether in
      Cangjie3, Cangjie 5 or even with our "short code" system for
      punctuation.
      
      These are impossible to obtain with libcangjie, so there is really no
      point in keeping them.
      
      They can always be added back if someone complains about their absence
      and provides a code for them.
      4d8a41fa
  9. 06 May, 2018 7 commits
    • Mathieu Bridon's avatar
      data: Remove the enclosed CJK numerals · dc2bff29
      Mathieu Bridon authored
      These characters were added to Unicode for compatibility with legacy
      standards.
      
      However, their usage is rather unclear, and everything points towards
      the fact that they are not necessary at all.
      
      These characters were identified by finding the characters whose Unicode
      code point is between U+3200 and U+32FF.
      dc2bff29
    • Mathieu Bridon's avatar
      data: Remove the small form variants · 136709d6
      Mathieu Bridon authored
      These characters were added to Unicode for compatibility with the
      Chinese National Standard CNS 11643.
      
      However, their usage is rather unclear, and everything points towards
      the fact that they are not necessary at all.
      
      These characters were identified by finding the characters whose Unicode
      code point is between U+FE50 and U+FE6F.
      136709d6
    • Mathieu Bridon's avatar
      data: Remove the enclosed numerals · 52aa0808
      Mathieu Bridon authored
      These characters were originally intended for use as bullets for lists.
      
      However, people will generally use the list facilities from the
      applications (e.g word processors), so these aren't actually used.
      
      They are included in the Unicode standard "for interoperability with the
      legacy East Asian character sets and for the occasional text context
      where such symbols otherwise occur."
      
      That is, they are only useful when they need to be displayed, but not in
      an input method.
      
      These characters were identified by finding the characters whose Unicode
      code point is between U+2460 and U+24FF.
      52aa0808
    • Mathieu Bridon's avatar
      data: Remove the Roman numerals · 24cd9833
      Mathieu Bridon authored
      This is a Chinese input method, so let's tighten our focus a bit.
      
      These characters were identified by getting the name of each character
      with the Python `unicodedata.name()` function, searching for those
      characters whose name start with "ROMAN NUMERAL " or "SMALL ROMAN
      NUMERAL ".
      24cd9833
    • Mathieu Bridon's avatar
      data: Remove the Cyrillic letters · d118fe6e
      Mathieu Bridon authored
      This is a Chinese input method, so let's tighten our focus a bit.
      
      These characters were identified by getting the name of each character
      with the Python `unicodedata.name()` function, searching for those
      characters whose name start with "CYRILLIC CAPITAL LETTER " or "CYRILLIC
      SMALL LETTER ".
      d118fe6e
    • Mathieu Bridon's avatar
      data: Remove the Greek letters · b7f7f73c
      Mathieu Bridon authored
      This is a Chinese input method, so let's tighten our focus a bit.
      
      These characters were identified by getting the name of each character
      with the Python `unicodedata.name()` function, searching for those
      characters whose name start with "GREEK CAPITAL LETTER " or "GREEK SMALL
      LETTER ".
      b7f7f73c
    • Mathieu Bridon's avatar
      data: Remove the vertical presentation forms · 1f9bec41
      Mathieu Bridon authored
      Back at GUADEC 2015, Behdad told me we don't need those.
      
      The Unicode Standard describes them as "compatibility characters":
      
      > Conceptually, compatibility characters are characters that
      > would not have been encoded in the Unicode Standard except
      > for compatibility and round-trip convertibility with other
      > standards. [...] For the most part, these are widely used
      > standards that pre-dated Unicode
      >
      > [...]
      >
      > The purpose for the inclusion of compatibility characters
      > like these is not to implement or emulate alternative text
      > models, nor to encourage the use of plain text distinctions
      > in characters which would otherwise be better represented
      > by higher-level protocols or other mechanisms.
      
      Basically, people aren't expected to input those characters
      directly. Instead, they should always input the regular form, and it is
      the software displaying the character which has to choose either the
      horizontal vertical form, depending on the typing orientation.
      
      These characters were identified by getting the name of each character
      with the Python `unicodedata.name()` function, searching for those
      characters whose name start with "PRESENTATION FORM FOR VERTICAL ".
      
      Cleaning those out is long overdue.
      
      Fixes #88
      1f9bec41
  10. 29 Apr, 2018 1 commit
  11. 28 Apr, 2018 1 commit
    • Mathieu Bridon's avatar
      ci: Run the tests on Fedora Rawhide as well · e59bbe4b
      Mathieu Bridon authored
      Back when we moved to testing with our custom images, Rawhide was broken
      in Docker.
      
      This has since been resolved, so we can start building on Rawhide as
      well, which helps insuring we continue working on future versions of
      Fedora.
      e59bbe4b
  12. 21 Nov, 2017 1 commit
  13. 15 Sep, 2017 2 commits
  14. 14 Sep, 2017 1 commit
  15. 13 Sep, 2017 2 commits
    • Mathieu Bridon's avatar
      ci: Pretty print the OS from an existing file · ce0a5a02
      Mathieu Bridon authored
      The CI already shows us the content of the $OS veriable (which is what
      we put in the /var/version file), for example:
      
          $ docker pull "cangjians/build-essential:${OS}"
          ubuntu-lts: Pulling from cangjians/build-essential
      
          Status: Downloaded newer image for cangjians/build-essential:ubuntu-lts
      
      So we don't need the /var/version file at all.
      
      However, knowing this is `ubuntu-devel` is great, but in a few months,
      will we remember what actual version was the Ubuntu LTS?
      
      We can get a more interesting output from the /usr/lib/os-release file,
      which is a cross-distro standard:
      
          $ docker run --rm "cangjians/libcangjie:${OS}" bash -c '. /usr/lib/os-release && echo $PRETTY_NAME'
          Ubuntu 16.04.3 LTS
      ce0a5a02
    • Koala Yeung's avatar
      Merge pull request #100 from Cangjians/docker · d16aa1b6
      Koala Yeung authored
      Docker-based CI
      
      resolves #99 
      d16aa1b6