Re: ISSUE-54: doctype-legacy-compat

Lachlan Hunt wrote:
> ...
> Considering the fact that both you and DanC [1] mistakenly typed a 
> non-well-formed version of the DOCTYPE, and that on occasion, even I 
> have accidentally omitted the PUBLIC keyword while typing the HTML4 
> DOCTYPEs, I think this is evidence that it is more prone to errors and 
> that we should not encourage people to use it at all.  I think it is 
> better to actively discourage its use and this is one reason why the 
> perjorative alternatives are preferred.
> ...

The whole point in allowing a second doctype (or for changing it) is to 
allow more people to produce *valid* documents right now.

It's not a feature for people who do not care about validity.

And no, I disagree that the 2nd notation needs to be ugly. It should be 
as valid as the other one, so there's actually no reason to push people 
into one direction.

>> 2) DOCTYPE with an optional null quoted string
> 
> This alternative fails to discourage people from trying to use the 
> longer alternative.

Which, to me, is a non-goal.

> We know from experience that people will often use the longer version of 
> the HTML4 DOCTYPEs including the optional SYSTEM identifier, despite the 
> fact that it is optional.  It seems that there is a general perception 
> that the longer alternative is better.

Why does it matter?

> ...
> I'm fine with both of these alternatives, although I have a preference 
> for the latter because of the reason I pointed out in IRC yesterday [2]:
> 
> <Lachy> the only problem with legacy-compat is that it's not entirely
>         clear that it's meant for compat with legacy generator tools,
>         like XSLT, rather than legacy consumers like browsers
> ...

This has been clarified a number of times by now; it's an issue that not 
only applies to XSLT.

> Additionally, it's not clear to me which tools besides XSLT have 
> difficulty outputting <!DOCTYPE html> in HTML serialisations, and thus 
> what problem we are solving by changing "XSLT-compat" to 
> "legacy-compat".  (If such tools have been mentioned in previous 
> threads, then I may have missed them)

Apparently.

 > ...

Best regards, Julian

Received on Saturday, 17 January 2009 16:58:49 UTC