Jeremy Howard wrote:
> Make sure you read the interesting RFCs from Damian Conway on related
> issues:
>
> > * Built-ins: min() and max() functions and acceptors
> >
> > * Built-ins: reduce() function
Couldn't see these.
> >
> > * Data structures: Semi-finite (lazy) lists
> >
> > * Subroutines: higher order functions
Found those (RFC23&24)
> >
> > * Subroutines: lazy evaluation of argument lists
> >
> > * Superpositions: vector operations via superpositions
Couldn't see those either. Could you refer to the actual RFC #s, please?
> The excellent news is that Dan Sugalski (who's heading up the internals
> team) has promised that he'd make sure that the optimisations we need would
> be seen to. In particular, we talked about making sure that the following
> creates only one loop and no array copies:
> $a = sum(@b*@c+@d)
> ...and I did point out that @b et al might actually be columns of a sparse
> matrix accessed through an iterator.
Sounds interesting. From our point of view one question is: assuming
that there still is a need for compact numerical objects (PDLs) as
opposed to conventional perl arrays (for reasons of efficiency and
memory consumption) will those
compile time optimizations also be accessible to PDL objects?
> Anyway, all I can say is, give it a go! Start cross-posting your thoughts to
> perl6-language, and pop over to dev.perl.org to see what RFCs are around.
> I'm sure you want to keep up the discussion with PDL-P, but if it's about
> perl 6, why not have it on perl6-language where everyone can follow it?
Sure. One problem we AFAIK have on pdl-porters is that all developers I
know of use PDL and perl for real work in their day jobs and are hard
pressed to deal with all the mailing list traffic, upcoming problems and
RFC writing just in the time frame that seems available.
Christian
