I'd prefer you named your parallelization pragma parallelizable or such. I have similar interests in this flag except mine would be for explictly altering the execution order of blocks to allow pipelines like grep { ... } map { ... } ... to present a low profile instead of having to fully compute each section before passing on to the next operation.
Using the parallizable pragma I can transform the following code. The benefit is I only have the minimum amount of intermediate results present at any given moment rather than the maximum.
{ use parallizable; @ary = map { ... } # block 1 = grep { ... } # block 2 = map { ... } # block 3 LIST; } { for ( LIST ) { for ( do { block 1 } ) { next if not do { block 2 }; for ( do { block 3 } ) { push @ary, $_; } } } }
In reply to Re: RFC: Implicit Parallelization Pragma: use parallizable
by diotalevi
in thread RFC: Implicit Parallelization Pragma
by hardburn
For: | Use: | ||
& | & | ||
< | < | ||
> | > | ||
[ | [ | ||
] | ] |