gump-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Adam R. B. Jack" <>
Subject Re: directory-naming
Date Mon, 09 Feb 2004 16:47:23 GMT
Stefan wrote:

> I agree, but I'm currently not able to make Gumpy use an installed
> Maven as I'm neither familiar enough with Python nor the Gumpy code.
> The minimum requirement obviously would be that Gump(y) uses an
> installed version of Maven but this installed Maven will never
> download any jars and use the CLASSPATH or any other configurable
> source of jars when building a project - even if the project states
> some dependencies on specific released versions in its project.xml.
> I understand from you past posts that this should be possible, but I
> don't know whether we have all the facts on how to make it possible.

I believe I got a fair way along with this, I create the Maven project
properties override file, and in it I try to override jar handling. I'll try
to focus on this again soon.

Truth though, I'm kinda scared to embark on this. There have been issues
with the 'simple' interface between Gump and Ant, and I'm very nervous about
subtleties in using Maven causing major grief.

That said, perhaps I just try it and see what occurs...

> Making Maven build within in Gump would be important as well IMHO, but
> if this isn't possible for any reason, so be it.

Yes. I feel passionately about that. It is partly that Maven doesn't build
on Gump that worries me about above. I'm not saying this is anything to do
with the Maven community --- it is dependencies:

... I'm just saying that the Maven community don't directly interact w/ the
Gump community (like that Ant community do.) That said,

Also, Gump uses latest cvs/svn/forrest (not self built) [despite me
rathering it didn't] so until I fix all them, I can't really see a reason
not to try w/ Maven.

> On my "traditional" Gump installation the project maven-bootstrap
> doesn't start a compilation because commons-jelly-tags-ant fails.
> This particular project fails as one of the unit tests fails because
> of a hard coded absolute path that doesn't exist (this is what it
> seems to be to me).

We have a bit of a ways to go [see states of dependencies], but I'm game to
work with folks to folcus on this goal:

Q: Do we know if the Maven descriptor is up to date? Since it hasn't been
used it could be stale. I wonder if we ought set failed dependencies to be
optional, to see what happens.



View raw message