>> I think, something like (thread (load "path/to/program.rkt"))
> This doesn't work unless program.rkt is written in #lang racket/load
> which is unsuitable for large applications.
Perhaps use racket/sandbox?
http://docs.racket-lang.org/reference/Sandboxed_Evaluation.html
`make-module-evalutor
At Thu, 27 Mar 2014 07:45:23 -0300, David Bremner wrote:
> Aha, that was indeed part of the problem, thanks. Now it works for
> absolute paths. For relative paths it gets hung up part way;
I've pushed a repair for relative paths. Thanks for the report!
Racket Users list:
On 29/03/2014 03:54, Yuhao Dong wrote:
Using accumulator+reverse won't really improve the runtime at all.
Every other benchmark has (unfortunately) shown the opposite.
I think that tail recursion doesn't help at all, and introduces
conceptual overhead. Racket doesn't use the stack, and conver
On Mar 29, 2014, at 8:45 AM, Yuhao Dong wrote:
> I think my point still holds that tail recursion in the case I posted
> does not improve performance
The purpose of transforming a program into tail-call shape is to save space. It
does NOT necessarily improve performance.
[[ I do not understa
This doesn't work unless program.rkt is written in #lang racket/load
which is unsuitable for large applications.
On Sat, 2014-03-29 at 15:17 +0400, Roman Klochkov wrote:
> I think, something like (thread (load "path/to/program.rkt"))
>
>
> Fri, 28 Mar 2014 19:12:03 -0400 от Yuhao Dong :
>
Thanks. I probably mixed it up also with the implementation of the
compiler targeting C of Chicken Scheme, which basically just converts
everything to continuation-passing style.
I think my point still holds that tail recursion in the case I posted
does not improve performance. At least the timing
Yuhao Dong wrote:
The thing is, Racket's "stack" is a list of continuations. I don't see
how explicitly keeping the "stack" in an accumulator would help. Stack
overflow won't be a problem since the "stack" is a list on the heap, and
unless memory runs out you wont overflow.
I think that tail
Some comments:
1) Never run benchmarks inside DrRacket (I suspect this is what you did
because the reported times are high)
2) Why do you include the garbage collections in the time measurment?
That's probably not what you want to measure (plus the gc should before the
calls)
3) The need to call `r
I think, something like (thread (load "path/to/program.rkt"))
Fri, 28 Mar 2014 19:12:03 -0400 от Yuhao Dong :
>So, I have a really, really cheap VPS. I need to run several racket
>programs on it; however, just the memory usage of the standard library,
>jit, etc of each racket instance means that
9 matches
Mail list logo
