I've been working on a mod_perl extension, tested with Apache::Test.
I was running 1.27 before and everything was working great. I upgraded to
1.28 because I wanted to run the 'testcover' action. Now if I run my tests
(using "./Build test", "testcover", or "perl t/TEST"), I get the followin
cc'ing david :)
Tyler MacDonald wrote:
> I've been working on a mod_perl extension, tested with Apache::Test.
> I was running 1.27 before and everything was working great. I upgraded to
> 1.28 because I wanted to run the 'testcover' action. Now if I run my tests
> (using "./Build test", "tes
Geoffrey Young <[EMAIL PROTECTED]> wrote:
> I've seen this recently as well, but I don't think it's primarily an
> Apache-Test issue. at least not a code-based one - I think it happens when
> stale A-T files are left lying around and are reused under various
> circumstances...
Aha! That d
Tyler MacDonald <[EMAIL PROTECTED]> wrote:
> > I've seen this recently as well, but I don't think it's primarily an
> > Apache-Test issue. at least not a code-based one - I think it happens when
> > stale A-T files are left lying around and are reused under various
> > circumstances...
> Aha
Tyler MacDonald wrote:
> Geoffrey Young <[EMAIL PROTECTED]> wrote:
>
>>I've seen this recently as well, but I don't think it's primarily an
>>Apache-Test issue. at least not a code-based one - I think it happens when
>>stale A-T files are left lying around and are reused under various
>>circums
Geoffrey Young <[EMAIL PROTECTED]> wrote:
> actually, we should probably embed the A-T version into
> apache_test_config.pm and reconfigure if the A-T in %INC is newer than the
> version in the file. any takers?
That reminds me, I still need to write up that documentation patch
for mod_pe
> actually, we should probably embed the A-T version into
> apache_test_config.pm and reconfigure if the A-T in %INC is newer than the
> version in the file. any takers?
done.
of course it would't have helped in this case anyway - we attempt to read in
the pidfile to kill the server before reco
> ulimit -c unlimited; /usr/bin/perl /home/faraway/dev/Apache2-AUS/t/TEST
> -bugreport -verbose=0
> /usr/sbin/apache2 -d /home/faraway/dev/Apache2-AUS/t -f
> /home/faraway/dev/Apache2-AUS/t/conf/httpd.conf -D APACHE2 -D
> PERL_USEITHREADS
you're missing an important switch there: -D ONE_PROCESS
David Wheeler wrote:
> On Mar 8, 2006, at 17:08, Geoffrey Young wrote:
>
>> using MakeMaker and 'make testcover' this switch is added. I'm not a
>> M::B
>> user, but judging from your output whatever Apache::TestMB is doing
>> isn't
>> sufficient. and I think we _do_ need david for that :)
>
David Wheeler wrote:
> On Mar 8, 2006, at 17:15, Geoffrey Young wrote:
>
>> no, I've never touched that file. I assumed it was all you :)
>
>
> If so, I just copied it from Module::Build for some reason. But it does
> have this line:
>
> local $ENV{APACHE_TEST_EXTRA_ARGS} = "-one-proces
David Wheeler wrote:
> On Mar 8, 2006, at 17:28, Geoffrey Young wrote:
>
>> it is. but I think the issue is that it isn't being passed over to the
>> actual process that calls t/TEST. in MM land we actually encode that
>> into
>> the makefile, so it's in the shell's environment and can be pi
Geoffrey Young <[EMAIL PROTECTED]> wrote:
> well, I don't even know how to get started...
>
> but I will say that, if we're committed to supporting M::B, and if it's
> possible to have both in the same tree, making it possible to build
> Apache-Test proper using M::B would at least give us an easy
> Is this the right repo to generate diffs
> against?
for this, it's
http://svn.apache.org/repos/asf/perl/Apache-Test/
--Geoff
> OK... Just doing a quick grep for EXTRA_ARGS in trunk/,
> it appears that only TestMM and TestMB.pm actually *have* it. The reason it
> works on the MM size is because it's makefile snippets actually use it. The
> following patch fixes *that*,
applied. thanks!
> but there's still some o
David Wheeler <[EMAIL PROTECTED]> wrote:
> > After the test shuts down completely, the apache process is still
> >there, and has to be killed before I can run the test again:
> Does it fail to be shut down only with testcover?
Nope, but it *does* only fail to shut down only with -one-p
Geoffrey Young <[EMAIL PROTECTED]> wrote:
> the only way to determine whether this is really an A-T issue would be to
> use MakeMaker's testcover target and see if it's behavior differs from that
> of M::B.
It's not a testcover or M::B issue, it's a -one-process issue, and
it's weird:
16 matches
Mail list logo
