[Rt-commit] rt branch, 4.0/oracle-testing, created. rt-4.0.7-59-g959226e

Thomas Sibley trs at bestpractical.com
Tue Sep 4 14:16:23 EDT 2012


The branch, 4.0/oracle-testing has been created
        at  959226ef2864c38bf7bb25e9b88a5fefed43a050 (commit)

- Log -----------------------------------------------------------------
commit c28b477a0084209ae35590858130b8bb67ae6e30
Author: Thomas Sibley <trs at bestpractical.com>
Date:   Fri Aug 31 17:10:50 2012 -0700

    Remove a stray signal handler used for debugging

diff --git a/t/api/rights_show_ticket.t b/t/api/rights_show_ticket.t
index 62f62c4..b438fde 100644
--- a/t/api/rights_show_ticket.t
+++ b/t/api/rights_show_ticket.t
@@ -208,7 +208,6 @@ diag "Testing with UseSQLForACLChecks => $option";
 }
 
 sub have_no_rights {
-    $SIG{'INT'} = $SIG{'TERM'} = sub { print STDERR Carp::longmess('boo'); exit 1 };
     local $Test::Builder::Level = $Test::Builder::Level + 1;
     foreach my $u ( @_ ) {
         foreach my $q (

commit 959226ef2864c38bf7bb25e9b88a5fefed43a050
Author: Thomas Sibley <trs at bestpractical.com>
Date:   Fri Aug 31 17:11:13 2012 -0700

    Send all test servers TERM instead of TERM to apache and INT to plack
    
    Plack handles TERM just fine.  When testing under Oracle, SIGINT is
    trapped and ignored by the Oracle client libraries, leading to a
    deadlock for any test files which specify actual_server => 1 (or set a
    plack web handler explicitly).  SIGTERM is respected, however, and
    avoids the deadlock.

diff --git a/lib/RT/Test.pm b/lib/RT/Test.pm
index 3e7c910..68a3415 100644
--- a/lib/RT/Test.pm
+++ b/lib/RT/Test.pm
@@ -1443,9 +1443,7 @@ sub stop_server {
     my $in_end = shift;
     return unless @SERVERS;
 
-    my $sig = 'TERM';
-    $sig = 'INT' if $ENV{'RT_TEST_WEB_HANDLER'} eq "plack";
-    kill $sig, @SERVERS;
+    kill 'TERM', @SERVERS;
     foreach my $pid (@SERVERS) {
         if ($ENV{RT_TEST_WEB_HANDLER} =~ /^apache/) {
             sleep 1 while kill 0, $pid;

-----------------------------------------------------------------------


More information about the Rt-commit mailing list