Re: [hybi] frame length encoding

[Pieter Hintjens <[e-mail]>]

On Sat, Aug 21, 2010 at 9:54 PM, Shelby Moore <[e-mail]> wrote:

> No because for any file on disk larger than 4GB, we will have to move the
> entire file through the CPU, when we wouldn't had to otherwise...

More pragmatically it means every single WebSocket implementation,
both readers and writers, have to either impose a 4GB limit on frames,
or else have extra functionality to detect large frames and fragment
and defragment them.  It creates significant additional cost and risk
and interoperability concerns, and the benefits are to win 4 bytes
with negligible benefit.

It's not just disk files, my box has 8GB and it's more than a year
old.  Loading, processing, and sending 8GB chunks of memory will be
banal in a few years' time.  That really should remain as simple as
sending a 8KB chunk of memory.

-- 
-
Pieter Hintjens
iMatix - www.imatix.com