Commits on Aug 28, 2009
  1. Set revisions for 1.020 release.

    committed Aug 27, 2009
Commits on Aug 27, 2009
  1. Fixes to POE::Filter::HTTPD:

        - RFC2616 says a body is implied if we have Content-Encoding or
          Content-Length;
        - C-E w/o C-L is an error for HTTP 1.1;
        - Added copious comments to justify the above;
        - Created ->_reset();
        - If possible, include the request in an error response object;
        - Set ->message on an error response object;
        - Updated documentation.
    Unit tests for above.
    Philip Gwyn committed Aug 27, 2009
Commits on Aug 26, 2009
  1. Comment on the likely cause of the deep recursion noted in rt.cpan.org

    ticket 45109.  Waiting for confirmation that Philip Gwyn's
    USE_SIGNAL_PIPE changes haven't fixed this particular problem.
    committed Aug 26, 2009
  2. Explain why wheels can't be altered outside their creator sessions.

    Potentially explains an oblique side effect reported by Sean Pieper in
    rt.cpan.org ticket 45408.
    committed Aug 26, 2009
Commits on Aug 25, 2009
  1. Create new project directories for POE::Loop::{Event,Tk,Gtk}. Moved

    the loops to their new project directories.  Configured the new
    projects for CPAN distribution.  Removed the event loops from POE's
    distro configuration.
    committed Aug 25, 2009
  2. Document POE::Wheel::Run coderef execution side effects. Disable

    the POE::Kernel->run() not-called warning if POE::Kernel->stop() is
    called.
    committed Aug 25, 2009
Commits on Aug 22, 2009
  1. Mainly rewrite POE::Filter::HTTPD's parser. May support get_one()

    now, so some tests also had to change.  I hope this will resolve
    rt.cpan.org ticket 48802, reported by a very patient zerhash.
    committed Aug 22, 2009
Commits on Aug 15, 2009
  1. Remove $Id$ and $Revision$ tags. No significant code changed despite

    most of the distribution being touched.
    committed Aug 15, 2009
Commits on Aug 13, 2009
  1. Added experimental code to allow postbacks and callbacks to be

    weakened, explicitly reducing their reference counts.  Untested.
    Resolves rt.cpan.org #48670, submitted by Adam Kennedy.
    committed Aug 13, 2009
Commits on Aug 10, 2009
  1. Hiding Event's exceptions is no longer necessary

    Philip Gwyn committed Aug 10, 2009
Commits on Aug 9, 2009
  1. Defer rethrowing exceptions until after run() cleans up. Previously

    exceptions would be thrown farther inside POE's dispatcher, preventing
    some important cleanup.  This only affects a few people who re-run()
    POE::Kernel after an exception or using $kernel->stop().
    committed Aug 9, 2009
Commits on Aug 8, 2009
  1. Quiet Event's exception handler.

    I'm not entirely sure this is a good idea, but it's worth a shot.
    committed Aug 8, 2009
  2. Increase filehandle trace verbosity. Add more diagnostic information

    to Event watchers, in case they fail.  Slightly optimize
    POE::Wheel::Run destruction.  There's no need to unselect file handles
    in modes that were never watched.
    committed Aug 8, 2009
  3. Move Philip Gwyn's signals test to POE::Test::Loops, so it can

    exercise everyone's event loops.
    committed Aug 8, 2009
Commits on Aug 7, 2009
  1. Added test to make sure we get the 3 "completion" events from

    POE::Wheel::Run.  Multiplied by the 4 conduit types.
    Philip Gwyn committed Aug 7, 2009
Commits on Aug 6, 2009
  1. Roll back previous patch. We need to call ->close_slave() on the _read

    handles.  Do that instead.
    Philip Gwyn committed Aug 6, 2009
  2. Fix case where POE::Wheel::Run CloseEvent doesn't fire on Linux.

    Thanks to Philip Gwyn's investigation and test cases, we found out
    that close() wasn't sufficient to close and destruct slave ptys, so
    the parent processes were keeping them open.  Replaced close() with
    $io_pty->close_slave() as needed.
    committed Aug 6, 2009
Commits on Aug 4, 2009
  1. Small comment tweak re: top/bottom handlers

    Philip Gwyn committed Aug 4, 2009
Commits on Jul 31, 2009
  1. @bingos

    Updated agaran-filter-httpd.t to bind server poco specifically to the…

    … loopback as per RT #48341
    bingos committed Jul 31, 2009
Commits on Jul 29, 2009
  1. !!! This change breaks backward compatibility on a relatively unused

    !!! feature.  You are affected if you use ARG0 or ARG1 in a
    !!! POE::Component::Server::TCP ClientConnected callback.
    
    ClientArgs promised more than it could deliver, and people finally
    noticed.  This change backs off supplying the socket in $_[ARG0], and
    it expands ClientArgs' arrayref into @_[ARG0..$#_].  Thanks to Michael
    Fowler for rt.cpan.org #47855 (which this resolves), and POE's mailing
    list for advice on which way this change should go.
    committed Jul 29, 2009
  2. Initialize the output masks for sigprocmask. Undefined output masks

    may trigger warnings in some versions of Perl.  At least, I hope
    http://www.nntp.perl.org/group/perl.cpan.testers/2009/07/msg4809122.html
    warnings will be resolved by this change.  I wasn't able to reproduce
    them myself.
    committed Jul 29, 2009
Commits on Jul 28, 2009
  1. Some people may be using POE without actually reading the docs.

    Furthermore, they may go on to rant about the side effects in obscure
    places.  Recommend reading the documentation in one of POE's most
    "infamous" warnings.
    committed Jul 28, 2009
Commits on Jul 27, 2009
  1. Rely on POE::Test::Loops 1.021, which should be everywhere by now.

    Update links in documentation.
    committed Jul 27, 2009
  2. @bingos
Commits on Jul 26, 2009
Commits on Jul 22, 2009
  1. Explain the likely cause and solution for the "Child process (pid)

    reaped" messages, and why it's important for programs to avoid leaking
    processes.
    committed Jul 22, 2009
  2. Add error number to signal pipe errors. Windows doesn't always have a…

    …n error string for some of winsock's wacky error numbers.
    committed Jul 22, 2009
  3. Wait for longer for all children to exit

    Philip Gwyn committed Jul 22, 2009
  4. Try to speed up the tests by triggering the children to exit upon

    demand rather than after a certain amount of time.  Doesn't seem
    significant on my G4 laptop, but it might be on faster machines.
    committed Jul 22, 2009