Re: Versioning re-visited (was : mixed signals on "Writing HTML documents", tutorial, etc.)

On Sat, 23 Jun 2007, Craig Francis wrote:

> Adding a version does not necessarily mean that a browser needs to look 
> at that information... but it does mean that information is there if 
> required (for what ever reason I cannot think of at the moment).

Should we also therefore include a way of having the postal address of the 
author, and the author's mother's maiden name? Just in case that 
information is required, for whatever reason that we can't think of yet? 
Or should we be sure to include the length of time the document took to 
write? Or should we include the domain name of the origin domain?

Those pieces of metadata are as useful if not more useful than the version 
of the language that was contemporary when the author wrote the document. 
I think we're going to need far better arguments than "in case it's 
useful" to include a feature. Versioning isn't special.


> Personally, I am taking the view that HTML should be parsed by the 
> browser as HTML, irrispective of version... its up to them if they want 
> to act on the version number... the real benefit of the version though 
> is for the author to claim compliance with a specification (not one 
> which may or may not exist in the future... like HTML-100).

I understand that you believe this is a benefit. I argue it's not.


> Given your example of the <credit> tag, where the meaning was to change 
> from a money value to the name of the author... well I should hope that 
> this will never happen, as it would show huge incompetentcy on the part 
> of those writing the specification.

Strongly agreed.


> A better example is perhaps over the table @summary attribute... 
> currently writing for HTML4, this is perfectly valid, and even 
> encouraged... but at the moment, HTML5 will not include this 
> attribute... does that mean that all tables (with tabular data) that use 
> the @summary attribute, are now invalid... even though they are written 
> in HTML4?

No, they are still valid HTML4. But if we omit an attribute like that, 
there's presumably a reason; wouldn't we want the author to hear about it? 
If there's version information and the validators just use that, the 
author won't know.


> How about we just go with the "if it ain't broke, don't fix it" theory, 
> and keep the version number...

I argue that it is heavily broken. And in fact positively dangerous.

-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'

Received on Monday, 25 June 2007 07:04:27 UTC