Re: HTML Streaming

Albertfine@aol.com
Mon, 8 Sep 1997 08:06:30 -0400 (EDT)


From: Albertfine@aol.com
Date: Mon, 8 Sep 1997 08:06:30 -0400 (EDT)
Message-ID: <970908080629_907653494@emout04.mail.aol.com>
To: www-html@w3.org
cc: darsal@tezcat.com
Subject: Re: HTML Streaming

darsal@tezcat.com (Dave Salovesh) wrote:

>I've been thinking about that part for a couple days - why does this
>idea have anything at all to do with HTML?  It seems to me that the most
>successful approach to this type of thing would be to have the -server-
>generate a description on the fly, but only when the request is made by
>a UA which can benefit from it.

It is HTML. Sending hints to the UA on how an element will display is not 
my invention. It has been used for everything from images to tables. The 
events element is just a different way of doing it.

>As part of the HTTP request a browser could announce its current
>settings, and then there would be no guessing or errors.  This also has
>the advantages of working equally well on any document and requiring no
>effort or special tools for authors.

This would be very inefficient. The isp/server would need to have a copy 
every user's browser and operating system. It would then need to set each 
copy to the settings of the user. It would then load the page and create a 
preview for each preferences or have a copy on hand. You would have to delay
the upload of the HTML file to the user or he/she would get the HTML file 
before the preview. You would need a pretty fast processor or the delay would
be signifiant. You may still have compression problems. If the user makes any
changes to the operating system or the browser while the file was being 
download it would not work and you would need to upload another file. The 
agent would also need to send updates on its os and browser preferences. The 
files would probably be useless after a single change and you might have 
copyright and privacy problems. All this to smooth out what is just a couple
seconds of load time?

>Albert, while I've enjoyed the discussiion so far and I intend to
>participate in any other discussions that may arise on this topic, I
>suggest you take the time to develop your idea a little further before
>attempting to defend it as it stands.  It would be most helpful if you
>could prepare a document where you describe what you're trying to
>acheive and how you think it will work - in detail - and perhaps return
>here to provide a URL where we could examine that proposal in full.  As
>it is now we're going around in circles without even really knowing
>where we're trying to go.  Before we beat this to death, it would be
>nice to have something a little more concrete to point at.

I am glad you enjoyed this discussion. As I said before, I thought coming to
this board would help in the development of the idea. It is not just the
events element I am concerned about. It is really how a page loads. This is a
very broad topic and their are many ways of approaching it. I don't think I 
will return to this list to discuss my ideas. I will respond to any questions
on my previous posts but I don't think I gained any insight by coming here.

Albert Fine