[whatwg] Re: some issues (SGML)

[Malcolm Rowe]

Jim Ley writes:
>> And you're worried that this will continue once the DOCTYPE is
>> finalised?
> I understand from posts that part of the process here is to fix the
> stuff in HTML 4.01 that was never implemented and cause problems.

What Ian said (and what I understood anyway) was that WF2 would fix "a few 
relatively minor exceptions where the spec is out of touch with reality". I 
didn't understand that to imply that the WF2 spec would proceed on a similar 
basis to CSS2.1 -- removal of the parts of the spec that were never 
implemented -- as you are suggesting. 

This isn't about removing SGML's NET feature from the spec simply because no 
UA gets it right; it's about codifying parts of the spec that aren't 
well-defined (taking the de facto behaviour and making it de jure, 
essentially), and extending the forms behaviour of HTML to make it more 
useful to users. 

>> and don't support some valid SGML SHORTTAG features either (NET, for
>> example), 
> This isn't specified in the DTD, it's in the SGML Declaration:
> http://www.w3.org/TR/REC-html40/sgml/sgmldecl.html

Ok, my mistake. I did say I wasn't an SGML expert. I assumed that the SGML 
declaration was either included in the DTD by reference, or was somehow 
identifiable from the DOCTYPE. Anyway, however HTML4 does it, HTML4+WF2 
should do the same. You seem to know more about SGML than I do - is there 
any reason this isn't possible? (for example: how would W3C do it for 
'HTML4.1', because that's essentially what we're talking about here). 

One thing I do wonder about is the SGML attribute minimisation feature. If I 
understand it correctly, this is the ability to write <IMG ISMAP> and have 
that be parsed as <IMG ISMAP="ISMAP"> (the provided 'ISMAP' is the attribute 
*value* and not the name). Strictly speaking, you need the DTD in order to 
properly determine which attribute is being set when you encounter a value. 

Does WF2 cause any problems with attribute minimisation? (which is one of 
the few parts of SHORTTAG which *is* supported by HTML UA's) Specifically, 
are the attribute values unique (so that, for example, <INPUT OFF> has a 
single expansion, to <INPUT AUTOCOMPLETE="OFF">). If the values are not 
unique, does that cause any new problems? I've not got enough experience 
with SGML to be able to evaluate this - but perhaps you have. 

>> Anyway, I would be *extremely* surprised if a WF2 DTD were to be produced
>> which caused valid HTML4 documents not to be valid 'HTML4 with WF2'
>> documents. Is there any reason you seriously believe that this might be
>> the case, 
> Hixie's sensible dislike of those things in specs which were never
> implemented, and various other things said on the list.

Fair enough. I don't think that that is the case here, though. The WHATWG 
charter isn't to produce a version of HTML4/DOM/etc with bits stripped out 
(unless they're 'bugs' in the specification, essentially, because the real 
world conflicts with the specification so much so that implementing the spec 
would either be impossible, or would make you incompatible with virtually 
every other HTML UA). 

In HTML4, the W3C HTML WG provide the following: 

"As of the 24 December version of HTML 4.01, the HTML Working Group commits 
to the following policy: 

* Any changes to future HTML 4 DTDs will not invalidate documents that 
conform to the DTDs of the present specification. The HTML Working Group 
reserves the right to correct known bugs.
* Software conforming to the DTDs of the present specification may ignore 
features of future HTML 4 DTDs that it does not recognize."
 -- http://www.w3.org/TR/html4/struct/global.html#version-info 

Now, while the WHATWG isn't the W3C HTML WG, and while the W3C may not be 
the eventual destination for a WF2 spec, I would suggest that this policy is 
one that also makes sense for WF2, as a 'successor' to HTML4. 

Regards.
Malcolm