W3C home > Mailing lists > Public > public-html@w3.org > November 2008

Re: The compatibility DOCTYPE

From: Leif Halvard Silli <lhs@malform.no>
Date: Tue, 11 Nov 2008 20:00:42 +0100
Message-ID: <4919D65A.8070905@malform.no>
To: Jirka Kosek <jirka@kosek.cz>
CC: HTMLWG <public-html@w3.org>

Jirka Kosek 2008-11-11 14.52:
> Leif Halvard Silli wrote:
>>  The proposal had 2 variants.  Isn't "DOCTYPE html" proper enough?
> No

What does it lack? What does it hamper?

>>> nor it shows that its in HTML5 only to accommodate
>>> legacy content producers.
>> The message of a recursive acronym is: For God's sake, use the acronym.
>> In the case of <!DOCTYPE html public "DOCTYPE html">, the message would
>> be to just use <!DOCTYPE html>.
> I don't think that this will work. You are assuming that average web
> author who uses copy'n'paste approach 

Till that day when they ask about the DOCTYPE at all, those 
authors are irrelevant to us - they will not touch a DOCTYPE 
whether it looks that way or another.

But on that day, what then?  Do you think "xslt-compat" will have 
them ask *more* questions or ask them earlier? Why not have them 
ask the *right* question instead? "Why that longer variant, when 
it appears to repeat what is allready said and which the much more 
common short variant says too?"

> is clever enough to get this and
> at the same time he/she is too dumb to use the shortest preferred
> variant of !DOCTYPE.

So you *do* want it to appear as a choice between short and long?

If so, then you don't pick 'xslt-compat' as identifier. For as 
long as the long DOCTYPE appears to hide a "secret", the choice 
will *not* appear to authors as a choice between short and long.

>> Wheras for a proper, but purposeless identifier, such as "compat", and
>> which still is supposed to tell for whom it is intended (!), it is
>> simple to make up a nonintended purpose. Better to let the message be in
>> the form rather than in the content. After all, it is the form that
>> those content producers have problem with.
> If you think that average web developer is so dumb (I don't know) 

Recursive means: The answer is in the name.

With "xslt-compat", the answer is not in the name, but outside it, 
in specs and Web history. Without that background even a smart ass 
will feel uncertain about whether <!DOCTYPE HTML public 
"xslt-compat"> can be swapped with <!DOCTYPE HTML>.

With a recursive identifier, the history is still there, but then 
there is also a satisfactying explanation inside it: Yes, it is 
unecessary, chose the short variant, to which the identifier in 
fact points. I think this is trusting that authors are not dumb.

>then you can always use something very explicit:
> <!DOCTYPE HTML PUBLIC "Human, do not use this in your document! This is
> just to accommodate legacy computer applications.">

In essence, this is exactly what "xslt-compat" tries to do.  Or as 
Ian probably would have described the ideal reaction: "this 
DOCTYPE is so ugly that it is impossible - let me check the spec".

The "recursive" approach, however, is to hint to the authors that 
"this is superfluous", and make them draw the conclusions. I think 
the answer will "click" much more easily then.
leif halvard silli
Received on Tuesday, 11 November 2008 19:01:26 UTC

This archive was generated by hypermail 2.4.0 : Saturday, 9 October 2021 18:44:39 UTC