1. 14 Mar, 2019 3 commits
  2. 12 Mar, 2019 1 commit
  3. 05 Mar, 2019 7 commits
  4. 04 Mar, 2019 6 commits
    • Chun-wei Fan's avatar
      win32/README.txt: Update SpiderMonkey build instructions · 169e21fc
      Chun-wei Fan authored
      As there are some gotcha's/pitfalls for building SpiderMonkey for use
      with GJS on Windows, note in the README file in win32 telling people
      -They could obtain the SpiderMonkey sources (which is not easy,
      -They know the flags and envvar's that must be passed into
       SpiderMonkey's configure line for GJS to build and run correctly.
      -That they should not try to link to mozglue.lib at any point during the
       GJS build.
      Fixes issue #228
    • Chun-wei Fan's avatar
      Visual Studio builds: Refine things a bit · 5a5d8eb4
      Chun-wei Fan authored
      Use separate output directories for different Visual Studio versions,
      and ensure that we do indeed wipe out the vcxxx.pdb files and the object
      directories upon clean.  Make the compiler generate vcxxx.pdb files
      under the respective intermediate object directories as far as possible.
      Also avoid polluting the source tree during the build by copying
      config.h.win32 to libgjs' intermediate directory and also create
      generate the filelist for introspection in libgjs' intermediate
    • Chun-wei Fan's avatar
      win32/NMake: Require Visual Studio 2017 15.6 or later · e3492ffd
      Chun-wei Fan authored
      The code, in addition to SpiderMonkey 60, requires C++-14 features that
      are supported in Visual Studio 2017 or later, plus we need to use the
      /Zc:externConstexpr compiler flag, which is provided in Visual Studio 2017
      15.6.x or later, so update the NMake Makefiles to check for that.
    • Philip Chimento's avatar
      CI: Remove "Total errors found" line from cpplint output · b1b9f3b9
      Philip Chimento authored
      The newly built Docker image must have gotten a newer version of cpplint
      which prints a total error count at the end. We need to remove this line
      or the error counts will be flagged as a diff in the linter output.
    • Philip Chimento's avatar
      gi: Initialize pointer members of wrapper classes · 6c00884d
      Philip Chimento authored
      Cppcheck is newly complaining about this, perhaps because of the changes
      to the `friend class` declarations.
    • Chun-wei Fan's avatar
      gjs: Work around Visual Studio 2017 bug · c455da15
      Chun-wei Fan authored
      It is known that codecvt_utf8_utf16 does not work when converting to
      char16_t because the symbols are somehow missing from the C++ runtime
      .lib.  So, we use the Windows APIs to do that for us, and use the
      Windows-specific std::u16string constructor using the wstring that we
      obtain using the Windows APIs.
      See: https://social.msdn.microsoft.com/Forums/en-US/8f40dcd8-c67f-4eba-9134-a19b9178e481/vs-2015-rc-linker-stdcodecvt-error?forum=vcgeneral
  5. 03 Mar, 2019 6 commits
  6. 02 Mar, 2019 6 commits
  7. 01 Mar, 2019 1 commit
  8. 26 Feb, 2019 2 commits
  9. 25 Feb, 2019 4 commits
  10. 24 Feb, 2019 4 commits
    • Philip Chimento's avatar
      CI: Always build HEAD in Flatpak manifest · 554e3d7a
      Philip Chimento authored
      The commit being built for CI will already be checked out, so if we
      specify HEAD in the manifest flatpak-builder will check out that commit
      The previous way of specifying this only worked for branches, not tags.
    • Philip Chimento's avatar
      boxed: Mark pointer as unowned when discarding boxed object · 7b330751
      Philip Chimento authored
      Sometimes, in the constructor of a boxed object, we will delegate to
      another constructor, such as GLib.Error() or GLib.Variant.new_internal()
      and replace the object under construction by the object returned by
      the delegated constructor.
      When that happens, the internal GBoxed pointer of the BoxedInstance
      remains null. That causes problems when eventually the discarded boxed
      object is garbage collected, and we try to free the pointer. In those
      cases, mark it as unowned so it is not freed.
    • Philip Chimento's avatar
      boxed: Clarify lifecycle log messages · cf71e3ba
      Philip Chimento authored
      To aid in debugging, we want to log a lifecycle message whenever the
      internal boxed pointer is created or allocated. This was only logged in
      some cases, and the message confusingly said "JSObject created".
    • Philip Chimento's avatar
      context: Trace atoms · 6d25dbeb
      Philip Chimento authored
      Apparently I forgot to trace these. It didn't affect most of the members
      of GjsAtoms because string atoms are pinned in memory anyway. But the
      symbol atoms only had their symbol descriptions pinned in memory, not
      the symbols themselves. So garbage collections would collect these
      This wasn't obvious because the symbols seem only to be used early in
      the GJS process anyway. It became apparent when running with JS_GC_ZEAL
      (extra garbage collection after every 100 allocations.)