- From: Dan Kolis <dank@idirect.com>
- Date: Fri, 22 Nov 1996 03:23:34 -0500
- To: www-talk@w3.org
Greetings, I'm looking at HTTP1.1 HTTP/NG andam really interested in CATV style 100K seat multicast and non multicast. I keep looking for hooks in your spec for HTML so the following is totally reliable. Very often a HTML author will reuse lots of components, graphics; (typical JPG GIF), etc. Now Each uses its own GET. I gather this will be 'fixed' this in http1.1. However, if the local machine (or set top converter) could absolutely KNOW the objects inside the HTML are the same as the last time, it could use the last sent instance, over any number of sessions, (not just one session). The point being the larger bandwidth eaters could stay really local. for instance? <img src="redball.gif"> could be <img src="redball.gif" absobject=1215231341555671> The local machine could scan its local registry, and if the object ID AND number are the same display the local copy. The author would change the absobject number, (but not nessessarily filename). Of course, the entire message should be marked to refetch fresh, (not from the proxy). Seems to me the MIME elements SHOULD be able to do this with some wildcard attribute, but I have never seen a browser exhibit this behavior. Envision each time you get the New York Times, you get a graphic of the banner, edges, buttons that say "forward", "back", etc. Interestingly, if the html creator 'steals' a filename and keeps the number, your local machine could even display am item from a previous session from another source. If this is possible now, why isn't it done, I wonder? If you think its useful I could cobble together server software and a plug-in to Netscape to show this work. Important objects might be betther off not using the absobject tag, for forgery issues beyond the scope of this initial message. (If someone else comes to mind regarding this, please feel more than free to forward it to them without further comment from yourself(s)). Regards/ Thanks Dan Kolis dank@tanda.com Tuttle and Associates
Received on Friday, 22 November 1996 03:23:53 UTC