Re: [Geotools-devel] Using the SDK build, trampled by that there ol' CLASSPATH

[Jody G. <jga...@re...>]

Martin Desruisseaux wrote:
> Jody Garnett a =E9crit :
>>> Any other ideas? Any wisdom? Any magic spells to restore sanity to th=
e
>>> class loader mechanism?  =20
>> Ant!
>
> Not needed, provided that next geotools-bin*.zip zipball are produced=20
> with Maven 2 (in which case the user would not be required to use=20
> Maven 2 himself). The trick is:
The original email spelled out that maven (of any description) was out=20
of bounds, and I suspect that Ant is too heavyweight as well. That said=20
I am not sure we *can* get geotools into a bite size gulp for Mr Custer.=20
Of course not sure if we should either ...
> Note: is is also possible to instruct Maven 1 to fill the "Class-path"=20
> entry in META-INF/MANIFEST.MF file, but it is a little bit more=20
> complicated than Maven 2. Do we want to put more energy into Maven 1=20
> build, or just try to finish the switch to Maven 2?
The maven 1 build is dead Martin, we are switching uDig over to grab=20
jars from the maven 2 Repo etc...  GeoServer is switching over this week=20
as well, I did run into a bunch of trouble that only occurred when=20
*using* geotools jars, mostly in terms of "excludes" that became=20
required when importing more then one geotools jars. Suspect that we=20
need some "dummy" modules that just catch common geotools configurations.=
..

Cheers,
Jody