> > > That's the problem at the moment. I'm still undecided as to what goes > where. > > > The 'secure' flag may be Catalina specific (and the processor has no > clue > > > whether or not the connection is secure; it's the connector which knows > > > that), so I moving it to the adapter. > > > > I was just asking. :) 3.3.x has the flag as well, since isSecure() is also > in 2.2. It's just that I wasn't planning on exposing the o.a.c.Request to > the rest of Tomcat so if nobody was ever going to look at the flag, I could > ignore it.
Maybe I'll put it back then :-) > > Catalina uses the "secure" property of the current HTTP connector to give > > the right answer to ServletRequest.isSecure(). I imagine the same thing > > will need to be done for Coyote (and the "secure" flag would also need to > > be sent along for a JK request because it can vary each time???). > > > > > The API may change a bit. Maybe it will become closer to the one used by > TC > > > 3.3. I hope you don't mind the API stability problems too much ... > > > Hopefully, it will be stable soon. > > > > I don't mind at all. :) Good. I did a few changes, so that the 3.3 adapter can use the tc-util provided cookie and parameter parsing. The TC 4 adapter doesn't use that cookie parser, because there's no incentive performance wise (the object would have to be wrapped by facades, which would create some GC). Given the tester results (only one 'real' failure, on a really nasty test), I'd say the new HTTP stack is ready for some testing. The next TC 4 nightly should have a usable version. To use it, install the nightly, and access port 8081 instead on 8080. Remy -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
