Feedback for the network kioslave, please

Two things that cheered me up today:

1. A really interesting interview with two developers of Étoilé. I think they are so right with their approach. The application-centric approach sucks a lot and is one of the biggest culprits of the sad state of the personal computer interface still after four decades since the so-called mother of all demos, from 1968 (sic!). Since I first heard from that project three (already?) years ago I have been checking their homepage for news every week 🙂 Definitely a project to watch and learn from. If you look closely at the Okteta sources, the Kakao part, you might even see where I try to lay the base for something remotely similar to CoreObject. And a project to join? Hm, if I would start with FLOSS I would definitely try to work in that project. But I am now engaged with KDE, which also is going some good directions 🙂

2. A really big Multiseat deployment in Brazil, with up to ten(!) workplaces per mainboard. No idea, if they are using KDE. But I feel so inspired by the technical approach, and still feel sorry I once failed to setup a multiseat myself (due to X drivers of the graphic cards surprised to not be the only graphics card on the same PCI bus and by this confused, or something, memories are fading), so I could never show it at the local Linux-Info-Tag Dresden (BTW: Will you be at the Chemnitzer Linuxtage, 14.-15. March? KDE will be there, thanks to Palapeli-Stefan and others).

If they are using KDE, I am wondering what the output of the network kioslave would be and how it could be improved.

Now, you can cheer me up, too:
Just give some ideas how you would want network kioslave to work like. Go and try it, should also compile with KDE 4.2:

svn co svn://
mkdir build
cd build
cmake ../networkkio
make install

This will install a kded-module for kio-updates on changes of the network, a solidnetwork library, the network-kioslave and some mimetypes for known services.
Some of the services are for now simply subtypes of the types directory or symlink, so clicking on it should forward to the url that is attached to them. You can also redefine the handling with the usual file association dialog.

Even better, join the coding, I welcome everyone to write the backends for UPnP, WebServices, Lisa-like scanning and what else there is that could be shown in the network.
I also would like better code for the device type detection. Are there any services which would help to see if a host is a big server, a laptop, a netbook, a router? Perhaps even the model, like Producer Cooldevice 2010? Currently I plan to go for heuristics by found services, but I wonder how it is done in Windows or OS X if listing other devices in the network.
Then a config would be nice, to limit the number of service shown (and hide the unknown optionally). Also to define which other domains than the local network should be scanned.


6 thoughts on “Feedback for the network kioslave, please

  1. It’s so true about Etoile’s idea of just have lots of little bits composing an app, which is really just a workflow. Being stuck on islands of software really sucks! I hope more KDE apps pick up on this, mind you, the underlying libraries don’t entirely help matters.

  2. The only obvious problems I see are that it does not offer a way to launch applications in a customized manner for protocols not supported by KIO slaves.

    It also displays the computer under its own entry, leaving a useless entry.

    Otherwise, it is an excellent KIO slave I can use to connect to machines on my LAN, without even having to remember their names.

    Maybe support for more than just SLP as the backend ( UPnP maybe ) would be a idea for future expansion.

  3. @bcooksley: Yes, support for protocols not matched by kioslaves needs to be added somewhere in the KIO system (vnc:/ as a prominent example). For that purpose I’m already subscribed on kfm-devel, just need to come up with a good proposal, so I am still studying KIO.

    What do you mean with “the computer under its own entry”?

    And the only backend currently is DNS-SD, not SLP. The latter has already some code, but that is crashing (threading problems).

  4. The computers MAC address is a file underneath the computer “folder” itself is what I meant.

    I assumed the backend was SLP when it required me to install openslp-devel.

  5. Eh, so I forgot about the delay with SLP should be optionally as of that blog entry at least on, i.e. it worked for me. And the name of the device, if constructed from the name of the servicetype “workstation”, has now the [MAC] part stripped for it.
    So I hope this two problems can be tagged “solved” 🙂

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s