[PATCH: 5.005_61] an EBCDIC friendly cgi-html.t test

This is a build failure report (with a patch) for os/390 R 2.6.


FWIW I obtained a warning on util.c that mentioned:

`sh  cflags libperl.a util.o`  util.c
          CCCMD =  c89 -DPERL_CORE -c -DMAXSIG=38 -DOEMVS -D_OE_SOCKETS -D_XOPE
WARNING CBC3193 ./util.c:3502  The size of this type is zero.
FSUM3065 The COMPILE step ended with return code 4.

but the build proceeded.  

I cannot build SDBM_File since the compiler balks at ext/SDBM_File/sdbm.c.
Until I get some tuits I'll leave that one be but I do note that on another 
list someone complained of trouble with `make test` and SDBM_File on 
Tru 64 Unix 4.0F recently.

Here is a summary of the failures that I saw with a modified cgi-html.t
recently:


op/regexp...........CEE5213S The signal SIGPIPE was received.
FAILED at test 483
op/regexp_noamp.....CEE5213S The signal SIGPIPE was received.
FAILED at test 483

pragma/locale.......CEE3703I In HPCB Control Block, the Eye Catcher is damaged.
CEE3704I Expected data at 00000001 : HPCB.
CEE0802C Heap storage control information was damaged.
         From entry point XS_POSIX_setlocale at compile unit offset +000002B6 
a.
FAILED at test 99
pragma/overload.....CEE5213S The signal SIGPIPE was received.
FAILED at test 178

lib/bigfloat........CEE5213S The signal SIGPIPE was received.
FAILED at test 38
lib/bigfloatpm......CEE5213S The signal SIGPIPE was received.
FAILED at test 354

lib/charnames.......CEE5213S The signal SIGPIPE was received.
FAILED at test 1

lib/dumper..........CEE5213S The signal SIGPIPE was received.
FAILED at test 43

lib/io_unix.........Can't call method "getline" on an undefined value at 
lib/io.
FAILED at test 3

I note that nowadays:

Failed 11 test scripts out of 206, 94.66% okay.
[snip]
u=8.03  s=2.67  cu=145.72  cs=48.57  scripts=206  tests=10184
                                                        ^^^^^
the test count at 10K is in excess of the java portability test suite
that I've only read about - pretty cool.

OK here is the patch to get cgi-html.t running on this platform (as well as 
vmesa and posix-bc I hope).  For some alignment reason I omitted the 
indent-by-4 in the if else blocks but I'd be perfectly happy to redo this 
with indents if that would be preferred.

--- t/lib/cgi-html.t.orig	Fri Aug 27 16:17:34 1999
+++ t/lib/cgi-html.t	Fri Aug 27 17:23:05 1999
@@ -17,6 +17,8 @@
 
 ######################### End of black magic.
 
+my $Is_EBCDIC = $Config{'ebcdic'} eq 'define';
+
 # util
 sub test {
     local($^W) = 0;
@@ -37,10 +39,17 @@
     local($") = '-'; 
     test(8,h1('fred','agnes','maura') eq '<H1>fred-agnes-maura</H1>',"open/close tag \$\" interpolation");
 }
+if (!$Is_EBCDIC) {
 test(9,header() eq "Content-Type: text/html\015\012\015\012","header()");
 test(10,header(-type=>'image/gif') eq "Content-Type: image/gif\015\012\015\012","header()");
 test(11,header(-type=>'image/gif',-status=>'500 Sucks') eq "Status: 500 Sucks\015\012Content-Type: image/gif\015\012\015\012","header()");
 test(12,header(-nph=>1) eq "HTTP/1.0 200 OK\015\012Content-Type: text/html\015\012\015\012","header()");
+} else {
+test(9,header() eq "Content-Type: text/html\r\n\r\n","header()");
+test(10,header(-type=>'image/gif') eq "Content-Type: image/gif\r\n\r\n","header()");
+test(11,header(-type=>'image/gif',-status=>'500 Sucks') eq "Status: 500 Sucks\r\nContent-Type: image/gif\r\n\r\n","header()");
+test(12,header(-nph=>1) eq "HTTP/1.0 200 OK\r\nContent-Type: text/html\r\n\r\n","header()");
+}
 test(13,start_html() ."\n" eq <<END,"start_html()");
 <!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML//EN">
 <HTML><HEAD><TITLE>Untitled Document</TITLE>
@@ -61,8 +70,13 @@
     ;
 test(16,($cookie=cookie(-name=>'fred',-value=>['chocolate','chip'],-path=>'/')) eq 
      'fred=chocolate&chip; domain=localhost; path=/',"cookie()");
+if (!$Is_EBCDIC) {
 test(17,header(-Cookie=>$cookie) =~ m!^Set-Cookie: fred=chocolate&chip\; domain=localhost; path=/\015\012Date:.*\015\012Content-Type: text/html\015\012\015\012!s,
      "header(-cookie)");
+} else {
+test(17,header(-Cookie=>$cookie) =~ m!^Set-Cookie: fred=chocolate&chip\; domain=localhost; path=/\r\nDate:.*\r\nContent-Type: text/html\r\n\r\n!s,
+     "header(-cookie)");
+}
 test(18,start_h3 eq '<H3>');
 test(19,end_h3 eq '</H3>');
 test(20,start_table({-border=>undef}) eq '<TABLE BORDER>');
End of Patch.

Peter Prymmer

0
pvhp
8/28/1999 1:51:04 AM
perl.perl5.porters 48227 articles. 1 followers. Follow

4 Replies
4792 Views

Similar Articles

[PageSpeed] 4

Peter Prymmer writes:
> Failed 11 test scripts out of 206, 94.66% okay.
> [snip]
> u=8.03  s=2.67  cu=145.72  cs=48.57  scripts=206  tests=10184
>                                                         ^^^^^
> the test count at 10K is in excess of the java portability test suite
> that I've only read about - pretty cool.

This count is fake.  The real number of subtests is significantly
more: to not skew the counting in the favor on massive table-driven
tests some *groups* of subtests are counted as one subtest.  Say, 2400
or somesuch REx tests are counted in pairs (giving the count of 1200),
similarly, 20000 or somesuch number-convertion tests are counted as
1440.

So the real count is closer to 32000.

Ilya
0
ilya
8/28/1999 5:18:19 AM
Ilya Zakharevich wrote:

> Peter Prymmer writes:
> > Failed 11 test scripts out of 206, 94.66% okay.
> > [snip]
> > u=8.03  s=2.67  cu=145.72  cs=48.57  scripts=206  tests=10184
> >                                                         ^^^^^
> > the test count at 10K is in excess of the java portability test suite
> > that I've only read about - pretty cool.
> 
> This count is fake.  The real number of subtests is significantly
> more: to not skew the counting in the favor on massive table-driven
> tests some *groups* of subtests are counted as one subtest.  Say, 2400
> or somesuch REx tests are counted in pairs (giving the count of 1200),
> similarly, 20000 or somesuch number-convertion tests are counted as
> 1440.
> 
> So the real count is closer to 32000.

Indeed.  Having failed only 11 out of more than even 10k tests and
the harness has the nerve to report a 95% failure rate!  But 11/32k
is even worse reporting - good result though :-)

Peter Prymmer

0
pvhp
8/28/1999 5:56:26 AM
Peter Prymmer wrote:

> Indeed.  Having failed only 11 out of more than even 10k tests and
> the harness has the nerve to report a 95% failure rate!  But 11/32k

Oops - it reported a 95% success rate.  The reporting of `make test` 
is not as bad as my own.

Peter Prymmer

0
pvhp
8/28/1999 6:02:52 AM
Peter Prymmer writes:
 > 
 > Peter Prymmer wrote:
 > 
 > > Indeed.  Having failed only 11 out of more than even 10k tests and
 > > the harness has the nerve to report a 95% failure rate!  But 11/32k
 > 
 > Oops - it reported a 95% success rate.  The reporting of `make test` 
 > is not as bad as my own.

Another complicating factor when counting the number of tests is
that the locale.t subtests 99..115 are run for each locale found
in your system.

-- 
$jhi++; # http://www.iki.fi/jhi/
        # There is this special biologist word we use for 'stable'.
        # It is 'dead'. -- Jack Cohen
0
jhi
8/30/1999 8:50:52 PM
Reply:

Similar Artilces:

[ID 20000825.012] [PATCH@6822] t/lib/cgi-html.t produces ugly cruft during 'make test'
This is a bug report for perl from spider@leggy.zk3.dec.com, generated with the help of perlbug 1.32 running under perl v5.7.0. ----------------------------------------------------------------- [Please enter your report here] 'make test' spits this out: lib/cgi-html........."my" variable $CRLF masks earlier declaration in same scope at lib/cgi-html.t line 48. ok It seems to be cause a patch was applied twice. It's fixable, of course, by deleting the second copy: --- t/lib/cgi-html.t.DIST Fri Aug 25 10:02:00 2000 +++ t/lib/cgi-html.t Fri Aug 25 16:43:...

[PATCH lib/CGI/t/carp.t] Tests for CGI::Carp
Andrew Wilson strikes again! --- /dev/null Wed Sep 5 19:15:49 2001 +++ lib/CGI/t/carp.t Wed Sep 5 19:15:49 2001 @@ -0,0 +1,263 @@ +# -*- Mode: cperl; coding: utf-8; cperl-indent-level: 2 -*- +#!/usr/local/bin/perl -w + +use strict; +use lib qw(t/lib); +use Test::More tests => 42; +use IO::Handle; + +BEGIN { use_ok('CGI::Carp') }; + +#----------------------------------------------------------------------------- +# Test id +#----------------------------------------------------------------------------- + +# directly invoked +my $expect_f = __FILE__; +my $expect_...

[PATCH MANIFEST lib/CGI/t/push.t] New Test for CGI::Push
Just about everything else on the untested list scares me. There are a few modules in ExtUtils::* that need attention, and a handful of Pod::* modules. Someone previously mentioned writing a Pod::Parser test that could be subclassed or extended to cover its children. I might have a look at that. -- c --- ~MANIFEST Mon Sep 10 15:04:38 2001 +++ MANIFEST Mon Sep 10 15:05:00 2001 @@ -807,6 +807,7 @@ lib/CGI/t/function.t See if CGI.pm works lib/CGI/t/html.t See if CGI.pm works lib/CGI/t/pretty.t See if CGI.pm works +lib/CGI/t/push.t See if CGI::Push works lib/CGI/t/reque...

[PATCH MANIFEST, lib/CGI/t/fast.t] Added Test for CGI::Fast
This one's a little trickier, as it requires the non-core FCGI module. It passes both with and without FCGI installed. It doesn't really exercise FCGI-specific bits, just makes a couple of quick passes over things. <can type="worms"> Is it likely that FCGI will be installed in an appropriate location for a new Perl installation to find? Perhaps this module should be distributed with FCGI. </can> I used the -w flag instead of warnings to make it easier, should Lincoln wish to add this to his distribution. In retrospect, using Test::More may make his...

[PATCH t/TEST lib/Test.pm t/lib/Test/*.t] Syncing with Test-1.17
Here's a patch that brings Test.pm in sync with version 1.17 on CPAN. It needs the deep directory t/TEST patch I posted earlier and does its own TEST patch so it recognizes old-style todo tests. Here's the changes: 2001-05-29 Michael G Schwern <schwern@pobox.com> * Release 1.17 - protected against stray $/ and $, * documented ok() * made the regex diagnostics a bit more obvious * Backported to 5.004 (if you're using the new Test::Harness) * Included the VMS fix for ok() 2001-02-16 Michael G Schwern <schwern@pobox.com> * Release 1.16 * T...

Re: Change 26165 broke ext/threads/t/stress_re.t test on Win32 (and patch to t/test.pl and/or Test::Harness)
------=_Part_7656_32239749.1134836603310 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline On 12/16/05, Steve Hay <steve.hay@uk.radan.com> wrote: > The real bummer, though, is that I'm now away until Jan 3rd and I'm > switching my machine off now, so you can't see the fruits of your > efforts in my overnight smokes until next year :-( If its any help to you guys I built and tested just now on Win32 (VC7+Win2k) and it passes all tests + 2 todo tests. This is at patchlevel 26386 I get ...

[PATCH t/comp/proto.t] Test (5.9.x)
--- t/comp/proto.t.bak Mon Oct 27 13:46:43 2003 +++ t/comp/proto.t Mon Oct 27 14:42:20 2003 @@ -16,7 +16,7 @@ use strict; -print "1..140\n"; +print "1..141\n"; my $i = 1; @@ -629,3 +629,10 @@ eval "sub good (\$\t\$\n\$) { 1; }"; print "not " if $@; print "ok ", $i++, "\n"; + + +# Ought to fail, doesn't in 5.8.x. +eval 'sub bug (\[%@]) { } my $array = [0 .. 1]; bug %$array;'; +print "not " unless $@ =~ /Not a HASH reference/; +print "ok ", $i ++; +print "\n"; ...

[PATCH t/io/argv.t vms/test.com t/test.pl] argv.t cleanup & fixes for VMS
Plowing through the list of excepted tests in vms/test.com, we start with t/io/argv.t. This version now works under VMS and Linux. It should work under Windows as well, but I'd like someone to give it a shot. Here's what's been done: The mixed logic about running programs has been consoldated into runthis(). At least if I got something wrong it'll be easier to fix. Switched from ad hoc tests to t/test.pl Eliminated as many warnings as I could. There seems to be a genuine problem with the combination of $^X, STDIN and pipes in...

[PATCH 5.8.x] Allow more tries for t/op/time.t test 2.
I had a chance to fiddle with OpenBSD on a fast x86 system, and it was failing test 2 in t/op/time.t a little more than half the time. I suspect it's a combination of the granularity of times reported by the system and the speed of the system. (Linux, on similar hardware, didn't have any problems.) This patch ups the maximum number of iterations; with this, I got no failures. Most systems should still bail out of this test very quickly so that the maximum number of iterations doesn't matter. All systems should also still honor the 20-second bailout time limit. In s...

[PATCH t/comp/proto.t] TODO tests for bug #24276 (5.8.x only)
A TODO test for bug #24276. This one is fixed in blead - I'll submit a patch to add a similar test, but without the TODO. Abigail --- t/comp/proto.t.bak Mon Oct 27 13:46:43 2003 +++ t/comp/proto.t Mon Oct 27 14:42:20 2003 @@ -16,7 +16,7 @@ use strict; -print "1..140\n"; +print "1..141\n"; my $i = 1; @@ -629,3 +629,10 @@ eval "sub good (\$\t\$\n\$) { 1; }"; print "not " if $@; print "ok ", $i++, "\n"; + + +# Ought to fail, doesn't in 5.8.x. +eval 'sub bug (\[%@]) { } my $array = [...

ext/Digest/MD5/t/files.t of perl5.9.5 failed on EBCDIC platform
--0__=C7BBF980DFBF84F18f9e8a93df938690918cC7BBF980DFBF84F1 Content-type: text/plain; charset=GB2312 Content-transfer-encoding: base64 DQpIaSwNCg0KSSByYW4gZXh0L0RpZ2VzdC9NRDUvdC9maWxlcy50IG9mIHBlcmw1LjkuNSBvbiBF QkNESUMgcGxhdGZvcm0sIHRoZSByZXN1bHQNCmlzIGFzIHRoZSBmb2xsb3dpbmc6DQo9PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PQ0KDQpvayAxDQpvayAyDQpvayAzDQpNRDUueHM6IEJh ZCBkaWdlc3QNCk1ENS54czogQmFkIGhleGRpZ2VzdDogZ290IGQ0YjYxZmQ2Zjg3NWRjZDNlMzA2 NmEzZDY3NTAwMTBlIGV4cGVjdGVkDQozYTkyODlmZjUwM2Q2NzZjNjdjNzJiZmFkOTAyO...

[PATCH t/io/dup.t ext/IO/lib/IO/t/io_dup.t vms/test.com] Recoving dup tests for VMS
This fixes up the two dup tests so they work on VMS. The basic problem was they expected echo to exist. So instead we just use perl. In the process a VMS bug was uncovered. system() does not honor the redirection of STDOUT, so that test has been made TODO on VMS. --- ext/IO/lib/IO/t/io_dup.t Mon Jun 18 01:21:15 2001 +++ ext/IO/lib/IO/t/io_dup.t Mon Nov 12 19:25:00 2001 @@ -39,14 +39,15 @@ print $stdout "ok 2\n"; print $stderr "ok 3\n"; -if ($^O eq 'MSWin32' || $^O eq 'NetWare') { - print `echo ok 4`; - print `echo ok 5 1>&am...

[PATCH t/test.pl ext/POSIX/t/posix.t vms/test.com] POSIX cleanup
Ok, here's the fixups to get ext/POSIX/t/posix.t working on VMS. It also does a few other things. t/test.pl's skip message was a little broken by my last bit of fiddling. posix.t has been cleaned up to use test.pl The problem with autoloading munging POSIX::errno() on VMS is declared TODO. vms/test.com has been updated to remove all the tests that are known good or well-behaved (ie. they skip themselves) -- Michael G. Schwern <schwern@pobox.com> http://www.pobox.com/~schwern/ Perl6 Quality Assurance <perl-qa@perl....

[PATCH t/pod/find.t t/TEST] fixing t/pod/find.t, running t/pod
I just noticed t/pod and that its not run with "make test". This appears to have come up before and was an oversight. http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2000-09/msg00730.html Looks like people just gave up on fixing t/pod/find.t and I don't blame them, its a mess. I'm tempted to just chuck it and start over. Well, the fundemental problem is that Pod::Find is doing all sorts of gyrations to look at the source code. This is fundementally unmaintainable (tests should not break when we add new files to the dist). Instead, I've added t/pod/test...

Web resources about - [PATCH: 5.005_61] an EBCDIC friendly cgi-html.t test - perl.perl5.porters

190th Fighter Squadron, Blues and Royals friendly fire incident - Wikipedia, the free encyclopedia
The 190th Fighter Squadron, Blues and Royals friendly fire incident was a friendly fire incident involving two United States Air Force (USAF) ...

Wildfire Shows How (And Why) To Make Your Facebook Campaign Influencer-Friendly
On Facebook, not all fans are created equal . Certain people have a little more influence than others — when they recommend a brand’s page or ...

Enhanced Family and Children-Friendly Inflight Zone on Our New A380 Economy Main Cabin
Malaysia Airlines will soon join the league of A380 operators, when its first A380 takes to the skies on 1 July 2012. Although the national carrier ...

Friendly, Facepad Grab 2 Million Monthly Actives on the iPad as Facebook Holds Off on Official App
Facebook’s deliberate approach to designing and launching its long-anticipated iPad app has created an opening for enterprising developers. Two ...

Mr Friendly (@MisterFriendly) on Twitter
Sign in Sign up To bring you Twitter, we and our partners use cookies on our and other websites. Cookies help personalize Twitter content, tailor ...

A friendly reminder about password security - Twitter Blogs
... moment to remind you of best practices... Skip to main content Blog Have an account? Sign in Search form Search All Twitter Blogs &nbsp A friendly ...

Friendly Plus for Facebook for iPad on the iTunes App Store
Read reviews, get customer ratings, see screenshots, and learn more about Friendly Plus for Facebook on the App Store. Download Friendly Plus ...

Bomb threat leads to cancellation of Germany-Netherlands 'freedom' football friendly
An international football friendly between Germany and The Netherlands is called off and the stadium evacuated prior to the match because of ...

AMD announces powerful and wallet-friendly Radeon R9 380X graphics card
While Intel and Nvidia produce some amazing hardware, I always have a place in my heart for AMD. True, the company often lags behind Intel for ...

12 travel-friendly tech gifts you'll want to take with you
The travel-friendly tech in this gift guide can save your favorite frequent flier some stress on their next trip. Whether it's a clip-on camera ...

Resources last updated: 12/1/2015 2:51:08 PM