Re: short-tag considered unhealthy

Hi

At 10:23 AM 9/13/96 CDT, you wrote:
>On Fri, 13 Sep 1996 07:56:39 +0100 Martin Bryan said:
>>Just to make it clear to everyone the consequences of what you are
>>saying, you will be forcing me to enter <OL compact="compact"> rather
>>than <OL compact>.  (Note that <OL "compact"> is invalid SGML!)
>
>Yes.
>
>Since the main application area of XML is expected to be program to
>program communication, losing <OL compact> does not seem to me to be a
>big problem.  Even if users do want or need to type XML documents in
>ASCII editors, I think it's better this way.
>
>The posts on HTML discussion groups in which users try to explain to
>each other the meaning of constructs like <OL compact> have convinced
>me, at least, that requiring all attribute value specifications to take
>the form attribute-name, value-indicator, quoted-string would result in
>much less confusion and would be a net gain for everyone concerned.

No, not really. As one who has tried to explain this intricacy many times
on comp.infosystems.www.authoring.html, I will rather immodestly claim to
know the *real* problem. Essentially, <OL compact="compact"> is a kludge.

SGML lacks the "boolean" datatype.

A boolean defaulted to false (by absence) is the concept many learners
*naturally* associate with this form of minimization. I'll also add that
no popular HTML browser actually groks the minimization concept: all of
them barf on things like <P center>, i.e. minimization applied to name 
token groups with more than one member. Judging from the source code I've
seen (and some that I've written myself) HTML browsers actually treat
things like ISMAP etc as boolean attribute names, i.e. a parse like this

%%
attlist :
        | attlist attr
        ;
attr    : NAME
        | NAME EQ VALUE
        ; 
%%

Allowing boolean attributes "solves" the practical problem, IMHO. And
anyone who says it will complicate parsing shouldn't write a parser:-)


Regards,
Arjun

Received on Saturday, 14 September 1996 01:30:37 UTC