gmane.lisp.scheme.scheme48

Re: finalizers vs. weak pointers

Ivan Shmakov — 2008-07-22T11:42:23

>> Seemingly, weak pointers are invalidated before the finalizers for >> the object are run. I wonder, what is the reason behind such a >> behaviour? > The rule is that when a finalizer is called it is passed the only > existing refererence to the finalized object. If the weak pointers > were not invalidated first there would be other references to the > finalized object, allowing it to be accessed after finalization. I've got it. Thanks!

Re: finalizers vs. weak pointers

Richard Kelsey — 2008-07-22T09:32:14

From: Ivan Shmakov theory.asu.ru> Date: Tue, 22 Jul 2008 09:14:02 +0700 Seemingly, weak pointers are invalidated before the finalizers for the object are run. I wonder, what is the reason behind such a behaviour? The rule is that when a finalizer is called it is passed the only existing refererence to the finalized object. If the weak pointers were not invalidated first there would be other references to the finalized object, allowing it to be accessed after finalization. -Richard Kelsey

Re: finalizers vs. weak pointers

Ivan Shmakov — 2008-07-22T08:13:15

>> Seemingly, weak pointers are invalidated before the finalizers for >> the object are run. I wonder, what is the reason behind such a >> behaviour? > While weak pointers are invalidated by the GC itself, finalizers are > run *after* the GC, I. e., after all the garbage objects were ``marked'' as such, but before they're actually removed from memory? > so this order is pretty much intrinsic. (Programs are generally > ill-advised to rely on the ordering of finalizer-like GC actions.) Indeed, I was quite surprised that the issue like this has ever appeared on my way.

Re: finalizers vs. weak pointers

Michael Sperber — 2008-07-22T07:39:02

Ivan Shmakov theory.asu.ru> writes: While weak pointers are invalidated by the GC itself, finalizers are run *after* the GC, so this order is pretty much intrinsic. (Programs are generally ill-advised to rely on the ordering of finalizer-like GC actions.)

c/external.c:727:8: warning: extra tokens at end of #endif directive

Ivan Shmakov — 2008-07-22T04:35:32

$ nl -ba c/external.c ... 723 for (i = 0; i < nargs; i++) { 724 s48_ref_t ref = va_arg(arguments, s48_ref_t); 725#ifdef DEBUG_FFI 726 fprintf(stderr, "call_scheme_2: pushing arg %d ref %x\n", i, ref); 727#endif DEBUG_FFI Shouldn't there be no DEBUG_FFI on this line? 728 s48_push(s48_deref(ref)); 729 } ... $

finalizers vs. weak pointers

Ivan Shmakov — 2008-07-22T02:14:02

Seemingly, weak pointers are invalidated before the finalizers for the object are run. I wonder, what is the reason behind such a behaviour? foo> (define weak-pointer (let* ((s (string #\f #\o #\o)) (w (make-weak-pointer s))) (add-finalizer! s (lambda (obj) (write `(finalizing: ,obj)) (newline) (write `(weak-pointer => ,(weak-pointer-ref w))) (newline))) w)) ; no values returned foo> ,collect Before: 2009301 out of 3000000 words available After: 2038984 out of 3000000 words available foo> (finalizing: "foo") (weak-pointer => #f) foo>

Re: Message archive?

Taylor R Campbell — 2008-07-21T13:57:16

Date: Mon, 21 Jul 2008 10:43:02 +0200 From: "Michael Lesniak" uni-kassel.de> do you have a message archive of old messages from this mailing list? The list is archived by Gmane: see .

Message archive?

Michael Lesniak — 2008-07-21T08:43:02

Hello, do you have a message archive of old messages from this mailing list? Best regards, Michael Lesniak

Internals of the VM

Michael Lesniak — 2008-07-21T08:41:24

Hello, my name is Michael Lesniak and I am a research associate in the parallel programming languages group of the university of Kassel. I'd like to take a deeper look at the (automatic) parallelization of virtual machines and think that scheme48 code, esp. its virtual machine, seems to be quite good for this[1]. Do you have any resources beside the papers "A tractable scheme implementation" and "Pre-Scheme: A Scheme dialect..."? I've overlooked and which are good start with? Is chapter five of SICP a good starting point (even scheme48's VM uses a stack?) (read it a couple of years ago but forgot most of the details). Any general suggestions before starting to hack on scheme48's VM? Best regards and thanks for your help, Michael Lesniak [1] ...and I like Scheme in general and can therefore connect work with fun :)

Re: bug in CPS conversion

Michael Sperber — 2008-07-11T06:01:54

"Dimitris Vardoulakis" ccs.neu.edu> writes: Dirk Hüsken just fixed this. (Thanks, Dirk!)

Re: File too large

Michael Sperber — 2008-07-05T16:26:13

Ivan Shmakov theory.asu.ru> writes: I've just pushed this---thanks!

Re: scheme48-1.8 and --as-needed, DESTDIR, -D_GNU_SOURCE

Michael Sperber — 2008-07-05T15:59:55

Panagiotis Christopoulos gentoo.org> writes: This should go away by itself in a future version, as the networking code has been rewritten already in the development version to not even make use of it. Yes, this will automatically go into the next release. I've just applied and pushed this. Thanks (and thanks for the explanation of --as-needed)!

Re: scheme48-1.8 and --as-needed, DESTDIR, -D_GNU_SOURCE

Panagiotis Christopoulos — 2008-07-01T08:03:32

Hi Mike, On 08:40 Tue 01 Jul , Michael Sperber wrote: :) From GNU linker's man page: --as-needed --no-as-needed This option affects ELF DT_NEEDED tags for dynamic libraries mentioned on the command line after the --as-needed option. Normally, the linker will add a DT_NEEDED tag for each dynamic library mentioned on the command line, regardless of whether the library is actually needed. --as-needed causes DT_NEEDED tags to only be emitted for libraries that satisfy some symbol reference from regular objects which is undefined at the point that the library was linked. --no-as-needed restores the default behaviour. In reality, it is an experimental flag, used by some of our fellow users, and from time to time, we fix bugs related with linker failures when it is used. But it is not considered safe, for production use and it's buggy in earlier versions of binutils. However, theoritically, the idea behind it, is not bad (the use of the --as-needed flag allows the linker to avoid linking

Re: Fixed select bug in devel

Michael Sperber — 2008-07-01T06:43:13

Andrew Lentvorski allcaps.org> writes: Applied and pushed. Thanks for the report!

Re: scheme48-1.8 and --as-needed, DESTDIR, -D_GNU_SOURCE

Michael Sperber — 2008-07-01T06:40:35

Thanks for the report! Panagiotis Christopoulos gentoo.org> writes: Could you briefly explain what --as-needed does, and why it's in Gentoo's LDFLAGS? Should we add it generally?

Fixed select bug in devel

Andrew Lentvorski — 2008-07-01T00:28:16

The select system got broken in the last round of devel changes. The patch is very minor: $ hg diff diff -r 9f6b8596d82b c/unix/event.c --- a/c/unix/event.cFri Jun 27 08:35:39 2008 +0200 +++ b/c/unix/event.cMon Jun 30 17:23:11 2008 -0700 < at >< at > -841,7 +841,7 < at >< at > return errno; } } -#elif defined HAVE_SIGNAL +#elif defined HAVE_SELECT /* * Call select() on the pending ports and move any ready ones to the ready * queue. If wait is true, seconds is either -1 (wait forever) or the

scheme48-1.8 and --as-needed, DESTDIR, -D_GNU_SOURCE

Panagiotis Christopoulos — 2008-06-30T18:49:00

Re: Where is the source code repository?

Andrew Lentvorski — 2008-06-28T08:31:08

Errr ... ... thanks. Now I feel like a complete moron. Note to self: use Google less and project website more. Thanks, -a

Re: Where is the source code repository?

Ivan Shmakov — 2008-06-28T04:11:17

AL> Where is the s48 repository located? Given what we discussed AL> before, it would be best if I used the most recent code for stuff. Did you try the ones mentioned on the Scheme48 Web pages? --cut: http://www.s48.org/development.html-- Checking out the code The repositories for the development and the stable code are separate. Both are reachable via http (browsable in a regular browser, but read-only). Stable/release repository: http://www.deinprogramm.de/cgi-bin/hgs48-stable.cgi Development: http://www.deinprogramm.de/cgi-bin/hgs48.cgi --cut: http://www.s48.org/development.html-- PS. ... And it seems to me that the here links are the real evil.

Where is the source code repository?

Andrew Lentvorski — 2008-06-28T00:52:25

Where is the s48 repository located? Given what we discussed before, it would be best if I used the most recent code for stuff. Thanks, -a

Re: File too large

Ivan Shmakov — 2008-06-27T07:28:57

[...] > Error: File too large > (&primitive-i/o-error (status . 27) (operation . #{Procedure 2490 > (call-with-input-file in channel-ports)}) (arguments "/work/gene/ > xml/Danio_rerio.xml")) > make: *** [/work/gene/ttl/Danio_rerio.ttl] Error 1 [...] The following patch should fix the problem on 32-bit platforms where -D_FILE_OFFSET_BITS=64 setting is supported. The standard `--disable-largefile' option is added to `configure' as well. I guess it wouldn't be too hard to backport this patch to 1.8 or 1.4. $ uname -m i586 $ LC_ALL=C dd if=/dev/null of=bigfile bs=1M seek=8192 0+0 records in 0+0 records out 0 bytes (0 B) copied, 0.000154792 seconds, 0.0 kB/s $ wc -c bigfile 8589934592 bigfile $ scheme48 Welcome to Scheme 48 1.9T (made by ivan on Fri Jun 27 13:43:34 NOVST 2008) Copyright (c) 1993-2008 by Richard Kelsey and Jonathan Rees. Please report bugs to scheme-48-bugs< at >s48.org. Get more information at http://www.s48.org/. Type ,? (comma question-mark) for help. #\nul diff -r c96b54