Hi Julien,
I agree with all your points, especially about dependencies. There are
several very old embedded-source style dependencies that would benefit from
using Maven.
> I had to restructure the project a bit to follow the Maven directory
> structure.
I think we can avoid doing this sort of restructuring, given the current
directory structure of VisAD. Thanks to Bruce's earlier efforts, components
are already nicely isolated in src folders, and it is very easy to override
the default Maven directory structure with one line of XML; e.g.:
<sourceDirectory>${project.basedir}/src</sourceDirectory>
I would suggest we do things this way, to minimize disruption of the VisAD
codebase.
Regards,
Curtis
On Wed, Jun 6, 2012 at 11:23 AM, Julien Chastang <chastang@xxxxxxxx> wrote:
> Hi.
>
> Yes, I maintain a Mavenized VisAD mirror at
> https://github.com/julienchastang/VisAD. I had to restructure the project
> a bit to follow the Maven directory structure. I also had to prune out some
> (I suspect obsolete) source in order for VisAD to build with Maven. See the
> note at the bottom of the README. This effort took < 2 hours of work.
>
> VisAD development seems to move in "waves", and I wait for a wave to
> complete before resyncing the projects. (In other words, there are a few
> recent commits that I have not grabbed yet. Waiting for things to settle a
> bit.)
>
> I think a more official move to Maven and github would benefit the VisAD
> community for various reasons.
>
> . Giving new and existing VisAD users an easy avenue to incorporate VisAD
> into their project via Maven mechanisms (both through mvn build and access
> through Maven Central). (The THREDDS project does this now, internally in
> Unidata)
> . Maven is the de facto standard project management tool in the Java
> ecosystem. Once a project has been Mavenized, other tools like continuous
> integration software can take advantage of this common baseline. It also
> gives developers outside the project an idea of what they are dealing with
> and where to start poking around.
> . Opportunity to clean out cruft. For example, I think there is a
> netcdf-java dependency in VisAD that is extremely out of date. Also,
> current VisAD dependencies are brought in with the raw source, rather than
> external dependencies (e.g. Jama). This would have to be cleaned up.
> . In my experience github has a more vibrant open source community,
> leading to greater collaborative development. Many many open source
> projects have been moving to github.
>
> I would be very happy to help out in this area if there is any or
> sufficient interest. Having a Mavenized VisAD available on Maven central
> would be a good goal.
>
> In the meantime, I will continue to maintain the VisAD mirror at
> https://github.com/julienchastang/VisAD.
>
> Just my $0.02
>
> -Julien
>
> On 6/5/12 6:08 PM, Tomas Pluskal wrote:
>
> Hi Curtis,
>
> Another option would be to use Unidata's existing Maven repository (
> https://artifacts.unidata.ucar.edu/). Then you could deploy JARs there
> with a little less bureaucracy than Maven Central has. Of course, artifacts
> would also be less visible, as they would not be discoverable on search
> aggregators such as mvnrepository.com.
>
>
> That is of course an option. However, I cannot help with that, so the
> effort has to be taken by someone who has access to that repository.
>
> On the other hand, if you decide to deploy to Maven Central, I can do
> the setup and provide the complete ant tasks for that.
>
> Best,
>
> Tomas
>
>
> ===============================================
> Tomáš Pluskal
> G0 Cell Unit, Okinawa Institute of Science and Technology Graduate
> University
> 1919-1 Tancha, Onna-son, Okinawa 904-0495, Japan
> TEL: +81-98-966-8684
> Fax: +81-98-966-2890
>
>
>
> _______________________________________________
> visad mailing listvisad@xxxxxxxxxxxxxxxx
> For list information, to unsubscribe, visit:
> http://www.unidata.ucar.edu/mailing_lists/
>
>
>
> _______________________________________________
> visad mailing list
> visad@xxxxxxxxxxxxxxxx
> For list information, to unsubscribe, visit:
> http://www.unidata.ucar.edu/mailing_lists/
>
>
