kio_giobridge IO slave
nf2 at scheinwelt.at
Mon Jan 7 15:08:45 GMT 2008
Allan Sandfeld Jensen wrote:
> On Thursday 03 January 2008 03:35, nf2 wrote:
>> Over the Christmas holidays i have written an experimental kio_gio/gvfs
>> bridge IO slave for KDE4:
>> The kprotocolinfo system has been extended to allow a "multi protocol"
>> slave (protocol = "*") which receives all calls for protocols not known
>> to KIO. Furthermore there is a "OverrulesProtocols" parameter to
>> override some protocols which KIO already provides.
> Rather than making it a multiprotocol slave, why not let it handle one slave
> at a time? The fallback model is going to create a lot of confusion. There is
> no reason to install all of kdebase if you are not going to use it all, most
> distributions split it up by now. That way you could install gio based
> version of SMB or native kio based, users choice. Not the surprise of not
> having installed the native kio based, and then have odd interaction bugs in
> the smb protocol.
Let me explain the reasons for having a multiprotocol handler:
* It's not known, which protocols GIO/GVFS actually handles. You might
install additional GVFS protocol handlers which KIO doesn't know about
and probably expect them to be available in KDE apps also. Therefore i
think that delegating all unknown protocols to the "kio_giobridge" makes
* As kio_giobridge can handle multiple protocols - why launch multiple
instances? You only need one kio_giobridge process, unless there are
I understand that KIO hasn't be designed for the situation of having
more than one slave for the same protocol. I agree that protocol=* in
combination with OverrulesProtocols= is not a beautiful solution but it
makes it possible to develop and test kio_giobridge. In the long run i
agree that it would be better to fix that with packaging (install one or
the other). Though that's difficult, cause not all handlers which GIO
should replace are in kdebase. ftp - for instance - is in kdelibs.
More information about the kde-core-devel