[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-core-devel
Subject:    Re: PATCH: slavebase.cpp [Correct patch]
From:       Dawit Alemayehu <adawit () kde ! org>
Date:       2001-07-23 22:37:40
[Download RAW message or body]

On Monday 23 July 2001 15:39, Waldo Bastian wrote:
> On Monday 23 July 2001 06:34 am, Dawit Alemayehu wrote:
> > Sorry to reply to my own post.  But  I cannot figure out why the ssl-*
> > metadata info gets lost.  Following the trail back to the TransferJob, it
> > looks like all incoming data is actually accumulated properly in there.
> >
> > Here is what happens:  The ssl-* metadata gets set before we send the
> > mimetype signal to the application since it is obtained during the
> > process of connecting to the remote machine.  However, in khtml we do not
> > look for these ssl-* meta-data until we receive the actual data from the
> > remote server, i.e. in KHTMLPart::slotData. Anyways, between these two
> > processes, i.e. the mimetype signal being called and the actual data
> > being sent, the ssl-* meta data gets lost!  I just cannot figure out
> > where since in the application side it looks like the incoming meta-data
> > is accumlated properly.  Any ideas Waldo ?  This was what I was trying to
> > resolve with my previous patch which of course was incorrect, but the
> > problem still remains...
>
> What might happen is that the slave is passed over to another job. In that
> case the meta-data will have to be transmitted again from the slave to the
> application, to make sure that the new job has the metadata as well.

Ahh... I now see why it did not matter when I connected to the mimetype signal 
of the job in KHTMLPart.  You got to love all the built-in gottcha's :)

> But if you look at SlaveBase::mimeType() you will see that that is exactly
> what happens, the metaData is sent each time the mime-type is send, and
> after the slave is transferred to another job, both the meta-data and
> mimetype are send again.

Hmm... how can that be possible ?  I mean the meta-data being re-sent when
the slave is transferred to another job ? The outgoing meta-data list gets cleared 
not once, but twice in SlaveBase::mimeType().  It gets cleared in ::sendMetaData() 
the first time and then again when the do loop terminates in ::mimeType().

> I will try to see if I can reproduce the problem. Do you have a SSL
> test-site that I can use?

Nope, but just go to any SSL site like https://sourceforge.net and you will see.  I put 
enough debug statements in all these classes to see that the meta-data indeed reaches 
the application correctly which then should result in it being cached.  However, it is not 
there when we actually get around to looking for it in KHTMLPart::slotData().  So to me
your above explanation about the slave being passed over to another job makes sense
since the patch I reverted fixed the problem. :)

Regards,
Dawit A.

[prev in list] [next in list] [prev in thread] [next in thread]