Re: bit.lshift and performance - luabitop v.s. lua-5.2.0-work4

[Mike Pall <mikelu-1010@...>]

Miles Bader wrote:
> Mike Pall <mikelu-1010@mike.de> writes:
> >>   bit.lshift (NUM, bit.band (COUNT, 31))
> >> 
> >> [instead of plain "bit.lshift (NUM, COUNT)"]
> >
> > Oh, sure. ;-) Because absolutely everyone is just waiting to
> > rewrite all of their code with that bloat.
> 
> How much highly speed-critical Lua bitop-shift-using code is out there
> anyway?  Probably not all that much, I'd wager.

There certainly is. But that's completely missing the point.
Nobody is going to change all of their existing code and nobody is
going to use your workaround for all shift operations in new code.

But everybody whose code silently breaks with Lua 5.2 will
complain. Probably to me, because they don't realize that
require("bit") is not loading the bit.so/bit.dll, anymore.

--Mike