Permalink
Switch branches/tags
request_handler_with_pipes release-5.1.11 release-5.1.10 release-5.1.9 release-5.1.8 release-5.1.7 release-5.1.6 release-5.1.5 release-5.1.4 release-5.1.3 release-5.1.2 release-5.1.1 release-5.1.0 release-5.0.30 release-5.0.29 release-5.0.28 release-5.0.27 release-5.0.26 release-5.0.25 release-5.0.24 release-5.0.23 release-5.0.22 release-5.0.21 release-5.0.20 release-5.0.19 release-5.0.18 release-5.0.17 release-5.0.16 release-5.0.15 release-5.0.14 release-5.0.13 release-5.0.11 release-5.0.10 release-5.0.9 release-5.0.8 release-5.0.7 release-5.0.6 release-5.0.5 release-5.0.4 release-5.0.3 release-5.0.2 release-5.0.1 release-5.0.0.rc2 release-5.0.0.rc1 release-5.0.0.beta3 release-5.0.0.beta2 release-5.0.0.beta1 release-4.0.60 release-4.0.59 release-4.0.58 release-4.0.57 release-4.0.56 release-4.0.55 release-4.0.53 release-4.0.52 release-4.0.51 release-4.0.50 release-4.0.49 release-4.0.48 release-4.0.46 release-4.0.45 release-4.0.44 release-4.0.43 release-4.0.42 release-4.0.41 release-4.0.40 release-4.0.39 release-4.0.38 release-4.0.37 release-4.0.36 release-4.0.35 release-4.0.34 release-4.0.33 release-4.0.32 release-4.0.31 release-4.0.30 release-4.0.29 release-4.0.28 release-4.0.27 release-4.0.26 release-4.0.25 release-4.0.24 release-4.0.23 release-4.0.21 release-4.0.20 release-4.0.19 release-4.0.18 release-4.0.17 release-4.0.16 release-4.0.14 release-4.0.13 release-4.0.10 release-4.0.8 release-4.0.7 release-4.0.6 release-4.0.5 release-4.0.4 release-4.0.3 release-4.0.2 release-4.0.1
Nothing to show
Commits on Dec 5, 2008
  1. Bump version to 2.0.5.

    FooBarWidget committed Dec 5, 2008
  2. setupSyscallInterruptionSupport: use sigaction() instead of signal() …

    FooBarWidget committed Dec 5, 2008
    …for better portability.
  3. Backport an ApplicationPool race condition and invariant violation fix.

    FooBarWidget committed Dec 5, 2008
    Some people reported that 'count' can go over 'max'. This is the cause,
    I suspect.
Commits on Dec 4, 2008
  1. Detect the Rake installation that belongs to the Ruby interpreter tha…

    FooBarWidget committed Dec 4, 2008
    …t was used to run the installer, instead of the Rake that's in PATH. Update the gem installation instructions so that the RubyGems command that belongs to the Ruby interpreter that was used to run the installer is preferred.
  2. Make sure that the installer runs Rake with the same Ruby interpreter…

    FooBarWidget committed Dec 4, 2008
    … that was used to run the installer.
    
    It makes installation on Ruby Enterprise Edition less error-prone.
Commits on Dec 3, 2008
Commits on Dec 1, 2008
Commits on Nov 24, 2008
Commits on Nov 23, 2008
Commits on Nov 22, 2008
  1. Fix compatibility with Rails edge, in which application.rb was rename…

    FooBarWidget committed Nov 22, 2008
    …d to application_controller.rb.
Commits on Nov 20, 2008
  1. Fix indentation.

    FooBarWidget committed Nov 20, 2008
  2. When compiles fail, the final output is from rake

    pgollucci committed with FooBarWidget Nov 17, 2008
    (See full trace by running task with --trace)
    
    We could do this by inspecting ARGV, or the ENV.  I chose
    the latter.  Setting ENV['TRACE'] will now pass --trace to rake.
    
    Should probably document that somewhere.
  3. Both apr and apr-util are required to build passenger.

    pgollucci committed with FooBarWidget Nov 17, 2008
    The are not neccessarily installed in the same --prefix
    (see ./configure)
    (note, the apu-1-config does not accept the --cppflags option
    
    With my pgollucci@apache.org HTTPD PMC hat on:
    
    Technicallly, this is still wrong, this should be using apxs alone,
    but that would have required me to refactor this and I did not have
    time at the moment.   For an example of the correct way to do this,
    see the apreq (libapreq2) httpd module (c-binding portion).
    [apreq-dev@httpd.apache.org]
    
    Note, Eventually, httpd will have an httpd-config similiar to
    gnome-config, neon-config, serf-config, etc.... at which point
    it should switch to that.
Commits on Nov 19, 2008
Commits on Nov 16, 2008
  1. Set the FD_CLOEXEC flag on various IO objects that we create, so that…

    FooBarWidget committed Nov 15, 2008
    … fork()+exec() or a system() inside the application properly closes the connection to the web server.
  2. Fix the stub application that doesn't specify a Rails version.

    FooBarWidget committed Nov 15, 2008
    cache_template_loading is removed from Rails 2.2.
Commits on Nov 15, 2008
  1. Fix Utils#close_all_io_objects_for_fds and its calls.

    FooBarWidget committed Nov 15, 2008
    The old implementation appears to be broken. It also traverses the entire
    object space and calls #is_a? on each object, causing Rails deprecation
    warnings to be printed if said object is an ActiveSupport::Deprecation.
  2. Fix a crash on MacOS X.

    FooBarWidget committed Nov 15, 2008
Commits on Nov 14, 2008
  1. Set Rake verbosity to true so that compilation commands are still sho…

    FooBarWidget committed Nov 14, 2008
    …wn on newer Rake versions.
Commits on Oct 28, 2008
  1. Implement global queueing for ApplicationPool, to fight long-running …

    FooBarWidget committed Oct 27, 2008
    …requests. Feature sponsored by 37signals.
  2. The non-blocking I/O stuff in Bucket causes a lot of read problems, s…

    FooBarWidget committed Sep 24, 2008
    …o disable non-blocking I/O completely.
Commits on Sep 24, 2008
  1. Properly treat EAGAIN errors when reading data from the backend appli…

    FooBarWidget committed Sep 24, 2008
    …cation. This fixes some mysterious freeze bugs.