Current desktop detection / app access

[Michael Meeks]

On Thu, 2004-08-12 at 13:19 +0100, Dave Cridland wrote:
> There is a standard. Simply refer to it. Please don't invent a new 
> one.

	Ah; excellent - of course, being abnormally ignorant I didn't find it;
clearly if someone else has done the work, it's excellent to cite them.

> Traditional style would mean that one arbitrary 'to' recipient was 
> encoded directly

	That seems to make most sense; we should recommend using the
'traditional style' then (?)

> The only additional thing you're trying to add is the attachment. 
> IIRC, there is nothing in the specification for this. An 
> 'x-attachment' parameter with a URI would probably work, and have the 
> failure condition (assuming a proper mailto handler) of inserting an 
> 'X-Attachment' header with the URI(s) present. You might also want 
> x-body-uri, which'd allow you to supply text/html bodies and other 
> evils. (via data scheme URIs, if you wanted it inlined.)

	Ok; since it seems you're the domain expert here - any chance of a
small block of spec. verbiage to define this ? Given that netscape 6 ->
mozilla, evolution and kmail (with patch) all support attachment='<uri>'
it's perhaps unwise to specify something different; I guess we should
include that.

	With that, assuming people are happy, we can at least up-load it and
encourage people to adopt the tiny patches to make this work right.

> Note that multiple parameter values are a little vague - I can't see 
> anything mentioned in the specification. I assume that you'd do it 
> via multiple parameters with the same name, thus:

	I believe the existing practice is to separate them with various
delimiters - I'm basing this all on a crude reading of:

http://ooo.ximian.com/lxr/source/util/officecfg/registry/data/org/openoffice/Office/Common.xcu

	(search for evolution eg.).

	Thanks,

		Michael.	

-- 
 michael at ximian.com  <><, Pseudo Engineer, itinerant idiot