On Mon, Sep 21, 2009 at 12:28, Felix Meschberger <[email protected]> wrote: > I agree also with Ian's POV (and have also argumented the same way in > the past), the unless there is an actual resource around we cannot tell > where the name ends and the selectors/extension/suffix starts ...
Of course this is just a "guess", but it is the same guess that the non-existent servlet handler (which is one single-big thing at the moment) would have to do itself. And in 99% cases this guess will be the right one - for the other 1% the handler can do it himself. But I wonder how such a case (the 1%) should be handled: it's all about resources that contain a dot inside (because the path will force path info parsing starting at the first dot), and programming against that case sounds a bit crazy to me: if this post to a non-existing URL starts with xyz.ext.bla (which does not exist!), then cut off at the third dot... So far I haven't seen a real use-case where this patch would introduce a problem (that is larger than the very useful gain it brings, namely to allow for selectors/extensions on those requests). > Create a generic servlet handling requests to non-existing resources. Introducing another servlet/mechanism for that only introduces more complexity that isn't needed IMHO. Regards, Alex -- Alexander Klimetschek [email protected]
