Alex Zepeda wrote:
> On Mon, 14 Jun 1999, Kurt Granroth wrote:
> > then it is almost surely a POST request.  That's what I was referring
> > to -- typically, POST requests are not 'arguments to the URL' while
> > GET requests are.
> 
> How can I put this.  I know.  But I'd be hesitant to support a change to
> KURL (say adding KURL::post(name, data)) that would only be useful to
> http.

Right!  Anyway.. I wasn't suggesting doing anything with KURL.  AND it's
all a huge moot point now -- with your and Torben's help, I think I
finally have this entire thing licked.  A few minor changes to libkio
(committing as we speak...er.. write.. er.. read.. whatever) and the
entire framework will work like a charm.

However, kio_http can't support Put (in general) or POST (in specific)
as it is now.  I am in the process of re-writing it in a semi-"proper"
way.  It works for me now... but I'm going to test it some more before
submitting it here for review.

Minor digression: what do you think ready() should be used for?  It's
typically used like so:

  ready();
  data(..);
  dataEnd();
  finished();

In other words, "I'm ready to SEND data".  However, emails with Torben
and digging through the code AND implementing PUT lead me to wonder if
that is wrong!  I'm thinking now that ready() means "I'm ready to RECEIVE
data".

This is what it looks like to me: an ioslave calls ready() to signify that
it needs some data from the client.  This emits a sigReady() (at least it
does now!) that will be intercepted by the client.  The client will then
do

  data(...);
  dataEnd();

and the ioslave will get that data in slotData(..).  When the ioslave's
slotDataEnd() is called, then it should send back a finished().

Thoughts?
-- 
Kurt Granroth            | granroth@kde.org
KDE Developer/Evangelist | http://www.pobox.com/~kurt_granroth

        KDE -- Putting a Friendly Face on Linux