|
|
Posted by Joel Shepherd on 11/15/06 17:04
Toby Inkster <usenet200611@tobyinkster.co.uk> wrote:
> Joel Shepherd wrote:
> > I.e., what difference can it make whether the include functionality is
> > client-side or server-side?
>
> Because the client needs to build a DOM tree out of the HTML file, whereas
> the server doesn't. So it matters to the client whether an include could
> possibly break the DOM tree, whereas it doesn't matter to the server.
Sure, but the statement wasn't who it matters to, it was that dumb
includes on either end are just as likely to hork the page.
Also the server _could_ build a DOM tree, and _could_ be concerned about
whether it's structurally what is expected. It doesn't have to, but it
could.
> The obvious solution is that the client could perform the inclusion
> *before* building a DOM tree, but that requires a two-pass parse, which
> would slow down rendering.
That might sound obvious to you, but it sounds close to impossible to
me. Going back to your original example: the HTML post-include is valid
(IIRC), but not what the author intended semantically. If you can write
a client to draw that distinction and automatically make the appropriate
correction, I'll be _very_ impressed.
If anything, that sounds like something that could possibly be addressed
on the server, since the server could have knowledge of what the
semantic structure of the page should be. (Not saying that's easy or
practical: but it is possible.) The client can't possibly know what the
correct structure should be ... so how is your solution the obvious one?
--
Joel.
http://www.cv6.org/
"May she also say with just pride:
I have done the State some service."
Navigation:
[Reply to this message]
|