bug#75761: Acknowledgement (request make-bytevector/pointers)

Another option would be to add a procedure to add "watch location" in a bytevector. The GC could just look at those locations. Matt

bug#75761: request make-bytevector/pointers

make-bytevector allocates pointerless memory. For using bytevectors to store structures used with C libraries we really need pointer-search, I believe. Please add. Matt

bug#73685: [patch] to silence diagnostic compile messages and welcome message (was GUILE_QUIET)

The attached three patches provide modifications to Guile to provide users the ability to silence compile and loading messages that occur when using Guile interactively. In addition, there is a separate capability added to allow users to silence the welcome message. The first is performed

bug#71980: close 71980

close 71980

bug#71980: close

close

bug#71980: tmpnam

See also bug #71796, which has smaller delta.

bug#71980: [3.0.10] posix.test uses depcrecated tmpnam

With `--disable-tmpnam'  argument to configure, posix.test fails: missing `tmpnam`. patch attached --- test-suite/tests/posix.test-orig    2024-07-03 15:45:16.352132881 -0700 +++ test-suite/tests/posix.test    2024-07-07 09:21:53.342855356 -0700 @@ -361,20 +361,21 @@    (status:exit-val (sy

bug#71765: 3.0.10: make check fails when configure --disable-tmpnam

In 3.0.10, if I use the --disable-tmpnam argument to configure, then `make check' fails Running ports.test Running posix.test In execvp of something-that-does-not-exist: No such file or directory In execvp of something-that-does-not-exist: No such file or directory In execvp of something-that-do

bug#71742: [3.0.10] make check fails when configure --disable-tmpnam

In 3.0.10, if I use the --disable-tmpnam argument to configure, then `make check' fails Running ports.test Running posix.test In execvp of something-that-does-not-exist: No such file or directory In execvp of something-that-does-not-exist: No such file or directory In execvp of something-that-do

bug#69857: ports.c, "scm-set-current-output-port" vs "set-current-output-port"

In guile-3.0.9, in ports.c I believe scheme name for setting the error output is inappropriately named "scm-set-current-output-port".   It does not show in guile because set-current-output-port is defined using the parameter current-output-port. SCM scm_set_current_input_port (SCM port) #defin

bug#68703: update to lightning.texi to work w/ texi2any --pdf (with patch)

The following changes to libguile/lightening/lightning.texi will allow it to be used with texi2any --pdf: --- lightning.texi    2022-06-20 07:44:21.159880656 -0700 +++ lightning.texi-fixed    2024-01-24 12:07:27.621547717 -0800 @@ -83,22 +83,12 @@  @ifnottex  @node Top  @top @lightning{} - -@ift

bug#62125: ref manual errors in assembler insn descriptions

Hi All, I'm pretty sure the instruction desciprtions shown below in the manual are in error.   Please let me know if my corrections are correct. Matt --- vm.texi-orig    2023-03-11 08:40:59.000388237 -0800 +++ vm.texi    2023-03-11 08:48:57.940061512 -0800 @@ -1468,8 +1468,8 @@  @var{obj}, and

bug#27782: mman patch for v3.0.9

I think this is still not there. I have found additional issues with some suggested updates. Maybe we should have a branch in the guile repo for this.

bug#61788: [3.0.9] --disable-tmpnam build fails "make check"

I built gule-3.0.9 with "configure --disable-tmpnam" and get the following failure from posix.test while running "make check": Running ports.test warning: call to primitive-fork while multiple threads are running; further behavior unspecified.  See "Processes" in the manual, for

bug#27782: mman patch for v3.0.9

Note.  I have made more changes based on feedback from the mailing list. 1) removed use of scm_c_take_typed_bytevector 2) changed code to generate PAGE_SIZE I think the mmap finalizer still needs review from the Guile experts. I'm attaching a patch to the v3.0.9 release (commit 9b20ca). Sorry a

bug#27782: patch for mmap and friends

On 1/14/23 2:42 PM, Maxime Devos wrote:     {   /* Use the fd of the port under clobber protection from concurrency. As scm_dynwind_acquire_port assumes that FILE is a port, check that first. */   SCM_VALIDATE_PORT (SCM_ARG5, file);   scm_dynwind_acquire_port (fil

bug#27782: patch for mmap and friends

On 1/13/23 4:49 PM, Matt Wette wrote: Please consider this patch for adding mmap(), munmap() and msync()  to libguile/filesys.c.  Included is update for posix.texi and test file mman.test. Once included, feature 'mman should be #t. Matt Please add the attached file: test-suite/

bug#27782: patch for mmap and friends

Please consider this patch for adding mmap(), munmap() and msync()  to libguile/filesys.c.  Included is update for posix.texi and test file mman.test. Once included, feature 'mman should be #t. Matt From 6c944174d35d43f87340c8199d47f3f088fa6ca7 Mon Sep 17 00:00:00 2001 From: Matt Wette

bug#27782: patch to add support for mmap and friends

Please disregard previous patch.  I have more to do. I'll try to catch the next release cycle.

bug#27782: patch to add support for mmap and friends

Guile Maintainers: Please consider the atttached patch for mmap and friends. Includes mmap, mmap/shared, munmap, msync. Matt From 306570beb3d1895abd03700593cc342282e4ccd1 Mon Sep 17 00:00:00 2001 From: Matt Wette Date: Tue, 20 Dec 2022 17:15:27 -0800 Subject: [PATCH] Add support for mmap

bug#32040: bug in repl/server.scm

I can confirm this still exists in guile 3.0.8: The symptom that I see is when client (telnet localhost 37146) quits server responds with the following:    In thread:   Wrong type to apply: ()  Nice find Isaac!

bug#54478: unwanted source references

I think the error is in use of "src" arg to make-seq in the following loop, in ice-9/psyntax.scm, starting at line 1633: (let lp ((var-ids var-ids) (vars vars) (vals vals) (tail (expand-tail-expr))) (cond ((null? var-id

bug#54478: unwanted source references showing up w/ -O0 (with patch)

In guile-3.0.8, when I compile w/ optimiation-level 0, I am seeing references to the beginning let-form. The following illustrates the issue. Consider foo.scm: (define *a* (make-parameter 0)) (define (foo . args) (let* ((b (+ (*a*) 1)) (c (+ (*a*) b 2))) (simple-format #t

bug#42345: 3.0.4 says "wrong num' of arg's to 1" vs 2.2.4: "wrong num' of arg's to baz"

In Guile 3.0.4, the program below results in the error: /home/mwette/zz.scm:5:2: In procedure bar: Wrong number of arguments to 1 In Guile 2.2.4, the same program results in the error: ;;; /home/mwette/zz.scm:11:14: warning: wrong number of arguments to `bar' ;;; compiled /home/mwette/.cache/

bug#27782: new patch for mma

Attached is a patch against guile master (at 3.0.4), commit 5e1748f75128107e3a0707b66df5adb95d98437e It is a incomplete, but functional, implementation of a mmap-api, including 1) mmap : low-level mmap, returns a bytevector, not searched for roots 2) mmap/search : like mmap, but not marked w/ GC_

bug#21897: verified in 3.0.2

This still exists in 3.0.2 scheme@(guile-user)> ,use (ice-9 control) scheme@(guile-user)> (call/ec (lambda (c) (with-continuation-barrier (lambda () (c "through continuation"))) "c-w-b returned")) $1 = "through continuation" scheme@(guile-user)> (version) $2 = "3.0.2"

bug#27209: patch

attached is patch : added ", or @code{#f} if there are no keyword arguments" >From 2c3f7ea9ddd9d245a000ad8b19176f132607405b Mon Sep 17 00:00:00 2001 From: Matt Wette Date: Fri, 27 Mar 2020 10:19:53 -0700 Subject: [PATCH] * doc/ref/compiler.texi: for description of lambda-case

bug#38486: done

scm (preserve-eq?): New helper. (sigbits-union): Use the new helper. Fixes bugs.gnu.org/38486. Thanks to Zack Marvel for the bug report and Matt Wette for tracking it down.

bug#29001: git patch

Attached is the git patch against the following guile commit: bef5e0b3938cc88e3a1a1ac590b009875cc38162 >From 71ff7e79369a4514a961fc5cf76593b254c32d4c Mon Sep 17 00:00:00 2001 From: Matt Wette Date: Sun, 22 Mar 2020 09:12:37 -0700 Subject: [PATCH] 2020-03-22 Matt Wette * configure

bug#38486: specialize-numbers.scm: compute-significant-bits

I've narrowed it down to the named let loop "lp" in this routine in module/language/cps/specialize-numbers.scm (define (compute-significant-bits cps types kfun)   "Given the locally inferred types @var{types}, compute a map of VAR -> BITS indicating the significant bits needed for a variable.  B

bug#38486: hang

So I hacked modules/cps/optimize.scm to display the optimization phases. Here is what I got up to the point where guile hangs. It looks like it's hanging in specialize-numbers. running eliminate-dead-code running prune-top-level-scopes running simplify running contify running inline-constructors

bug#38486: try all options

The behavior is that guild compile 38486-1.scm hangs These optimization flags eliminate that behavior: the code compiles quickly:  -O0  -O1  -Ono-simplify  -Ono-contify  -Ono-cse  -Ono-specialize-numbers These optimization flags are ineffective: the behavior remains:  -Ono-partial-eval  -Ono-eli

bug#38486: compile livelock

I reproduced the behavior on 2.2.7.   I also found that -O0 and -O1 do not have a problem: $ meta/guild compile -O0 38486-1.scm -o 38486-1.go wrote `38486-1.go' $ meta/guild compile -O1 38486-1.scm -o 38486-1.go wrote `38486-1.go' $ meta/guild compile -O2 38486-1.scm -o 38486-1.go ^Cerror: int

bug#29001: patch

Note: The above patch makes tmpnam report that it is deprecated.

bug#29001: --disable-tmpnam

"make" and "make check" succeeds on 3.0.1 / Ubuntu 18.04 / x86_64 with 1) --disable-tmpnam 2) --enable-tmpnam 3) neither This updates libguile/posix.c and configure.ac to allow extra configuration option --disable-tmpnam.   This is made available for installations that don't want to allow the ins

bug#40075: patch

Below is a patch against guile-3.0.1. "make" works, "make check" works. null-threads.c was not seeing "SCM_USE_NULL_THREADS" so include libguile/scmconfig.h null-threads.c was not seeing def of SCM_API so include "libguile/scm.h" web-server.test used (if (provided? 'threads) ) so #undefined if

bug#40075: pthreads

I have duplicated this w/ 3.0.1 on ubuntu 18.04, x86_64. I will look into this today. Matt

bug#40075: oops

scratch that - 29001 is about the tmpnam issue

bug#21076: dynamic-link

I have encountered this issue with dynamic-link. I propose a small patch to Guile (works on v3.0.1) to address it. At least this allows me to load the file if I know the exact name. This patch fixes issue with libtool and OSes which use non-standard extensions for shared libraries.  The libltdl f

bug#29001: tmpnam

On 3/18/20 4:48 PM, Matt Wette wrote: This is related to 4075 at debbugs.gnu.org \ er  40075 at debbugs.gnu.org working on patch against 3.0.1 going better this time

bug#29001: tmpnam

This is related to 4075 at debbugs.gnu.org

bug#40075: tmpnam

This is related to 29001 at debbugs.gnu.org

bug#17940: (begin)

I have run into this also, in writing macros. My workaround is to always add (if #f #f) in my forms that allow no expressions in the body. See second-to-last line below. (define-syntax sx-match-1   (syntax-rules ()     ((_ v (pat exp ...) c1 ...) (let ((kf (lambda () (sx-match-1 v c1 ...

bug#39196: patch

Here is a patch to make it work. --- module/system/repl/command.scm-orig    2020-01-19 13:46:55.041867060 -0800 +++ module/system/repl/command.scm    2020-01-19 13:48:02.982264244 -0800 @@ -30,7 +30,7 @@    #:use-module (system vm program)    #:use-module (system vm trap-state)    #:use-module

bug#39196: ,L brainfuck : not working in 3.0.0

scheme@(guile-user)> (version) $4 = "3.0.0" scheme@(guile-user)> ,L brainfuck While executing meta-command: Unbound variable: language-title noting change to system/base/syntax.scm: < ;; Copyright (C) 2001, 2009, 2016 Free Software Foundation, Inc --- > ;; Copyright (C) 2001, 2009, 2016, 2019 Fr

bug#32367: sigaction hangs

On 12/15/18 8:46 AM, Matt Wette wrote: -s seems to end up using load-in-vicinity: I can re-create this way: mwette$ guile -L `pwd` -c '(load-in-vicinity (getcwd) "foo.scm")' ;;; note: auto-compilation is enabled, set GUILE_AUTO_COMPILE=0 ;;;   or pass the --no-auto

bug#32367: sigaction hangs

-s seems to end up using load-in-vicinity: I can re-create this way: mwette$ guile -L `pwd` -c '(load-in-vicinity (getcwd) "foo.scm")' ;;; note: auto-compilation is enabled, set GUILE_AUTO_COMPILE=0 ;;; or pass the --no-auto-compile argument to disable. ;;; compiling /home/mwette/proj/guile

bug#32429: scm_to_off_t

can you post your config.log file? That error is bizarre because scm_to_off_t should be #defined from libguile/syscalls.h

bug#29845: close

Not needed.

bug#30094: load-lang patch

I am now posting patch for this to github.com mwette guile-contrib patch-2.2.4 load.patch .

bug#30094: test script

The following test script works with the 2.2.4 patch provided. elisp tests don't work well but I think that is elisp issue. This stuff works with my own developed. ;;; load-lang.test --*- scheme -*- (define-module (test-suite test-load-lang) #:use-module (test-suite lib))

bug#30094: patch for adding external lang support

There is a left-over simple-format statement which should be removed.

bug#30094: patch for adding external lang support

Here is a patch against 2.2.4. It compiled and passed "make check". Still to go: some test-suite scripts. --- module/system/base/compile.scm-orig 2016-08-01 04:32:31.0 -0700 +++ module/system/base/compile.scm 2018-09-04 06:27:53.056330281 -0700 @@ -28,6 +28,7 @@ #:use-module (ic

bug#30094: proposed code for alt languages

Hey all, The attached code implements alternative languages. It is roughly a patch to (system base compile). It provides 1) extra procedures lang-from-port and lang-from-file 2) the global %file-extension-map 3) an altered version of compile-file Behavior: 1) if the first line of the file is `#

bug#32164: guile --language=xxx

This bug applies to version 2.2.3.

bug#32164: guile --language=xxx fails with outdated go file

If using ``--language='' option with `guile' and one of the language source files is newer than the .go file, guile will fail with "no such language". See below. Apparently the file is compiled. I'm guessing this could be an issue a circular dependency. mwette$ guile --language=javascript ,zz.j

bug#31154: Acknowledgement (predicate function for foreign-object not obvious)

Perhaps it belongs with the demo in Section 5.5 Defining New Foreign Object Types, with the image demo.

bug#31154: predicate function for foreign-object not obvious

The reference manual (guile-2.2.3) does not provide instruction on how to generate a predicate for a foreign object, and it is not obvious. Maybe it doesn't need to be in the manual but should then be in an example somewhere, IMO. The following patch to the reference manual is one option: --

bug#30869: regexp-substitute/global does not like #\nul in string

version = 2.2.3 I think this is a bug: regexp-substitute/global rejects strings with #\nul: scheme@(guile-user)> (regexp-substitute/global #f "b.d" "ab\x00de" 'pre 'post) ERROR: In procedure regexp-exec: string contains #\nul character: "ab\x00de"

bug#30276: typos in the manual (api-macros, vm)

A couple typos in the maual --- api-macros.texi-fix 2018-01-28 13:22:52.305712852 -0800 +++ api-macros.texi 2017-11-19 11:22:01.0 -0800 @@ -1280,7 +1280,7 @@ In this way @code{(macroexpand @var{foo})} is equivalent to @code{(macroexpand @var{foo} 'e '(eval))}. The second argumen

bug#29910: minor typo in doc/ref/api-data.texi: u8 vs u32 literal array syntax

patch is with respect to guile-2.2.2, in doc/ref I believe #u8(0 1 2) is U8 array not U32. --- api-data.texi.orig 2017-12-30 09:47:17.0 -0800 +++ api-data.texi 2017-12-30 09:47:49.0 -0800 @@ -7261,7 +7261,7 @@ is a non-uniform array of rank 2; a 2@cross{}3 matrix with in

bug#29845: Acknowledgement ([wishlist] bitvector-copy)

and I would like something like bitvector_move_left I need to dup bitvector to a new one with one extra bit on the end [1|0|1] ==copy==> [1|0|1|0]

bug#29845: [wishlist] bitvector-copy

please add bitvector-copy and scm_bitvector_copy

bug#27782: mmap for guile

here is a start on test-suite/tests/mmap.test + mmap.test --- test suite for Guile's mmap functions -*- scheme -*- + + +(define-module (test-mmap) + #:use-module (test-suite lib)) + +(use-modules (rnrs bytevectors)) + +(with-test-prefix "mmap" + + (pass-if "basics" +(let* ((siz #x10

bug#27782: mmap for guile

got it. > On Nov 24, 2017, at 8:22 AM, Nala Ginrut wrote: > > Thanks for the work! Could you please add MAP_POPULATE too? > > 2017年11月24日 下午11:55，"Matt Wette" <mailto:matt.we...@gmail.com>>写道： > > > > > > I did a little more on this.

bug#27782: mmap for guile

I did a little more on this. Here is the latest. It provides mmap (not searched) and mmap/search (searched for pointers to GC). --- libguile/filesys.c.orig 2017-03-01 10:54:31.0 -0800 +++ libguile/filesys.c 2017-10-28 10:05:10.0 -0700 @@ -1828,9 +1828,14 @@  +#inc

bug#27579: wchar_t needed also

We will also need wchar_t. And what about char16_t and char32_t ?

bug#27782: mmap for guile 2.2.2

I located GC_exclude_static_roots() call for the collector. Now mmap/search will run allocate a bytevector as in the previously provided patch. And mmap will call mmap/search and then apply GC_exclude_static_roots() to the mmap'd region.

bug#27782: mmap for guile 2.2.2

I worked on the code a bit more. It is currently implemented as a #include "mman.c" in filesys.c and another file "mman.c". I believe this needs to be reviewed by someone who understands the constraints on foriegn pointers and finalization better than I do. Also, there is a comment to add a

bug#29001: tmpnam option

OK I have it working. Complete patch here, assuming config.h.in is generated by your autotools. --- libguile/posix.c-orig 2017-10-24 05:28:30.0 -0700 +++ libguile/posix.c2017-10-24 05:28:51.0 -0700 @@ -1557,6 +1557,7 @@ } #undef FUNC_NAME +#ifdef ENABLE_TMPNAM #ifd

bug#29001: [wishlist] configure option to disable tmpnam

It would be nice to have a configure option to disable/enable including the POSIX function tmpnam. It may give Guile a more "code secure" option to have the default be disabled. I tried to generate a patch, but I could not get it to work both ways (disable and enable). My autoconf knowledge and

bug#27579: System foreign

While we are at it, I need long-long-int and unsigned-long-long-int. This is for my FFI helper, to auto-code FFI code. (Then my chore is to deal with varargs.)

bug#27782: Acknowledgement ([wishlist] scheme level mmap)

Works on guile-2.2.2: > (use-modules (system foreign)) > (define raw (mmap %null-pointer #x1000 (logior PROT_READ PROT_WRITE) (logior MAP_ANON MAP_PRIVATE) -1 0)) > (munmap raw #x1000) $ diff filesys.c.orig filesys.c + #include "mmap.c" + void scm_init_filesys () { + MMAP_DEFS;

bug#27782: [wishlist] scheme level mmap

There was an implicit request on the user-guile mailing list (20 Jul 2017) to provide a scheme language call to mmap. I am working on a prototype and will post when I get a simple case working. Here is non-working code so far: Currently I have this in a file “mmap.c” and #including into files

bug#27579: intptr_t and uintptr_t

I found the issue. I called the symbol intptr_t in foreign.c and intptr in foreign.scm Now both use intptr_t and uintptr_t, with patch below. Matt --- libguile/foreign.c-orig 2017-07-04 15:57:55.0 -0700 +++ libguile/foreign.c 2017-07-04 16:02:45.0 -0700 @@ -56,6 +56,8 @@

bug#27579: Intptr and uintptr

I sent this patch prematurely. The symbol intptr is not visible at the interpreter prompt via (use-modules (system foreign)). — Matt

bug#27579: [patch] add intptr uintptr to (system foreign)

I submit this patch for adding intptr and uintptr to (system foreign). This is with reference to guile-2.2.2 and changes libguile/foreign.c and module/system/foreign.scm. After application, I was albe to get guile-2.2.2 to complete “make” and “make check”. No specific tests for this patch have b

bug#27209: [minor] documentation of tree-il/lambda-case

In the documentation on lambda-case under Tree-IL there is a paragraph starting with “kw is a list of the form …”. The paragraph does not specify what to use if there are no keywords. I believe in this case the argument should be #f.

bug#25324: Make check fails on FreeBSD 11.0

“make check” works for guile 2.2.2 on FreeBSD 11.0.

bug#15227: Possible bug in (web server)

I tried this on guile 2.2.0 and the localhost queries seem to work OK. — Matt mwette$ ./15227 server ;;; note: auto-compilation is enabled, set GUILE_AUTO_COMPILE=0 ;;; or pass the --no-auto-compile argument to disable. ;;; compiling /Users/mwette/proj/scheme/guile/bugs-guile/./15227 ;;;

bug#26183: patch for guile-2.2.0 on FreeBSD 11.0

> On Apr 19, 2017, at 7:20 AM, Andy Wingo wrote: > > On Mon 20 Mar 2017 02:39, Matt Wette writes: > >> Attached is a patch to apply to guile-2.2.0 to compile on FreeBSD 11.0 >> Changes >> 1) updated README to indicate which extra FreeBSD packages need to be >

bug#22601: guile.texi section on ice-9 match

> On Mar 19, 2017, at 4:42 PM, Matt Wette wrote: > > Patch for guile-2.2.0, wrt doc/ref/match.texi, is attached. > > Updated patch. Upon proofread, I found an error wrt “ooo” in the original. match.texi.patch Description: Binary data

bug#26183: patch for guile-2.2.0 on FreeBSD 11.0

Attached is a patch to apply to guile-2.2.0 to compile on FreeBSD 11.0 Changes 1) updated README to indicate which extra FreeBSD packages need to be installed 2) minor change to configure.ac (implies autoconf needs to run) to use bdw-gc-threaded instead of bdw-gc With this, I had successful config

bug#24862: guile 2.0.13 build error under macOS 10.12 Sierra

This should be closed. — Matt

bug#22601: guile.texi section on ice-9 match

Patch for guile-2.2.0, wrt doc/ref/match.texi, is attached. match.texi.patch Description: Binary data

bug#22601: guile.texi section on ice-9 match

I can’t find my fixes. I will start working this again., — Matt

bug#25509: FreeBSD 11.0

HI saffron, I have been able to get guile-2.1.8 to build on FreeBSD 11.0. I had to “pkg install” : pkgconf, gmake, boehm-gc-threaded Then to work around following config issue, I “pkg install autotools” Edit autoconf.ac and find “bdw-gc”, change to “bdw-gc-threaded” PKG_CHECK_MODULES([BDW_GC], [

bug#22901: drain-input doesn't decode

> On Feb 26, 2017, at 9:46 AM, Matt Wette wrote: > > I put together a test and tried on 2.1.7 - my test fails. See attached. > > (pass-if "encoded input" >(let ((fn (test-file)) > (nc "utf-8") > (st "\u03b2\u03b1\u03b4

bug#22901: drain-input doesn't decode

I put together a test and tried on 2.1.7 - my test fails. See attached. (pass-if "encoded input" (let ((fn (test-file)) (nc "utf-8") (st "\u03b2\u03b1\u03b4 \u03b1\u03c3\u03c3 am I.") ;;(st "hello, world\n") ) (let ((p1 (open-output-file fn #:en

bug#25872: [minor] SXML Transform section in ref-man does not mention `(sxml transform)'

This is a minor nit. The reference manual does not direct user to do `(use-modules (sxml transform))' The following patch is wrt guile-2.1.7/doc/ref/sxml.texi --- sxml.texi.orig 2017-02-25 06:33:38.0 -0800 +++ sxml.texi 2017-02-25 06:36:00.0 -0800 @@ -457,6 +457,8 @@ @no

bug#25822: ref manual typo in "Shifts, Resets and All That"

So I was reading on prompts and came across this sentence in the section entitled “Shifts, Reset and All That”: The default handler accepts a procedure of one argument, which will called on the captured continuation, within a prompt. Something appears to be missing text or something; I have no

bug#25791: atomics on macOS

I saw this too and gave up trying to fix it. I think this needs to be worked bottom up: I believe the C99 standard is specific about the type qualifier _Atomic being present for the first argument. The workaround for now IMO, is to change atomics-internal.h from this: #ifdef HAVE_STDA

bug#25790: guile-2.1.7 SOCK_CLOEXEC etc

I see the same problem on Mac OS. If the following appeared in a header somewhere CLOEXEC would work: #define SOCK_CLOEXEC FD_CLOEXEC It would work. Note the implementation in lib/accept4.c: #ifndef SOCK_CLOEXEC #define SOCK_CLOEXEC 0 #endif Then later the flags to accept4 have SOCK_CLOEXEC set,

bug#25785: scm_c_make_polar broken

So, I may have missed the update for config.h.in. I am not sure this gets autogenerated or not. --- config.h.in.orig2017-02-18 13:27:19.0 -0800 +++ config.h.in 2017-02-18 13:27:45.0 -0800 @@ -2183,6 +2183,9 @@ /* Define to 1 if you have the `sincos' function. */ #undef HAV

bug#25786: [PATCH] use of abs() in numbers.c: should be labs()

In libguile/numbers.c abs() is used for long argument where labs() should be used. A patch is included below for guile-2.1.7. /* Returns log(n/d), for exact non-zero integers n and d */ static SCM log_of_fraction (SCM n, SCM d) { long n_size = scm_to_long (scm_integer_length (n)); long d_si

bug#25785: scm_c_make_polar broken on APPLE

numbers.c:scm_c_make_polar in guile-2.1.7 breaks on APPLE (macOS 10.12).Reason: APPLE does not have sincos(), and gcc will optimize sin/cos to use cexp(), and cexp() does not preserve sign of zero.Patch includes (and attached) for guile-2.1.7.  This uses macOS __sincos().Example:In scm_c_make_polar

bug#24677: Failure in automatic tests for guile 2.0.13

This may be related to the following, in which I have posted some fixes: https://debbugs.gnu.org/cgi/bugreport.cgi?bug=24862

bug#24677: Failure in automatic tests for guile 2.0.13

I did a make check, Mac OS Sierra, and saw it crash too. $ cd test-suite/lalr $ ../../meta/guile ... scheme@(guile-user)> (load "test-lr-basics-01.scm") ERROR: In procedure #: ERROR: Wrong type to apply: # Entering a new prompt. Type `,bt' for a backtrace or `,q' to continue. scheme@(guile-us

bug#24884: Segfault on (mkstemp! "XX" 0)

Please check 24862, which includes a patch: https://debbugs.gnu.org/cgi/bugreport.cgi?bug=24862 The issue is that (at least on Mac OS 10.12) mkstemp will only accept specific flags.

bug#24862: guile 2.0.13 build error under macOS 10.12 Sierra

Here is a patch that gets mine working. Not sure it solves all the Sierra issues. *** libguile/filesys.c-orig 2016-11-05 07:49:09.0 -0700 --- libguile/filesys.c 2016-11-10 05:38:53.0 -0800 *** *** 1486,1491 --- 1486,1494 mode_bits = scm_i_mode_b

  1   2   >