chemistry-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Klevenz, Stephan" <stephan.klev...@sap.com>
Subject Re: Issue release 0.5.0
Date Tue, 27 Sep 2011 13:07:05 GMT
Hi Florent,

the OSGi artifact names do not have classifier in their name:

chemistry-opencmis-osgi-client-0.6.0-SNAPSHOT.jar
chemistry-opencmis-osgi-server-0.6.0-SNAPSHOT.jar

Both bundles are OSGi wrappers including all the single artifacts required
to get a client or server bundle. The classifier becomes only visible in
the Bundle-ClassPath entry of the MANIFEST.MF where all the internally
used libraries are referenced. The libraries and also the classifier
tagged libraries are included in the lib directory of the bundle. Finally
the classifier usage is only an implementation detail for the build.

Does this clarify your concerns?

Regards,
Stephan



>Hi,
>
>Sorry I didn't have time to comment on this thread yesterday.
>
>If we have something that works that's great, but I'm curious as to
>why the scheme:
>  chemistry-opencmis-server-bindings // war
>  chemistry-opencmis-server-bindings-classes // lib
>was chosen, when for me the natural choice would be:
>  chemistry-opencmis-server-bindings // lib
>  chemistry-opencmis-server-bindings-war // war
>
>For me a WAR has always been a secondary artifact assembled from other
>stuff, some of them being one (or more) underlying libraries.
>This may be a pure maven issue of course.
>But for OSGi artifact and basic library it would be much cleaner if
>the artifact name didn't have any classifier, as is the norm for 99%
>of libraries.
>
>At some point in the future I may be working on this. Is it really a
>problem if for an hypothetical 0.6 or 0.7 we change the artifact names
>(again)? In any case people have to upgrade the <version> in their
>POM, it's probably not a huge deal to slightly change the artifact
>name as well, is it?
>
>Florent
>
>
>On Tue, Sep 27, 2011 at 12:50 PM, Klevenz, Stephan
><stephan.klevenz@sap.com> wrote:
>> A new release is not required for the OSGi use case because of only the
>> build procedure for artifacts has changed. The artifact itself is more
>>or
>> less the same.
>>
>> The issue that I see is that InMemory server artifact is accidentally
>> renamed in 0.5.0. This is maybe a reason for having a 0.5.1 version.
>>
>> Regards,
>> Stephan
>>
>>
>>
>>
>>
>>>Do yo think we should do a 0.5.1?
>>>
>>>Or we can live with this until 0.6.0 (not sure if it's too early to
>>>foresee that one)?
>>>
>>>What do the others think?
>>>
>>>Ciao!
>>>Gab
>>>
>>>On Sep 27, 2011, at 11:28 AM, Klevenz, Stephan wrote:
>>>
>>>> Gab,
>>>>
>>>> now it is perfect! The OSGi bundles look like before and I could
>>>>install
>>>> and activate it within the OSGi runtime.
>>>>
>>>> Regards,
>>>> Stephan
>>>>
>>>>> Hi Stephan,
>>>>>
>>>>> as I commented on [1], this is due to a regression in
>>>>>maven-war-plugin
>>>>> 2.1.1 [2].
>>>>>
>>>>> I reverted to the previously working version (2.1-beta-1) in the main
>>>>> pom.xml <pluginManagement> section so that all subprojects inherit
>>>>>that
>>>>> version.
>>>>>
>>>>> I closed [1] and tested a bit. Would you mind double checking? :)
>>>>>
>>>>> Thanks,
>>>>> Gab
>>>>>
>>>>> [1]
>>>>>
>>>>>https://issues.apache.org/jira/browse/CMIS-435?focusedCommentId=131153
>>>>>49
>>>>>#c
>>>>> omment-13115349
>>>>> [2] http://jira.codehaus.org/browse/MWAR-240
>>>>>
>>>>>
>>>>> On Sep 27, 2011, at 9:20 AM, Klevenz, Stephan wrote:
>>>>>
>>>>>> Hi Gab,
>>>>>>
>>>>>> All the "-classes.jar" do not include any classes. This is the case
>>>>>>for
>>>>>> all modules like server and  also test modules are affected, too.
I
>>>>>> assume
>>>>>> that this is a general remaining issue.
>>>>>>
>>>>>> The solution for the OSGi bundles is great. The "*-classes.jar" are
>>>>>> included and this should work. Unfortunately not really as long as
>>>>>>the
>>>>>> files are empty.
>>>>>>
>>>>>> Regards
>>>>>> Stephan
>>>>>>
>>>>>>
>>>>>>> Hey Stephan,
>>>>>>>
>>>>>>> this should be fixed in revision r1175977, I also changed
>>>>>>>references
>>>>>>>in
>>>>>>> all modules I could sensibly update.
>>>>>>>
>>>>>>> Do you mind double checking the proper functionalities of those
>>>>>>>modules
>>>>>>> and close the issue (when ASF JIRA scheduled upgrade is finished)
>>>>>>>if
>>>>>>> happy?
>>>>>>>
>>>>>>> HTH,
>>>>>>> Gab
>>>>>>>
>>>>>>>
>>>>>>> On Sep 26, 2011, at 2:34 PM, Gabriele Columbro wrote:
>>>>>>>
>>>>>>>> Hey Stephane,
>>>>>>>>
>>>>>>>> On Sep 26, 2011, at 2:07 PM, Klevenz, Stephan wrote:
>>>>>>>>
>>>>>>>>> Hi Gab,
>>>>>>>>>
>>>>>>>>> I'm not that Maven expert and was just not aware of how
to get a
>>>>>>>>> dependency reference to -classes.jar file. So basically
I agree
>>>>>>>>>and
>>>>>>>>> +1
>>>>>>>>> for
>>>>>>>>> your proposal.
>>>>>>>>
>>>>>>>> Nice!
>>>>>>>>
>>>>>>>> Classifiers are indeed not a very famous, but quite useful,
Maven
>>>>>>>> feature. Just opened this issue [1], will take a stab at
it today
>>>>>>>>and
>>>>>>>> ask your review.
>>>>>>>>
>>>>>>>> Thanks,
>>>>>>>> Gab
>>>>>>>>
>>>>>>>>>
>>>>>>>>> Regards,
>>>>>>>>> Stephan
>>>>>>>>
>>>>>>>>
>>>>>>>> [1] https://issues.apache.org/jira/browse/CMIS-435
>>>>>>>>
>>>>>>>
>>>>>>
>>>>>
>>>>>
>>>>
>>>
>>>Gabriele Columbro
>>>ECM Architect and Open Source Enthusiast
>>>columbro@gmail.com
>>>twitter: @mindthegabz
>>>
>>>I am raising money for Save the Children by walking up Mount Toubkal in
>>>Morocco. To support me and find out more visit my JustGiving site
>>>(http://www.justgiving.com/mindthegab-toubkal-alfresco-2011/)
>>>
>>>
>>>
>>>
>>
>>
>
>
>
>-- 
>Florent Guillaume, Director of R&D, Nuxeo
>Open Source, Java EE based, Enterprise Content Management (ECM)
>http://www.nuxeo.com   http://www.nuxeo.org   +33 1 40 33 79 87


Mime
View raw message