On 02/28/2012 02:45 PM, Robby Findler wrote:
On Tue, Feb 28, 2012 at 3:39 PM, Ryan Culpepper<r...@cs.utah.edu> wrote:
[A] topologically-sorted list might actually result in worse
scheduling. It may place dependencies close together and they might get
scheduled on different places.
The current strategy (alphabetically sorting things) guarantees to put
x/private near to x in the list. Wouldn't you expect almost anything
to be better than that, with regards to this consideration?
The list seems to include only top-level collections: x/private isn't on
the list. I guess it's only discovered once setup starts compiling x. In
any case, x/private always seems to be handled by the same place as x.
Ryan
Speaking of which, I've suggested trying out randomly sorting the list
before. Maybe I'll give that a try next.
If B depends on A, then B should be *started* only after A is *completed*.
But the current interpretation of the worklist is just the order in which to
*start* collections.
Right.
Robby
_________________________
Racket Developers list:
http://lists.racket-lang.org/dev
