Re: client side includes

Murray Wrote:
"Unless my math is weak, a page with 30 or so CSI files (not that
unrealistic) would take a minimum time of seven round trips (One for the
html file itself, and 30/5 = 6 for the CSI files.)  Imagine if it were only
one."

Geez! What makes you think that 30 csi files is not that unrealistic? I've
kept my ssi calls down to less than 5 all of the time! I usually pull it out
in only 3 ssi calls. I cannot imagine a realistic application of CSI taking
that many files! I could see maybe 10 tops, which would take two
connections...

But if you want to complain about how much longer it would take to load...
just don't. I've said it before: Currently, in the big two, if there is a
table (which there ALWAYS is), which sets the layout of the page, then the
page will not be displayed until the final </table> is recieved. Now, while
the UA is loading the hypertext in one connection, it's loading images in
the other four. On a modem (or on my modem, at least), it usually
successfully loads more than one image per http connection, before it
finally loads enough hypertext to get that final </table> tag, and finally
display all of these graphics it's loaded. How would CSI files be
signifficantly different from that, if future UA's are designed to load all
markup BEFORE loading images or other binary files?

And, again, we'll have to trust the Darwinian / Capitalist principle of
competition will prevent us from seeing any new feature of a w3c spec, be it
svg, or the hypothetical client-side includes, being abused by millions of
unscroupulous, moronic html authors. For example, how often to you see
people writing hole paragraphs of text using CSS's blinking text property?
(yes, I know IE doesn't support that property, and I auplaud the IE team for
that decision. :)  )

Oh well... have a good day everyone!!!

Daniel

Received on Sunday, 4 February 2001 15:29:12 UTC