On 15 Feb 2003 at 13:49, Nicholas Clark wrote: > Jarkko's view is that with hindsight we'd've been better of doing Unicode > internally in a fixed width representation (eg 8 bit, upgrade to UCS-2, > upgrade to UCS-4), as all the length,offset,etc algorithms would stay O(1) > The speed hit and extra complexity of variable length has more than > outweighed the space saving.
That sounds reasonable. (Hey, you can still change the internal encoding -- people aren't supposed to rely on the fact that it's UTF-8 behind the scenes, are they? </pipedream>) Cheers, Philip -- Philip Newton <[EMAIL PROTECTED]>