Linked by Thom Holwerda on Tue 29th Jan 2008 18:48 UTC, submitted by Michael Larabel
Thread beginning with comment 298255
To view parent comment, click here.
To read all comments associated with this story, please click here.
To view parent comment, click here.
To read all comments associated with this story, please click here.




Member since:
2005-07-06
GVFS/GIO are bottom-up rewrites of GnomeVFS by Alexander Larsson, I believe.
In a nutshell, GnomeVFS is one big, ugly hack. It's not very abstract, yet its place in the stack implies it's a high-level library ("gnome-*" librares are only useful in Gnome apps, not GTK or merely GLib apps). It's got a pretty crappy API (inconsistent, very few utilty functions, too dependent on Unix-isms). Many backends are broken and will bring down your whole app unexpectedly. It relies very heavily on the Unix filesystem model (too much for a VFS). It lacks such basics as atomicity, state (!!!), or a GObject-based model.
It was desperately in need of a rewrite.
GVFS will exist in the right place in the stack (GLib, the basic, framework libraries upon which others are built). It will have a generalized, abstract API. It will use GObject (this makes it easier to wrap for other languages, or use in an OO way in your GTK/GLib app). It will maintain state (for authentication, caching, data sharing, connection maintenance, etc). And perhaps most importantly, backends will no longer be in-process (they can't crash your app if they go down).
So yes, it is big, and it is cool. I, for one, can't wait.