I think classes should be first. (before test-classes)

-----Original Message-----
From: Daniel Kulp [mailto:[EMAIL PROTECTED] 
Sent: Thursday, November 15, 2007 11:14 PM
To: [email protected]
Cc: Brian E. Fox
Subject: Re: moving forward with 2.0.8

On Thursday 15 November 2007, Brian E. Fox wrote:
> Ultimately I think the intent was that the order be slightly
> controllable via the order in the dependencies list. If that's still
> the case, then it seems ok to me.

I completely agree with that.   If it makes it PREDICTABLE and 
CONTROLLABLE, that's a huge improvement.

I guess the question is what is the proper order for the classpath 
entries that are NOT defined via dependencies.  Basically, the classes 
and classes-test directories.     I think those two have reversed since 
2.0.7.

In anycase, I'm completely OK with the change.   I've fixed my testcase 
so it works properly either way now.

Dan


>
> -----Original Message-----
> From: Hervé BOUTEMY [mailto:[EMAIL PROTECTED]
> Sent: Thursday, November 15, 2007 7:49 AM
> To: Maven Developers List
> Subject: Re: moving forward with 2.0.8
>
> Perhaps I know the code that changed the order:
> while fixing "[MANTTASKS-22] ensure proper order of artifacts in
> ArtifactResolutionResult", the fix in trunk (r591996) was to transform
> a HashSet to a LinkedHashSet in
> o.a.m.artifact.resolver.ArtifactResolutionResult.
> But when porting the fix back to 2.0.x branch (r592009), I found that
> such changes in DefaultArtifactCollector had not been ported from
> trunk to branch, then I did it, which can have such a reordering
> consequence.
>
> I just checked in DefaultArtifactCollector history, and found the
> change that has not been back-ported to 2.0.x branch:
> http://svn.apache.org/viewvc/maven/artifact/trunk/src/main/java/org/ap
>ache/maven/artifact/resolver/DefaultArtifactCollector.java?r1=379650&r2
>=512339&pathrev=591996&diff_format=h
>
> Was this intentional to helps ensure compatibility in 2.0.x branch?
>
> HTH to make right choices
>
> Hervé
>
> Le mercredi 14 novembre 2007, Daniel Kulp a écrit :
> > Brian,
> >
> > 2.0.8 seems to have re-ordered the classpath or something such that
> > if I have a unit test that does:
> > System.out.println(getClass().getResource(".").toString());
> > it prints out the directory in the target/classes dir instead of the
> > target/test-classes.    With 2.0.5 - 2.0.7, it returned the
> > test-classes version.
> >
> > Now, I DO think the test is bogus and shouldn't depend on the
> > classpath ordering like that.  However, this is a change that could
> > cause issues for people.
> >
> > Dan
> >
> > On Wednesday 14 November 2007, Daniel Kulp wrote:
> > > Brian,
> > >
> > > CXF doesn't build with 2.0.8-SNAPSHOT.   It builds fine with
> > > 2.0.7, but I'm getting test failures with 2.0.8-SNAPSHOT.   I'll
> > > try and dig into it in a bit.
> > >
> > > Also, the README.txt needs updating before you release it.
> > >
> > > Dan
> > >
> > > > 2007/11/13, Brian E. Fox <[EMAIL PROTECTED]>:
> > > > > All,
> > > > >
> > > > > It's time to move forward with 2.0.8. I stopped to evaluate
> > > > > MNG-3259 but I think this is an edge case and the fix has a
> > > > > greater chance of breaking more stuff. I'd rather fix this
> > > > > early in 2.0.9 so there is plenty of time for any issues to
> > > > > surface. I've placed a new build on
> > > > > http://people.apache.org/~brianf/2.0.8
> > > > >
> > > > >
> > > > >
> > > > > This new build fixes MNG-2277 in addition to the previous
> > > > > build. Please try it out and report any new issues found. If
> > > > > no new showstoppers occur, then I'll do the official build
> > > > > later this week / early next week.
> > > > >
> > > > >
> > > > >
> > > > > Thanks,
> > > > >
> > > > > Brian
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]



-- 
J. Daniel Kulp
Principal Engineer
IONA
P: 781-902-8727    C: 508-380-7194
[EMAIL PROTECTED]
http://www.dankulp.com/blog

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to