Gambit-list Février 2007

gambit-list@iro.umontreal.ca

30 participants
37 discussions

Re: [gambit-list] open-tcp-client behavior
by schemeway＠sympatico.ca 19 Fév '07

19 Fév '07

François, > I am currently working on a multi-platform project and I stumbled on > a difference in the behavior of open-tcp-client between MacOSX and > Windows (MinGW) using Gambit-C 4.0 beta20. The difference occurs > when I open a connection to an address where no server is running and > try to read from it. In MacOSX I get an exception and in Windows I > get an infinite loop. I noticed this behaviour too a few days ago (on Cygwin). If you set a timeout on the input connection, read-line returns #!eof instead of throwing an exception. That's misleading. I think the correct behaviour would be for open-tcp-client to throw an exception, not wait until read-line is called. Dominique _______________________________________________ Gambit-list mailing list Gambit-list(a)iro.umontreal.ca https://webmail.iro.umontreal.ca/mailman/listinfo/gambit-list

1 0

open-tcp-client behavior
by François Magnan 19 Fév '07

19 Fév '07

Hi, I am currently working on a multi-platform project and I stumbled on a difference in the behavior of open-tcp-client between MacOSX and Windows (MinGW) using Gambit-C 4.0 beta20. The difference occurs when I open a connection to an address where no server is running and try to read from it. In MacOSX I get an exception and in Windows I get an infinite loop. Here is a trivial test: (define (test-tcpclient) (let ((p (open-tcp-client (list server-address: "127.0.0.1" port-number: 8087 eol-encoding: 'cr-lf )))) (display "test\n\n" p) (read-line p))) On MacOSX I get: *** ERROR IN ##read-line -- Connection refused (read-char '#<input-output-port #3 (tcp-client "127.0.0.1" 8087)>) On Windows I get an infinite loop waiting for the read-line to terminate. I know I can put some timeout on the read-line but I think the MacOSX behavior is the correct one. Thank you, Francois Magnan

1 0

termite moved?
by Phil Dawes 19 Fév '07

19 Fév '07

Hi Guillaume, I can't resolve toute.ca anymore. Has termite moved? $ host toute.ca Host toute.ca not found: 3(NXDOMAIN) Many thanks, Phil

2 1

Compiling Gambit-C to dll and loading into .NET
by James Long 13 Fév '07

13 Fév '07

Hey guys, I'm attempting to write a .NET stub for a Scheme program. I want to compile the gambit scheme code down to C and then compile it into a Windows DLL. Then I can load this DLL into .NET and write an interface for it. I've accomplished all of this (with a simple hello world program), but the .NET app always crashes. The first time I tried it, I used cygwin and compiled this scheme code: * (c-define (helloworld) () void "helloworld" "" * (display "Hello")) with gsc like this: * gsc -link func.scm and then compiled the corresponding c files with gcc like so: * gcc -Iinclude -shared -D___DYNAMIC func*.c -o func.dll -lgambc -lwsock32 -lws2_32 which compiled fine. That should produce a Windows DLL, right? I'm a little skeptical at that part because I've never used gcc to produce a windows dll before. But read on before you think that's the problem. As the final step I created the .NET interface with the following few lines: * [DllImport("func.dll")] * public static extern void helloworld(); And this compiled, but .NET threw an access violation at me when the function ran. Actually, I fixed some of the gcc flags and got it instead to throw a NullReference exception when the function returned. Since I was skeptical about gcc, I tried compiling a hand-rolled ansi c file that contained the equivalent helloworld function with basically the same gcc command as above, ran the .NET program and the function ran fine (printed "Hello World" to the console). It shouldn't be anything with the calling convention or anything either since it looks like the c file produced by scheme declares the helloworld function as __cdecl. I know that the Gambit-C code includes a bunch of other stuff though (note how I had to link the winsock libraries), so it must be somewhere in there that I'm not compiling it right or something. I found the Visual Studio solution files in the gambit directory and tried compiling gambit to a windows shared lib to see if that would help. I recompiled the Gambit-C code with visual studio and linked gambit in to a complete dll, ran the .NET app again, and now it's giving me an exception saying it can't find an entry point in the dll for the helloworld function. There's got to be something I'm missing, has anyone else gotten this to work? Or does anyone see anything wrong with how I'm compiling the dll (the loading part is relatively simple and I doubt the problem is in there)? Thanks, James

2 2

More on using Gambit Scheme on UNIX
by urrozalf＠vcn.bc.ca 13 Fév '07

13 Fév '07

First of all, thanks for the prompt reply of Marc Feeley. I tried that, according to his answer there are four steps. The first one: ./configure works fine Then when I type: $make the system answers: /bin/ksh: make: not found If I type: $whoami The system answers again: /bin/ksh: whoami: not found What am I missing? Any help will be really appreciated. Thanks. Alfonso

2 1

list ref counting with c interface
by Phil Dawes 13 Fév '07

13 Fév '07

Hi Marc, Hi List, I need a fast csv parser and so I plan to wrap a the C libcsv[1] code. My gambit interface will be doing a lot of consing and so I just wanted to check that I've correctly interpreted how the C scheme-object refcounting works (from previous posts[2]). Here's an example which returns a list of a million elements - have I done all the de-referencing right? (I'm assuming that everytime I put a scheme object into another one, I can release it - correct?) (define test-return-list (c-lambda () scheme-object #<<c-lambda-end ___SCMOBJ s,list,tmp; int i; char *p = "hello"; ___EXT(___UTF_8STRING_to_SCMOBJ)(p,&s,___STILL); list = ___NUL; for (i=0; i<1000000; i++){ tmp = ___EXT(___make_pair) (s,list, ___STILL); ___EXT(___release_scmobj) (list); list = tmp; } /* deref remaining ref counts */ ___EXT(___release_scmobj) (list); ___EXT(___release_scmobj) (s); ___result = list; c-lambda-end )) Many thanks, Phil [1] http://libcsv.sourceforge.net/ [2] https://webmail.iro.umontreal.ca/pipermail/gambit-list/2005-December/000510…

2 2

Problem with u8vector ports?
by schemeway＠sympatico.ca 13 Fév '07

13 Fév '07

Hi, I think I've just uncovered a bug with u8vector ports when the u8vector is more than 64 elements long. Here is the simplest way to reproduce this problem: > (define v (make-u8vector 64 32)) > (u8vector-set! v 62 65) > (with-input-from-u8vector v read) A > (define v (make-u8vector 65 32)) > (u8vector-set! v 63 65) > (with-input-from-u8vector v read) (u8vector-rbuf-fill #<input-port #11 (u8vector)> 1 #t) 1> The last two lines are produced by lines 3264-3265 of _io.scm, if this can be of any help. Dominique

2 1

open-process on Windows
by Aemon Cannon 13 Fév '07

13 Fév '07

Hey there, First off, thanks for the awesome scheme implementation, I'm really digging it so far. Here's my question: On my Windows XP system, running gambit - 4.0 b20, I'm having trouble with the following code: ********************************************************** (let* ((p (open-process "C:/Program Files/GnuWin32/bin/cat.exe"))) (display "1" p) (close-output-port p) (let ((bc (read-all p))) (pp bc))) ********************************************************** I expect the output to always be: (1) but about 1/3 of the time I get: () I can't figure why this code would be inconsistent. The 'cat' implementation is (I think) a direct port of GNU cat. In any case I've had the same problem with other binaries that read from STDIN and write to STDOUT. I've tried inserting calls to (newline), (force-output), using (write) or (write-char) instead of (display), all to no avail. Any ideas? Thanks, Aemon

2 1

Table resizing bug?
by TJ 13 Fév '07

13 Fév '07

Running the following code in gambit's repl causes its memory usage to shoot up and end up making the whole pc unresponsive. (define t (make-table min-load: 0.75 max-load: 0.95)) (table-set! t 0 0) (table-set! t 1 1) (table-set! t 2 2) (table-set! t 3 3) (table-set! t 4 4) I managed to interrupt it in gsi and get the following backtrace (I could never get it to interrupt in gsc): > (define t (make-table min-load: 0.75 max-load: 0.95)) > (table-set! t 0 0) > (table-set! t 1 1) > (table-set! t 2 2) > (table-set! t 3 3) > (table-set! t 4 4) *** INTERRUPTED IN ##equal?-hash 1> ,b 0 ##equal?-hash 1 ##table-access 2 ##table-resize! 3 ##table-resize! 4 ##table-resize! 5 ##table-resize! 6 ##table-resize! 7 ##table-resize! 8 ##table-resize! 9 ##table-resize! ... 654294 ##table-resize! 654295 ##table-resize! 654296 ##table-resize! 654297 (interaction) (console)@6:1 (table-set! t 4 4) 1> 600,000 resizes -.- I noticed that (- 0.95 0.75) gives .19999999999999996 on my system, so I set min-load to 0.74, which brings the difference between min and max to .20999999999999996 - above the stated minimum of 0.20. ut the problem persists. Setting min-load to 0.73 eliminates the problem, so perhaps it is merely an inaccuracy in the documentation. I'm running beta 20. Cheers, TJ

2 1

fast csv parser (libcsv wrapped for gambit)
by Phil Dawes 13 Fév '07

13 Fév '07

Hi Gambit List, I needed a fast csv parser for parsing large (multi-gig) files so I've wrapped the libcsv c code by Robert Gamble[1]. The first cut of csv.scm is here: http://phildawes.net/2007/gambit-csv/0.1/csv.scm Example usage: (define it (csv-make-iterator fname)) (it) ; returns the first row as a list (it) ; returns 2nd row ..etc.. ; (it) returns '() when it hits the end of the file. I'm very new to gambit/scheme - is this a reasonable interface or is there a more schemey idiom I should be presenting? Cheers, Phil [1] http://sourceforge.net/projects/libcsv

7 23

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

Gambit-list Février 2007