Re: HTML Streaming

Albertfine@aol.com
Mon, 25 Aug 1997 09:38:16 -0400 (EDT)


From: Albertfine@aol.com
Date: Mon, 25 Aug 1997 09:38:16 -0400 (EDT)
Message-ID: <970825093814_-735687089@emout14.mail.aol.com>
To: www-html@w3.org
cc: anthony@pce.net
Subject: Re: HTML Streaming

anthony@pce.net (Anthony DiPasquale) wrote:

>Elements of a page pop up as the file is downloaded, with the exception of
>tables, which must be fully loaded to display (and with the absence of
>WIDTH/HEIGHT attributes in IMG tags, any images in the tables must be
>loaded as well.)

HTML is not just formated text. It can start programs, contain programs etc. 
I have several protocols for dealing with these new elements. For example,
an events tag could list the types of tags in a page, giving the browser a 
better idea of what to expect. So it could load the java compiler but wait
to load the code until it has finished displaying the page. I think the 
events tag is probably overkill. Of course all this is very early in 
development. I have worked longer on the stream tag. It tells the browser 
when to load a tag; table, plugin, applet, script etc.

>Until now..the HTML 4 specs say (quote) "The ability of designers to
>recommend column widths allows user agents to display table data
>incrementally (as it arrives) rather than waiting for the entire table
>before rendering." 
>
>So basically, as long as nobody places their HEAD, TITLE, tags, etc in
>unorthodox places, the HTML in essence will 'stream'.

The head is more complicated. Everything in a HTML file is not what is 
displayed. If you go to any page, I am sure you will not see the title tag 
then all the things that are displayed. For example, their are keywords, 
program titles, signatures etc. My protocols would probably be added to 
HTML editors. The advantage would probably be negligible for many HTML files.

I think the difference is comparable to running code through an assembler.

Albert Fine