reduce-algebra-developers — Discussion of development, administration and support for Reduce

[Reduce-algebra-developers] Issue with num_min

[Arthur N. <ac...@ca...>]

Until this evening if you passed non-numeric values such as
(!:rd!: . 1.0e-36) to expt in CSL it crashed out in an embarassing way, 
but I have now made that function check its args a bit better. I hit this 
when trying to do a numerical minimization, and sometimes that seems to 
pass a Reduce wrapped up float not a number. Well also the value 
10^1.0e-36 looks a pretty odd thing to be calculating! I have not (yet) 
looked inside the code to understand what is going on - this is posted so 
that others can comment too....


1: on backtrace;

2: num_min(cos x^2 + cos y^2, {x=1, y=2}, accuracy=1.0e-36, 
iterations=100);
+++ Error: expt 10 (!:rd!: . 1.0e-36)
Inside: steepdeceval1
Arg1: (plus (expt (cos x) 2) (expt (cos y) 2))
Arg2: (x y)
Arg3: (1 2)
Arg4: num_min
Inside: rdmineval
Arg1: ((plus (expt (cos x) 2) (expt (cos y) 2)) (list (equal x 1) (equal y 
2)) (
equal accuracy (!:dn!: 10 . -37)) (equal iterations 100))
Inside: lispapply
...

Arthur

Re: [Reduce-algebra-developers] Building Reduce with current GCC 15

[Arthur N. <ac...@ca...>]

> Greetings,
>
> I'm not sure when the changes occurred, but there is some trouble using
> current GCC 15 (tested with 15.0.1 20250228 / Red Hat 15.0.1-0 which is
> distributed with Fedora 42 Beta and Fedora Rawhide).  This will be the
> new stable GCC version released in the next month or so, so it probably
> makes sense to get the fixes in now.
>
> g++: error: unrecognized command-line option ‘-fconstexpr-steps=32000000’
>
> The `-fconstexpr-steps` argument is not supported, and is a fatal error.
> This happens when building in `csl/cslbase`.  It seems the current
> incantation would be `-fconstexpr-ops-limit`.  I also was able to build
> without this flag at all and didn't notice any excessive compilation
> time, so it may no longer be strictly necessary.
>

In cslbase/configure.ac it says:
# Adjust the number of steps that clang can take while evaluating
# constexpr things at compile time. With the CSL code as of September
# 2022 the code uses almost 3.7 million steps while building cslread.o - I
# can see ways to reduce that at the code of having more very special code
# present. I believe that the clang default is 2^20 (about a million),
# while gcc limits at more like 32 million and that is why I set my
# explicit clang override here to such a huge value - it is in the hope
# that this will leave clang and gcc issues arising at around the same
# point.
AC_CHECK_DECL([__clang__],
   [CXXFLAGS="$CXXFLAGS -fconstexpr-steps=32000000"])
and the intent had been that that would only he used with clang and so it
ought not to be offered to g++.

Indeed the clang on my Macbook today gets through without it, but not I 
think by a huge margin - maybe clang increased their default limit.

So let me explain what is going on! I want a table that is a bit like
   {3, 7, 13, 31, 61, ...}
being primes a bit below powers of 2. These are used as sized for hash 
tables. Now I could have just written a table with the numeric values
explicitlyt tabulated. But I took the view that that was bad since
for instance it leaves 1073741789 in the code as a "magic number".
Using constexpr code to set up the list was going to feel like much better
in terms of documented code and robustness against typos. When I first did 
that I took the table of "primes just below a power of 2" up an absurd
way into values beyond 32-bits and to check for primality I used Miller-
Rabin. I am now backing off from support for having numbers of symbols in
one calculation that call for a symbol hashtable using over 16 or 32GB,
so I stop my table within 32-bits. That makes simple test division better
for the primality checking and today on a Mac with clang the table gets 
set up within 250000 "steps" of constexpr calculation. And with that I do
not need to pass special command line arguments to clang. I hope their 
limit is the same everywhere!


> The configure script should probably try to see if compiling using
> `-fconstexpr-steps` or `-fconstexpr-ops-limit` limit actually works
> before using it.
If I just avoid that I can relax a bit!

> Tangentially related, trying to configure reduce to build with Clang
> results in only some of the project actually building with Clang.
> I configured with: `env CC=clang CXX=clang++ ./configure --with-csl
> --with-clang` ... and at least `cslbase/flatten` was still being
> built with g++.  I'll save the compilation output and look into this
> more tonight.
>
The issue with "flatten" and a few others is cross compilation. It I am
cross ocmpiling then $CC and $CXX can be generating "foreign" binaries 
that I can not execute on the current platform, so I wanted a C++ compiler 
that I could use to build utilities that I run to assemble parts of the 
source code. Because it is very widely available I just say "gcc" or 
"g++". So this is not an accident or oversight. But alternative ideas
for what one might do would be welcome.
If I cross compile I can build the executables like "csl[.exe]" and
"reduce[.exe]" and in some cases I can use binfmt and qemu or some such to
run them under emulation, but even without that it is potentially valuable
because the *.img files are platform neutral and I can safely package
a locally build copy of them with the cross-built executable files. This 
is not deployed as part of the snapshot building etc but I hope you can 
see that the possibility will be good. Well in "not too long" it may dig 
me out of the issues of building a system for windows-11-on-arm in that 
there is a spin of clang that can target that in cross compilation....

> --
> Jeffrey H. Johnson
> tr...@po...
>
Arthur

Re: [Reduce-algebra-developers] Please unsubscribe me from this list, thanks.

[Francis W. <f.j...@li...>]

I have unsubscribed you. You should get an email from the mail system that you will need to respond to.

Francis

________________________________
From: Gary Knight
Sent: Sunday, March 30, 2025 12:08 AM
To: red...@li...
Subject: [Reduce-algebra-developers] Please unsubscribe me from this list, thanks.

Hi there at Reduce-algebra ;
  please remove me from your list of subscribing developers.
best regards, Gary Knight

Re: [Reduce-algebra-developers] Reduce (CSL, rev 7058) window on ubuntu 24.04 is very hard to read

[Arthur N. <ac...@ca...>]

The menu Edit/Font should let you reset the font and that popup should 
appear with alt-E alt-F. Then tha selected font should remain until you 
use Edit/reset-window.

Howeer the font-size adjustment only applies to the main contents of the 
windows not to the menus. The font  for me is DejaVu Sans Mono but if for 
any reason that is not good for you that too can be switched in Edit/Font. 
But you may need something like xzoom to view the menus since they do not 
readily change size without just altering overall display resolution and 
right now I am not certain what font is used for menus.

So is the key issue that things are just too small or are the characters 
missing or malformed potentially because of some missing system font?

Arthur

On Sun, 30 Mar 2025, Osman Buyukisik wrote:

> Hello Arthur,
>
> I had redcsl working fine some years ago (on linux machines), but last two 
> building from svn source resulted in very hard to read menus/windows. The 
> text content is readable but File/Edit etc. menus are not (am over 70 :) Is 
> there a font that I am missing that is needed or some init/config file?
>
> I can run redcsl with -w- option without any problems.
>
> Thanks in advance,
>
> Osman
>
>

Re: [Reduce-algebra-developers] compiling, for javascript

[Michael S. <sh...@sa...>]

The first command "make bootstrapreduce.img" at the top of
for-emscripten/Makefile gives me the same errors, as do "make webreduce"
and "make mini-webreduce".

On Sat, Mar 29, 2025 at 3:29 PM Francis Wright <f.j...@li...>
wrote:

> I'm fairly sure that the way I build wasm REDUCE doesn't involve
> makeall.sh. Try just following the instructions at the top of Makefile by
> hand. That's what I do.
>
> Francis
>
> On 29 Mar 2025 9:02 pm, Michael Shulman <sh...@sa...> wrote:
>
> Thanks everyone!  I got version 6745 from svn and installed emscripten
> 2.0.34.  The main program built fine, but for-emscripten/makeall.sh
> complained that
>
> ../../cslbase/headers.h:102:10: fatal error: 'crlibm.h' file not found
>
> I found a file called crlibm.h in
> cslbuild/x86_64-unknown-ubuntu22.04/include, so I tried copying that into
> cslbase, and makeall was then able to make some more progress.  But now I
> get a couple more errors that I have no idea what to do about:
>
> ../../cslbase/csl.cpp:2958:9: error: void function 'cslstart' should not
> return a value [-Wreturn-type]
>         THROW(LispStop);
>         ^~~~~~~~~~~~~~~
> ../../cslbase/lispthrow.h:890:35: note: expanded from macro 'THROW'
>     do { exceptionFlag = flavour; return SPID_Throw; } while(false)
>                                   ^      ~~~~~~~~~~
> ../../cslbase/csl.cpp:3653:5: error: use of undeclared identifier
> 'initThreadLocals'
>     initThreadLocals();
>     ^
>
>
> On Sat, Mar 29, 2025 at 8:22 AM Francis Wright <f.j...@li...>
> wrote:
>
> I haven't thought about Web REDUCE for a while, but I guess it's time that
> I did!
>
> According to "/web/htdocs/web-reduce/README.md":
> 15 April 2024: The version currently on the REDUCE web site is the latest
> revision of REDUCE built using Emscripten 2.0.34, because I have had
> problems using later Emscripten versions.
> And the header displayed by Web REDUCE is
>
> Web REDUCE (6745),  2-Apr-2024 ...
>
> So "/csl/new-embedded/for-emscripten/Makefile" should reliably build
> REDUCE 6745 using Emscripten 2.0.34.
> If somebody with the appropriate expertise can get the latest version of
> REDUCE to build with the latest version of Emscripten then that would be
> great; please let me know how to do it!
> In the meantime, I have more or less decided to stay with the build that
> currently works, but to update the version of REDUCE, i.e. the packages
> directory. I think that should work although I haven't actually done it yet.
> Once you have built REDUCE, you can play around with it using Node.js or
> the web files in the for-emscripten directory. And, of course, you can play
> around with Web REDUCE.
> As for "known problems", I list some on the web site, but I'll try to say
> a bit more later when I have more time, maybe tomorrow. But a key problem
> is the inherent asynchronicity of JavaScript/WebAssembly.
> Francis
> ------------------------------
> *From:* Michael Shulman via Reduce-algebra-developers
> *Sent:* Friday, March 28, 2025 6:28 PM
> *To:* Arthur Norman
> *Cc:* red...@li...
> *Subject:* Re: [Reduce-algebra-developers] compiling, for javascript
>
> Thanks, that's very helpful!  I assumed that because the github
> version describes itself as a 'mirror' it would have exactly the same
> code in it, but apparently not: I tried the svn version and it worked
> just fine (for the regular version).
>
> Now I want to try to build the emscripten version.  I am looking at
> the instructions at
>
> https://sourceforge.net/p/reduce-algebra/code/HEAD/tree/trunk/csl/new-embedded/for-emscripten/
> that are linked from
> https://reduce-algebra.sourceforge.io/web-reduce/about.php?start as
> "for details of how to build this version of REDUCE".  But whenever I
> run any of the 'make' commands I get errors like
>
> make: *** No rule to make target 'allocate.o', needed by 'reduce.js'.
> Stop.
>
> I know you said that isn't your corner of the project, but maybe
> someone else can help me with this?
>
>
>

[Reduce-algebra-developers] Reduce (CSL, rev 7058) window on ubuntu 24.04 is very hard to read

[Osman B. <os...@fu...>]

Hello Arthur,

I had redcsl working fine some years ago (on linux machines), but last 
two building from svn source resulted in very hard to read 
menus/windows. The text content is readable but File/Edit etc. menus are 
not (am over 70 :) Is there a font that I am missing that is needed or 
some init/config file?

I can run redcsl with -w- option without any problems.

Thanks in advance,

Osman

[Reduce-algebra-developers] Please unsubscribe me from this list, thanks.

[Gary K. <gdk...@gm...>]

Hi there at Reduce-algebra ;
  please remove me from your list of subscribing developers.
best regards, Gary Knight

Re: [Reduce-algebra-developers] Building Reduce with current GCC 15

[Arthur C. N. <ac...@ca...>]

There are some utilities that I just compile with gcc not $CC and maybe I should not. Arthur
________________________________
From: Arthur Charles Norman via Reduce-algebra-developers <red...@li...>
Sent: Saturday, March 29, 2025 11:29:41 PM
To: Jeff Johnson <tr...@po...>
Cc: red...@li... <red...@li...>
Subject: Re: [Reduce-algebra-developers] Building Reduce with current GCC 15

Thanks. There are some constexpr calculations that needed more steps than the gcc default, so I am sad they have altered the directive that supports this. So you are probably correct that a yet more complex configure script to allow for this gratuitous change has to be set up. Arthur
________________________________
From: Jeff Johnson <tr...@po...>
Sent: Saturday, March 29, 2025 9:29:24 PM
To: Arthur Charles Norman <ac...@ca...>
Cc: red...@li... <red...@li...>
Subject: Building Reduce with current GCC 15

Greetings,

I'm not sure when the changes occurred, but there is some trouble using
current GCC 15 (tested with 15.0.1 20250228 / Red Hat 15.0.1-0 which is
distributed with Fedora 42 Beta and Fedora Rawhide).  This will be the
new stable GCC version released in the next month or so, so it probably
makes sense to get the fixes in now.

g++: error: unrecognized command-line option ‘-fconstexpr-steps=32000000’

The `-fconstexpr-steps` argument is not supported, and is a fatal error.
This happens when building in `csl/cslbase`.  It seems the current
incantation would be `-fconstexpr-ops-limit`.  I also was able to build
without this flag at all and didn't notice any excessive compilation
time, so it may no longer be strictly necessary.

The configure script should probably try to see if compiling using
`-fconstexpr-steps` or `-fconstexpr-ops-limit` limit actually works
before using it.

Tangentially related, trying to configure reduce to build with Clang
results in only some of the project actually building with Clang.
I configured with: `env CC=clang CXX=clang++ ./configure --with-csl
--with-clang` ... and at least `cslbase/flatten` was still being
built with g++.  I'll save the compilation output and look into this
more tonight.

--
Jeffrey H. Johnson
tr...@po...

Re: [Reduce-algebra-developers] compiling, for javascript

[Arthur C. N. <ac...@ca...>]

Thank you very much. As I noted before I recognize that many people prefer git to subversion significantly. Artbur
________________________________
From: Jeff Johnson <tr...@po...>
Sent: Saturday, March 29, 2025 8:18:54 PM
To: Arthur Charles Norman <ac...@ca...>
Cc: red...@li... <red...@li...>
Subject: Re: [Reduce-algebra-developers] compiling, for javascript

Hi there,

I run the script that keeps the mirror updated.  I'm sorry that it
was causing an issue.

The problem was that the Subversion software itself is responsible
for expanding and rewriting the $Id$ keyword in the file - the
actual files themselves just contain '$Id'.

I've updated the mirroring script to expand the $Id$ where needed,
so this shouldn't be a problem going forward.

Normally the mirror updates against the upstream SVN repository once
a day, but I gave it a manual kick so you should see the fix now.

--
Jeffrey H. Johnson
tr...@po...

Re: [Reduce-algebra-developers] Building Reduce with current GCC 15

[Arthur C. N. <ac...@ca...>]

Thanks. There are some constexpr calculations that needed more steps than the gcc default, so I am sad they have altered the directive that supports this. So you are probably correct that a yet more complex configure script to allow for this gratuitous change has to be set up. Arthur
________________________________
From: Jeff Johnson <tr...@po...>
Sent: Saturday, March 29, 2025 9:29:24 PM
To: Arthur Charles Norman <ac...@ca...>
Cc: red...@li... <red...@li...>
Subject: Building Reduce with current GCC 15

Greetings,

I'm not sure when the changes occurred, but there is some trouble using
current GCC 15 (tested with 15.0.1 20250228 / Red Hat 15.0.1-0 which is
distributed with Fedora 42 Beta and Fedora Rawhide).  This will be the
new stable GCC version released in the next month or so, so it probably
makes sense to get the fixes in now.

g++: error: unrecognized command-line option ‘-fconstexpr-steps=32000000’

The `-fconstexpr-steps` argument is not supported, and is a fatal error.
This happens when building in `csl/cslbase`.  It seems the current
incantation would be `-fconstexpr-ops-limit`.  I also was able to build
without this flag at all and didn't notice any excessive compilation
time, so it may no longer be strictly necessary.

The configure script should probably try to see if compiling using
`-fconstexpr-steps` or `-fconstexpr-ops-limit` limit actually works
before using it.

Tangentially related, trying to configure reduce to build with Clang
results in only some of the project actually building with Clang.
I configured with: `env CC=clang CXX=clang++ ./configure --with-csl
--with-clang` ... and at least `cslbase/flatten` was still being
built with g++.  I'll save the compilation output and look into this
more tonight.

--
Jeffrey H. Johnson
tr...@po...

Re: [Reduce-algebra-developers] compiling, for javascript

[Francis W. <f.j...@li...>]

<meta http-equiv="Content-Type" content="text/html; charset=utf-8"><div dir="auto"><div>I'm fairly sure that the way I build wasm REDUCE doesn't involve makeall.sh. Try just following the instructions at the top of Makefile by hand. That's what I do.</div><div dir="auto"><br></div><div dir="auto">Francis<br><div dir="auto"><br><div class="elided-text">On 29 Mar 2025 9:02 pm, Michael Shulman &lt;sh...@sa...&gt; wrote:<br type="attribution"><blockquote style="margin:0 0 0 0.8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div>Thanks everyone!&nbsp; I got version 6745 from svn and installed emscripten 2.0.34.&nbsp; The main program built fine, but for-emscripten/makeall.sh complained that</div><div><br></div><div>../../cslbase/headers.h:102:10: fatal error: 'crlibm.h' file not found<br></div><div><br></div><div>I found a file called crlibm.h in cslbuild/x86_64-unknown-ubuntu22.04/include, so I tried copying that into cslbase, and makeall was then able to make some more progress.&nbsp; But now I get a couple more errors that I have no idea what to do about:</div><div><br></div><div>../../cslbase/csl.cpp:2958:9: error: void function 'cslstart' should not return a value [-Wreturn-type]<br>&nbsp; &nbsp; &nbsp; &nbsp; THROW(LispStop);<br>&nbsp; &nbsp; &nbsp; &nbsp; ^~~~~~~~~~~~~~~<br>../../cslbase/lispthrow.h:890:35: note: expanded from macro 'THROW'<br>&nbsp; &nbsp; do { exceptionFlag = flavour; return SPID_Throw; } while(false)<br>&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; ^ &nbsp; &nbsp; &nbsp;~~~~~~~~~~<br>../../cslbase/csl.cpp:3653:5: error: use of undeclared identifier 'initThreadLocals'<br>&nbsp; &nbsp; initThreadLocals();<br>&nbsp; &nbsp; ^<br><br></div></div><br><div><div dir="ltr">On Sat, Mar 29, 2025 at 8:22 AM Francis Wright &lt;<a href="mailto:f.j...@li...">f.j...@li...</a>&gt; wrote:<br></div><blockquote style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb( 204 , 204 , 204 );padding-left:1ex"><div>




<div dir="ltr">
<div style="font-family:'calibri' , 'helvetica' , sans-serif;font-size:12pt;color:rgb( 0 , 0 , 0 )">
I haven't thought about Web REDUCE for a while, but I guess it's time that I did!</div>
<div style="font-family:'calibri' , 'helvetica' , sans-serif;font-size:12pt;color:rgb( 0 , 0 , 0 )">
<br>
</div>
<div style="font-family:'calibri' , 'helvetica' , sans-serif;font-size:12pt;color:rgb( 0 , 0 , 0 )">
According to &quot;/web/htdocs/web-reduce/README.md&quot;:</div>
<div style="margin-top:1em;margin-bottom:1em;font-family:'calibri' , 'helvetica' , sans-serif;font-size:12pt;color:rgb( 0 , 0 , 0 )">
15 April 2024: The version currently on the REDUCE web site is the latest revision of REDUCE built using Emscripten 2.0.34,&nbsp;because I have had problems using later Emscripten versions.</div>
<div style="margin-top:1em;margin-bottom:1em;font-family:'calibri' , 'helvetica' , sans-serif;font-size:12pt;color:rgb( 0 , 0 , 0 )">
And the header displayed by Web REDUCE is</div>
<pre><div style="font-family:'calibri' , 'helvetica' , sans-serif;font-size:12pt;color:rgb( 0 , 0 , 0 )">Web REDUCE (6745), &nbsp;2-Apr-2024 ...</div></pre>
<div style="margin-top:1em;margin-bottom:1em;font-family:'calibri' , 'helvetica' , sans-serif;font-size:12pt;color:rgb( 0 , 0 , 0 )">
So &quot;/csl/new-embedded/for-emscripten/Makefile&quot; should reliably build REDUCE 6745 using Emscripten 2.0.34.</div>
<div style="margin-top:1em;margin-bottom:1em;font-family:'calibri' , 'helvetica' , sans-serif;font-size:12pt;color:rgb( 0 , 0 , 0 )">
If somebody with the appropriate expertise can get the latest version of REDUCE to build with the latest version of Emscripten then that would be great; please let me know how to do it!</div>
<div style="margin-top:1em;margin-bottom:1em;font-family:'calibri' , 'helvetica' , sans-serif;font-size:12pt;color:rgb( 0 , 0 , 0 )">
In the meantime, I have more or less decided to stay with the build that currently works, but to update the version of REDUCE, i.e. the packages directory. I think that should work although I haven't actually done it yet.</div>
<div style="margin-top:1em;margin-bottom:1em;font-family:'calibri' , 'helvetica' , sans-serif;font-size:12pt;color:rgb( 0 , 0 , 0 )">
Once you have built REDUCE, you can play around with it using Node.js or the web files in the for-emscripten directory. And, of course, you can play around with Web REDUCE.</div>
<div style="margin-top:1em;margin-bottom:1em;font-family:'calibri' , 'helvetica' , sans-serif;font-size:12pt;color:rgb( 0 , 0 , 0 )">
As for &quot;known problems&quot;, I list some on the web site, but I'll try to say a bit more later when I have more time, maybe tomorrow. But a key problem is the inherent asynchronicity of JavaScript/WebAssembly.</div>
<div style="margin-top:1em;margin-bottom:1em;font-size:12pt;color:rgb( 0 , 0 , 0 )">
<span style="font-family:'calibri' , 'helvetica' , sans-serif">Francis</span><span style="font-family:'calibri' , 'arial' , 'helvetica' , sans-serif"><br>
</span></div>
<hr style="display:inline-block;width:98%">
<span style="font-family:'calibri' , 'arial' , 'helvetica' , sans-serif;font-size:12pt;color:rgb( 0 , 0 , 0 )"><b>From:</b>&nbsp;Michael Shulman via Reduce-algebra-developers<br>
<b>Sent:</b>&nbsp;Friday, March 28, 2025 6:28 PM<br>
<b>To:</b>&nbsp;Arthur Norman<br>
<b>Cc:</b>&nbsp;<a href="mailto:red...@li...">red...@li...</a><br>
<b>Subject:</b>&nbsp;Re: [Reduce-algebra-developers] compiling, for javascript </span>
<div style="font-family:'calibri' , 'arial' , 'helvetica' , sans-serif;font-size:12pt;color:rgb( 0 , 0 , 0 )">
<br>
</div>
<div style="font-size:11pt">Thanks, that's very helpful!&nbsp; I assumed that because the github<br>
version describes itself as a 'mirror' it would have exactly the same<br>
code in it, but apparently not: I tried the svn version and it worked<br>
just fine (for the regular version).<br>
<br>
Now I want to try to build the emscripten version.&nbsp; I am looking at<br>
the instructions at<br>
<a href="https://sourceforge.net/p/reduce-algebra/code/HEAD/tree/trunk/csl/new-embedded/for-emscripten/">https://sourceforge.net/p/reduce-algebra/code/HEAD/tree/trunk/csl/new-embedded/for-emscripten/</a><br>
that are linked from<br>
<a href="https://reduce-algebra.sourceforge.io/web-reduce/about.php?start">https://reduce-algebra.sourceforge.io/web-reduce/about.php?start</a>&nbsp;as<br>
&quot;for details of how to build this version of REDUCE&quot;.&nbsp; But whenever I<br>
run any of the 'make' commands I get errors like<br>
<br>
make: *** No rule to make target 'allocate.o', needed by 'reduce.js'.&nbsp; Stop.<br>
<br>
I know you said that isn't your corner of the project, but maybe<br>
someone else can help me with this?<br>
</div>
</div>

</div></blockquote></div>
</blockquote></div><br></div></div></div>

[Reduce-algebra-developers] Building Reduce with current GCC 15

[Jeff J. <tr...@po...>]

Greetings,

I'm not sure when the changes occurred, but there is some trouble using
current GCC 15 (tested with 15.0.1 20250228 / Red Hat 15.0.1-0 which is
distributed with Fedora 42 Beta and Fedora Rawhide).  This will be the
new stable GCC version released in the next month or so, so it probably
makes sense to get the fixes in now.

g++: error: unrecognized command-line option ‘-fconstexpr-steps=32000000’

The `-fconstexpr-steps` argument is not supported, and is a fatal error.
This happens when building in `csl/cslbase`.  It seems the current
incantation would be `-fconstexpr-ops-limit`.  I also was able to build
without this flag at all and didn't notice any excessive compilation
time, so it may no longer be strictly necessary.

The configure script should probably try to see if compiling using
`-fconstexpr-steps` or `-fconstexpr-ops-limit` limit actually works
before using it.

Tangentially related, trying to configure reduce to build with Clang
results in only some of the project actually building with Clang.
I configured with: `env CC=clang CXX=clang++ ./configure --with-csl
--with-clang` ... and at least `cslbase/flatten` was still being
built with g++.  I'll save the compilation output and look into this
more tonight.

--
Jeffrey H. Johnson
tr...@po...

Re: [Reduce-algebra-developers] compiling, for javascript

[Michael S. <sh...@sa...>]

Thanks everyone!  I got version 6745 from svn and installed emscripten
2.0.34.  The main program built fine, but for-emscripten/makeall.sh
complained that

../../cslbase/headers.h:102:10: fatal error: 'crlibm.h' file not found

I found a file called crlibm.h in
cslbuild/x86_64-unknown-ubuntu22.04/include, so I tried copying that into
cslbase, and makeall was then able to make some more progress.  But now I
get a couple more errors that I have no idea what to do about:

../../cslbase/csl.cpp:2958:9: error: void function 'cslstart' should not
return a value [-Wreturn-type]
        THROW(LispStop);
        ^~~~~~~~~~~~~~~
../../cslbase/lispthrow.h:890:35: note: expanded from macro 'THROW'
    do { exceptionFlag = flavour; return SPID_Throw; } while(false)
                                  ^      ~~~~~~~~~~
../../cslbase/csl.cpp:3653:5: error: use of undeclared identifier
'initThreadLocals'
    initThreadLocals();
    ^


On Sat, Mar 29, 2025 at 8:22 AM Francis Wright <f.j...@li...>
wrote:

> I haven't thought about Web REDUCE for a while, but I guess it's time that
> I did!
>
> According to "/web/htdocs/web-reduce/README.md":
> 15 April 2024: The version currently on the REDUCE web site is the latest
> revision of REDUCE built using Emscripten 2.0.34, because I have had
> problems using later Emscripten versions.
> And the header displayed by Web REDUCE is
>
> Web REDUCE (6745),  2-Apr-2024 ...
>
> So "/csl/new-embedded/for-emscripten/Makefile" should reliably build
> REDUCE 6745 using Emscripten 2.0.34.
> If somebody with the appropriate expertise can get the latest version of
> REDUCE to build with the latest version of Emscripten then that would be
> great; please let me know how to do it!
> In the meantime, I have more or less decided to stay with the build that
> currently works, but to update the version of REDUCE, i.e. the packages
> directory. I think that should work although I haven't actually done it yet.
> Once you have built REDUCE, you can play around with it using Node.js or
> the web files in the for-emscripten directory. And, of course, you can play
> around with Web REDUCE.
> As for "known problems", I list some on the web site, but I'll try to say
> a bit more later when I have more time, maybe tomorrow. But a key problem
> is the inherent asynchronicity of JavaScript/WebAssembly.
> Francis
> ------------------------------
> *From:* Michael Shulman via Reduce-algebra-developers
> *Sent:* Friday, March 28, 2025 6:28 PM
> *To:* Arthur Norman
> *Cc:* red...@li...
> *Subject:* Re: [Reduce-algebra-developers] compiling, for javascript
>
> Thanks, that's very helpful!  I assumed that because the github
> version describes itself as a 'mirror' it would have exactly the same
> code in it, but apparently not: I tried the svn version and it worked
> just fine (for the regular version).
>
> Now I want to try to build the emscripten version.  I am looking at
> the instructions at
>
> https://sourceforge.net/p/reduce-algebra/code/HEAD/tree/trunk/csl/new-embedded/for-emscripten/
> that are linked from
> https://reduce-algebra.sourceforge.io/web-reduce/about.php?start as
> "for details of how to build this version of REDUCE".  But whenever I
> run any of the 'make' commands I get errors like
>
> make: *** No rule to make target 'allocate.o', needed by 'reduce.js'.
> Stop.
>
> I know you said that isn't your corner of the project, but maybe
> someone else can help me with this?
>

Re: [Reduce-algebra-developers] compiling, for javascript

[Jeff J. <tr...@po...>]

Hi there,

I run the script that keeps the mirror updated.  I'm sorry that it
was causing an issue.

The problem was that the Subversion software itself is responsible
for expanding and rewriting the $Id$ keyword in the file - the
actual files themselves just contain '$Id'.

I've updated the mirroring script to expand the $Id$ where needed,
so this shouldn't be a problem going forward.

Normally the mirror updates against the upstream SVN repository once
a day, but I gave it a manual kick so you should see the fix now.

--
Jeffrey H. Johnson
tr...@po...

Re: [Reduce-algebra-developers] compiling, for javascript

[Francis W. <f.j...@li...>]

I haven't thought about Web REDUCE for a while, but I guess it's time that I did!

According to "/web/htdocs/web-reduce/README.md":
15 April 2024: The version currently on the REDUCE web site is the latest revision of REDUCE built using Emscripten 2.0.34, because I have had problems using later Emscripten versions.
And the header displayed by Web REDUCE is

Web REDUCE (6745),  2-Apr-2024 ...

So "/csl/new-embedded/for-emscripten/Makefile" should reliably build REDUCE 6745 using Emscripten 2.0.34.
If somebody with the appropriate expertise can get the latest version of REDUCE to build with the latest version of Emscripten then that would be great; please let me know how to do it!
In the meantime, I have more or less decided to stay with the build that currently works, but to update the version of REDUCE, i.e. the packages directory. I think that should work although I haven't actually done it yet.
Once you have built REDUCE, you can play around with it using Node.js or the web files in the for-emscripten directory. And, of course, you can play around with Web REDUCE.
As for "known problems", I list some on the web site, but I'll try to say a bit more later when I have more time, maybe tomorrow. But a key problem is the inherent asynchronicity of JavaScript/WebAssembly.
Francis
________________________________
From: Michael Shulman via Reduce-algebra-developers
Sent: Friday, March 28, 2025 6:28 PM
To: Arthur Norman
Cc: red...@li...
Subject: Re: [Reduce-algebra-developers] compiling, for javascript

Thanks, that's very helpful!  I assumed that because the github
version describes itself as a 'mirror' it would have exactly the same
code in it, but apparently not: I tried the svn version and it worked
just fine (for the regular version).

Now I want to try to build the emscripten version.  I am looking at
the instructions at
https://sourceforge.net/p/reduce-algebra/code/HEAD/tree/trunk/csl/new-embedded/for-emscripten/
that are linked from
https://reduce-algebra.sourceforge.io/web-reduce/about.php?start as
"for details of how to build this version of REDUCE".  But whenever I
run any of the 'make' commands I get errors like

make: *** No rule to make target 'allocate.o', needed by 'reduce.js'.  Stop.

I know you said that isn't your corner of the project, but maybe
someone else can help me with this?

Re: [Reduce-algebra-developers] compiling, for javascript

[Arthur N. <ac...@ca...>]

Another thought if the build files etc in the csl/new-embedded directory 
are not in step with the current main version of the trunk is that you 
can easily go
    svn -r NNN update
on the whole tree to go backwards in time until you find a version that 
works for you. If I browse the sources on sourceforge I find that for
revision 6789 a checkin has the message "Web REDUCE Makefile" so around 
then might be a good place to start.
For experiments at the stage you are at thus far working with a version 
that does not include the last year of various updates would not be a huge 
issue.... and if you make good enough progress with that to feel that the 
project as a whole is going to work then (a) it will provide us with a 
strong incentive to bring that area up to date (b) the delay will mean 
that maybe some of us will be under less pressure from other directions 
and (c) you will have built up your Reduce skills and experience!

I do not guarantee that revision 6788 or 6789 is a good one! In cases like 
this I go sort of way back and then do a binary chop on revision numbers 
to find where key changes were made - and then I can often see just what 
they were and allow for them. That tends to be a bit tedious and I ususlly
make a script that goes basically
    svn -r $1 update
    [sometimes I need to tidy up or delete built files here]
    [sometimes I need to re-run a "configure" script]
    make
and have a picxe of paper where I keep score for how various revisions 
behave...

Good luck. Arthur

Re: [Reduce-algebra-developers] compiling, for javascript

[Michael S. <sh...@sa...>]

I submitted an issue
<https://github.com/reduce-algebra/reduce-algebra/issues/85> on the github
mirror.

Thanks for the suggestions about updating the makefile.  Changing the
references to allocate.o and cslgc.o does let it get further.  I added
gc-check.o too, although it didn't seem to have any noticeable effect yet.

Now it gets stuck on bytes1.o saying it has no rule to make
../../cslbase/opnames.cpp.  The only file I found with a similar name is
../../embedded/opnames.c, so I tried replacing the reference with that;
that silenced that particular complaint but I don't know if it might cause
problems later.

Next the compilation of ../../cslbase/u01.cpp fails with a bunch of errors
saying "unknown type name 'Save'" and "no matching function for call to
'reclaim'" (it needs 4 arguments).  I grepped through the sources and found
a definition of a 4-argument "reclaim" in csl/embedded/gc.c, which sounds
plausible, but I'm not a C/C++ programmer and I can't guess how to bring
this into scope when compiling the uxx files.  If I just add "gc.o" to the
"main_o_files", as I did with the others, it complains that it has no rule
to make it; and I don't see any references to gc.o or gc.c in the root
Makefile.am.



On Fri, Mar 28, 2025 at 3:35 PM Arthur Norman <ac...@ca...> wrote:

> On Fri, 28 Mar 2025, Michael Shulman wrote:
>
> > Thanks, that's very helpful!  I assumed that because the github
> > version describes itself as a 'mirror' it would have exactly the same
> > code in it, but apparently not: I tried the svn version and it worked
> > just fine (for the regular version).
> >
> It may well copy across to github nicely but the subversion repository may
> get updated several times in a day so keeping quite up with it may be a
> chalklenge. Most of the time that should not matter and I know that some
> people really prefer git over subversion. However if I look there right
> now if has unexpanded "$Id:$ stuff there which as you found out so I guess
> that the mirroring process "tidies that up" with in this particular case
> bad consequences. I know that git is not keen on simple sequential
> revision numbers but maybe you can contact the maintainer of the git
> mirror and see if they can get those preserved as they make thei copies?
>
>
>
> > Now I want to try to build the emscripten version.  I am looking at
> > the instructions at
> >
> https://sourceforge.net/p/reduce-algebra/code/HEAD/tree/trunk/csl/new-embedded/for-emscripten/
> > that are linked from
> > https://reduce-algebra.sourceforge.io/web-reduce/about.php?start as
> > "for details of how to build this version of REDUCE".  But whenever I
> > run any of the 'make' commands I get errors like
> >
> > make: *** No rule to make target 'allocate.o', needed by 'reduce.js'.
> Stop.
> >
> > I know you said that isn't your corner of the project, but maybe
> > someone else can help me with this?
> >
> As with various of the sections beyond the ones that do simple builds for
> Windows, Linux and Mac the Makefiles there are not routinely checked and
> updated even when there are serious changes elsewhere - the sort-of stance
> is that what is being provided is a starting point for a programmer who is
> going to develop on top of it and who hence is liable to be capable of
> discerning what has to be done - and where IDEALLY they will then propose
> fixes and we register them as additional people able to maintain stuff -
> with them taking a particular interest in that section of the code because
> they are using it!  I think that here you will find that a while back
> "allocate.cpp" was superceded by "newallocate.cpp"! and similarly
> newcslgc.cpp took over from cslgc.cpp (and possibly gc-check.cpp was
> added). There was a fairly extended time when both versions were present
> and a "./configure"-time option could enable the newer version while it
> was being tested a bit.
> So I suggest you compare the list of sources in csl/cslbase/Makefile.am
> with the ones in the new-embedded directory and that will reveal some
> fairly simple updates.
>
> I am a bit in the middle of other things and really do not want to break
> my trains of thought by investigating that this weekend. But if the
> scripts to make webreduce are somewhere they may also have this sorted
> out...
>
> Arthur
>
>

Re: [Reduce-algebra-developers] compiling, for javascript

[Arthur N. <ac...@ca...>]

On Fri, 28 Mar 2025, Michael Shulman wrote:

> Thanks, that's very helpful!  I assumed that because the github
> version describes itself as a 'mirror' it would have exactly the same
> code in it, but apparently not: I tried the svn version and it worked
> just fine (for the regular version).
>
It may well copy across to github nicely but the subversion repository may 
get updated several times in a day so keeping quite up with it may be a 
chalklenge. Most of the time that should not matter and I know that some 
people really prefer git over subversion. However if I look there right 
now if has unexpanded "$Id:$ stuff there which as you found out so I guess 
that the mirroring process "tidies that up" with in this particular case 
bad consequences. I know that git is not keen on simple sequential 
revision numbers but maybe you can contact the maintainer of the git 
mirror and see if they can get those preserved as they make thei copies?



> Now I want to try to build the emscripten version.  I am looking at
> the instructions at
> https://sourceforge.net/p/reduce-algebra/code/HEAD/tree/trunk/csl/new-embedded/for-emscripten/
> that are linked from
> https://reduce-algebra.sourceforge.io/web-reduce/about.php?start as
> "for details of how to build this version of REDUCE".  But whenever I
> run any of the 'make' commands I get errors like
>
> make: *** No rule to make target 'allocate.o', needed by 'reduce.js'.  Stop.
>
> I know you said that isn't your corner of the project, but maybe
> someone else can help me with this?
>
As with various of the sections beyond the ones that do simple builds for 
Windows, Linux and Mac the Makefiles there are not routinely checked and 
updated even when there are serious changes elsewhere - the sort-of stance 
is that what is being provided is a starting point for a programmer who is 
going to develop on top of it and who hence is liable to be capable of
discerning what has to be done - and where IDEALLY they will then propose 
fixes and we register them as additional people able to maintain stuff - 
with them taking a particular interest in that section of the code because 
they are using it!  I think that here you will find that a while back 
"allocate.cpp" was superceded by "newallocate.cpp"! and similarly 
newcslgc.cpp took over from cslgc.cpp (and possibly gc-check.cpp was 
added). There was a fairly extended time when both versions were present
and a "./configure"-time option could enable the newer version while it 
was being tested a bit.
So I suggest you compare the list of sources in csl/cslbase/Makefile.am 
with the ones in the new-embedded directory and that will reveal some 
fairly simple updates.

I am a bit in the middle of other things and really do not want to break 
my trains of thought by investigating that this weekend. But if the 
scripts to make webreduce are somewhere they may also have this sorted 
out...

Arthur

Re: [Reduce-algebra-developers] compiling, for javascript

[Michael S. <sh...@sa...>]

Thanks, that's very helpful!  I assumed that because the github
version describes itself as a 'mirror' it would have exactly the same
code in it, but apparently not: I tried the svn version and it worked
just fine (for the regular version).

Now I want to try to build the emscripten version.  I am looking at
the instructions at
https://sourceforge.net/p/reduce-algebra/code/HEAD/tree/trunk/csl/new-embedded/for-emscripten/
that are linked from
https://reduce-algebra.sourceforge.io/web-reduce/about.php?start as
"for details of how to build this version of REDUCE".  But whenever I
run any of the 'make' commands I get errors like

make: *** No rule to make target 'allocate.o', needed by 'reduce.js'.  Stop.

I know you said that isn't your corner of the project, but maybe
someone else can help me with this?

On Fri, Mar 28, 2025 at 9:47 AM Arthur Norman <ac...@ca...> wrote:
>
> OK you raise two issues. The first is javascript.
>
> Quite a few years ago there was a demo build using javascript and stuck in
> a directory "jslisp". Note that the README file there dates it to 2011 and
> has a comment: "While the code is stabilising here there really are NO
> guarantees that it will all even build!". So that is only potentially
> useful for individuals willing to dive deep into it and resolve problems
> mostly for themselves. And since that starts with stuff fetched from
> mathpiper around 14 years ago and I have no ideal at all where mathpiper
> has got to in the meanwhile, anybody wanting to build on that needs to
> check there as well as with Reduce people.
> More recently a spin of Reduce has been built using "emscripten" which
> turns the C++ code of the CSL kernel for Reduce into webassembly stuff
> that is expected to hook onto javascript and the web version of Reduce
> uses that. The source directory relevant is csl/new-embedded but mosly you
> need to look at
> https://reduce-algebra.sourceforge.io/web-reduce/about.php?start
> and I do not know how to rebuild it since that is not my corner of the
> whole project - Francis Wright is the guru there. But trying that out may
> give you an idea of whether things would behave and do what you need.
>
> Re building a regular version, I note that the main place where Reduce
> sources live is https://sourceforge.net/projects/reduce-algebra/ and I do
> not know how any github mirror is maintained or whether files from there
> will always be complete or up to date. I hope they are but againb because
> there are things that are "my business" and others that are not I do not
> even know who set up a github mirror. So I can not provide support for
> anything fetched from there rathar than using
> svn checkout https://svn.code.sf.net/p/reduce-algebra/code/trunk reduce-algebra
>
> I have just now been configuring and building everything on an Ubuntu
> Linux based on a sources synchronised with sourceforge and I do not
> observe the odd diagnostic that you see.... but the complaint you see
> suggests that in version.h the symbol VERSION_ID has not got set properly.
> The value that gets is derived from the way that subversion updates an
> "$Id:" record when one makes a checkin so as to record the revision
> number, date and the identity of the person making the checkin, so you may
> like to look at your version of csl/cslbase/version.h to see if it looks
> good. For MINE I see
> #define VERSION_ID "$Id: version.h 7055 2025-03-25 21:40:11Z arthurcnorman $"
>
>
> Apologies if this does not instantly resolve all your issues, but I hope
> it helps you in the right direction!
>
> Arthur
>
>
>
> On Fri, 28 Mar 2025, Michael Shulman via Reduce-algebra-developers wrote:
>
> > Hi,
> >
> > My goal is to use the javascript version of reduce in a web
> > application; so if there are known problems with that at the moment,
> > please let me know.  However, at the moment I think I'm stuck on even
> > just compiling the base system, before converting it to JS.  I have
> > cloned the github mirror from
> > https://github.com/reduce-algebra/reduce-algebra and run the commands
> >
> > scripts/ubuntu-sanity-check.sh
> > ./configure --with-csl
> > scripts/csl-sanity-check.sh
> > make
> >
> > and I get the error message
> >
> > In file included from /home/shulman/reduce-algebra/csl/cslbase/fns1.cpp:40:
> > /home/shulman/reduce-algebra/csl/cslbase/version.h:60:6:   in
> > ‘constexpr’ expansion of ‘<lambda closure
> > object><lambda()>{}.<lambda()>()’
> > /home/shulman/reduce-algebra/csl/cslbase/version.h:60:7: error: array
> > subscript value ‘5’ is outside the bounds of array type ‘const char
> > [5]’
> >   60 |     }();
> >      |       ^
> >
> > This is a weird message and doesn't suggest a solution like installing
> > some missing library.  Do I have the wrong version of some compiler or
> > library perhaps?
> >
> > Thanks for any help,
> > Mike
> >
> >
> > _______________________________________________
> > Reduce-algebra-developers mailing list
> > Red...@li...
> > https://lists.sourceforge.net/lists/listinfo/reduce-algebra-developers

Re: [Reduce-algebra-developers] compiling, for javascript

[Arthur N. <ac...@ca...>]

OK you raise two issues. The first is javascript.

Quite a few years ago there was a demo build using javascript and stuck in
a directory "jslisp". Note that the README file there dates it to 2011 and
has a comment: "While the code is stabilising here there really are NO 
guarantees that it will all even build!". So that is only potentially 
useful for individuals willing to dive deep into it and resolve problems 
mostly for themselves. And since that starts with stuff fetched from 
mathpiper around 14 years ago and I have no ideal at all where mathpiper 
has got to in the meanwhile, anybody wanting to build on that needs to 
check there as well as with Reduce people.
More recently a spin of Reduce has been built using "emscripten" which 
turns the C++ code of the CSL kernel for Reduce into webassembly stuff 
that is expected to hook onto javascript and the web version of Reduce 
uses that. The source directory relevant is csl/new-embedded but mosly you 
need to look at
https://reduce-algebra.sourceforge.io/web-reduce/about.php?start
and I do not know how to rebuild it since that is not my corner of the 
whole project - Francis Wright is the guru there. But trying that out may 
give you an idea of whether things would behave and do what you need.

Re building a regular version, I note that the main place where Reduce 
sources live is https://sourceforge.net/projects/reduce-algebra/ and I do 
not know how any github mirror is maintained or whether files from there 
will always be complete or up to date. I hope they are but againb because 
there are things that are "my business" and others that are not I do not 
even know who set up a github mirror. So I can not provide support for 
anything fetched from there rathar than using
svn checkout https://svn.code.sf.net/p/reduce-algebra/code/trunk reduce-algebra

I have just now been configuring and building everything on an Ubuntu 
Linux based on a sources synchronised with sourceforge and I do not 
observe the odd diagnostic that you see.... but the complaint you see 
suggests that in version.h the symbol VERSION_ID has not got set properly. 
The value that gets is derived from the way that subversion updates an 
"$Id:" record when one makes a checkin so as to record the revision 
number, date and the identity of the person making the checkin, so you may 
like to look at your version of csl/cslbase/version.h to see if it looks 
good. For MINE I see
#define VERSION_ID "$Id: version.h 7055 2025-03-25 21:40:11Z arthurcnorman $"


Apologies if this does not instantly resolve all your issues, but I hope 
it helps you in the right direction!

Arthur



On Fri, 28 Mar 2025, Michael Shulman via Reduce-algebra-developers wrote:

> Hi,
>
> My goal is to use the javascript version of reduce in a web
> application; so if there are known problems with that at the moment,
> please let me know.  However, at the moment I think I'm stuck on even
> just compiling the base system, before converting it to JS.  I have
> cloned the github mirror from
> https://github.com/reduce-algebra/reduce-algebra and run the commands
>
> scripts/ubuntu-sanity-check.sh
> ./configure --with-csl
> scripts/csl-sanity-check.sh
> make
>
> and I get the error message
>
> In file included from /home/shulman/reduce-algebra/csl/cslbase/fns1.cpp:40:
> /home/shulman/reduce-algebra/csl/cslbase/version.h:60:6:   in
> ‘constexpr’ expansion of ‘<lambda closure
> object><lambda()>{}.<lambda()>()’
> /home/shulman/reduce-algebra/csl/cslbase/version.h:60:7: error: array
> subscript value ‘5’ is outside the bounds of array type ‘const char
> [5]’
>   60 |     }();
>      |       ^
>
> This is a weird message and doesn't suggest a solution like installing
> some missing library.  Do I have the wrong version of some compiler or
> library perhaps?
>
> Thanks for any help,
> Mike
>
>
> _______________________________________________
> Reduce-algebra-developers mailing list
> Red...@li...
> https://lists.sourceforge.net/lists/listinfo/reduce-algebra-developers

[Reduce-algebra-developers] compiling, for javascript

[Michael S. <sh...@sa...>]

Hi,

My goal is to use the javascript version of reduce in a web
application; so if there are known problems with that at the moment,
please let me know.  However, at the moment I think I'm stuck on even
just compiling the base system, before converting it to JS.  I have
cloned the github mirror from
https://github.com/reduce-algebra/reduce-algebra and run the commands

scripts/ubuntu-sanity-check.sh
./configure --with-csl
scripts/csl-sanity-check.sh
make

and I get the error message

In file included from /home/shulman/reduce-algebra/csl/cslbase/fns1.cpp:40:
/home/shulman/reduce-algebra/csl/cslbase/version.h:60:6:   in
‘constexpr’ expansion of ‘<lambda closure
object><lambda()>{}.<lambda()>()’
/home/shulman/reduce-algebra/csl/cslbase/version.h:60:7: error: array
subscript value ‘5’ is outside the bounds of array type ‘const char
[5]’
   60 |     }();
      |       ^

This is a weird message and doesn't suggest a solution like installing
some missing library.  Do I have the wrong version of some compiler or
library perhaps?

Thanks for any help,
Mike

Re: [Reduce-algebra-developers] Updating gnuplot

[Eberhard S. <esc...@ca...>]

That's fine with me.

Eberhard

On 23.03.25 10:23, Rainer Schöpf via Reduce-algebra-developers wrote:
> We are distributing gnuplot 4.6.0 (released 2012) as part of the Windows
> snapshot. I believe we should upgrade to release 5.4(.10). The 
> examples run OK;
> I don't see anything critical in the in the manual section listing
> the differences between versions 4 and 5.
>
> Objections, anyone?
>
>    Rainer
>
>
> _______________________________________________
> Reduce-algebra-developers mailing list
> Red...@li...
> https://lists.sourceforge.net/lists/listinfo/reduce-algebra-developers

Re: [Reduce-algebra-developers] Updating gnuplot

[Francis W. <f.j...@li...>]

I think upgrading is a good idea. For what its worth, Web REDUCE has been using gnuplot 5.4.10 for a while. I've also been running gnuplot 5.4.2 in Ubuntu 22 on WSL and not noticed any problems. It would be good to have the same version (more or less) across all platforms.

Francis

________________________________
From: Rainer Schöpf via Reduce-algebra-developers
Sent: Sunday, March 23, 2025 9:23 AM
To: red...@li...
Subject: [Reduce-algebra-developers] Updating gnuplot

We are distributing gnuplot 4.6.0 (released 2012) as part of the Windows
snapshot. I believe we should upgrade to release 5.4(.10). The examples run OK;
I don't see anything critical in the in the manual section listing
the differences between versions 4 and 5.

Objections, anyone?

    Rainer


_______________________________________________
Reduce-algebra-developers mailing list
Red...@li...
https://lists.sourceforge.net/lists/listinfo/reduce-algebra-developers

Re: [Reduce-algebra-developers] Updating gnuplot

[Arthur N. <ac...@ca...>]

No objections.

On Sun, 23 Mar 2025, Rainer Schöpf via Reduce-algebra-developers wrote:

> We are distributing gnuplot 4.6.0 (released 2012) as part of the Windows
> snapshot. I believe we should upgrade to release 5.4(.10). The examples run 
> OK;
> I don't see anything critical in the in the manual section listing
> the differences between versions 4 and 5.
>
> Objections, anyone?
>
>   Rainer
>
>
> _______________________________________________
> Reduce-algebra-developers mailing list
> Red...@li...
> https://lists.sourceforge.net/lists/listinfo/reduce-algebra-developers
>

[Reduce-algebra-developers] Updating gnuplot

[Rainer S. <rai...@gm...>]

We are distributing gnuplot 4.6.0 (released 2012) as part of the Windows
snapshot. I believe we should upgrade to release 5.4(.10). The examples run OK;
I don't see anything critical in the in the manual section listing
the differences between versions 4 and 5.

Objections, anyone?

    Rainer