Hello Experts,

Our TCK developer discovered a corner case in the 73-MappingDiscovery
spec and impl: what to do about the case of getNamedDispatcher()?

I propose we do the same approach in this case as we do for the "9.3.1
Included Request Parameters" and "9.4.2 Forwarded Request Parameters"
sections: declare we must not produce a HttpServletMapping instance if
the "servlet was obtained by using the getNamedDispatcherMethod". This
will be specified in the javadoc for
HttpServletRequest.getHttpServletMapping() by adding a paragraph after
the first paragraph that states:

If the servlet corresponding to this request was obtained using the
getNamedDispatcherMethod method, this method must return null.

ACTION: Since we are now after Pubilc Review, I would like to request a
quick response on this. Please reply by COB GMTs0500 Friday 16 June
2017. No response indicates you accept this proposal.

Thanks,

Ed

--
| edward.burns@... | office: +1 407 458 0017

Another alternative would be to add a MappingMatch of NAMED, use
getServletName() to return the name, and have pattern and matchValue
return null (or the empty string).

I think this could be more useful, however it is a larger change which
may be an issue this late in the process.

Stuart

On Thu, 15 Jun 2017 02:55:07 +0200, "Stuart Douglas" <sdouglas@...> said:

SD> Another alternative would be to add a MappingMatch of NAMED, use
SD> getServletName() to return the name, and have pattern and matchValue
SD> return null (or the empty string).

SD> I think this could be more useful, however it is a larger change which
SD> may be an issue this late in the process.

Yes, it is too late. I stick by my existing proposal.

Ed
--
| edward.burns@... | office: +1 407 458 0017

--