From the recent bug report about wrapped responses vs .forward() I did this query: Search on description "wrapped request forward"
http://nagoya.apache.org/bugzilla/buglist.cgi?bug_status=UNCONFIRMED&bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&bug_status=RESOLVED&bug_status=VERIFIED&bug_status=CLOSED&email1=&emailtype1=substring&emailassigned_to1=1&email2=&emailtype2=substring&emailreporter2=1&bugidtype=include&bug_id=&changedin=&votes=&chfieldfrom=&chfieldto=Now&chfieldvalue=&product=Tomcat+4&short_desc=&short_desc_type=allwordssubstr&long_desc=wrapped+request+forward&long_desc_type=allwordssubstr&bug_file_loc=&bug_file_loc_type=allwordssubstr&keywords=&keywords_type=anywords&field0-0-0=noop&type0-0-0=noop&value0-0-0=&cmdtype=doit&newqueryname=&order=Reuse+same+sort+as+last+time
Turned up each of the bugs in question ... http://nagoya.apache.org/bugzilla/show_bug.cgi?id=8566
The resolution and comments are correct, apparently. We don't quite do that right now, and it seems fixed if you looks at 4.1.x behavior.
http://nagoya.apache.org/bugzilla/show_bug.cgi?id=11335
That one seems bad with respect to the spec (no wrapping is supposed to occur, so we reflect whatever path the wrapped request wants to). There's a contradiction, though. That bug probably doesn't exist anymore with 4.1.x since we wrap.
http://nagoya.apache.org/bugzilla/show_bug.cgi?id=16032
That's probably fixed now (not too sure).
http://nagoya.apache.org/bugzilla/show_bug.cgi?id=9754
That's a dupe.
http://nagoya.apache.org/bugzilla/show_bug.cgi?id=11366
This likely doesn't exist in 4.1.x.
http://nagoya.apache.org/bugzilla/show_bug.cgi?id=20314
Are all of there dups of one another?
The spec contradicts itself, basically (I'm reading Servlet 2.4 PFD 3). I posted messages complaining about it, and (unsurprisingly) got ignored. Sun spec folks are really good about ignoring people (even if you're a coworker) unless:
- you're on their expert group, and start complaining really loud (I'm not on the expert group - no time)
- you have access to the Sun Bugtraq and file a P1 bug (that one works so great, it's amazing :-D unfortunately, I have no access to it :-( )
There are RD requirements that would be hard to implement without wrapping (all those special attributes, and parameter isolation). IMO the parameter isolation is impossible if the request is not ours (or I don't know, maybe we have to extract everything before invoking the pipeline if a request, and then restore them one by one). So having a query string will make the RD dead slow (but the general case will be faster if we can get rid of wrapping).
Section's 6.2.2 wording is bad with respect to the new spec: the request dispatcher now invokes a filter pipeline, not the servlet itself.
I'm -1 for trying to change the RD behavior in TC 4.1.x (reaslistically, we'd break things trying to fix it, and it is clearly not worth the pain for the stable branch). If the spec is clarified, then we can implement the specified behavior in TC 5, otherwise I'd prefer keeping the current behavior.
Remy
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
