Re: Proposal: gnome-user-share

[Havoc Pennington <hp redhat com>]

On Sun, 2004-12-05 at 17:20 +0000, Alan Cox wrote:
> 
> Things like ".directory " files don't break badly on people. It's
> invisible to apps that don't care. ls doesn't show me different gnome
> magic for gnome file types and labels. It could if we had an fdo library
> but the failure case is not problematic.

I don't understand how .directory is different - in fact from reading
that OS X link, "display names" on OS X are implemented in pretty much
the same way as .directory files.

The end result is a separate display name vs. name on the disk displayed
by the shell.

> You have to interoperate and things like magic renaming leak everywhere
> and perform badly. It actually degrades worse than just having a gconf
> key and a couple of script helpers or environment variables. Its not
> perfect either.
> 
> Thats the biggest problem I have with this - you are (IMHO anyway)
> digging a large trench that will have to be filled in again if you take
> this hack approach. 

Hey, if you want to argue the technical merits, see Alex's post for all
the details. I'm simply refuting the claim that "different from
Motif/shell" is a trump card argument.

If we can get past that a priori litmus test, on the pragmatic technical
issue itself: we have already tried both of these approaches - display
names and configurable directory. Alex is one of the people who would
have seen most of the bugs and discussion threads for both.

There's no perfect solution. I personally think display names are more
robust, both in terms of ensuring people see the right i18n name on the
folder, and having apps and scripts put stuff in the correct folder.
Whether to do it as a hack or some generic extensible mechanism I don't
care really.

We already have a ton of examples of the GUI user model not mapping to
the shell model; look at your Desktop, Home, and Computer as displayed
by nautilus; look at the file selector; look at non-file: URIs in gnome-
vfs. This was fait accompli long ago. And as I mentioned in my previous
post, GNOME is doing this to a lesser extent than any of its competitors
(OS X, Longhorn, Workplace). The point here isn't that we currently have
it right, only that it's hardly a new or radical idea to diverge from
the shell on what the universe of files looks like.

I certainly agree that some spec or library on freedesktop.org should
define said universe of files, and I've mentioned that a number of times
in conference talks and so forth.

Havoc