El 11 de febrero de 2011 14:48, Ray Ryan <[email protected]> escribió:

> If Dave has already LGTM'd one of the approaches as maven friendly, is
> there any reason not to go with that? What are the trade offs?


No LGTM from David Chandler yet.

But aside from that, from what David and I have discussed, both approaches
are Maven friendly. The only remaining points of discussion seem to be
maintainability and possible (yet unknown) conflicts with external versions
of  the validation API jars. Bundling the classes as done here forces every
GWT user to compile with our version of javax.validation (so far the only
one AFICT, and the user can still use any implementation on the server).

The other Issue (1342803) touches more places in our ant build, but keep the
validation jars separate and thus allow our users to easily choose to use
other validation APIs (as long as the API is backwards compatible with
validation-api-1.0.0.GA).

I personally believe in modularity and de-bundling, so I lean toward the
later option (Issue 1342803). In any case I'd be happy with either an LGTM
here (issue1323803) from David or an unopposed LGTM in issue1342803.

-- 
http://groups.google.com/group/Google-Web-Toolkit-Contributors

Reply via email to