Thanks Curtis.

Cheers,

Jay


On Oct 3, 2014, at 12:53 PM, Curtis Rueden <ctrue...@wisc.edu> wrote:

> Hi Jay,
> 
> > In TrackMate v 2.5 pom there is...
> > 
> > <parent>
> >     <groupId>sc.fiji</groupId>
> >     <artifactId>pom-fiji</artifactId>
> >     <version>2.0.0-beta-1-SNAPSHOT</version>
> >     <relativePath/>
> > </parent>
> > <artifactId>TrackMate_</artifactId>
> > <version>2.5.0</version>
> > <name>plugins/TrackMate_.jar</name>
> 
> OK, I finally made a little time to look at this.
> 
> Yes, this is a major problem with TrackMate's releases. Version 2.5.0 
> inherits from sc.fiji:pom-fiji:2.0.0-beta-1-SNAPSHOT. This is forbidden in 
> Maven terms -- a release must depend only on other releases.
> 
> This is a consequence of the fact that Fiji artifacts do not use the 
> maven-release-plugin for releases, and hence nothing ever actually enforces 
> that stipulation. The solution will be for all Fiji plugins to extend a 
> _release_ version of the pom-fiji parent, not a SNAPSHOT version.
> 
> We are moving in that direction... but until we complete that migration, 
> downstream artifacts such as JEX will _not_ be stable when they depend on a 
> Fiji project such as TrackMate, _even if_ the dependency purports to be a 
> "release" version! Caveat emptor.
> 
> Regards,
> Curtis
> 
> On Sun, Sep 14, 2014 at 7:40 AM, Jay Warrick <jay.w.warr...@gmail.com> wrote:
> I changed the subject line from Re: [ImageJ-devel] Maven TrackMate Dependency 
> Question because of the following...
> 
> Oh boy, really at the extent of my knowledge... but I think I may have it, is 
> this what is happening??? Is maven pulling TrackMate v 2.3 via TrackMate v2.5 
> and fiji?
> 
> In TrackMate v 2.5 pom there is...
> 
> <parent>
>       <groupId>sc.fiji</groupId>
>       <artifactId>pom-fiji</artifactId>
>       <version>2.0.0-beta-1-SNAPSHOT</version>
>       <relativePath/>
> </parent>
> <artifactId>TrackMate_</artifactId>
> <version>2.5.0</version>
> <name>plugins/TrackMate_.jar</name>
> 
> which thus pulls in pom-fiji 2.0.0-beta-1-SNAPSHOT where there is the 
> following...
> 
> <fiji-compat.version>2.0.0-SNAPSHOT</fiji-compat.version>
> 
> and 
> 
> <dependency>
>       <groupId>sc.fiji</groupId>
>       <artifactId>fiji-compat</artifactId>
> </dependency>
> 
> The pom for fiji-compat uses pom-fiji 2.0.0-SNAPSHOT as parent
> 
> <parent>
>       <groupId>sc.fiji</groupId>
>       <artifactId>pom-fiji</artifactId>
>       <version>2.0.0-SNAPSHOT</version>
>       <relativePath/>
> </parent><
> artifactId>fiji-compat</artifactId>
> <name>jars/fiji-compat.jar</name>
> 
> The pom-fiji 2.0.0-SNAPSHOT has the following
> 
> <TrackMate.version>2.3.0-SNAPSHOT</TrackMate.version>
> 
> and 
> 
> <dependency>
>       <groupId>sc.fiji</groupId>
>       <artifactId>TrackMate_</artifactId>
>       <version>${TrackMate.version}</version>
> </dependency>
> 
> Thus, ?maybe? with TrackMate v 2.5's pom parent being pom-fiji 
> 2.0.0-beta-1-SNAPSHOT, fiji-compate 2.0.0-SNAPSHOT gets pulled in which pulls 
> pom-fiji 2.0.0-SNAPSHOT which tries to pulls in TrackMate v 2.3 which then 
> tries to use <groupId>${jama.groupId}</groupId> instead of 
> <groupId>gov.nist.math</groupId> for jama. 
> 
> Is this counter to what is intended or is there some maven magic which should 
> right this "wrong"? If it is not intended and could be affecting things, what 
> might be the fix? 
> 
> Thanks all,
> 
> Jay
> 
> 
> On Sep 14, 2014, at 6:51 AM, Jay Warrick <jay.w.warr...@gmail.com> wrote:
> 
>> Ok. So this makes more sense...  TrackMate v 2.3 uses the 
>> <groupId>${jama.groupId}</groupId> which causes the error, which was fixed 
>> by TrackMate v2.5 which uses <groupId>gov.nist.math</groupId>.
>> 
>> So the whole issue boils down to, "Why is maven trying to do ANYTHING with v 
>> 2.3 when I can't find a single dependency on this version? If I can get 
>> Maven to stop looking for v 2.3 my issue will be solved.
>> 
>> Thoughts on where to search for hidden/residual links to v 2.3 given what 
>> I've checked already?
>> 
>> Thanks,
>> 
>> Jay
>> 
>> On Sep 14, 2014, at 6:45 AM, Jay Warrick <jay.w.warr...@gmail.com> wrote:
>> 
>>> I think never mind. Still learning about Maven, but it looks like the 
>>> groupID is assumed the same as the parent so you TrackMate doesn't need to 
>>> define it. Should have guessed this based on how everything else seems to 
>>> work in Maven.
>>> 
>>> On Sep 14, 2014, at 6:15 AM, Jay Warrick <jay.w.warr...@gmail.com> wrote:
>>> 
>>>> If this might be on the right track, is there any way to test this on my 
>>>> end? Use offline mode and go into my .m2 repository and muck with poms or 
>>>> something?
>>>> On Sep 14, 2014, at 6:11 AM, Jay Warrick <jay.w.warr...@gmail.com> wrote:
>>>> 
>>>>> Hi JY,
>>>>> 
>>>>> Could it be a problem with the TrackMate groupId? Searching the internet 
>>>>> with the error string brought up this past issue with jama and SPIM...
>>>>> 
>>>>> https://groups.google.com/forum/#!topic/fiji-devel/zEhBIf1bZEA
>>>>> 
>>>>> When I look at the TrackMate pom, I don't think I see a "groupId" defined 
>>>>> for the TrackMate plugin. Hard for me to completely decipher in the issue 
>>>>> discussion if this would cause the problem I'm seeing or if it would 
>>>>> merely prevent the imageJ/Fiji folks from "seeing" that such a problem 
>>>>> exists.
>>>>> 
>>>>> Stephan Preibisch, Curtis, and Dscho - Does that sound like the right 
>>>>> track? Thoughts?
>>>>> 
>>>>> Thanks,
>>>>> 
>>>>> Jay
>>>>> 
>>>>> On Sep 13, 2014, at 9:20 AM, Jean-Yves Tinevez <tine...@pasteur.fr> wrote:
>>>>> 
>>>>>> On 13-Sep-14 16:01, Jay Warrick wrote:
>>>>>>> Hi Jean-Yves,
>>>>>>> 
>>>>>>> Thanks for the quick reply! Here are the poms for my multi module 
>>>>>>> project. I added to the default "pom.xml" names to make things easier 
>>>>>>> to follow. The Master pom is the parent pom for each of my project 
>>>>>>> modules. The parent of Master is scijava. The JEX module has 
>>>>>>> essentially one class, Main which calls the main class in Core. Core 
>>>>>>> contains essentially the whole program including the dependency on 
>>>>>>> TrackMate. Broken contains broken code for one reason or another but 
>>>>>>> doesn't really come into play as the Broken module is not part of the 
>>>>>>> build process. It is "disabled" using a profile within the JEX module. 
>>>>>>> Despite this, I still confirmed it didn't depend on TrackMate 2.3 using 
>>>>>>> the Eclipse dependency hierarchy viewer, just in case.
>>>>>> 
>>>>>> Ok thank you for the crystal clear report. 
>>>>>> We are on the verge of what I know of maven. However I remember having a 
>>>>>> similar issue that was fixed by making the parent of the master pom the 
>>>>>> fiji plugin pom, like for TrackMate pom.
>>>>>> 
>>>>>> Any chance you could just try that first?
>>>>>> jy
>>>>>> 
>>>>>> 
>>>>>>          
>>>>>> This email is free from viruses and malware because avast! Antivirus 
>>>>>> protection is active.
>>>>>> 
>>>>>> 
>>>>>> _______________________________________________
>>>>>> ImageJ-devel mailing list
>>>>>> ImageJ-devel@imagej.net
>>>>>> http://imagej.net/mailman/listinfo/imagej-devel
>>>>> 
>>>>> _______________________________________________
>>>>> ImageJ-devel mailing list
>>>>> ImageJ-devel@imagej.net
>>>>> http://imagej.net/mailman/listinfo/imagej-devel
>>>> 
>>>> _______________________________________________
>>>> ImageJ-devel mailing list
>>>> ImageJ-devel@imagej.net
>>>> http://imagej.net/mailman/listinfo/imagej-devel
>>> 
>>> _______________________________________________
>>> ImageJ-devel mailing list
>>> ImageJ-devel@imagej.net
>>> http://imagej.net/mailman/listinfo/imagej-devel
>> 
> 
> 
> _______________________________________________
> ImageJ-devel mailing list
> ImageJ-devel@imagej.net
> http://imagej.net/mailman/listinfo/imagej-devel
> 
> 
> _______________________________________________
> ImageJ-devel mailing list
> ImageJ-devel@imagej.net
> http://imagej.net/mailman/listinfo/imagej-devel

_______________________________________________
ImageJ-devel mailing list
ImageJ-devel@imagej.net
http://imagej.net/mailman/listinfo/imagej-devel

Reply via email to