Permalink
Commits on Jan 25, 2011
  1. fixed a small typo

    slact committed Jan 25, 2011
Commits on Nov 30, 2010
  1. added a new option push_channel_timeout which controls how long a cha…

    …nnel should be kept after accepting the last subscriber in the channel with no queued messages and no active subscriber.
    
    this option should be used along with push_subscriber_timeout, and push_channel_timeout should be greater than push_subscriber_timeout
    
    push_channel_timeout should appear along with push_publisher (or parent context); if push_authorized_channels_only is off, push_channel_timeout should also be set in push_subscriber context (or parent context)
    liucougar committed Nov 30, 2010
Commits on Oct 25, 2010
  1. ngx_http_push_broadcast_status_locked would never return NGX_OK, so c…

    …hecking the return value against NGX_OK is meaningless, instead always return NGX_OK in case of CONCURRENT_LASTIN
    liucougar committed Oct 25, 2010
  2. merged @yanjian timeout cleanup refactor

    set discard_body=0 in timeout handler to prevent nginx worker process segfault (at least with nginx 0.8.6x)
    liucougar committed Oct 25, 2010
Commits on Oct 23, 2010
Commits on May 31, 2010
  1. a dummy commit to ammend the last erroneous commit msg. amends: A cor…

    …rectly working example test.py. Last commit was a bit stupid one, in excitement, never checked that it was not working right. :) Thanks to Leo, this version has the correct headers too.
    jaivikram committed with slact Feb 13, 2010
  2. A correctly working example test.py. Last commit was a stupid one, in…

    … excitement, never checked that its not working right. :) Thanks to leo for pointing out the correct headers to use.
    jaivikram committed with slact Feb 13, 2010
  3. added test.py

    jaivikram committed with slact Jan 7, 2010
  4. Merge erikh push_ignore_queue_on_no_cache.

    I don't buy the cors_allow feature, because arbitrary headers can already be
    set up through the nginx config file.
    jedisct1 committed with slact May 21, 2010
  5. Handle OPTIONS differently. Only the headers are really needed.

    There's no point in returning whatever lies in the queue.
    jedisct1 committed with slact May 20, 2010
  6. Typo :)

    jedisct1 committed with slact May 20, 2010
  7. Refuse to push empty messages.

    jedisct1 committed with slact May 18, 2010
Commits on Feb 4, 2010
  1. fix: error log reported failed close() for some publisher requests w…

    …ith large messages,
    
     fix: occasional memory leak during message deletion, 
     fix: worker messages intended for dead worker processes were not deleted
    slact committed Feb 4, 2010
Commits on Feb 2, 2010
  1. teststuff

    slact committed Feb 2, 2010
  2. 0.691 -- tiny bugfix release

    slact committed Feb 2, 2010
Commits on Jan 28, 2010
  1. too many files!!!

    slact committed Jan 28, 2010
  2. some more explorative SIGHUP work

    slact committed Jan 28, 2010
Commits on Jan 27, 2010
Commits on Jan 26, 2010