Re: Writing a gvfs backend

[Mark Ellis <mark mpellis org uk>]

On Sat, 2008-05-10 at 09:11 +0200, Alexander Larsson wrote:
> On Fri, 2008-05-09 at 11:29 -0600, Boyd Timothy wrote:
> > Hey Alex,
> > 
> > I wondered if you could give me a couple pointers about writing a gvfs
> > backend.  I'd like to write one using something like python which only
> > exposes its interface via DBus.  Where can I look to see which
> > specific DBus Interfaces I need to implement, complete with their
> > method signatures?
> 
> You can't do this at the moment. The dbus interfaces are not totally
> stable, and additionally file i/o is done using a custom binary protocol
> over a pipe. We have deliberately not stabilized either the dbus
> protocol or the daemon side C interface because we're still tweaking it
> as we learn how things work.
> 
> The way you currently implement gvfs backends is by subclassing
> GVfsBackend and implement the operations there. I'd recommend you look
> at the current ones to see some examples.
> 

Hi All

This is something I'd been meaning to ask myself.

I've recently implemented a backend for the synce project, to access the
filesystem of a Windows Mobile device. To do this I've had to reference
a built gvfs tree for libdaemon, the headers, and daemon-main*. It works
fine, but is obviously a little clumsy, and highly inappropriate for
release to an automated build system.

Are there any long term plans for 3rd party backends, such as exposing
the headers and libdaemon in an install ? The daemon-main* code is small
and boilerplate enough to replicate into a 3rd party project, but
obviously the rest would be excessive and a nightmare to keep in sync.

Mark