Hi,
Considering the axiom SNAPSHOT dependency puzzle has been figured out,
should we go ahead to push M2 release out?
Thanks,
Raymond
----- Original Message -----
From: "Raymond Feng" <[EMAIL PROTECTED]>
To: <tuscany-dev@ws.apache.org>
Sent: Monday, December 04, 2006 12:10 PM
Subject: Re: Much ado about nothing (Re: WSCOMMONS-131 and options for
Tuscany SCA Java M2 release)
> Hi,
>
> I did some further investigations with Ant and here are some
observations:
>
> 1) maven2 build will pull in axis2 artifacts from maven1 or maven2
repo
> randomly. I even see a mixture that the pom is downloaded from a
m1 repo
> while the jar from a m2 repo.
>
> 2) axis2-kernel 1.1 pom from maven2 repo is invalid. It has a
dependency
> on "org.apache.ws.commons.neethi:neethi"
> <dependency>
> <groupId>org.apache.ws.commons.neethi</groupId>
> <artifactId>neethi</artifactId>
> </dependency>
>
> But the axis2-parent pom has "org.apache.neethi:neethi" instead.
>
> <dependency>
> <groupId>org.apache.neethi</groupId>
> <artifactId>neethi</artifactId>
> <version>2.0</version>
> </dependency>
>
> As a result, the pom fails to pass the validation and it's treated
as an
> invalid pom. Transative dependency is not activated at all for
> axis2-kernel 1.1.
>
> To conclude, no matter axis2 1.1 is downloaded from m1 or m2 repo,
the
pom
> is invalid and transative dependency is not activated at all for
> axis2-kernel. The bug becomes good news to us: the SNAPSHOT
version of
> axiom-api will NOT be used by Tuscany runtime.
>
> Do we feel safe to release our M2 now without waiting for further
releases
> from axis2?
>
> Thanks,
> Raymond
>
> ----- Original Message -----
> From: "ant elder" <[EMAIL PROTECTED]>
> To: <tuscany-dev@ws.apache.org>
> Sent: Monday, December 04, 2006 12:31 AM
> Subject: Re: Much ado about nothing (Re: WSCOMMONS-131 and options
for
> Tuscany SCA Java M2 release)
>
>
>> I'm starting to wonder if the subject line is even more apt than we
>> realize
>> and we're just assuming this is a significant problem without
>> investigating.
>>
>>
>> AFAICT the only thing thats happening is what I originally
reported -
you
>> see log messages about the AXIOM API SNAPSHOT when using the WS
>> samples...but it doesn't look like that SNAPSHOT jar is actually
getting
>> used by anything.
>>
>> - If you delete if from your repository the samples run fine. You
can
run
>> the samples in offline mode and not have that SNAPSHOT jar in your
>> repository and the samples still run fine.
>>
>> - If you build the sample webapps with loadExtensionDependencies
set to
>> true
>> then the SNAPSHOT jar is *not* included in the WAR repository and
the
>> samples still run fine even in offline mode.
>>
>> - If you make incompatible changes to classes in the SNAPSHOT jar in
your
>> local repository, eg change the OMElement interface to have no
methods,
>> the
>> samples still run fine.
>>
>> Additionally debugging in the MavenHelper class you can see the
SNAPSHOT
>> messages are coming from the artifactResolver.resolveTransitively
call
on
>> line 264 but the ArtifactResolutionResult thats returned from the
call
>> *does
>> not* include the axiom-api SNAPSHOT artifact.
>>
>> Based on that it doesn't look like the SNAPSHOT jar is even added to
the
>> classpath used by the samples. Could anyone else confirm these
>> observations?
>> If this is true the I think we could safely cut the M2 release
as-is,
>> maybe
>> just add a comment to the release notes about the SNAPSHOT messages.
>>
>> ...ant
>>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]