[whatwg] Endianness of typed arrays

Boris Zbarsky bzbarsky at MIT.EDU
Wed Mar 28 01:50:45 PDT 2012


On 3/28/12 1:22 AM, Charles Pritchard wrote:
> You are making the assertion that "deliberately misleading in terms of
> what it takes to achieve interoperable behavior".

Yep.

> And you and Robert have given reasons why you believe developers will
> make many mistakes.

No, I'm just saying developers are, today, writing code that assumes 
little endian typed arrays.  Some are doing it deliberately, some are 
doing it because they don't know better, some are not doing it it all. 
But the net effect is that there is some code on the web.

> I brought up adding some additional text to aide developers...

This won't change the dynamic...

> What would it take, without changing the behavior of the current spec,
> for it to contain terms that you do not feel are "deliberately misleading"?

I don't think that's possible.

> Do you feel the spec is misleading implementers?

Yes, precisely.  The point of a spec is that it can be implemented and 
then the implementation would interoperate with other implementations. 
An implementation of the typed array spec that actually followed the 
spec and happened to run on big-endian hardware would not interoperate 
with other implementations and would not work with existing content.

-Boris



More information about the whatwg mailing list