a common VFS - a different approach

Ken Deeter ktdeeter at alumni.princeton.edu
Tue Sep 23 12:03:55 EEST 2003

> So, we are not in fact talking about a common vfs library, but rather a
> vfs implementation separate for each project, and a way to implement
> third party modules that can be used by both desktops?

Well, I don't entirely disagree. If we could just simply get
interoperable modules, then that would be a huge step forward in my

One difference though: the imagined development (or at least how i
imagined it) would have 'most' modules having a common single
implementation, and having things like 'start-here:' and 'preferences:'
be desktop-specific add-ons. The goal should be to share as much as
possible. If we wan't URI's that only one desktop would use, then
there's no reason to not support them. They are desktop specific
add-ons, and I don't think that means we need two separate
implementations just to support those.

I still think there are certain areas where more than just backend
modules would help. Vfs chaining, supporting high-level operation
through the automatic composition of lower-level ones, file monitoring.
These kinds of things I think are still hard to do, and so we should
strive to only require that they be done once, and correctly. But I
suppose there's nothing that says we can't do this after we have the
basic backend modules going.

Also, having a non-desktop specific implementation lets non-desktop
specific apps use the functionality. Personally I think it would be
great to have something like bash that worked with URI's. I don't think
I should have to depend on gnome or qt to write that though.. (although
if there are only glib and qt library wrappers available, then maybe
that is unavoidable)

Maybe d-bus isn't the answer, but if we're gonna get the asynchronous
stuff right, then it has to be separate for both qt/glib, or it has to
be duplicated like it is with d-bus. I just hope we don't have to
reinvent this wheel again. Or perhaps create the freedesktop threading
and signal/slot library. /me runs

Seems like the activation stuff, and the service ownership stuff is
perfect for trying to address some of the features we need (like a
flexible authentication mechanism).



Ken Deeter (Kentarou SHINOHARA)
Graduate Student, Computer Science
University of British Columbia

  "If only God were alive to see this.. "
                              -Homer Simpson

More information about the xdg mailing list