Just to let you know: yes, that fixes the problem. The CVS version reports the error cleanly (well, still with a big stack trace, but that's not your fault ;-). Thanks for the pointer!

Something it doesn't seem to fix (which is not so much a bug as an 'I thought it would work this way' thing) is that 'name' attributes on ojb.field tags are ignored. I know you can infer a name based on the name of the Java property being mapped, but if I explicitly specify a different name I'd like it to be used. Is there a way to get that to happen?

L.

Thomas Dudziak wrote:

On Thu, 11 Mar 2004, Laurie Harper wrote:


Using the xdoclet stuff packaged in db-ojb-1.0.rc5-contrib:

If the element-class-ref attribute is missing from the @ojb.collection tag, an internal error occurs which is not handled and reported clearly. You should get an error message like 'Could not determine collection element type; missing element-class-ref attribute. Instead you get a huge stack trace which doesn't communicate the problem:


Please use the CVS version of the XDoclet module which is way improved
over the rc5 version (you don't have to use OJB from CVS, only the XDoclet
module). You can get it from the ViewCVS access from the OJB website
(http://cvs.apache.org/viewcvs/db-ojb/). Download the three jars
xdoclet-1.2.jar, xdoclet-ojb-module-1.2.jar, and xjavadoc-1.0.2.jar from
the lib folder, and also the documentation (xdoclet.xml) from the xdocs
folder (html is build via 'ant htmldoc').
If the problem persists, please post the stacktrace that the CVS
module version generates.
As for the stacktrace, I can't do much about it, XDoclet 1 is not known
for its user-friendly error messages. A conversion to XDoclet 2 is
planned for OJB 1.1 though.

Tom



--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]



Reply via email to