On 22/01/10 9:41 PM, Andrey Razumovsky wrote:
> 2010/1/22 Aristedes Maniatis<ar..aniatis.org>
>
>> On 22/01/10 8:40 PM, andre..pache.org wrote:
>>
>>> URL:http://svn.apache.org/viewvc?rev=902045&view=rev
>>> Log:
>>> trying to make clover work
>>>
>>
>> Yeah, I've got some uncommitted changes to try and do the same. But I
>> suspect the problem is that all the maven poms don't fall into a nice
>> hierarchical tree. I know Andrus has goals of not sucking in too many poms
>> for end users, but I still think it is causing problems.
>>
>>
> No, maven reactor build order is always the same. Just somewhy same
> generate-sources tasks in main and clover lifecycle produces different
> number of files (specifically, Node.java). I see someone already faced this
> trouble and that's why there is Ant delete task. I only changed its
> directory to correct one and now Cayenne-doc builds successfully
But you'll notice that Clover didn't find a single instrumented class so nothing looks like it is tested. I'm guessing this is because each pom which doesn't have a valid parent is pulling the parent code from the jars installed in ~/.m2 and not from the code compiled into target folders.
http://hudson.zones.apache.org/hudson/job/Cayenne-doc/clover
This is probably why you always have to perform 'install' in order to ensure that the project builds properly and you don't get wierd random bugs. Clover instruments into target folders, but never installs instrumented code to a jar. Parents know about children, but children never know about the parent.
I know we've been over this ground before, but as I understand maven better and how it expects things to work, I see that it is very sensitive to doing things the 'maven way'.
Ari
-- --------------------------> Aristedes Maniatis GPG fingerprint CBFB 84B4 738D 4E87 5E5C 5EFA EF6A 7D2E 3E49 102A
This archive was generated by hypermail 2.0.0 : Fri Jan 22 2010 - 05:55:22 EST