Read about our upcoming Code of Conduct on this issue

  1. 16 Nov, 2021 1 commit
    • Georges Racinet's avatar
      Merged upstream GDK as of roughly GitLab 14.4 · 673b82a891e0
      Georges Racinet authored
      This entails an update of the Ruby version to 2.7.4
      (totally necessary as it can conflict witht the one for Rails).
      With this change also comes the split of the Makefile, a good thing in
      the absolute, yet it could take us some time to stabilize.
  2. 18 Oct, 2021 1 commit
  3. 26 Aug, 2021 1 commit
    • Georges Racinet's avatar
      Makefile: restoring the `install` target · 171929442264
      Georges Racinet authored
      It was inadvertly removed with the `update` target (we don't support
      the latter).
      Because it provides more feedback than the `all` default target,
      we suggest to use `make install` in the documentation.
  4. 19 Aug, 2021 1 commit
  5. 31 May, 2021 1 commit
  6. 19 May, 2021 1 commit
  7. 17 May, 2021 1 commit
  8. 04 May, 2021 1 commit
  9. 08 Apr, 2021 1 commit
  10. 30 Mar, 2021 1 commit
  11. 16 Mar, 2021 1 commit
  12. 14 Mar, 2021 1 commit
  13. 03 Mar, 2021 1 commit
  14. 16 Mar, 2021 1 commit
  15. 10 Feb, 2021 4 commits
    • Georges Racinet's avatar
      Release series: heptapod-switch-new-stable and documentation · bd215f39ec2b
      Georges Racinet authored
      Now that the names of release series are fixed, we can use them
      in the heptapod-switch-new-stable script, making the development
      cycle idea more that a feeling.
      The case of release HDK workspaces wasn't documented at all until
    • Georges Racinet's avatar
      doc: explained the development cycle idea · 8ef0e15ac7e9
      Georges Racinet authored
      So we now have cycles (Heptapod 0.18), series (stable) and
      The doc is rewritten for emphasis on how a HDK workspace typically
      starts on the default series, then moves on to stable, and can
      eventually die out.
      If at some point we extend the (currently very short) support time,
      we could add, to paraphrase Debian, `oldstable` and even `oldoldstable`
      or someone would come with better namings.
    • Georges Racinet's avatar
      doc: fixed blockquote · c5e9cb4f8d75
      Georges Racinet authored
      Must have thought in RST while writing this.
    • Georges Racinet's avatar
      doc: mention `gdk help` and shorten its output · 90a6fecf03db
      Georges Racinet authored
      The shortening prevents the doc to become out of sync, even
      if it doesn't matter much.
      `gdk help` is really useful.
  16. 13 Jan, 2021 2 commits
  17. 11 Jan, 2021 6 commits
    • Georges Racinet's avatar
      Merge from heptapod-stable branch · cb512dd165f8
      Georges Racinet authored
      Improvements in global scripts and documentation
    • Georges Racinet's avatar
      README: two typos · a3fb8b0f82d6
      Georges Racinet authored
    • Georges Racinet's avatar
      README: upgraded part about virtualenv · cf0f0e504d87
      Georges Racinet authored
      The `virtualenv` executable itself is no longer used (our
      Makefile only uses `$python -m venv`)
    • Georges Racinet's avatar
      README: Python 2 is no longer a requirement · 331c50b02726
      Georges Racinet authored
    • Georges Racinet's avatar
    • Georges Racinet's avatar
      Merged and adapted upstream GDK at the time of v13.7.0 · 311e99cc3eee
      Georges Racinet authored
      More recent commits in GDK have preparations for Ruby 3.0 which
      we don't want to look at right now.
      Notable adaptations:
      - GDK::Config: non-setting Heptapod-specific methods
        The merge brings in the `ConfigSettings#validate!` method,
        which does not understand methods with arguments.
        Renaming our methods with `__` to match the ignore rule
      - GDK::Config: `heptapod_revisions`
        The `config` instance is no longer magically present
        in that kind of context, but it is in each nested `string`
        We could fix by introducing another helper function, but there
        is now a `hash_setting` type. It forces us to change calling
        sites, but that's still much less annoying boilerplate for what
        should be a simple reexposition.
      - hdk.hgrc generation
        The pre-escaped Unix Domain socket URI is not available anymore,
        and it seems the current norm is to go through the workhorse URI
        (seen something similar in Omnibus).
  18. 04 Jan, 2021 1 commit
  19. 15 Dec, 2020 1 commit
  20. 01 Dec, 2020 1 commit
  21. 27 Nov, 2020 2 commits
  22. 07 Oct, 2020 1 commit
  23. 03 Sep, 2020 1 commit
  24. 31 Aug, 2020 1 commit
  25. 02 Aug, 2020 1 commit
  26. 01 Aug, 2020 1 commit
    • Georges Racinet's avatar
      heptapod-tests: launcher script, with documentation · cb83f6c697b9
      Georges Racinet authored
      We settled for a top level script that passes the needed
      arguments rather than fiddling with `tox.ini`, because it's
      simpler to setup and (slightly) simpler to use with additional
      user options than `tox -- [OPTIONS]`.
      The launcher script is installed by default, and will complain
      explicitely if `tox` isn't available, so that makes tox a soft
      Since tox is itself (among other things) a virtualenv manager,
      we thought it would only be more brittle to preinstall it in our
      `venv3/`, and unnecessarily heavy to make a dedicated virtualenv
      (that latter point of view may change if we
      end up needing a specific tox version in the future).
  27. 18 Jul, 2020 2 commits
    • Georges Racinet's avatar
      Consequences of `.ruby-version` being under version control · 9c683f7ecea0
      Georges Racinet authored
      As previously noted `.ruby-version` is now under version control.
      Upstream stopped linking it from the Rails application,
      so `.ruby-version` it's no longer a make target.  We were making it
      a prerequisite of the HDK specific `init` target because it's
      needed to run `gdk`. We wouldn't want to make the choice to fix
      it for the HDK alone (as opposed to the GDK). This is now clearly
      This gives us the occasion to provide more in our `init` make
      - install of rake if needed (actually running on each `make`
      - install of the `gdk` command (provided by the `gitlab-development-kit`
      That doesn't prevent using a different Ruby version in the various
      components, but the HDK for Heptapod 0.15 would need some fixing
      to work with Heptapod 0.14 anyway.
    • Georges Racinet's avatar
      Merged upstream GDK for Heptapod 0.15 / GitLab 13.1 · fc8f2ebcbb92
      Georges Racinet authored
      Developpment of Heptapod 0.15 has up to now been done
      with the HDK as it was made for 0.14 (GitLab 12.10), with
      a bunch of manual modifications. We need to make it work
      out of the box now.
      An interesting difference is that `.ruby-version` is now versioned
      directly in the GDK repo. While this certainly will simplify our
      bootstrap, it may have the consequence that we'd need two branches
      of the HDK (heptapod and stable), but we'll see.
  28. 10 Jun, 2020 1 commit
  29. 03 Jun, 2020 1 commit