Cause timeouts in test1.sh to be honoured in the Common Lisp tests.
Further work towards integration of comon lisp support.
This is just a start towards arranging that Common Lisp based builds
For a mac snapshot do not (any more) put a copy of the source archive
Try to exclude "*.wc" files from source archives that will go into macintosh
Fix dependencies so that things are configured before being built.
Common Lisp: fix for CCL REDUCE
debianbuild: include PSL for raspberry pi etc.
libraries/core-math is (probably?) the successor to crlibm amd I may
I am trying to get Mac snapshots building working on a rather old
Apologies! Of course r5259 should be replaced by r7258.
Hi Thomas, you are right! When I tested bash version number I did it by using the command bash --version (which corresponds to /opt/local/bin/bash --version) instead of using the command /bin/bash --version. I verified it just now that using #!/opt/local/bin/bash instead of #!/bin/bash as first line in the file r5259 build.sh works fine. Best regards Marco
Just informational: The current bash version on MacOS 26.2 is: sturm@vir-2/1[~] /bin/bash --version GNU bash, version 3.2.57(1)-release (arm64-apple-darwin25) Copyright (C) 2007 Free Software Foundation, Inc. It seems that this is not updated anymore for license reasons. I do no not remember details - it might be problems with GPL 3. My personal solution is to have an up-to-date /opt/local/bin/bash via Macports in my path first but still use the old Apple bash as a login shell. sturm@vir-2/1[~] bash...
Common Lisp: updates for macOS and CLISP
OK, I'll stop using the -v operator. However, I think you've found a bug in the version of bash that you're using, which would be worth reporting. The -v operator works for me in bash 5.2 on Cygwin and Ubuntu, and it's described in the current online documentation for bash 5.3. Francis From: bugs@reduce-algebra.p.re.sourceforge.net bugs@reduce-algebra.p.re.sourceforge.net on behalf of Marco Ferraris marcoferraris@users.sourceforge.net Sent: Friday, January 16, 2026 4:46 PM To: [reduce-algebra:bugs]...
Building r7257 and r5278 Common Lisp Reduce on macOS fails, because 'bash' does not digest the following commands: [ -v lisp ] || ... if [ -v clean ]; then ... [ -v debug ] ... if [ ! -v reduce ]; then ... if [ ! -v revision ] if [ -v bootstraponly ] if [ -v coreonly ] I changed them as follows [ -n "$lisp" ] || if [ -n "$clean" ]; then [ -n "$debug" ] ... if [ -z "$reduce" ] if [ -z "$revision" ] if [ -n "$bootstraponly" ] if [ -n "$coreonly" ] and there was not any problem in building. The 'bash'...
Common Lisp: minor updates
This improves (I hope) the file that arranges macports for you. Run
Check if quick_exit is available before trying to use it.
Wrong Common Lisp Reduce header message
My latest commit should fix both the issues about revision numbers and committing versions with debugging enabled.
Common Lisp: revision numbers and debugging
I think it would be reasonable to parse the revision number out of the root directory name, so I'll think about doing that as a fall-back. I occasionally commit a version of sl-on-cl configured for debugging by mistake. (It should work, but it will be slow and balk on errors more than you might want.) I think I can provide a better way of controlling this, which will mean I can't commit the wrong version, so I'll also think about that. Thanks for a couple of useful ideas. Francis
Hi Francis, I did not understand why you insisted in asking me the result of 'svnversion'. Now I understand why: you want to use the command 'svnversion' to set the value of the variable 'revision!*'. Using 'svnversion' to set the value of the variable 'revision!*' as you do in the (e.g. r7253) file 'build.sh' works only if you are using a 'trunk' directory obtained by downloading reduce-algebra code using the standard checkout command svn checkout svn://svn.code.sf.net/p/reduce-algebra/code/trunk...
Common Lisp: fix recently introduced bugs
It would be useful to know what output is produced by running svnversion. (Please see my previous message.) What does the top of the output from a build look like? It should look something like this: franc@Centaur23 /c/REDUCE/reduce-algebra-code/common-lisp $ ./build.sh -l sbcl -c +++++ Clean build +++++ REDUCE revision number set to 7246 If you don't see the line about the revision number then it is not being set automatically. (There is no warning if it is not set automatically, but I will fix...
On my 'Windows 11 Home (25H2)' X86_64 laptop I tested the old revions of Reduce I found in a backup disk. Compiling and running Common Lisp Reduce I got the following results: i) SBCL Common Lisp Reduce r7204, r7205, r7213, r7246, r7251 REDUCE (Free SBCL version, revision ???), build date 11-Jan-2026 ... 1: bye; This seems to suggest that there is some difference between the 7246 revision I downloaded today and the one you used. ii) CLISP Common Lisp Reduce r7204, r7213, r7246, r7251 REDUCE (Free...
Get rid of residual use of MacOS DEPLOYMENT_TARGET stuff.
PSL kernel: more modernizing and cleanup
Wrong Common Lisp Reduce header message
I don't see this problem on Windows. For a version I built yesterday on SBCL I see REDUCE (Free SBCL version, revision 7246), build date 10-Jan-2026 ... What platform are you running on? revision ??? means that the build script could not automatically determine the Subversion revision number. What happens if you go into the packages directory that you are using for your build and run svnrevision I currently get (on Cygwin) franc@Centaur23 /c/REDUCE/reduce-algebra-code/packages $ svnversion 7245:7246...
Unwind a misbegotten patch where I had put some C++ code in a file
Wrong Common Lisp Reduce header message
I think the short answer is that you should not use wincsl_6860.cab. To install the latest prebuilt version of REDUCE on Windows you should download and execute Reduce-Setup_6860.exehttps://sourceforge.net/projects/reduce-algebra/files/snapshot_2024-08-12/windows/Reduce-Setup_6860.exe/download. If I recall correctly, wincsl_6860.cab is an alternative packaging of the files necessary to run CSL REDUCE, which was provided because a user asked for it (whereas Reduce-Setup_6860.exehttps://sourceforge.net/projects/reduce-algebra/files/snapshot_2024-08-12/windows/Reduce-Setup_6860.exe/download...
PSL: cleanup
PSL: update wfloat.sl source for changes in C kernel
Update PSL sparsify program (Maybe no longer needed)
I am Reduce newbie, and have a quick question - how should I use wincsl_6860.cab file which can be downloaded in the Windows file section? I see it contains some fonts and bunch of executables, but no idea what is purpose of them.
Add functions sinh,cosh,tanh,hypot to PSL kernel for all relevant systems.
Common Lisp: declaim all function types
OSL kernel: extra function interfaces uxsinh, uxcosh, uxhypot for Linux and FreeBSD
PSL/CSL comptibility: define function symbol!-argcount and prinhex in support/pslrend.red
How to set up a build environment for Window on ARM. I can
Small fixes
Update BPSL kernel for Linux x86_64 (forgot one build step)
Update the Reduce bibliography from the web site books page.
Web site: add new book
PSL for x86_64: Add sinh, ocsh, hypot to kernel (as uxsinh, uxcosh, uxhypot)
PSL Kernel: fix a misplaced curly brace in the last commit
Make builds of crlibm more robust for WoA by arranging that tbx_timing.h does
Cleanup of kernel code for AMD64 on Linux:
Common Lisp: fix tracing on Linux
I subversion was being fussy about replacing symbolic links with
Avoid the use of symbolic links between the cslbase and fox directories by
This is basically a follow-on from introducing namespaces in the CSL source
Detect the latest version of MacOS so its name is used rather than the
Fix typo in scripts/test1.sh
Once a year I update the copyright date in all the CSL files. Well here in the
Common Lisp: minor improvements
For the using-clang stuff it is convenient to have these
The new directory csl/new-embedded/using-clang contains a Makefile
In CSL I allocate memory in "Page"s, and I need each such Page to be
Avoid use of make_unsigned to get the unsigned version of 128 bit integers
Comments on compiling r7222 Reduce on macOS
Here I rename arithlib.hpp to arithlib.cpp, and then obviously need
Fix fairly longstanding typos here.
Compiling r7220 Common lisp Reduce (SBCL, CLISP, CCL) under x86_64 macOS Big Sur v11.7.10 and under M1 macOS Tahoe v26.2 works fine. Marco REMARK. Compiling CCL on M1 Macs works well and produces X86_64 code which can be run under Rosetta 2. The file CCL64 does not recognize the Apple Silicon chip so that one has to modify it by adding the code arm64) OPENMCL_KERNEL=dx86cl64 ;; in the Dawin) case section.
CLISP problem with #_getpid in sl-on-cl.lisp
I have implemented getpid in CCL REDUCE as suggested. The version I have just committed (r7220) builds REDUCE without errors for me on SBCL, CLISP and CCL on Windows. Francis
Common Lisp: better tracing and CCL support
Work on tuning thresholds for transitions between multiplication methods.
Make arithlib.hpp use my cleaned-up code for running subtasks in a thread
Fixed to previous checkin now I have tested ffttrune on Raspberry pi, Macbook
That seems to work for me today. I don't know why it didn't before. But I want to test this thoroughly, and perhaps fix a few more issues, before I commit an update. Thanks, Francis From: bugs@reduce-algebra.p.re.sourceforge.net bugs@reduce-algebra.p.re.sourceforge.net on behalf of Marco Ferraris marcoferraris@users.sourceforge.net Sent: Sunday, December 07, 2025 12:11 PM To: [reduce-algebra:bugs] 186@bugs.reduce-algebra.p.re.sourceforge.net Subject: [reduce-algebra:bugs] #186 CLISP problem with...
Whenon on x86_64 the arithlib.hpp multiplier now tries to check of avx or sse4.2
Yet more work on FFT multiplication - the hardest part towards making
On M1 macOS Tahoe 26.1 and X86_64 macOS Big Sur 11.7.10 I use CCL64 Version 1.13 DarwinX8664 GNU CLISP 2.49.95+ (2024-11-03) On Windows 11 I use CCL64 Version 1.12.1 WindowsX8664 GNU CLISP 2.49.92 (2018-02-18) The function 'ccl::getpid' is defined in the file 'level-1/linux-files.lisp' (lines 918 through 922). Using both Clozure Common Lisp Version 1.12.1 (v1.12.1) WindowsX8664 and Clozure Common Lisp Version 1.13 WindowsX8664 I get ? (apropos 'getpid) GETPID CCL::GETPID, Def: FUNCTION ? (quit) and...
CLISP problem with #_getpid in sl-on-cl.lisp
I'm using the latest version of sl-on-cl.lisp that has near the top ;; Time-stamp: <2025-12-02 16:50:12 franc> For me, this file compiles with no warnings or errors on Windows 11 using GNU CLISP 2.49+ (2010-07-17) and on Ubuntu using GNU CLISP 2.49.93+ (2018-02-18). As far as I can see, the current version of CCL, which is 1.13 released on Aug 14, 2024 (so over a year old) does not define ccl::getpid. Instead, it uses the non-standard reader macro #_ to look up getpid in a table of foreign functions,...
CLISP problem with #_getpid in sl-on-cl.lisp
Common Lisp: fix reader macro #_
Get arithlib to compile better on Windows where previously there were
core-math is (I believe) the successor to crlibm, and it seems probable that
Common Lisp: updates for CLISP and CCL
Fixes to the very recent checkin when I check using clang on the Macbook.
This should cause arithlib to transition to FFT-based multiplicatioon for
Hi Francis, compiling both r7207 and r7208 Common Lisp Reduce (on x86_64 macOS 11.7.10 Big Sur) is now OK with SBCL , CCL and CLISP. Marco
Test scripts
Compiling r7205 Common Lisp Reduce fails while using CCL and CLISP
The code I have just committed builds REDUCE on SBCL and CLISP on Windows for me. SBCL REDUCE runs well; CLISP REDUCE runs less well and needs a bit more work. I have not yet tested on CCL or on Linux, which I will try to do soon.
Common Lisp: CLISP REDUCE builds again
packages/ellipfn/efellint.red: Correct a number of errors in the evaluation of
Compiling r7205 Common Lisp Reduce fails while using CCL and CLISP
Thanks for the bug report. I only tested the current code on SBCL on Windows. I committed it so that I could easly test it on Linux and it seems to work fine on SBCL on Ubuntu. I then discovered the problem building on CLISP. I think it's caused by a very recent change I made to getd. Common Lisp is fussier about what constitutes a function than is Standard Lisp, but SBCL seems to be more flexible than CLISP (or probably CCL). Anyway, I think I know where the problem is. Francis
Compiling r7205 Common Lisp Reduce fails while using CCL and CLISP
Common Lisp improvements
Improve CRACK on Common Lisp
Change to account for different kernel page sizes on aarch64 linux: query system page size with sysconf(_SC_PAGESIZE) and align the BPS memory to a page boundary. If the call to sysconf returns -1, align to 16kb boundary.