I too think it might be interesting implementing pipelining, but don't know how hard it may be.
By the way, HttpClient package is a nice work, thanks. > -----Message d'origine----- > De : Evert Hoff [mailto:[EMAIL PROTECTED]] > Envoy� : vendredi 7 juin 2002 09:02 > � : [EMAIL PROTECTED] > Objet : [HttpClient] Pipelining > > > Hi, > > Is there a specific reason why pipelining hasn't been > implemented, other > than not having round tuits? In other words, can anyone think of a > reason why it shouldn't be implemented? > > I am referring to pipelining as described in section 8.1.2.2 of the > HTTP/1.1 RFC2616. > > At the moment the HttpMultiClient waits until a connection is > available > and then the method executes on that connection, doing its reading of > the response immediately after writing its request. > > An alternative would be to let it function as it presently does until > all connections are busy, and then to start queuing requests. Then, as > soon as a connection becomes available, it looks in the > queue. If there > are queued requests, ALL the requests are written onto the connection, > and thereafter the responses are read in the order in which > the requests > were made. > > I'll give this further thought if you think this is something > we should > do. > > Regards, > > Evert > > > > -- > To unsubscribe, e-mail: > <mailto:[EMAIL PROTECTED]> > For additional commands, e-mail: > <mailto:[EMAIL PROTECTED]> > -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
