maxima-discuss — Maxima discussion list

Re: [Maxima-discuss] Return a funcion with an if, from block

[Gunter K. <gu...@pe...>]

On Samstag, 2. Juli 2022 15:54:00 CEST Me Self wrote:
> Everything works, but just wanted to say two things:
> 
> @Gunther
> 
> That trick doesn't seem to work? If I write y:'makelist() in the
> block() then it does return this:
> 
> (%i2) f(x+1);
> (%o2)   makelist(if equal(k,0) then 1.618 else -1.0*k,k,0,n)
> 

'makelist() makes the makelist a noun. But '(makelist()) might work. The manual says:

Applied to a parenthesized expression, the single quote prevents evaluation of all symbols 
and function calls in the expression. E.g., '(f(x)) means do not evaluate the expression f(x). 
'f(x) (with the single quote applied to f instead of f(x)) means return the noun form of f 
applied to [x]. 

Kind regards,

  Gunter.

Re: [Maxima-discuss] Intro to Special Functions section and other issues.

[Raymond T. <toy...@gm...>]

On Wed, Jul 6, 2022 at 10:20 AM Gunter Königsmann <gu...@pe...>
wrote:

>
>
> Am 6. Juli 2022 18:46:24 MESZ schrieb Raymond Toy <toy...@gm...>:
>
> >
> >I thought specint might be better in the same place as laplace, but
> laplace
> ><https://maxima.common-lisp.dev/docs/maxima_95.html#index-laplace> is
> >weirdly in the Functions and Variables for Differentiation
> ><https://maxima.common-lisp.dev/docs/maxima_95.html> section.
> >
> >Not sure what to do about that.  Of course, this only matters if you're
> >reading the manual all at once and not just as a single help entry in
> which
> >case the section the entry is in doesn't really matter so much.
>
> Laplace for me is a method of solving differential equations, not for
> Differentiation...
>

Yes, but laplace is basically an integral so being in the differentiation
section is just plain weird to me.

>
>


-- 
Ray

Re: [Maxima-discuss] Intro to Special Functions section and other issues.

[Robert D. <rob...@gm...>]

I kind of like the table of special functions, although clearly it
could be fixed up a lot -- cutting out obsolete stuff, moving stuff
into more appropriate sections, etc.

What counts as a special function isn't well defined, and it's
probably not terribly important exactly what's in the table. But we
can try to list whatever is conventionally recognized. Other stuff can
be omitted from the table.

I don't think it's important for all functions in the table to be in
the same Texinfo node (Functions and Definitions for Special Functions
or whatever). Ideally we would link from the table to the item.

Also ideally we would format the table using Texinfo @ macros (I don't
know what's best, @itemize, @enumerate, something else?) instead of
fixed width text.

Anyway just some opinions about that, thanks for looking at it.

best,

Robert

Re: [Maxima-discuss] Intro to Special Functions section and other issues.

[Gunter K. <gu...@pe...>]

Am 6. Juli 2022 18:46:24 MESZ schrieb Raymond Toy <toy...@gm...>:
>In the section Introduction to Special Functions
><https://maxima.common-lisp.dev/docs/maxima_78.html>, there's a table of
>special functions.  It's a bit strange.  It lists assoc_legendre_p and q,
>but they're not in the Special functions section.  All of the elliptic
>functions and integrals are also missing.  There's erfc, but no erf.
>There's a reference to kelliptic, but that doesn't exist, probably replaced
>by ellpitic_kc.
>
>Would it be better to just rip out this table?  And maybe add some links to
>the elliptic functions/integrals section, the orthopoly section with some
>appropriate text?
>

I believe that table makes the manual longer without providing additional information.

>Also, Functions and Variable for Special Functions
><https://maxima.common-lisp.dev/docs/maxima_87.html> has specint, which is
>not what I would call a special function.  hypergeometric_simp and hgfred
>are here, but perhaps that's better in the hypergeometric section.  If
>these are rmoved, we would only have the lambert functions, kbateman, and
>nzeta (plasma dispersion function).
>
>I thought specint might be better in the same place as laplace, but laplace
><https://maxima.common-lisp.dev/docs/maxima_95.html#index-laplace> is
>weirdly in the Functions and Variables for Differentiation
><https://maxima.common-lisp.dev/docs/maxima_95.html> section.
>
>Not sure what to do about that.  Of course, this only matters if you're
>reading the manual all at once and not just as a single help entry in which
>case the section the entry is in doesn't really matter so much.

Laplace for me is a method of solving differential equations, not for Differentiation...
>

[Maxima-discuss] Intro to Special Functions section and other issues.

[Raymond T. <toy...@gm...>]

In the section Introduction to Special Functions
<https://maxima.common-lisp.dev/docs/maxima_78.html>, there's a table of
special functions.  It's a bit strange.  It lists assoc_legendre_p and q,
but they're not in the Special functions section.  All of the elliptic
functions and integrals are also missing.  There's erfc, but no erf.
There's a reference to kelliptic, but that doesn't exist, probably replaced
by ellpitic_kc.

Would it be better to just rip out this table?  And maybe add some links to
the elliptic functions/integrals section, the orthopoly section with some
appropriate text?

Also, Functions and Variable for Special Functions
<https://maxima.common-lisp.dev/docs/maxima_87.html> has specint, which is
not what I would call a special function.  hypergeometric_simp and hgfred
are here, but perhaps that's better in the hypergeometric section.  If
these are rmoved, we would only have the lambert functions, kbateman, and
nzeta (plasma dispersion function).

I thought specint might be better in the same place as laplace, but laplace
<https://maxima.common-lisp.dev/docs/maxima_95.html#index-laplace> is
weirdly in the Functions and Variables for Differentiation
<https://maxima.common-lisp.dev/docs/maxima_95.html> section.

Not sure what to do about that.  Of course, this only matters if you're
reading the manual all at once and not just as a single help entry in which
case the section the entry is in doesn't really matter so much.

But it makes for weird reading if you want to know about what maxima can do
with differentiation and you find the laplace transform there.

-- 
Ray

[Maxima-discuss] A&S references for assoc_legendre_q?

[Raymond T. <toy...@gm...>]

In the section for assoc_legendre_q
<https://maxima.common-lisp.dev/docs/maxima_306.html#index-assoc_005flegendre_005fq>,
the text references A&S 8.5.3
<https://personal.math.ubc.ca/~cbm/aands/page_334.htm> and 8.1.8
<https://personal.math.ubc.ca/~cbm/aands/page_333.htm>

8.5.3 is a relationship for assoc_legendre_p of different degrees and 8.1.8
is the Wronskian.

Perhaps this is supposed to be 8.6.7 (assoc_legendre_q in terms of
legendre_q) and maybe 8.1.3 (assoc_legendre_q as a hypergeometric for |z| >
1).

I'm not really sure what the right references should be here, but 8.6.7
seems to be correct.

-- 
Ray

[Maxima-discuss] Javascript from Maxima, was: Probably a Simple Query

[Robert D. <rob...@gm...>]

On Tue, Jul 5, 2022 at 7:44 PM Joel Scott <joe...@uq...> wrote:

> Just wondering if there is a way to force maxima not to simplify, specifically, can I make x*x stay as that rather than be simplified to x^2 (lack of exponentials in javascript).

Great question, exporting Maxima to other systems is an important topic.

Instead of suppressing the simplification x*x --> x^2 (it turns out
that turning off just that one identity is more or less impossible),
try to output Math.pow(x, 2) instead. Here's a way to rewrite an
expression with Math.pow.

subst ("**" = lambda ([a, b], Math.pow (a, b)), myexpr);

e.g.

myexpr: expand ((x - a*y)^3);
mynewexpr: subst ("**" = lambda ([a, b], Math.pow (a, b)), myexpr);

output:

(-3*Math . pow(x,2)*a*y)+3*Math . pow(a,2)*Math . pow(y,2)*x
                        -Math . pow(a,3)*Math . pow(y,3)+Math . pow(x,3)

Note that Maxima wants to display "." with a space on either side -- I
believe this has no effect on JS interpretation of the expression. It
is possible to suppress the extra spaces, but let's let it go for now.

To print that expression in a format which is interpretable as JS, try
the  function f90, which outputs Fortran 90 code. For math
expressions, I believe that's acceptable as JS as well.

load (f90);
f90_output_line_length_max:1000000;
f90 (mynewexpr);

output:

(-3*Math . pow(x,2)*a*y)+3*Math . pow(a,2)*Math . pow(y,2)*x-Math .
pow(a,3)*Math . pow(y,3)+Math . pow(x,3)

That's printed to the console. If you want to capture it to a file or
output stream, see with_stdout.

There are various problems to solve when working with going back and
forth from Maxima to another system -- if you say more about what
you're trying to do, we can probably give some specific advice.

best,

Robert Dodier

Re: [Maxima-discuss] Probably a Simple Query

[Gunter K. <gu...@pe...>]

That's what I meant: If we tell gentran to generate C there is a high probability we'll get valid JavaScript. I believe the "==" operator might be misleading, if present, though...

Re: [Maxima-discuss] Probably a Simple Query

[Michel T. <ta...@lp...>]

Gentran does more. It acts as a compiler for the maxima language, that is

it knows the grammar of maxima and recognizes the maxima constructs. For 
each construct it

builds an equivalent fortran or C construct, the same as a C compiler 
recognizes C constructs

and transform them to assembly language constructs.


Le 06/07/2022 à 06:05, Gunter Königsmann via Maxima-discuss a écrit :
>
> What I don't know is if gentran does something similar when generating 
> C code.

-- 
Michel Talon

Re: [Maxima-discuss] Probably a Simple Query

[Michel T. <ta...@lp...>]

In javascript i see that there exists a function "pow" like in C.  From 
Google:

"In JavaScript, *pow()* is a function that is used to return m raised to 
the nth power. Because the pow() function is a static function of the

Math object, it must be invoked through the placeholder object called Math."


So you can use the program to convert to C syntax (similar to the 
program to convert to fortran syntax, see the

fortran command) that you can find under share/contrib/cgrind.lisp. So 
simply do load(cgrind); and then

cgrind(expression) like you do grind(expression) or fortran(expression).




Le 06/07/2022 à 04:26, Joel Scott a écrit :
> Hi All,
>
> Just wondering if there is a way to force maxima not to simplify, 
> specifically, can I make x*x stay as that rather than be simplified to 
> x^2 (lack of exponentials in javascript).
>
> I can't find examples (since it seems like a pretty trivial example 
> and most times people seem to be looking to simplify).
>
> Thanks so much,
>
> Joel Scott
>
>
>
> _______________________________________________
> Maxima-discuss mailing list
> Max...@li...
> https://lists.sourceforge.net/lists/listinfo/maxima-discuss

-- 
Michel Talon

Re: [Maxima-discuss] Probably a Simple Query

[Richard F. <fa...@gm...>]

Write a power () routine.

On Tue, Jul 5, 2022, 10:40 PM Joel Scott <joe...@uq...> wrote:

> Hi All,
>
> Just wondering if there is a way to force maxima not to simplify,
> specifically, can I make x*x stay as that rather than be simplified to x^2
> (lack of exponentials in javascript).
>
> I can't find examples (since it seems like a pretty trivial example and
> most times people seem to be looking to simplify).
>
> Thanks so much,
>
> Joel Scott
>
> _______________________________________________
> Maxima-discuss mailing list
> Max...@li...
> https://lists.sourceforge.net/lists/listinfo/maxima-discuss
>

Re: [Maxima-discuss] Probably a Simple Query

[Gunter K. <gu...@pe...>]

You can set

simp:false;

which turns off nearly all of maxima's intelligence. But as maxima isn't meant to be used this way and isn't prepared to deal with unsimplified data expect maxima not to be of much help afterwards.

I guess, though, the other list members will know ways to convert maxima's output to a more java-like expression.

What I don't know is if gentran does something similar when generating C code.

Kind regards,

   Gunter.

Am 6. Juli 2022 04:26:45 MESZ schrieb Joel Scott <joe...@uq...>:
>Hi All,
>
>Just wondering if there is a way to force maxima not to simplify, specifically, can I make x*x stay as that rather than be simplified to x^2 (lack of exponentials in javascript).
>
>I can't find examples (since it seems like a pretty trivial example and most times people seem to be looking to simplify).
>
>Thanks so much,
>
>Joel Scott
>

[Maxima-discuss] Probably a Simple Query

[Joel S. <joe...@uq...>]

Hi All,

Just wondering if there is a way to force maxima not to simplify, specifically, can I make x*x stay as that rather than be simplified to x^2 (lack of exponentials in javascript).

I can't find examples (since it seems like a pretty trivial example and most times people seem to be looking to simplify).

Thanks so much,

Joel Scott

Re: [Maxima-discuss] proposal to omit call to radcan in changevar

[Richard F. <fa...@gm...>]

Since sqrt(1) is the set {  -1, +1 }, the rhs is mathematically wrong.
Radcan says it chooses one root, and so it does.


On Tue, Jul 5, 2022, 7:53 PM Barton Willis via Maxima-discuss <
max...@li...> wrote:

>
>    - What does anyone think about revising changevar to not call radcan?
>
>
> This is almost surely a good thing to do:
>
>
>
> Radcan is a great tool to use interactively, but it can be a troublemaker
> when used programmatically:
>
>
>
> (%i20) domain : complex$
>
>
>
> (%i21)          xxx : (x^(2))^(1/2)$
>
>
>
> (%i22)          subst(x=-1, radcan(xxx)) = subst(x=-1, xxx);
>
> (%o22)         -1=1
>
>
>
>
>
> Also for *changevar,* I think it calls solve. The setting of *solveradcan*
> might alter the result of solve.  Maybe the testsuite should set the value
> of *solverradcan *before calling *changevar?*
>
>
>
> --Barton
>
>
>
> PS  Don’t some rtest3 expected values need to be updated due to expunging
> *absconvtest*?
>
>
>
> _______________________________________________
> Maxima-discuss mailing list
> Max...@li...
> https://lists.sourceforge.net/lists/listinfo/maxima-discuss
>

Re: [Maxima-discuss] proposal to omit call to radcan in changevar

[Barton W. <wi...@un...>]

  *   What does anyone think about revising changevar to not call radcan?

This is almost surely a good thing to do:

Radcan is a great tool to use interactively, but it can be a troublemaker when used programmatically:

(%i20) domain : complex$

(%i21)          xxx : (x^(2))^(1/2)$

(%i22)          subst(x=-1, radcan(xxx)) = subst(x=-1, xxx);
(%o22)         -1=1


Also for changevar, I think it calls solve. The setting of solveradcan might alter the result of solve.  Maybe the testsuite should set the value of solverradcan before calling changevar?

--Barton

PS  Don’t some rtest3 expected values need to be updated due to expunging absconvtest?

Re: [Maxima-discuss] proposal to omit call to radcan in changevar

[Kris K. <kat...@gm...>]

On Tue, Jul 05, 2022 at 03:44:10PM -0700, Raymond Toy wrote:
> IIRC, maxima was using some special feature (or implementation of let) that
> did destructuring.  I think I renamed it to destructuring-let to make it
> clearer that this is not the normal CL let.  It was confusing to have
> destructuring in let (to me at least).  It's possible I messed that up
> somehow.

In Maclisp LET supported destructuring.  Later someone (presumably
Schelter) wrote a LET macro in CL that supported destructuring, and that
was later renamed to DESTRUCTURING-LET.  (Likewise for LET*.)

Cheers,
Kris Katterjohn

Re: [Maxima-discuss] proposal to omit call to radcan in changevar

[Raymond T. <toy...@gm...>]

On Tue, Jul 5, 2022 at 3:41 PM Richard Fateman <fa...@gm...> wrote:

> I fiddled with the code in nrat4 to replace  pfactor with psqfr.  No other
> change.   It makes the example work, but fails to simplify
> sqrt(1-t)/sqrt1-t^2).  I'm more convinced someone went over the code after
> me, because it uses features that I don't think I used in 1970 .. like
> destructuring-let.  But maybe I used it then?  Did it exist in 1970?
>

IIRC, maxima was using some special feature (or implementation of let) that
did destructuring.  I think I renamed it to destructuring-let to make it
clearer that this is not the normal CL let.  It was confusing to have
destructuring in let (to me at least).  It's possible I messed that up
somehow.

> RJF
>
>
> On Tue, Jul 5, 2022 at 4:07 PM Raymond Toy <toy...@gm...> wrote:
>
>> I think this makes sense.  I know I've removed calls to radcan in other
>> places exactly for the same reason.  IIRC, I basically replaced it with
>> ratsimp or something like that.
>>
>> On Tue, Jul 5, 2022 at 12:58 PM Robert Dodier <rob...@gm...>
>> wrote:
>>
>>> We've identified at least one problematic example for changevar in
>>> which the call to radcan causes trouble. Given radcan's potential for
>>> unexpected results, maybe we can do without the call in changevar.
>>>
>>> Cutting out the call to radcan from changevar and running the test
>>> suite, I see there's one failed test, which can be fixed up by
>>> changing the expected result to an equivalent form.
>>>
>>> There are only a few test cases which call changevar, maybe it
>>> wouldn't hurt to create more.
>>>
>>> What does anyone think about revising changevar to not call radcan?
>>>
>>> best,
>>>
>>> Robert
>>>
>>>
>>> _______________________________________________
>>> Maxima-discuss mailing list
>>> Max...@li...
>>> https://lists.sourceforge.net/lists/listinfo/maxima-discuss
>>>
>>
>>
>> --
>> Ray
>> _______________________________________________
>> Maxima-discuss mailing list
>> Max...@li...
>> https://lists.sourceforge.net/lists/listinfo/maxima-discuss
>>
>

-- 
Ray

Re: [Maxima-discuss] URL links

[Raymond T. <toy...@gm...>]

I've merged the branch to the main branch.  I've added links to A&S and
DLMF in the Special functions section.

Since we use macros to define the links, we can change the format if needed.

On Sun, Jul 3, 2022 at 9:06 PM Raymond Toy <toy...@gm...> wrote:

> I've implemented these two macros and updated a few links in the special
> function section.  See
> https://common-lisp.net/project/cmucl/downloads/maxima/maxima_singlepage.html#Bessel-Functions
> (Note:  this page loads really slowly now because of all the equations
> there.  Perhaps we can look into making it a bit faster so that rendering
> isn't completely blocking page navigation.  I don't have any problems going
> to a page with raw tex and eventually gets replaced with a nice formula.).
>
> And also https://common-lisp.net/project/cmucl/downloads/maxima/maxima.pdf
>
> Let me know what you think of the format of the links.
>
>
> On Fri, Jun 24, 2022 at 1:56 PM Raymond Toy <toy...@gm...> wrote:
>
>> I looked around the manual a bit and we basically use "A&S x.y.z" in most
>> places, but orthopoly has the nice "Abramowitz and Stegun, equations
>> 22.5.37, page 779".
>>
>> Similarly, I see "DLMF x.y.z" or for stirling1, there is already a link.
>>
>> Perhaps it would be best to define two macros such as the following:
>>
>> @macro urlaands {eqn, page}
>> @url{https://personal.math.ubc.ca/~cbm/aands/page_\page\.htm, Abramowitz
>> and Stegun eqn \eqn\}
>> @end macro
>>
>> @macro urldlmf {link}
>> @url{https://dlmf.nist.gov/\link\, DLMF \link\}
>> @end macro
>>
>> so that we'll have consistent links and link text everywhere.
>>
>> So @urlaands{9.1.3, 358} produces Abramowitz and Stegun eqn 9.1.3
>> <https://personal.math.ubc.ca/~cbm/aands/page_358.htm> and
>> @urldlmf{26.8.E17} produces DLMF 26.8.E17
>> <https://dlmf.nist.gov/26.8.E17>.
>>
>> The info (text manual) just has the link text: "Abramowitz and Stegun eqn
>> 9.1.3 (https://personal.math.ubc.ca/~cbm/aands/page_358.htm)" and "DLMF
>> 26.8.E17 (https://dlmf.nist.gov/26.8.E17)".
>>
>> I think this looks good.  The HTML and pdf docs have clickable links.
>> The manual shows the link so you can cut and paste it to a browser.
>>
>>
>>
>> --
>> Ray
>>
>
>
> --
> Ray
>


-- 
Ray

Re: [Maxima-discuss] proposal to omit call to radcan in changevar

[Richard F. <fa...@gm...>]

I fiddled with the code in nrat4 to replace  pfactor with psqfr.  No other
change.   It makes the example work, but fails to simplify
sqrt(1-t)/sqrt1-t^2).  I'm more convinced someone went over the code after
me, because it uses features that I don't think I used in 1970 .. like
destructuring-let.  But maybe I used it then?  Did it exist in 1970?
RJF


On Tue, Jul 5, 2022 at 4:07 PM Raymond Toy <toy...@gm...> wrote:

> I think this makes sense.  I know I've removed calls to radcan in other
> places exactly for the same reason.  IIRC, I basically replaced it with
> ratsimp or something like that.
>
> On Tue, Jul 5, 2022 at 12:58 PM Robert Dodier <rob...@gm...>
> wrote:
>
>> We've identified at least one problematic example for changevar in
>> which the call to radcan causes trouble. Given radcan's potential for
>> unexpected results, maybe we can do without the call in changevar.
>>
>> Cutting out the call to radcan from changevar and running the test
>> suite, I see there's one failed test, which can be fixed up by
>> changing the expected result to an equivalent form.
>>
>> There are only a few test cases which call changevar, maybe it
>> wouldn't hurt to create more.
>>
>> What does anyone think about revising changevar to not call radcan?
>>
>> best,
>>
>> Robert
>>
>>
>> _______________________________________________
>> Maxima-discuss mailing list
>> Max...@li...
>> https://lists.sourceforge.net/lists/listinfo/maxima-discuss
>>
>
>
> --
> Ray
> _______________________________________________
> Maxima-discuss mailing list
> Max...@li...
> https://lists.sourceforge.net/lists/listinfo/maxima-discuss
>

Re: [Maxima-discuss] [Maxima-commits] [git] Maxima CAS branch master updated. branch-5_46-base-224-gec3813eeb

[Wolfgang D. <wol...@da...>]

Am 05.07.22 um 22:20 schrieb Raymond Toy:
> Would it help if I changed the call to maxima-local to use the default 
> lisp instead?  I think that's possible, but I'd have to investigate how 
> to invoke it.

Hi Ray,

Building the Windows installer works with my commit now. No need to do 
any changes.

Best regards, Wolfgang


P.S. Some minor issues, which I found here:

I get some warnings (when trying the build on Linux (the same for the 
crosscompiled windows installer) here e.g. using SBCL):

make[3]: Entering directory '/home/dauti/Software/maxima-code/doc/info'
./build_html.sh
./build_html.sh: abs_srcdir=/home/dauti/Software/maxima-code/doc/info, 
pwd=/home/dauti/Software/maxima-code/doc/info, language=en, 
languagedir=/home/dauti/Software/maxima-code/doc/info, 
texiinitfile=/home/dauti/Software/maxima-code/doc/info/texi2html.init
./build_html.sh: execute category program: "sbcl" --script 
make-categories.lisp
makeinfo: warning: set_from_cmdline: unknown variable HTML_MATH
./Special.texi:91: unknown command `displaymath'
./Special.texi:94: warning: unknown @end displaymath
./Special.texi:135: unknown command `displaymath'
./Special.texi:137: warning: unknown @end displaymath

This is just the beginning, there are many more warnings about 
displaymath/@end displaymath.



And if some Lisp expert knows a fast solution: Maxima can be build using 
ABCL 1.8.0, but fails with the current release 1.9.0. (both on Linux and 
when building the crosscompiled Windows installer):

;        - Compiling source file
; 
"/home/dauti/Software/maxima-code/src/numerical/f2cl-package.lisp"; 
Compiling 
/home/dauti/Software/maxima-code/src/numerical/f2cl-package.lisp ...

Caught FILE-ERROR while processing --eval option "(funcall (intern 
(symbol-name :operate-on-system) :mk) "maxima" :compile :verbose t)":
   Unable to open 
#P"/home/dauti/Software/maxima-code/src/numerical/binary-abcl/numerical/f2cl-package.abcl-tmp".
make[1]: *** [Makefile:1450: binary-abcl/maxima] Error 2


I have no idea, how to fix that - and if the issue is on the Maxima side 
or ABCL side.
I don't believe, that Maxima with ABCL is really used, so it is not 
really 'high priority', but I include it in the nightly builds of the 
Windows installer (just because it works) and since it is 'supported', 
it should probably work with the recent release too.

Re: [Maxima-discuss] [Maxima-commits] [git] Maxima CAS branch master updated. branch-5_46-base-224-gec3813eeb

[Raymond T. <toy...@gm...>]

On Tue, Jul 5, 2022 at 1:43 PM Wolfgang Dautermann <wol...@da...>
wrote:

> Am 05.07.22 um 22:20 schrieb Raymond Toy:
> > Would it help if I changed the call to maxima-local to use the default
> > lisp instead?  I think that's possible, but I'd have to investigate how
> > to invoke it.
>
> Hi Ray,
>
> Building the Windows installer works with my commit now. No need to do
> any changes.
>

Ok.  I'll leave it then.

>
> Best regards, Wolfgang
>
>
> P.S. Some minor issues, which I found here:
>
> I get some warnings (when trying the build on Linux (the same for the
> crosscompiled windows installer) here e.g. using SBCL):
>
> make[3]: Entering directory '/home/dauti/Software/maxima-code/doc/info'
> ./build_html.sh
> ./build_html.sh: abs_srcdir=/home/dauti/Software/maxima-code/doc/info,
> pwd=/home/dauti/Software/maxima-code/doc/info, language=en,
> languagedir=/home/dauti/Software/maxima-code/doc/info,
> texiinitfile=/home/dauti/Software/maxima-code/doc/info/texi2html.init
> ./build_html.sh: execute category program: "sbcl" --script
> make-categories.lisp
> makeinfo: warning: set_from_cmdline: unknown variable HTML_MATH
> ./Special.texi:91: unknown command `displaymath'
> ./Special.texi:94: warning: unknown @end displaymath
>

You need texinfo 6.8 or later.  I think that's when support for MathJax was
added and also @displaymath.

I wish we could have stayed with an earlier version, but 6.8 adds support
so that @math gets converted to Mathjax when enabled.  This helps a lot,
but does require some care because the syntax isn't always right.  For
example @math{x^(n-1)} doesn't produce the right html output.  You need to
write @math{x^{n-1}).  But then the info file says x^{n-1}, which is ok, I
guess.

> ./Special.texi:135: unknown command `displaymath'
> ./Special.texi:137: warning: unknown @end displaymath
>
> This is just the beginning, there are many more warnings about
> displaymath/@end displaymath.
>
>
>
> And if some Lisp expert knows a fast solution: Maxima can be build using
> ABCL 1.8.0, but fails with the current release 1.9.0. (both on Linux and
> when building the crosscompiled Windows installer):
>
> ;        - Compiling source file
> ;
> "/home/dauti/Software/maxima-code/src/numerical/f2cl-package.lisp";
> Compiling
> /home/dauti/Software/maxima-code/src/numerical/f2cl-package.lisp ...
>
> Caught FILE-ERROR while processing --eval option "(funcall (intern
> (symbol-name :operate-on-system) :mk) "maxima" :compile :verbose t)":
>    Unable to open
>
> #P"/home/dauti/Software/maxima-code/src/numerical/binary-abcl/numerical/f2cl-package.abcl-tmp".
> make[1]: *** [Makefile:1450: binary-abcl/maxima] Error 2
>
>
> I have no idea, how to fix that - and if the issue is on the Maxima side
> or ABCL side.
> I don't believe, that Maxima with ABCL is really used, so it is not
> really 'high priority', but I include it in the nightly builds of the
> Windows installer (just because it works) and since it is 'supported',
> it should probably work with the recent release too.
>
>
>
>
>

-- 
Ray

Re: [Maxima-discuss] [Maxima-commits] [git] Maxima CAS branch master updated. branch-5_46-base-224-gec3813eeb

[Robert D. <rob...@gm...>]

On Tue, Jul 5, 2022 at 1:25 PM Raymond Toy <toy...@gm...> wrote:

> Would it help if I changed the call to maxima-local to use the default lisp instead?  I think that's possible, but I'd have to investigate how to invoke it.

Default Lisp is executed as part of the build in doc/info (replacing
Python which was formerly used there). Maybe that can give some
inspiration.

best,

Robert

Re: [Maxima-discuss] [Maxima-commits] [git] Maxima CAS branch master updated. branch-5_46-base-224-gec3813eeb

[Raymond T. <toy...@gm...>]

Would it help if I changed the call to maxima-local to use the default lisp
instead?  I think that's possible, but I'd have to investigate how to
invoke it.

On Tue, Jul 5, 2022 at 12:08 PM Wolfgang Dautermann via Maxima-commits <
max...@li...> wrote:

> This is an automated email from the git hooks/post-receive script. It was
> generated because a ref change was pushed to the repository containing
> the project "Maxima CAS".
>
> The branch, master has been updated
>        via  ec3813eeb7d0b397e03bea810fb15a754695b675 (commit)
>       from  61c00fd721df4f78dad8bffea8ae7325b1fd18a8 (commit)
>
> Those revisions listed above that are new to this repository have
> not appeared on any other notification email; so we list those
> revisions in full, below.
>
> - Log -----------------------------------------------------------------
> commit ec3813eeb7d0b397e03bea810fb15a754695b675
> Author: Wolfgang Dautermann <da...@us...>
> Date:   Tue Jul 5 21:04:28 2022 +0200
>
>     Fix the crosscompiled Maxima build.
>
>     It seems, maxima-local is now called during the build process.
>
>     This resulted in error messages like
>     fatal error encountered in SBCL pid 106107:
>     more than one core file specified
>
>     We call during the build sbcl.exe --core sbcl.core (using wine),
>     and a maxima specific maxima.core image is one core file too much.
>
>     So check if, the "--core" option is used (just a quick check, if it is
>     the first parameter, there might be other options before, but for now
>     that works).
>
> diff --git a/crosscompile-windows/wine-sbcl.sh.tmpl
> b/crosscompile-windows/wine-sbcl.sh.tmpl
> index 09b107091..ff8563370 100755
> --- a/crosscompile-windows/wine-sbcl.sh.tmpl
> +++ b/crosscompile-windows/wine-sbcl.sh.tmpl
> @@ -9,4 +9,8 @@ export WINEDLLOVERRIDES
>  SBCL_HOME=Z:@CMAKE_BINARY_DIR@/sbcl/
>  export SBCL_HOME
>  unset DISPLAY
> -@WINE_EXE@ @CMAKE_BINARY_DIR@/sbcl/sbcl.exe --core @CMAKE_BINARY_DIR@/sbcl/sbcl.core
> "$@"
> +if [ "$1" = "--core" ]; then
> +    @WINE_EXE@ @CMAKE_BINARY_DIR@/sbcl/sbcl.exe "$@"
> +else
> +    @WINE_EXE@ @CMAKE_BINARY_DIR@/sbcl/sbcl.exe --core @CMAKE_BINARY_DIR@/sbcl/sbcl.core
> "$@"
> +fi
>
> -----------------------------------------------------------------------
>
> Summary of changes:
>  crosscompile-windows/wine-sbcl.sh.tmpl | 6 +++++-
>  1 file changed, 5 insertions(+), 1 deletion(-)
>
>
> hooks/post-receive
> --
> Maxima CAS
>
>
> _______________________________________________
> Maxima-commits mailing list
> Max...@li...
> https://lists.sourceforge.net/lists/listinfo/maxima-commits
>


-- 
Ray

Re: [Maxima-discuss] proposal to omit call to radcan in changevar

[Raymond T. <toy...@gm...>]

I think this makes sense.  I know I've removed calls to radcan in other
places exactly for the same reason.  IIRC, I basically replaced it with
ratsimp or something like that.

On Tue, Jul 5, 2022 at 12:58 PM Robert Dodier <rob...@gm...>
wrote:

> We've identified at least one problematic example for changevar in
> which the call to radcan causes trouble. Given radcan's potential for
> unexpected results, maybe we can do without the call in changevar.
>
> Cutting out the call to radcan from changevar and running the test
> suite, I see there's one failed test, which can be fixed up by
> changing the expected result to an equivalent form.
>
> There are only a few test cases which call changevar, maybe it
> wouldn't hurt to create more.
>
> What does anyone think about revising changevar to not call radcan?
>
> best,
>
> Robert
>
>
> _______________________________________________
> Maxima-discuss mailing list
> Max...@li...
> https://lists.sourceforge.net/lists/listinfo/maxima-discuss
>


-- 
Ray

[Maxima-discuss] proposal to omit call to radcan in changevar

[Robert D. <rob...@gm...>]

We've identified at least one problematic example for changevar in
which the call to radcan causes trouble. Given radcan's potential for
unexpected results, maybe we can do without the call in changevar.

Cutting out the call to radcan from changevar and running the test
suite, I see there's one failed test, which can be fixed up by
changing the expected result to an equivalent form.

There are only a few test cases which call changevar, maybe it
wouldn't hurt to create more.

What does anyone think about revising changevar to not call radcan?

best,

Robert