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

List:       kde-core-devel
Subject:    Re: KRASH release
From:       Reginald Stadlbauer <reggie () troll ! no>
Date:       1999-09-14 20:45:05
[Download RAW message or body]

On Tue, 14 Sep 1999, Dirk A. Mueller wrote:
>David Faure <faure@kde.org> wrote:
>
>>> e) Examine how much of KIO is obsoleted by Qt 2.1
>> I heard of "more network transparent", but do you think
>> that's in such a way as to make kio obsolete ??
>
>Well, the kioslave-concept maybe.

As I just said, both ways are too different. In KDE we have the kioslaves (each
protocol is a program). In Qt you will implement a protocol as a class derived
from QNetworkProtocol and e.g. to tell Qt to use that you say something like
(QFtp is a FTP implementation)

QNetworkProtocol::registerNetworkProtocol( "ftp", new QNetworkProtocolFactory<QFtp> )

to add e.g. a FTP implementation to the network protocol dict. If now Qt needs
an instance of a protocol it gets a new one using the network protocol factory.

So, e.g. if you need a instance of an FTP protocol, you do 

QNetworkProtocol *p = QNetworkProtocol::getNetworkProtocol( "ftp" );
which does besides some other things

QNetworkProtocolFactoryBase *factory = qNetworkProtocolRegister->find( protocol );
if ( factory )
    return factory->createObject(); // creates a new instance of the protocol
                                                // implementation
return 0;

So the getNetworkProtocol() returns now in our example an instance of QFtp or
NULL if no ftp protocol is registered.

So, you see these things are quite different. I don't see any reason for
dropping kio. Also kio doesn much more than just the ioslave stuff. 

But as I already said, the design of the Qt stuff is not finished yet and also
not the implementation. But in the Qt 2.1 release everything where network
transparence is useful it will be possible.

-- 
Reggie

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