Permalink
Commits on Aug 26, 2016
Commits on Aug 25, 2016
  1. Merge pull request #84 from leonid-shevtsov/fix-jruby

    JRuby support
    committed on GitHub Aug 25, 2016
  2. replace fork with Process.spawn

    committed Aug 25, 2016
  3. remove mis-required mkmf

    committed Aug 25, 2016
Commits on Aug 24, 2016
  1. Merge pull request #83 from leonid-shevtsov/smart-destroy-at-exit-def…

    …ault
    
    Smart destroy at exit default (re-submitted without conflicts)
    committed on GitHub Aug 24, 2016
  2. Merge branch 'master' into smart-destroy-at-exit-default

    committed Aug 24, 2016
  3. Fix stupid typo

    committed Aug 24, 2016
  4. Merge pull request #82 from leonid-shevtsov/ffmeg-devices-options

    add draw_mouse support (re-submitted with resolved conflicts)
    committed on GitHub Aug 24, 2016
  5. be more explicit about ruby versions, add 2.1.10 to build matrix

    committed Aug 24, 2016
  6. refactor video recorder command line construction

    committed Aug 24, 2016
  7. add Ruby 2.3 to Travis build matrix and fix rspec command

    committed Aug 24, 2016
  8. Cache bundler on Travis

    committed Aug 24, 2016
Commits on Mar 17, 2016
  1. Release version 2.2.3

    committed Mar 17, 2016
  2. clean up gemspec file

    committed Mar 17, 2016
  3. Merge pull request #76 from NfNitLoop/fix-launch-race

    Fix race condition when starting Xvfb
    committed Mar 17, 2016
Commits on Feb 8, 2016
Commits on Dec 23, 2015
  1. Fixes for test failures.

     * Revert to the previous behavior/implementation of xvfb_running?
     * Documented the behavior of xvfb_running?
     * rename ensure_xvfb_is_running to ensure_xvfb_launched.
       (Distinguishes it from xvfb_running?)
     * ensure_xvfb_launched makes sure the process *we* launched won the race
       to the lock file.
    NfNitLoop committed Dec 23, 2015
  2. Fix race condition when starting new Xvfb processes.

    Previously, this could happen:
    
     * Two processes, A and B, try to use Headless.new().
     * Process A "wins" and gains the lock on /tmp/.X-#{display}-lock
     * Process B loses. Xvfb errors and exits immediately.
     * ensure_xvfb_is_running calls xvfb_running?
     * xvfb_running? finds a display lock file, and thinks it's ours. It
       erroneously reports that our process is running.
    
    This change saves the pid from Process.spawn() and verifies that *that*
    process is still running, *and* we have the lock on our display number.
    NfNitLoop committed Dec 23, 2015
  3. Don't close the stderr pipe while we're reading from it.

    Use 'ensure' to ensure it's closed before we leave launch_xvfb.
    NfNitLoop committed Dec 23, 2015
Commits on Sep 21, 2015
  1. Merge pull request #70 from reiz/patch-1

    Add license info to the gemspec.
    committed Sep 21, 2015
  2. Add license info to the gemspec.

    reiz committed Sep 21, 2015
Commits on Aug 10, 2015
  1. more sensible change. :devices is an option passed to :video hash if …

    …needed. fixed spec & lib. Now plays to ruby 1.9 as well
    atzorvas committed Aug 10, 2015
  2. fix video devices method. fix ffmpeg string so it doesnt add an extra…

    … space after provider, renamed test case
    atzorvas committed Aug 10, 2015