Re: Supporting MathML and SVG in text/html, and related topics

On Apr 16, 2008, at 4:08 AM, Paul Libbrecht wrote:

>
> Le 16 avr. 08 à 12:49, Henri Sivonen a écrit :
>> On Apr 16, 2008, at 13:30, Paul Libbrecht wrote:
>>> I don't know how big the holy name of backwards compatibility is
>>
>> *Very* big and *very* holy.
>
> Unfortunately, that holyness can only be seen from the measured by  
> the eyes of the billions of users' experiences.
> That is... backwards compatibility could mean, for a portion of  
> them, backwards compatibility to the many bugs of Netscape 4 or  
> Internet Explorer 5.
>
> I don't think this is what you want although it seems that the  
> statistics seem to be the hammer argument here.

In practice, bug-compatibility with IE 5 and Netscape 4 are not very  
relevant to today's web. IE6+ and Firefox (1 and higher) are more  
relevant.  You have to keep in mind that the primary kind of  
compatibility we care about here is what's identified in the HTML  
Design Principles as "Support Existing Content". In other words,  
changes to UA processing requirements in HTML5 should not cause UAs to  
misrender a significant proportion of actual existing web content.  
Bugs in old UAs that content does not depend on don't matter. But  
actual content does.

This rule is very important because implementors are generally  
unwilling to implement things that break the Web to a significant  
extent.

Regards,
Maciej

Received on Thursday, 17 April 2008 07:32:01 UTC