Re: Not OK computer:/// [was Re: Should Desktop = Home?]
- From: Alexander Larsson <alexl redhat com>
- To: Gabriel Bauman <gabe bravenet com>
- Cc: Rui Miguel Seabra <rms 1407 org>, "desktop-devel-list gnome org" <desktop-devel-list gnome org>
- Subject: Re: Not OK computer:/// [was Re: Should Desktop = Home?]
- Date: Fri, 08 Oct 2004 09:01:14 +0200
On Thu, 2004-10-07 at 14:09 -0700, Gabriel Bauman wrote:
> On Thu, 2004-10-07 at 08:46 +0200, Alexander Larsson wrote:
> > a) It'll require large changes to the gnome-vfs internals for little
> > reason.
>
> I didn't think it actually would require large changes to the vfs
> internals, since the computer:, trash: and applications: -style
> locations look like they are pretty much defined and provided by
> Nautilus. You can't actually access a file via the computer:/// protocol
> using the VFS, for instance - what would be the point? However, you are
> probably intimately familiar with the implementation, and I could very
> well be wrong.
gnome-vfs is modular on the method level, not inside one method.
Changing it is surely possible, but a large change.
computer:/// is very much accessible via the VFS. So it
applications:///. trash is atm a nautilus-side thing, but thats
changing.
> > b) The proposed items don't even look like uris that gnome-vfs can use.
> > Is that second part the hostname? Why is making up hostnames that are
> > not hostnames any better than making up uri methods? At least the gnome-
> > vfs uris work like rfc uris, even if some are not rfc defined.
>
> Actually, you're incorrect. In a Nautilus location box, try typing
> "computer:///". Now try typing "computer:". There is no difference here
> - both specify exactly the same protocol. The "///" actually refers to a
> *path* to be accessed via the "computer" protocol. This is all exactly
> in line with the RFC.
>
> For your reference, have a look at http://www.ietf.org/rfc/rfc1738.txt
> section 2.1, where the simplest valid form of a URL is defined as
>
> <scheme>:<scheme-specific-part>
>
> It is entirely up to the scheme (protocol) to define what appears in the
> scheme-specific part... that is, everything after the colon. Try typing
> about:plugins in Mozilla.
URIs as defined by the ietf can look however they like, however, the
form of uris used by gnome-vfs have other expectations, both by gnome-
vfs and the apps that use it.
> > c) We're trying to avoid exposing uris to users anyway, so this
> > shouldn't matter to users. And in the cases they are visible, we should
> > work to fix that. (But if you do know the uri schemes, surely its a lot
> > harder to remember/type these new versions.)
>
> The idea I am putting forward here would simplify things for users to an
> extent, but the change would really be more of an organizational one.
> Having a single 'scheme' namespace for shell-specific folders does make
> sense from a structural perspective, in my view. Better to encourage the
> use of a common addressing scheme than make up a new scheme every time
> we need a new shell location. Whether or not the user actually sees a
> mess of schemes or not doesn't strike me as a good reason to leave
> things the way they are.
You'd just make up the sub-scheme every time you need a new location.
Not a big difference. Certainly shouldn't affect users much (other than
all their links breaking when we change).
Anyway, I'm the gnome-vfs maintainer, and I say no. I refuse to make
large changes like this which are of little practical use. We've finally
gotten gnome-vfs to a state where it seems to sort of work. Now we need
to start using it in the desktop, not do random changes to the
internals.
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Alexander Larsson Red Hat, Inc
alexl redhat com alla lysator liu se
He's an obese zombie barbarian looking for 'the Big One.' She's a beautiful
mutant research scientist with the power to see death. They fight crime!
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]