Re: A List Apart: Articles: Prefix or Posthack

At 5:58 PM -0700 7/7/10, Brad Kemper wrote:

>Agreed. I thought it was interesting that he mentioned the raging
>debate about shadow blurs, as that one seems most in danger of
>becoming unprefixed while still largely inconsistent (esp. if MS
>follows the current spec and no one else does).

    The shadow-blur debate got mentioned because it seemed to 
perfectly capture the dangers of not prefixing and the advantages of 
requiring prefixing.  And also because it was the trigger that pushed 
me to propose and then write the article that had been rattling 
around my head for a few months.
    It does bother me (rather a lot) that there are no interoperable 
implementations of shadows and yet there's a chance of the prefix 
being dropped.  That's just wrong.  There's a very real possibility 
that there could be three public and inconsistent implementations. 
If Microsoft ships an unprefixed version that perfectly follows the 
spec, and is thus incompatible with everyone else, what happens if 
the spec changes to match one of the released implementations?  Now 
MS is once again in the position of "released unprefixed, spec 
changed, can't change our implementation" (see: 'clip').  But it's 
not just them: it could happen to anyone.
    By releasing with a prefix, they can more easily change to match 
any spec changes.  Even if the spec doesn't change, then once it's 
decided that the spec is the correct implementation and the other 
released implementations will have to change to match it, then MS 
gets to drop the prefix first, once the WG gives the go-ahead.
    I even believe MS should prefix its 'border-radius' implementation 
in the next beta, even though it doesn't prefix now (I believe).  The 
only exception I would make is if between now and their next beta, 
the WG deems it (along with all the other released implementations) 
to be interoperable.  If not, then prefix it.
    And I'm not picking on MS here.  The same could happen to Opera or 
the Gecko team or the WebKit team or whoever.  There are a whole 
class of traps that prefixes help to avoid, and even more they can 
avoid if their use is formalized and used as a metric for the 
advancement of a module.
    I was planning to post here about the piece once the piece was 
published and public feedback was received, but you beat me to it!  I 
should've known that would happen.

-- 
Eric A. Meyer (eric@meyerweb.com)     http://meyerweb.com/

Received on Thursday, 8 July 2010 16:22:31 UTC