[Sigia-l] Word HTML - money were my mouth is (was When Should a Manual be Web-based?)

Jon Hanna jon at spin.ie
Mon Mar 3 06:57:39 EST 2003


> It is one thing to generate web pages without HTML error. But it is
> quite another to have the generated HTML deemed "valid" with regard to
> a published version of HTML.

No, that's one and the same thing. How else do you define "HTML error"
except by saying it differs from what HTML allows?

> Mind you, web browsers do not require web pages with DTD declaration.
> Unless Word acted up, Internet Explorer, Netscape Navigator, and Opera
> have no problem with the Word-produced HTML.

Not strictly true. Web browsers are required to make best attempts to render
HTML when they encounter something that is an error. This is partly to
enable them to be compatible with later versions that contain features they
do not recognise (though hopefully the move towards modularising HTML and it
being based upon XML's clearer rules about what is "well-formed" will enable
more mistakes to be recognised as such), and partly because the
internationally distributed nature of the web makes an error message of
little value once something is released.

Relying on error-correction in this way is like using a lifeboat for a
sea-crossing.

It's worth noting that there is no way to define exactly how browsers should
do this (although some particular errors have particular responses defined,
for instance non-existent elements or attributes should always be ignored,
and be replaced by their content). Generally it all boils down to "reading
tea-leaves".

It's also worth noting that some HTML-processing tools (especially those
restricted to processing XHTML) don't have this behaviour.

> Regardless of how much people may want to bash Word, I hope they can
> be fair-minded.

Word is brilliant at many things (everyone, please don't send me thousands
of mails about how word processor X on *nix format Y by open-source project
Z is better :). I use it all the time. However it is not a HTML editor, does
not produce HTML, and probably shouldn't be used as such even if it produced
perfectly valid HTML (it does after all have a WYSIWYG metaphor that is
inherently mismatched with HTML's model).

No doubt it will improve (earlier versions' output is simply bizarre so it
has steadily improved) for which reason anyone wanting to get their claim in
for the €80 must do so by 5pm GMT today (I'd like to think that if I left it
open indefinitely someone would eventually win it). No claims by then and my
portion of the €80 is going to the charity of *my* choice instead.




More information about the Sigia-l mailing list