Permalink
Commits on Mar 26, 2010
  1. fix example

    benoitc committed Mar 26, 2010
  2. resync

    benoitc committed Mar 26, 2010
  3. s/Installing/Installation

    benoitc committed Mar 26, 2010
  4. typo

    benoitc committed Mar 26, 2010
  5. sync

    benoitc committed Mar 26, 2010
  6. update doc

    benoitc committed Mar 26, 2010
  7. Ignore SIGWINCH in worker. Fix crash on OpenBSD

    benoitc
    benoitc committed Mar 26, 2010
  8. Fix example config in docs.

    davisp committed Mar 26, 2010
Commits on Mar 24, 2010
Commits on Mar 23, 2010
  1. fix response

    benoitc committed Mar 23, 2010
  2. Fixed async responses.

    davisp committed Mar 23, 2010
  3. Fix async req/resp handlers.

    davisp committed Mar 23, 2010
Commits on Mar 22, 2010
  1. fix requirement

    benoitc committed Mar 22, 2010
Commits on Mar 21, 2010
  1. add headers_sent guard.

    benoitc committed Mar 21, 2010
Commits on Mar 20, 2010
  1. next version will be 0.7.0

    benoitc
    benoitc committed Mar 20, 2010
Commits on Mar 19, 2010
  1. change the way we handle last chunk, on't break but instead test if last

    benoitc
    benoitc committed Mar 19, 2010
    was an empty string. So we allows idiot content to send empty string at
    first.
Commits on Mar 18, 2010
Commits on Mar 17, 2010
  1. fix worker. status is a response thing. instead test if we got a status

    benoitc
    benoitc committed Mar 17, 2010
    line or if environ isn't empty.
  2. remove useless examples

    benoitc committed Mar 17, 2010
Commits on Mar 16, 2010
  1. merge grainbows & gunicorn

    benoitc
    benoitc committed Mar 16, 2010
  2. oups revert this change in util

    benoitc
    benoitc committed Mar 16, 2010
  3. allows custom arbiter

    benoitc
    benoitc committed Mar 16, 2010
  4. Be explicit about blocking status.

    davisp committed Mar 16, 2010
    Apparently the assumption that sockets don't inherit their blocking
    status from the server socket is invalid. Or python sets the flags
    to match. Either way, we need clients to use blocking sockets so
    we're now explicitly setting the blocking status.
  5. Be more considerate to future me.

    davisp committed Mar 16, 2010
    Sergey points out that its more proper to |= with the current value of
    the fd flags. I concur.
  6. Fix close_on_exec bug.

    davisp committed Mar 16, 2010
    Reported by Sergey Shepelev. I changed his patch slightly because of the
    man page language. As I read things, the close on exec flag is separate
    from the F_SETFL flags.
Commits on Mar 13, 2010
  1. revert previous change

    benoitc committed Mar 13, 2010