Hi,
> It sure is interesting that make-adder works with the old Rebol
1
> but not with the newer ones. What exactly did we lose?
> What work-arounds are feasible?
My work-arounds are collected in %highfun.r and %sfun.r
> I assume that the change was some deal with the devil that
> RT made for more speed.
Bind was easier to implement too..., but the speed may even
suffer, because the speed-up for the interpreter may mean a
slow-down for the compiled code (JIT compiling may be interesting
for Rebol and binding a block 1000 times may be more
time-consuming, then the more common approach - bind once, do many
times)
>... a while back somebody was asking for a something like a
whitepaper
> on the language and at the time people either denied it's
usefulness, or claimed
> it would take
> important developer time away from improving rebol so it
couldn't be
> written, or that nobody there seemed to understand the request.
> And really we are talking here about the core functionality of
the language,
> not trying to get Rebol to cough up source code nor to dwell on
the details of
> the implementation of web protocols. We just want to know how
to think
> about the process so we can write correct programs based on
understanding.
The problem may have something in common with the feeling, that
the language is still "under development".
>
> I still don't even use parse for any but the simplest things.
> If I had a burning need that nothing else could satisfy I
probably would sit
> down and grind through the details, but short of that I may
never go any deeper.
> The examples available for parse are certainly better today than
they used to
> be,
> but they are not that easy. This really means that only expert
users with a
> penchant for playing with parse will get much out of it.
Regular users will
> merely boggle. I suspect however that it is actually possible
to document
> parse better so that more people would be able to make it over
the hump.
>
> Please chime in, everyone!
>
> -galt
>
>
>
>
>
>
