On Fri, Aug 19, 2011 at 10:28 PM, Eli Barzilay e...@barzilay.org wrote:
9 minutes ago, Robby Findler wrote:
On Fri, Aug 19, 2011 at 12:50 PM, Eli Barzilay e...@barzilay.org wrote:
Three hours ago, Robby Findler wrote:
Maybe Eli can say more?
I have no idea what the problem is, or how
I can't seem to make typed racket happy about using 'sort':
#lang typed/racket
((ann sort ((Listof Integer) (Integer Integer - Boolean) - (Listof Integer)))
'(3 5 4) (ann (Integer Integer - Boolean)))
=
Type Checker: Expected ((Listof Integer) (Integer Integer - Boolean) -
(Listof
Have you tried using inst instead of ann? Telling TR what to fill
in for a and b will probably be more helpful to it.
Carl Eastlund
On Sat, Aug 20, 2011 at 11:55 AM, John Clements
cleme...@brinckerhoff.org wrote:
I can't seem to make typed racket happy about using 'sort':
#lang typed/racket
On Aug 20, 2011, at 12:06 PM, Carl Eastlund wrote:
Have you tried using inst instead of ann? Telling TR what to fill
in for a and b will probably be more helpful to it.
Yay! Many thanks.
John
smime.p7s
Description: S/MIME cryptographic signature
Okay I've narrowed things down to something strange about the context in
which racket calls the planet module name resolver. In particular, if you
change the planet resolver by inserting this code right as the first thing
it does:
(parameterize ([current-namespace (make-base-namespace)])
(eval
The FFI now includes support for C arrays and unions.
As Eli has noted, libffi (which Racket uses for calls to foreign
functions) doesn't exactly support arrays within structs (which is one
the main places to use arrays), but we're trying something. If you run
into problems, be at least a little
So, it doesn't look like this is related to the sandbox, and perhaps
also unrelated to planet -- unless the resolver is hooked in some bad
way. I managed to minimize it to this:
--- tmp.rkt
#lang racket/base
(require (planet neil/numspell))
Running `racket tmp.rkt' shows the failure, and
On Aug 20, 2011, at 1:55 PM, John Clements wrote:
I was editing a file, and this occurred. My reading of this suggests that
this involves garbage collection inside of an FFI call, but perhaps *all* GC
is inside an FFI call?
Anyhow, there's no evidence to suggest that I can reproduce
Yes, the code below is what I meant with my message. Sorry I wasn't
clear enough and apparently made you waste your time.
The standard-module-name-resolver is in
plt/src/racket/src/startup.rktl, but it doesn't really seem to do
anything before handing control to planet.
Robby
On Sat, Aug 20,
Matthew pointed out that current-module-declare-name might play a role
and, sure enough, that was the problem. I've pushed a fix to planet
(since the resolver should be setting that back to #f before starting
the setup process), but probably the sandbox should also work when
that is set to
10 matches
Mail list logo
