shared mime info: URI scheme handlers for non browsers

[Rémi Denis-Courmont]

Le lundi 31 octobre 2011 18:26:33 Bastien Nocera, vous avez écrit :
> On Mon, 2011-10-31 at 18:16 +0200, Rémi Denis-Courmont wrote:
> > Le lundi 31 octobre 2011 18:05:41 Bastien Nocera, vous avez écrit :
> > > There are many more problems with that than just knowing whether the
> > > application supports the protocol (or a subset of the protocol). What
> > > about passing cookie that the website might use for authentication?
> > > 
> > > In any case, it's not what x-scheme-handler is trying to solve.
> > 
> > I got that. I'm concerned that every file manager will lack this (except
> > KDE), just because one (?) desktop refuses to standardize it though.
> 
> Because it's broken, mostly.

It's broken if you don't use it correctly.

> David already explained the reasons why it
> was broken, and nothing changed since then:
> http://article.gmane.org/gmane.linux.xdg.devel/12370/

Sure, the credentials may need to be retyped. Though the browser could pass 
the username and password in the URL. Or better yet, the MIME handling 
application could query the XDG session secret store. And yeah, in some cases 
the (lack of) cookies will cause failure.

But for things like video streaming or DAV, it usually works because you 
rarely need cookies. On the other hand, using GVFS, FUSE or whatever, will 
fail because the applications will need to know the URL as a base (e.g. Apple 
HLS) and/or will need to make usage-specific HTTP queries, that the browser 
I/O abstraction cannot do.

Of course, if you're trying to download a document from some cookie-
authenticated service, passing the URL will fail miserably. But should 
LibreOffice or GIMP really register for HTTP anyway? I don't really see the 
benefit for those applications and their MIME types to use "progressive" 
download.

-- 
Rémi Denis-Courmont
http://www.remlab.net/
http://fi.linkedin.com/in/remidenis