1. 24 Oct, 2017 2 commits
  2. 23 Oct, 2017 1 commit
  3. 22 Oct, 2017 1 commit
    • Friedrich W. H. Kossebau's avatar
      Restore NavigationToolTips for file items in projects manager · 07bcde0d
      Friedrich W. H. Kossebau authored
      Summary:
      With the introduction of ProjectModelItemDelegate the tooltip events for
      items have been handled by its inherited QItemDelegate::helpEvent(), thus
      somehow no longer bubbling to ProjectTreeView::event().
      As a result no longer the NavigationToolTips were shown for supported file
      items, only the full path as delivered by ProjectModel.
      
      This patch restores that behaviour, by moving the respective code to the
      helpEvent handler of ProjectModelItemDelegate.
      
      Reviewers: #kdevelop, mwolff
      
      Reviewed By: #kdevelop, mwolff
      
      Subscribers: mwolff, kdevelop-devel
      
      Differential Revision: https://phabricator.kde.org/D8410
      07bcde0d
  4. 20 Oct, 2017 3 commits
  5. 19 Oct, 2017 1 commit
  6. 18 Oct, 2017 2 commits
    • Christoph Roick's avatar
      Allow abortion of starting another NativeAppJob · d896e748
      Christoph Roick authored
      Summary:
      - offers to cancel an accidental startup of an already running job
      - kill all running instances instead of one by one
      - fix: do not ask again if "No" was already selected
      
      Test Plan:
      - start second instance of an already running job
      - select
          * Kill All Instances (No) (kills all running instances)
          * Start Another (Yes) (just starts a new instance)
          * Cancel (closes the dialog and drops the new job)
      
      Reviewers: #kdevelop, mwolff
      
      Reviewed By: #kdevelop, mwolff
      
      Subscribers: kfunk, mwolff, kdevelop-devel
      
      Differential Revision: https://phabricator.kde.org/D8199
      d896e748
    • Aaron Puchert's avatar
      Don't try to guess the language from given arguments · 9ec89cdc
      Aaron Puchert authored
      Summary:
      Before this change, the compiler provider looked up the arguments for a
      given language, and passed those to the compiler, where the language was
      guessed from the arguments.
      
      Instead we now pass the language together with the arguments into the
      function, and no more guessing is needed. Besides being cleaner, it may
      not always be possible to guess the language from the given arguments
      anyway.
      
      Reviewers: #kdevelop, mwolff
      
      Reviewed By: #kdevelop, mwolff
      
      Subscribers: mwolff, apol, kdevelop-devel
      
      Differential Revision: https://phabricator.kde.org/D8317
      9ec89cdc
  7. 17 Oct, 2017 3 commits
    • Friedrich W. H. Kossebau's avatar
      Fix registration of the enabled state of plugins · 35f3f710
      Friedrich W. H. Kossebau authored
      Summary:
      The old code was slightly broken:
      At start-up plugin controller updated the config storage only for the
      enabled state of the global-category plugins (and indirectly their
      dependency plugins). Project-category plugins would only get an update
      indirectly once loadProjectPlugins() was run, other plugins only once
      the were requested by their interface.
      Thus in a new session with no project yet loaded the config storage would
      not yet cover the enabled state of the project-category plugins and those
      other.
      
      PluginPreferences code queried the initial enabled state of each plugin
      from the plugin controller, which then has extra code in the isEnabled()
      method to deal with that. Though that information is dropped and replaced
      with the one fetched from the config storage, multiple times even:
      * the plugins are added to the selector using the flag
        KPluginSelector::ReadConfigFile, which results in the data manually set
        to each KPluginInfo instance with kpi.setPluginEnabled(...) being
        overwritten with the config storage data
      * selector->load() was called after plugins were added, which results
        in the same update from the config storage data
      * the PluginPreferences instance like each KDevelop::ConfigPage gets an
        initial call of the reset() method on setup, which again calls
        selector->load()
      Which results in the plugin selector showing wrong enabled state at least
      for new sessions, where no project has been loaded yet and not all plugins
      at least been tried to load once.
      While the first two times of data drop could be fixed, the third would be
      more complicted to do.
      
      So instead of keeping the current on-the-fly enabled state calculation for
      non-global-category plugins, on start-up the config storage is updated for
      any known plugin (and then consecutively updated if failing to load a
      plugin when needed). This gives the plugin selector and any other code
      consistent data and simplifies things.
      
      Next to this, this patch also restores ShellExtension::defaultPlugins()
      power to control which plugins should be loaded by default. Thus any
      tests which (try to) make use of the AutoTestShell::init plugin list
      feature can (and have to) now explicitly define all userselectable plugins
      which should be used (thus also using the correct plugin id :) ).
      
      Test Plan:
      Create a new session, open directly the plugin selection settings and see
      that all build and projectmanager plugins but also the QML plugin are shown
      as enabled, as they should be.
      
      Reviewers: #kdevelop, mwolff
      
      Reviewed By: #kdevelop, mwolff
      
      Subscribers: mwolff, kdevelop-devel
      
      Differential Revision: https://phabricator.kde.org/D8156
      35f3f710
    • Friedrich W. H. Kossebau's avatar
    • Script Kiddy's avatar
      SVN_SILENT made messages (.desktop file) - always resolve ours · 5aca467f
      Script Kiddy authored
      In case of conflict in i18n, keep the version of the branch "ours"
      To resolve a particular conflict, "git checkout --ours path/to/file.desktop"
      5aca467f
  8. 16 Oct, 2017 2 commits
  9. 15 Oct, 2017 3 commits
  10. 13 Oct, 2017 3 commits
  11. 12 Oct, 2017 1 commit
    • René J.V. Bertin's avatar
      Don't set up DocumentationViews when shutting down · 4b4806d0
      René J.V. Bertin authored
      In rare situations a DocumentationView can be created when shutting
      down, which can lead to race conditions and crashing. Prevent this by
      returning early from the class ctor when shutdown is in progress.
      
      In addition, prevent errors from QWidget::setTabOrder() by checking its
      input arguments.
      
      BUG: 385640
      4b4806d0
  12. 11 Oct, 2017 1 commit
  13. 10 Oct, 2017 1 commit
  14. 09 Oct, 2017 4 commits
  15. 08 Oct, 2017 6 commits
    • Kevin Funk's avatar
      Another compile fix for Qt 5.5 · fd60b4b6
      Kevin Funk authored
      Problem:
        kdevplatform/shell/sourceformattercontroller.cpp:602:13: error:
      invalid use of incomplete type ‘class QAbstractButton’
      fd60b4b6
    • Kevin Funk's avatar
      Merge remote-tracking branch 'origin/5.2' · c4a10dbb
      Kevin Funk authored
      c4a10dbb
    • Kevin Funk's avatar
      Fix Qt 5.5 compatibility -- no constFirst/Last methods. · 8da8d1d2
      Kevin Funk authored
      Summary:
      QList and QVector both gained these in Qt 5.6, but we claim to support 5.5 (because it's found in Ubuntu 16.04 LTS).
      
      None of these are in performance-relevant code, so I don't believe it's worthwhile adding ifdefs; just revert this patch whenever 5.5 support is dropped.
      
      Reported by 'blaze' on IRC.
      
      Reviewers: #kdevelop, mwolff
      
      Subscribers: blaze, kfunk, mwolff, kossebau, kdevelop-devel
      
      Differential Revision: https://phabricator.kde.org/D8202
      8da8d1d2
    • Guillaume Racicot's avatar
      Filter out warnings as error from parser's compile flags · 0a29544f
      Guillaume Racicot authored
      Summary:
      The highlighting in my project was all screwed up like in the kdevelop codebase last week.
      
      Now kdevelop seems fine parsing it's own code, but my codebase highlighting was still screwed up. Removing warnings as error in the parser fixed the problem.
      
      Since  warnings as error can be enabled with `-Werror` for all of them, and individually with `-Werror=`, I thought disabling both. So a list of argument like that:
      ```
       $ clang++ -Wall -Werror -Werror=unreachable-code
      ```
      Becomes:
      ```
       $ clang++ -Wall -Wunreachable-code
      ```
      Now the parser is no longer reporting warnings as error, but still reporting unreachable code.
      
      Reviewers: #kdevelop, kfunk, mwolff, brauch
      
      Reviewed By: brauch
      
      Subscribers: mwolff, brauch, kdevelop-devel
      
      Tags: #kdevelop
      
      Differential Revision: https://phabricator.kde.org/D7979
      0a29544f
    • Kevin Funk's avatar
      Merge remote-tracking branch 'origin/5.2' · 9b226d07
      Kevin Funk authored
      9b226d07
    • Script Kiddy's avatar
      SVN_SILENT made messages (.desktop file) - always resolve ours · 277c7f62
      Script Kiddy authored
      In case of conflict in i18n, keep the version of the branch "ours"
      To resolve a particular conflict, "git checkout --ours path/to/file.desktop"
      277c7f62
  16. 07 Oct, 2017 2 commits
  17. 06 Oct, 2017 2 commits
  18. 05 Oct, 2017 2 commits