Re: Removal (Time for XMail?)

On Mon, 2 Oct 2000, Aaron Swartz wrote:

> Laird Popkin <laird@io.com> wrote:
> 
> > Yep, and there is a spec for allowing people to send URI's over the web. ICE
> > (see http://www.icestandard.org) has both push and pull content syndication
> > for this reason.
> 
> Thanks for the URL. I've looked at this before, didn't look in too much
> detail because the spec was rather complicated. I'll take another look now,
> however.

As an outgrowth of supporting various choices (push/pull, pass by
value/pass by reference, scheduling, etc.) it is a fairly complex
document. That being said, if you pick a particular combination and ignore
everything else the spec gets a bit "skinnier". We've recently published
an "ICE Cookbook" of recipe's that provide simple examples of how to use
ICE, specifically to make the protocol more approachable. It should be up
on the ICE web site shortly.

> > Given all of the various possibilities, I'm curious which combinations are
> > most interesting to you?
> 
> Personally, I like applications that feel like pull and really like those
> that actually are pull. While I check my email using POP currently, I'd much
> prefer it if my computer could notify my email server when I had a net
> connection, and have it forward all the email to my machine, so I'd be
> notified as soon as it was received. Then, of course, it would stop doing
> this when it discovered I had gone off line.

Makes sense. Sounds a little like Backweb's "polite" technology.

> In addition, as mentioned before, there is a need for push (in protocol
> form) over the Web. Formats like RSS, etc. would save lots of bandwidth and
> resources if push were available. Services that track pages (like Mind-It!
> and SpyOnIt) would become much more accurate and useful, if they notified
> you immediately once a page had changed (or close too it) rather than every
> hour, two hours, or however long it takes them to sweep. The problem is that
> pull protocols don't scale as well as push for these types of apps.

Indeed -- I agree that push is prefereble. Unfortunatly it requires a
degree of sophistication on the part of a content provider to implement
push notification of changes, while it's no work at all to put files on a
web server that are checked using "if if modified since" in HTTP, so I
suspect that true push will be limited to people/systems that are fairly
serious about optimizing content flow.

Received on Monday, 2 October 2000 16:05:03 UTC