- From: Bullard, Claude L (Len) <clbullar@ingr.com>
- Date: Fri, 5 Dec 2003 16:13:36 -0600
- To: 'Tim Bray' <tbray@textuality.com>, algermissen@acm.org
- Cc: "'www-tag@w3.org'" <www-tag@w3.org>
1. Subscribe is a better name for it as you suggest
although it could confuse people who associate that with
mailing lists.
2. People who click on things are used to getting
back a page or opening a dialog. Autosubscribing
based on a click seems like a bad idea. It doesn't
pass the Don't Shock The Monkey test. It seems like
a better idea not to subscribe, but to open a dialog
with that value with a Subscribe option on it. Otherwise,
accidental clicks cause problems. And since some
browsers render www.t as a hypertext link control as well,
a guessing game goes on regards defaults.
3. It seems that what you are after is a
control to pass a value to the right application and
the only way to do that via pushing the data is to
push the URI and insist that browsers implement a
new control (actually, a new switch in the
URI processing code that does something surprising),
plus some means to take care of the default behaviors.
It seems like a lot of hassle to get around a cut and
paste operation to push the feed:/ or subscribe:/ or
RSS itself to some first class citizen level. And if
that exception is made for RSS, why not do it for mailing
lists?
len
From: Tim Bray [mailto:tbray@textuality.com]
On Dec 5, 2003, at 1:45 PM, Jan Algermissen wrote:
> Hmm...this confuses me - isnt't dispatching to the right application
> one of the primary uses/purposes of mime-types?
>
>> because the RSS client doesn't need the
>> representation, it needs the URI.
>
> What do you mean by "it needs the URI"?
I mean it needs the URI, because it wants to *subscribe* to the feed,
i.e. stash the URI away somewhere and do a GET on it regularly to see
if anything's changed. I've suggested that in fact this proposed
scheme shouldn't be named "feed:" it should be named "subscribe:" -Tim
Received on Friday, 5 December 2003 17:13:47 UTC