http://www.perlmonks.org?node_id=930622


in reply to Why does Test-WWW-Mechanize-Catalyst-0.55\t\multi_content_type.t hang on win32?

Test-WWW-Mechanize-Catalyst-0.55\t\multi_content_type.t includes

use_ok 'ExternalCatty';
my $pid = ExternalCatty->background($PORT);

[download]
ExternalCatty does:
sub background {
    my $self  = shift;
    my $port  = shift;
    my $child = fork;
    die "Can't fork Cat HTTP server: $!" unless defined $child;
    return $child if $child;

    if ( $^O !~ /MSWin32/ ) {
        require POSIX;
        POSIX::setsid() or die "Can't start a new session: $!";
    }

    $self->run($port);
}

[download]
Maybe the fork is failing?

Replies are listed 'Best First'.
Re^2: Why does Test-WWW-Mechanize-Catalyst-0.55\t\multi_content_type.t hang on win32?
by Anonymous Monk on Oct 10, 2011 at 13:16 UTC

    Um, no, its hanging, if fork failed, program would end 

    $ perl -Mblib t/multi_content_type.t
1..8
#
# ###################################################################
# Starting an external Catalyst HTTP server on port 7357
# To change the port, please set the TWMC_TEST_PORT env variable.
# (The server will be automatically shut-down right after the tests).
# ###################################################################
ok 1 - use ExternalCatty;

    [download]
[reply]
[d/l]

      Without looking further into this - my experience with fork() and sockets on Windows has led me to always spawn external processes (that is, HTTP servers) for module self-testing instead. Launching an external process via pipe-open works for HTTP::Server::Simple, WWW::Mechanize, WWW::Mechanize::Firefox and WWW::Mechanize::Shell.

      I would not bother with investigating why fork and sockets don't interact like on Unix, because the fork-emulation never handled that well anyway.

[reply]
[d/l]

        Thanks

[reply]

In Section Seekers of Perl Wisdom