[whatwg] Endianness of typed arrays

[Boris Zbarsky]

On 3/28/12 2:22 AM, Jonas Sicking wrote:
> Except if the data was written in 32bit units you do a different byte
> swapping than if the data was written as 16bit units.

Hmm.  I just read the webgl spec more carefully and discovered that 
bufferData() actually takes a byte array whose structure is opaque, 
which would mean that in that case it does become very difficult to 
figure out what the swapping pattern should be.  :(

> The typed-array spec was specifically designed for use cases like
> sending buffers containing data in patterns like "32bit data, 16bit
> data, 16bit data, 32bit data, 16bit data, 16bit data...".

With all due respect, if it were really designed for those use cases it 
would allow declaring a "type" made of "32-bit data, 16-bit data, 16-bit 
data" and then creating an array of such types....  I understand we 
might get such APIs eventually, and then WebGL may end up usable again 
on big-endian platforms if developers use those APIs to fill the array 
buffer.  But as things stand, I think you're right: creating a browser 
implementation on big-endian hardware that has working webgl and works 
correctly with existing code that gets data using XHR arraybuffer is 
impossible as far as I can see.

-Boris