Thilo Goetz wrote:
Hi Marshall,
Marshall Schor wrote:
OK. Am implementing tests and then the fixes :-)
please don't commit anything to trunk until we're done
with this release. Else we might have to branch now,
and I'd like to avoid that if we can.
oops - sorry - didn't see this note, until after I had committed. For
these kinds of things,
best to send an instant message :-)...
2 more cases have come up:
1) One descriptor specifies multipleReferencesAllowed as false, the other
"omits" this. The spec says omitting is the same as false. So this
will be OK.
2) One descriptor specifies an element Range Type restriction, the other
doesn't specify.
I'm making this throw an exception, per the logic for #2 below in
Thilo's note.
Not specifying a component type restriction is the same as specifying
uima.cas.TOP.
OK - I guess this will need to be added to the test - to allow no type
restriction to "match" one which specifies explicitly uima.cas.TOP.
I'll re-open the Jira issue.
-Marshall
