[buckminster-dev] Re: Maven integration available in public site

[Carlos Sanchez <carlos.sanchez@xxxxxxxxxxx>]

Thomas Hallgren wrote:
> Carlos Sanchez wrote:
>> Thomas Hallgren wrote:
>>> Hi Carlos,
>>> I'm not able to connect to the SVN repo. It requires a password.
>>
>> sorry it's without the s
>> http://q4e.googlecode.com/svn/trunk
>>
> Thanks. I was able to get it into my workspace and compile it. Any
> chance I can get committer access to this repo so that I can keep it up
> to date with the changes in Buckminster 0.2.0?

I have to look into this because as we are trying to move it to Eclipse
there may be some concerns with the IP

> 
>>> Does this mean that all of Maven 2.1 is finally released? The IP-process
>>> for getting it all into Eclipse still awaits that release.
>>
>> no it uses the revision 549510 in apache svn. It is an "internal"
>> release that we did after some testing.
>>
> Any word on when the final release will be out?

unfortunately nothing is clear yet

> 
> 
>>> What Buckminster version is this running with?
>>
>> I built it against the latest trunk, haven't tried it for several weeks
>> already but at least now other people can give it a try easily and
>> provide feedback
>>
> OK. Here's some feedback. I tried running our demo:
> 
> http://www.eclipse.org/buckminster/samples/queries/demo.cquery
> 
> but it fails on:
> 
>   Provider maven2(http://repo1.maven.org/maven2): No match found for
> component se.tada.util.sax using version designator 1.0.0
> 
> The demo uses this rmap:
> 
>   http://www.eclipse.org/buckminster/samples/rmaps/demo.rmap
> 
> Since that particular component uses name mapping, I also tried to
> explicitly resolve the component name "se.tada/tada-sax" and an rmap
> that does not use name-mapping:
> 
>   http://www.eclipse.org/buckminster/samples/rmaps/dogfood.rmap
> 
> but that did not solve the problem so I did some debugging and found two
> issues:
> 
> First issue:
> Running the demo I could see that the
> Maven2VersionFinder.getComponentVersions() correctly finds the entry
> groupId "se.tada", artifactId "tada-sax". The call artifactVersions and
> availableVersions on the artifact correctly reports  [0.2.1, 1.0.0].
> 
> The subsequent call to resolve() on that artifact results in a
> ArtifactNotFoundException which is silently caught.
> 
> The version matching still takes place though, and it fails. This is
> because the MavenComponentType.createVersion() returns a version of type
> "String". The query is for a "Triplet" and they don't match.
> 
> At the end of the createVersion() method I see the following comment:
> 
> /* Maven versions are not exactly triplets, that's why we can't use the
> triplet type */
> //return VersionFactory.TripletType.fromString(versionStr);
> return VersionFactory.StringType.fromString(versionStr);
> 
> The Triplet type was created especially to accommodate the Maven
> versions, so obviously something needs to be done here :-)
> 
> What is it that made you reject the Triplet? How can we fix that?

IIRC there were problems with Maven snapshots
For instance some changes we made was that any snapshot was considered
"latest" when they'd be more accurately branches. You could have
1.0-SNAPSHOT and 1.1-SNAPSHOT

> 
> Second issue:
> When trying to resolve the "se.tada/tada-sax" component directly (i.e.
> without use of name mapping), I discovered that your MavenProvider
> implementation is outdated. After meeting you guys at EclipseCon we
> realized that a viable alternative to the RMAP name mapping was high
> prio and our solution was to allow component names like
> <groupId>/<artifactId> all the way. That in turn, affected the
> getDefaultGroupAndArtifact() method.

I think it was changed here to use dots as separator to be the same as
eclipse usual plugin names

> 
> Kind Regards,
> Thomas Hallgren
> 
> 
>