Is someone working on #41620 ?

[Germain Garand]

Le Vendredi 19 Juillet 2002 18:22, David Faure a écrit :
> -----BEGIN PGP SIGNED MESSAGE-----
> > 1) bool KHTMLPart::processObjectRequest() should be in sync with this
> > change, since that's what is used to create parts in frames (see
> > BR#43393) (same "if" test on mimetype at khtml_part.cpp -l.2664)
>
> One can't ask for another viewer than the preferred one, in khtml.
> Ah, but if the current viewer handles the new mimetype but isn't
> preferred.... I see. Ok, please test attached patch.
>

Works perfectly ... #43393 is gone !

> > 2) What about the case where the new mimetype is different, but the
> > preferred part for the new mimetype finally proves to be the same as the
> > current one ? e.g: you display a "text/plain" file, then a "text/css"
> > file...
> >
> > /me think konqueror should then reuse the current view, instead of
> > creating a new one of the same kind ?
>
> You are very right!
> Please test attached patch for konq_view too ;)
> (I'm busy with so many other things right now....)
>

It works also very well and crushes #41620 :-)
Though to nittpick like crazy,  the:
 
   kdDebug(1202) << "Giving focus to new part " << m_pPart << endl;
    m_pMainWindow->viewManager()->setActivePart( m_pPart );

statements that follows should probably be displaced into the "else" 
(otherwise you get a "Warning! part is already active")

> Ah, a similar test for KHTML is missing. A bit harder. I think we need to
> pass the current serviceName to createPart so that it skips creating a new
> part if the old one is of the same service. Hmm, to be checked.

Mhhh true... but then you'd have a hard time knowing afterward if the returned 
pointer is a new part, the same part or an invalid pointer ?
Konqui's implementation seems cleaner to me: separate the trading from the 
actual creation...

Bon, au lit, je meurs

G.