You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@maven.apache.org by Nigel Magnay <ni...@gmail.com> on 2008/03/18 12:51:45 UTC
Maven Integration Tests
Help...
I started to run the integration tests to work out why my patch breaks
them, and have moved on to trying to work out why my integration tests
work on some of my machines, but not others, to a bit of confusion. My
patch is in r638310
r638310 of 2.0.x,
(https://ci.sonatype.org/job/Maven-2.0.x-ITs/30/console) - Fails 19
tests
r638310 of 2.0.x, Java 1.5.0_13, linux : 0 Fails
r638310 of 2.0.x, Java 1.5.0_13, mac os : 3 Fails (2744,26,23) [1].
r638310 of 2.0.x, Java 1.6.0_02, windows XP : Fails many, many tests
[2] (>than the hudson build)
HEAD of 2.0.x, Java 1.6.0_02, windows XP : Passes most, Fails
mng3220ImportScope(testitMNG3220b)
So I'm having a hard time nailing down a consistent set of failures
even in the same code. I'm concerned that starting the core-it-tests
sometimes seems to trigger downloads of the actual tests themselves.
So -
Anything here look familiar as I go bughunting..?
What's the platform that ci.sonatype.org is running, so I can try and
get the same failure list?
Shouldn't the actual ITs build as a part of the running of the tests?
It feels a bit of a race condition to compile tests in one tree, then
run them in another, and seeing arbitary snapshots being downloaded
makes me think they're not properly reproduceable. Can't they be
linked into the 2.0.x tree with a svn:external (preferrably at a
*specific* revision number) ?
[1] Added bonus : 12 hours ago, these worked for me on the same machine...
[2] Looking in the IT test directories, this looks like there's a
"${maven.repo.local}" directory being created which is why things are
getting confused.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@maven.apache.org
For additional commands, e-mail: dev-help@maven.apache.org
Re: Maven Integration Tests
Posted by Brett Porter <br...@apache.org>.
On 19/03/2008, at 9:04 AM, Brian E. Fox wrote:
> Make sure you have installed the latest core integration test folder.
> I'll redeploy everything, but it's possible it is out of date. The
> Hudson at ci.sonatype.org is building the Its locally everytime it
> changes.
That's what I did... I haven't looked closely at the failures yet
though.
- Brett
--
Brett Porter
brett@apache.org
http://blogs.exist.com/bporter/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@maven.apache.org
For additional commands, e-mail: dev-help@maven.apache.org
RE: Maven Integration Tests
Posted by "Brian E. Fox" <br...@reply.infinity.nu>.
Make sure you have installed the latest core integration test folder.
I'll redeploy everything, but it's possible it is out of date. The
Hudson at ci.sonatype.org is building the Its locally everytime it
changes.
-----Original Message-----
From: Nigel Magnay [mailto:nigel.magnay@gmail.com]
Sent: Tuesday, March 18, 2008 10:29 AM
To: Maven Developers List
Subject: Re: Maven Integration Tests
Ok - I'll compare to what I get.
It still worries me that you've got failures when
https://ci.sonatype.org/job/Maven-2.0.x-ITs/ is showing a clean bill
of health since Mar 17, 2008 3:32:29 PM..
On Tue, Mar 18, 2008 at 2:24 PM, Brett Porter <br...@apache.org> wrote:
> These are the failures I'm getting today with 2.0.9-SNAPSHOT:
>
>
> testitMNG2861
>
(org.apache.maven.integrationtests.MavenITmng2861RelocationsAndRanges)
>
> testitMNG2744
> (org
>
.apache.maven.integrationtests.MavenITmng2744checksumVerificationTest)
>
> - Brett
>
>
> On 18/03/2008, at 10:51 PM, Nigel Magnay wrote:
>
>
>
> > Help...
> >
> > I started to run the integration tests to work out why my patch
breaks
> > them, and have moved on to trying to work out why my integration
tests
> > work on some of my machines, but not others, to a bit of confusion.
My
> > patch is in r638310
> >
> > r638310 of 2.0.x,
> > (https://ci.sonatype.org/job/Maven-2.0.x-ITs/30/console) - Fails 19
> > tests
> > r638310 of 2.0.x, Java 1.5.0_13, linux : 0 Fails
> > r638310 of 2.0.x, Java 1.5.0_13, mac os : 3 Fails (2744,26,23) [1].
> > r638310 of 2.0.x, Java 1.6.0_02, windows XP : Fails many, many
tests
> > [2] (>than the hudson build)
> > HEAD of 2.0.x, Java 1.6.0_02, windows XP : Passes most, Fails
> > mng3220ImportScope(testitMNG3220b)
> >
> > So I'm having a hard time nailing down a consistent set of failures
> > even in the same code. I'm concerned that starting the
core-it-tests
> > sometimes seems to trigger downloads of the actual tests
themselves.
> >
> > So -
> > Anything here look familiar as I go bughunting..?
> > What's the platform that ci.sonatype.org is running, so I can try
and
> > get the same failure list?
> > Shouldn't the actual ITs build as a part of the running of the
tests?
> > It feels a bit of a race condition to compile tests in one tree,
then
> > run them in another, and seeing arbitary snapshots being downloaded
> > makes me think they're not properly reproduceable. Can't they be
> > linked into the 2.0.x tree with a svn:external (preferrably at a
> > *specific* revision number) ?
> >
> > [1] Added bonus : 12 hours ago, these worked for me on the same
> > machine...
> > [2] Looking in the IT test directories, this looks like there's a
> > "${maven.repo.local}" directory being created which is why things
are
> > getting confused.
> >
>
>
> >
---------------------------------------------------------------------
> > To unsubscribe, e-mail: dev-unsubscribe@maven.apache.org
> > For additional commands, e-mail: dev-help@maven.apache.org
> >
>
> --
> Brett Porter
> brett@apache.org
> http://blogs.exist.com/bporter/
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@maven.apache.org
> For additional commands, e-mail: dev-help@maven.apache.org
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@maven.apache.org
For additional commands, e-mail: dev-help@maven.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@maven.apache.org
For additional commands, e-mail: dev-help@maven.apache.org
Re: Maven Integration Tests
Posted by Nigel Magnay <ni...@gmail.com>.
Ok - I'll compare to what I get.
It still worries me that you've got failures when
https://ci.sonatype.org/job/Maven-2.0.x-ITs/ is showing a clean bill
of health since Mar 17, 2008 3:32:29 PM..
On Tue, Mar 18, 2008 at 2:24 PM, Brett Porter <br...@apache.org> wrote:
> These are the failures I'm getting today with 2.0.9-SNAPSHOT:
>
>
> testitMNG2861
> (org.apache.maven.integrationtests.MavenITmng2861RelocationsAndRanges)
>
> testitMNG2744
> (org
> .apache.maven.integrationtests.MavenITmng2744checksumVerificationTest)
>
> - Brett
>
>
> On 18/03/2008, at 10:51 PM, Nigel Magnay wrote:
>
>
>
> > Help...
> >
> > I started to run the integration tests to work out why my patch breaks
> > them, and have moved on to trying to work out why my integration tests
> > work on some of my machines, but not others, to a bit of confusion. My
> > patch is in r638310
> >
> > r638310 of 2.0.x,
> > (https://ci.sonatype.org/job/Maven-2.0.x-ITs/30/console) - Fails 19
> > tests
> > r638310 of 2.0.x, Java 1.5.0_13, linux : 0 Fails
> > r638310 of 2.0.x, Java 1.5.0_13, mac os : 3 Fails (2744,26,23) [1].
> > r638310 of 2.0.x, Java 1.6.0_02, windows XP : Fails many, many tests
> > [2] (>than the hudson build)
> > HEAD of 2.0.x, Java 1.6.0_02, windows XP : Passes most, Fails
> > mng3220ImportScope(testitMNG3220b)
> >
> > So I'm having a hard time nailing down a consistent set of failures
> > even in the same code. I'm concerned that starting the core-it-tests
> > sometimes seems to trigger downloads of the actual tests themselves.
> >
> > So -
> > Anything here look familiar as I go bughunting..?
> > What's the platform that ci.sonatype.org is running, so I can try and
> > get the same failure list?
> > Shouldn't the actual ITs build as a part of the running of the tests?
> > It feels a bit of a race condition to compile tests in one tree, then
> > run them in another, and seeing arbitary snapshots being downloaded
> > makes me think they're not properly reproduceable. Can't they be
> > linked into the 2.0.x tree with a svn:external (preferrably at a
> > *specific* revision number) ?
> >
> > [1] Added bonus : 12 hours ago, these worked for me on the same
> > machine...
> > [2] Looking in the IT test directories, this looks like there's a
> > "${maven.repo.local}" directory being created which is why things are
> > getting confused.
> >
>
>
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: dev-unsubscribe@maven.apache.org
> > For additional commands, e-mail: dev-help@maven.apache.org
> >
>
> --
> Brett Porter
> brett@apache.org
> http://blogs.exist.com/bporter/
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@maven.apache.org
> For additional commands, e-mail: dev-help@maven.apache.org
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@maven.apache.org
For additional commands, e-mail: dev-help@maven.apache.org
Re: Maven Integration Tests
Posted by Brett Porter <br...@apache.org>.
These are the failures I'm getting today with 2.0.9-SNAPSHOT:
testitMNG2861
(org.apache.maven.integrationtests.MavenITmng2861RelocationsAndRanges)
testitMNG2744
(org
.apache.maven.integrationtests.MavenITmng2744checksumVerificationTest)
- Brett
On 18/03/2008, at 10:51 PM, Nigel Magnay wrote:
> Help...
>
> I started to run the integration tests to work out why my patch breaks
> them, and have moved on to trying to work out why my integration tests
> work on some of my machines, but not others, to a bit of confusion. My
> patch is in r638310
>
> r638310 of 2.0.x,
> (https://ci.sonatype.org/job/Maven-2.0.x-ITs/30/console) - Fails 19
> tests
> r638310 of 2.0.x, Java 1.5.0_13, linux : 0 Fails
> r638310 of 2.0.x, Java 1.5.0_13, mac os : 3 Fails (2744,26,23) [1].
> r638310 of 2.0.x, Java 1.6.0_02, windows XP : Fails many, many tests
> [2] (>than the hudson build)
> HEAD of 2.0.x, Java 1.6.0_02, windows XP : Passes most, Fails
> mng3220ImportScope(testitMNG3220b)
>
> So I'm having a hard time nailing down a consistent set of failures
> even in the same code. I'm concerned that starting the core-it-tests
> sometimes seems to trigger downloads of the actual tests themselves.
>
> So -
> Anything here look familiar as I go bughunting..?
> What's the platform that ci.sonatype.org is running, so I can try and
> get the same failure list?
> Shouldn't the actual ITs build as a part of the running of the tests?
> It feels a bit of a race condition to compile tests in one tree, then
> run them in another, and seeing arbitary snapshots being downloaded
> makes me think they're not properly reproduceable. Can't they be
> linked into the 2.0.x tree with a svn:external (preferrably at a
> *specific* revision number) ?
>
> [1] Added bonus : 12 hours ago, these worked for me on the same
> machine...
> [2] Looking in the IT test directories, this looks like there's a
> "${maven.repo.local}" directory being created which is why things are
> getting confused.
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@maven.apache.org
> For additional commands, e-mail: dev-help@maven.apache.org
>
--
Brett Porter
brett@apache.org
http://blogs.exist.com/bporter/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@maven.apache.org
For additional commands, e-mail: dev-help@maven.apache.org
Re: Maven Integration Tests
Posted by Nigel Magnay <ni...@gmail.com>.
Hokay.. I've had a bit more time for this.
I've re-rolled the patch against SVN, as I'm worried that the JIRA
issue was being confused with various different versions flying around
and changes happening in the test code.
Could someone try it against their local copy and see if the ITs work
the same? On my home machine (which has never compiled anything up to
yesterday) it works fine.
If it doesn't work fine, could you mail me the IT output so I can try
diagnosing why my world is different! :-) The hudson ci threw away the
earlier output of the non-working patch before I could get to it..
On Tue, Mar 18, 2008 at 11:41 PM, Brian E. Fox <br...@reply.infinity.nu> wrote:
> There are two ways you can run the Its. First is in the maven build
> tree, use mvn clean install -Prun-its. This will build and then unpack
> the maven and use it to run the Its from a jar. If you do this, make
> sure you have already installed the it tree.
>
> The other way is to build and install the maven and then go into the it
> tree and run mvn clean install -Prun-its. This will use your installed
> maven to execute the tests directly in the it project.
>
> Hudson is running on Redhat linux, but I'm running on Windows and I got
> almost the exact same errors.
>
>
> -----Original Message-----
> From: Nigel Magnay [mailto:nigel.magnay@gmail.com]
>
> Sent: Tuesday, March 18, 2008 7:52 AM
> To: Maven Developers List
>
>
> Subject: Maven Integration Tests
>
> Help...
>
> I started to run the integration tests to work out why my patch breaks
> them, and have moved on to trying to work out why my integration tests
> work on some of my machines, but not others, to a bit of confusion. My
> patch is in r638310
>
> r638310 of 2.0.x,
> (https://ci.sonatype.org/job/Maven-2.0.x-ITs/30/console) - Fails 19
> tests
> r638310 of 2.0.x, Java 1.5.0_13, linux : 0 Fails
> r638310 of 2.0.x, Java 1.5.0_13, mac os : 3 Fails (2744,26,23) [1].
> r638310 of 2.0.x, Java 1.6.0_02, windows XP : Fails many, many tests
> [2] (>than the hudson build)
> HEAD of 2.0.x, Java 1.6.0_02, windows XP : Passes most, Fails
> mng3220ImportScope(testitMNG3220b)
>
> So I'm having a hard time nailing down a consistent set of failures
> even in the same code. I'm concerned that starting the core-it-tests
> sometimes seems to trigger downloads of the actual tests themselves.
>
> So -
> Anything here look familiar as I go bughunting..?
> What's the platform that ci.sonatype.org is running, so I can try and
> get the same failure list?
> Shouldn't the actual ITs build as a part of the running of the tests?
> It feels a bit of a race condition to compile tests in one tree, then
> run them in another, and seeing arbitary snapshots being downloaded
> makes me think they're not properly reproduceable. Can't they be
> linked into the 2.0.x tree with a svn:external (preferrably at a
> *specific* revision number) ?
>
> [1] Added bonus : 12 hours ago, these worked for me on the same
> machine...
> [2] Looking in the IT test directories, this looks like there's a
> "${maven.repo.local}" directory being created which is why things are
> getting confused.
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@maven.apache.org
> For additional commands, e-mail: dev-help@maven.apache.org
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@maven.apache.org
> For additional commands, e-mail: dev-help@maven.apache.org
>
>
RE: Maven Integration Tests
Posted by "Brian E. Fox" <br...@reply.infinity.nu>.
There are two ways you can run the Its. First is in the maven build
tree, use mvn clean install -Prun-its. This will build and then unpack
the maven and use it to run the Its from a jar. If you do this, make
sure you have already installed the it tree.
The other way is to build and install the maven and then go into the it
tree and run mvn clean install -Prun-its. This will use your installed
maven to execute the tests directly in the it project.
Hudson is running on Redhat linux, but I'm running on Windows and I got
almost the exact same errors.
-----Original Message-----
From: Nigel Magnay [mailto:nigel.magnay@gmail.com]
Sent: Tuesday, March 18, 2008 7:52 AM
To: Maven Developers List
Subject: Maven Integration Tests
Help...
I started to run the integration tests to work out why my patch breaks
them, and have moved on to trying to work out why my integration tests
work on some of my machines, but not others, to a bit of confusion. My
patch is in r638310
r638310 of 2.0.x,
(https://ci.sonatype.org/job/Maven-2.0.x-ITs/30/console) - Fails 19
tests
r638310 of 2.0.x, Java 1.5.0_13, linux : 0 Fails
r638310 of 2.0.x, Java 1.5.0_13, mac os : 3 Fails (2744,26,23) [1].
r638310 of 2.0.x, Java 1.6.0_02, windows XP : Fails many, many tests
[2] (>than the hudson build)
HEAD of 2.0.x, Java 1.6.0_02, windows XP : Passes most, Fails
mng3220ImportScope(testitMNG3220b)
So I'm having a hard time nailing down a consistent set of failures
even in the same code. I'm concerned that starting the core-it-tests
sometimes seems to trigger downloads of the actual tests themselves.
So -
Anything here look familiar as I go bughunting..?
What's the platform that ci.sonatype.org is running, so I can try and
get the same failure list?
Shouldn't the actual ITs build as a part of the running of the tests?
It feels a bit of a race condition to compile tests in one tree, then
run them in another, and seeing arbitary snapshots being downloaded
makes me think they're not properly reproduceable. Can't they be
linked into the 2.0.x tree with a svn:external (preferrably at a
*specific* revision number) ?
[1] Added bonus : 12 hours ago, these worked for me on the same
machine...
[2] Looking in the IT test directories, this looks like there's a
"${maven.repo.local}" directory being created which is why things are
getting confused.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@maven.apache.org
For additional commands, e-mail: dev-help@maven.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@maven.apache.org
For additional commands, e-mail: dev-help@maven.apache.org
Re: Maven Integration Tests
Posted by Brett Porter <br...@apache.org>.
On 18/03/2008, at 10:51 PM, Nigel Magnay wrote:
> [1] Added bonus : 12 hours ago, these worked for me on the same
> machine...
> [2] Looking in the IT test directories, this looks like there's a
> "${maven.repo.local}" directory being created which is why things are
> getting confused.
This looks related to Brian's recent change (though I thought it was
in a profile).
I'd suggest making sure you have all the svn updates, clean install
from the core-integration-testing module, then run the its.
- Brett
--
Brett Porter
brett@apache.org
http://blogs.exist.com/bporter/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@maven.apache.org
For additional commands, e-mail: dev-help@maven.apache.org