FriCAS on windows

[Waldek Hebisch]

I thought a little about what can we write to Windows users.
IIUC:

- FriCAS release builds using mingw + sbcl (assuming utilites
from Mysys and ???).
- FriCAS builds using cygwin + ecl
- There is now Docker image by Kurt Pagani
- WSL, do I undertand correctly that FriCAS works (builds ???)
here
--
Waldek Hebisch

[Kurt Pagani]

Good idea! Would be desirable, indeed.

Another (IMO good) option for Windows users is abusing the awesome SageMath
Windows installer which provides (besides SageNotebook) a fully configured
Cygwin with ECL already installed. All one has to do then is download/
configure/make fricas-1.3.5-full.tar.bz2 (as reported below).
As a side effect one may use FriCAS from within SAGE etc. ;)

Get it:
-------
https://www.sagemath.org/download-windows.html
ftp://ftp.fu-berlin.de/unix/misc/sage/win/index.html

SageMath-8.6-Installer-v0.4.1.exe    1197.17 MB    2019-01-24 09:23


Build Fricas:
-------------
Open/(click) "SageMath 8.6 Shell":

(sage-sh) kfp@NUC:~$ uname -a
CYGWIN_NT-6.1 NUC 2.11.2(0.329/5/3) 2018-11-08 14:34 x86_64 Cygwin

(sage-sh) kfp@NUC:~$ which ecl
/opt/sagemath-8.6/local/bin/ecl

(sage-sh) kfp@NUC:~$ ecl --version
ECL 16.1.2

(sage-sh) kfp@NUC:~$ wget https://sourceforge.net/projects/fricas/files/fricas/1.3.5/fricas-1.3.5-full.tar.bz2

(sage-sh) kfp@NUC:~$ tar xvf fricas-1.3.5-full.tar.bz2
(sage-sh) kfp@NUC:~$ cd fricas-1.3.5

(sage-sh) kfp@NUC:fricas-1.3.5$ ./configure --with-lisp=ecl --without-x

config.status: creating contrib/emacs/Makefile
config.status: creating config/fricas_c_macros.h
extracting list of SPAD type definitions
Type 'make' (without quotes) to build FriCAS
(sage-sh) kfp@NUC:fricas-1.3.5$
(sage-sh) kfp@NUC:fricas-1.3.5$ make

...

   preloading /home/sage/fricas-1.3.5/target/x86_64-x-cygwin/algebra/OUTFORM.fas..loaded.

before fricas-restart
openServer result -2
                       FriCAS Computer Algebra System
                            Version: FriCAS 1.3.5
                    Timestamp: Mo, 11. Feb 2019 22:31:16
-----------------------------------------------------------------------------
   Issue )copyright to view copyright notices.
   Issue )summary for a summary of useful system commands.
   Issue )quit to leave FriCAS and return to shell.
-----------------------------------------------------------------------------

   Using local database /home/sage/fricas-1.3.5/target/x86_64-x-cygwin/algebra/compress.daase..   Using local database /home/sage/fricas-1.3.5/target/x86_64-x-cygwin/algebra/interp.daase..
   Using local database /home/sage/fricas-1.3.5/target/x86_64-x-cygwin/algebra/operation.daase..
   Using local database /home/sage/fricas-1.3.5/target/x86_64-x-cygwin/algebra/category.daase..
   Using local database /home/sage/fricas-1.3.5/target/x86_64-x-cygwin/algebra/browse.daase..
(1) ->
;;;
;;; Compiling /home/sage/fricas-1.3.5/contrib/emacs/gen-cpl.lisp.
;;; OPTIMIZE levels: Safety=0, Space=0, Speed=3, Debug=0
;;;
;;; End of Pass 1.
;;; Finished compiling /home/sage/fricas-1.3.5/contrib/emacs/gen-cpl.lisp.
;;;
(1) -> cat ./fricas-cpl.hd fricas-cpl.in ./fricas-cpl.tl > fricas-cpl.el
make[1]: Verzeichnis „/home/sage/fricas-1.3.5/contrib/emacs“ wird verlassen
(sage-sh) kfp@NUC:fricas-1.3.5$

(sage-sh) kfp@NUC:fricas-1.3.5$ make install

...
...
/usr/bin/install -c efricas '/usr/local/bin/efricas'
make[1]: Verzeichnis „/home/sage/fricas-1.3.5/contrib/emacs“ wird verlassen
79 FriCAS contrib installation finished.
(sage-sh) kfp@NUC:fricas-1.3.5$


before fricas-restart
openServer result 0
                       FriCAS Computer Algebra System
                            Version: FriCAS 1.3.5
                    Timestamp: Mo, 11. Feb 2019 22:31:16
-----------------------------------------------------------------------------
   Issue )copyright to view copyright notices.
   Issue )summary for a summary of useful system commands.
   Issue )quit to leave FriCAS and return to shell.
-----------------------------------------------------------------------------

   Function declaration sixel : TexFormat -> Void has been added to
      workspace.
                                                                   Type: Void

(1) -> D(x^n,x,12)

   (1)
        12       11         10          9           8            7             6
       n   - 66 n   + 1925 n   - 32670 n  + 357423 n  - 2637558 n  + 13339535 n
     +
                   5              4              3              2
       - 45995730 n  + 105258076 n  - 150917976 n  + 120543840 n  - 39916800 n
  *
      n - 12
     x
                                                    Type: Expression(Integer)
(2) -> )version

Value = "FriCAS 1.3.5 compiled at Mo, 11. Feb 2019 22:31:16"
(2) -> )lisp (lisp-implementation-type)

Value = "ECL"
(2) -> )lisp (lisp-implementation-version)

Value = "16.1.2"
(2) -> )q




-- PS
Use --with-x if Hyperdoc/Graphics is wished, however, you have to load the
X-libs before (see FricAS INSTALL), e.g. by using "apt-cyg"
(which is already installed).

[Martin R]

"sage -i fricas" is enough as soon as sage 8.7 is out.

If you are willing to use the sage develop branch, then fricas 1.3.5 is in fact already available.

Best wishes,

Martin

[Kurt Pagani]

On 11.02.2019 23:27, 'Martin R' via FriCAS - computer algebra system wrote:
> "sage -i fricas" is enough as soon as sage 8.7 is out.

Oh, good to know! I had no idea.

>
> If you are willing to use the sage develop branch, then fricas 1.3.5 is in fact
> already available.

Well, do you mean
http://doc.sagemath.org/html/en/reference/interfaces/sage/interfaces/fricas.html
(as a ECL lib?)

or will it be built standalone?

I'm not using the former very often (I admit :(
>
> Best wishes,
>
> Martin
>
Thanks

[Martin R]

I am sorry, I think I wrote nonsense.  After 'sage -i fricas' you can use fricas from *within* sage, but I guess this doesn't really make sense.

Please excuse the noise,

Martin

[Kurt Pagani]

On 11.02.2019 23:50, 'Martin R' via FriCAS - computer algebra system wrote:
> I am sorry, I think I wrote nonsense.  After 'sage -i fricas' you can use fricas
> from *within* sage, but I guess this doesn't really make sense.

Not at all. If "sage: fricas_console()" will work, then it's virtually the same.
I guess there are some who will use it this way, although not many of them on
Windows (just a bold claim ;)

>
> Please excuse the noise,
No worries!

>
> Martin
>
Greetings
Kurt

[oldk1331]

On 2/12/19 2:18 AM, Waldek Hebisch wrote:
> I thought a little about what can we write to Windows users.
> IIUC:
>
> - FriCAS release builds using mingw + sbcl (assuming utilites
> from Mysys and ???).

There are still problems for this method:

1. in function "obey" in vmlisp.lisp, the "lib/obey.bat" doesn't exist:

#+:sbcl
(defun OBEY (S)
#-:win32 (sb-ext::process-exit-code
(sb-ext::run-program "/bin/sh"
(list "-c" S) :input t :output t :error t))
#+:win32 (sb-ext::process-exit-code
(sb-ext::run-program (make-absolute-filename "/lib/obey.bat")
(list S) :input t :output t :error t))
)

2. Last time I tried, sbcl on windows can not compile texmacs.lsp
(heap exhausted). I haven't looked into it yet.

> - FriCAS builds using cygwin + ecl> - There is now Docker image by Kurt Pagani
> - WSL, do I undertand correctly that FriCAS works (builds ???)

Yes, it builds and works. And it should be fastest on windows.

> here
>

I'd like to include something to make a windows installer, like Kurt
did here: https://github.com/nilqed/fricas_w32

[Grégory Vanuxem]

Hi,

Le mar. 12 févr. 2019 à 01:57, oldk1331 <oldk...@gmail.com> a écrit :

> - WSL, do I undertand correctly that FriCAS works (builds ???)

Yes, it builds and works.  And it should be fastest on windows.

>   here
>

I'd like to include something to make a windows installer, like Kurt
did here: https://github.com/nilqed/fricas_w32

Yes completely faster. As an example :

WSL on left, MinGW64 on the right (same machine) :

https://drive.google.com/file/d/16cq6a_d2jDdGTiIFr7S4nu_eJL-gr-5a/view?usp=sharing

[Waldek Hebisch]

oldk1331 wrote:
>
> On 2/12/19 2:18 AM, Waldek Hebisch wrote:
> > I thought a little about what can we write to Windows users.
> > IIUC:
> >
> > - FriCAS release builds using mingw + sbcl (assuming utilites
> > from Mysys and ???).
>
> There are still problems for this method:
>
> 1. in function "obey" in vmlisp.lisp, the "lib/obey.bat" doesn't exist:
>
> #+:sbcl
> (defun OBEY (S)
> #-:win32 (sb-ext::process-exit-code
> (sb-ext::run-program "/bin/sh"
> (list "-c" S) :input t :output t :error t))
> #+:win32 (sb-ext::process-exit-code
> (sb-ext::run-program (make-absolute-filename "/lib/obey.bat")
> (list S) :input t :output t :error t))
> )
>
> 2. Last time I tried, sbcl on windows can not compile texmacs.lsp
> (heap exhausted). I haven't looked into it yet.

IIUC "obey" is not needed for buld from release tarball. Concerning
'texmacs.lsp' I at loss now, from earlier mails by Greg and Kurt
I thougt that it builds.

--
Waldek Hebisch

[Waldek Hebisch]

Gregory Vanuxem wrote:
>
> Hi,
>
> Le mar. 12 f=C3=A9vr. 2019 =C3=A0 01:57, oldk1331 <oldk...@gmail.com> a =
> =C3=A9crit :

>
> >
> > > - WSL, do I undertand correctly that FriCAS works (builds ???)
> >
> > Yes, it builds and works. And it should be fastest on windows.
> >
> > > here
> > >
> >
> > I'd like to include something to make a windows installer, like Kurt
> > did here: https://github.com/nilqed/fricas_w32
> >
> >
> Yes completely faster. As an example :
>
> WSL on left, MinGW64 on the right (same machine) :
>
> https://drive.google.com/file/d/16cq6a_d2jDdGTiIFr7S4nu_eJL-gr-5a/view?usp=

> =3Dsharing

Interesting. 90s versus 120s. I wonder why, do both build use
GMP?
--
Waldek Hebisch

[Kurt Pagani]

> IIUC "obey" is not needed for buld from release tarball.  Concerning
> 'texmacs.lsp' I at loss now, from earlier mails by Greg and Kurt
> I thougt that it builds.
> 
>

It builds fine (32bit), however, one has to increase DSS:

$ ./configure --with-lisp="sbcl --dynamic-space-size 1048"

kfp@NUC MINGW32 ~/fricas-1.3.5
$ uname -a
MINGW32_NT-6.1 NUC 2.9.0(0.318/5/3) 2017-09-13 23:16 x86_64 Msys



echo "else " >> efricas
echo "  echo You must have installed emacs to run efricas.;" >> efricas
echo "fi" >> efricas
make[1]: Zirkuläre Abhängigkeit gen-cpl.lisp <- gen-cpl.lisp wird nicht verwendet.
echo ')read gen-cpl.lisp' | \
            AXIOM=C:/msys64/home/kfp/fricas-1.3.5/target/i686-w64-mingw32 DAASE=C:/msys64/home/kfp/fricas-1.3.5/target/i686-w64-mingw32 \
            FRICAS_INITFILE='' C:/msys64/home/kfp/fricas-1.3.5/target/i686-w64-mingw32/bin/AXIOMsys
Checking for foreign routines
AXIOM="C:/msys64/home/kfp/fricas-1.3.5/target/i686-w64-mingw32"
spad-lib="C:/msys64/home/kfp/fricas-1.3.5/target/i686-w64-mingw32/lib/libspad.so"
foreign routines found

openServer result -2
                       FriCAS Computer Algebra System
                            Version: FriCAS 1.3.5

                    Timestamp: Di, 12. Feb 2019 20:56:11

-----------------------------------------------------------------------------
   Issue )copyright to view copyright notices.
   Issue )summary for a summary of useful system commands.
   Issue )quit to leave FriCAS and return to shell.
-----------------------------------------------------------------------------

   Using local database C:/msys64/home/kfp/fricas-1.3.5/target/i686-w64-mingw32/algebra/compress.daase..   Using local database C:/msys64/home/kfp/fricas-1.3.5/target/i686-w64-mingw32/algebra/interp.daase..
   Using local database C:/msys64/home/kfp/fricas-1.3.5/target/i686-w64-mingw32/algebra/operation.daase..
   Using local database C:/msys64/home/kfp/fricas-1.3.5/target/i686-w64-mingw32/algebra/category.daase..
   Using local database C:/msys64/home/kfp/fricas-1.3.5/target/i686-w64-mingw32/algebra/browse.daase..
(1) ->
; compiling file "C:/msys64/home/kfp/fricas-1.3.5/contrib/emacs/gen-cpl.lisp" (written 03 FEB 2019 01:36:26 PM):

; wrote C:/msys64/home/kfp/fricas-1.3.5/contrib/emacs/gen-cpl.fasl
; compilation finished in 0:00:00.005

(1) -> cat ./fricas-cpl.hd fricas-cpl.in ./fricas-cpl.tl > fricas-cpl.el

make[1]: Verzeichnis „/home/kfp/fricas-1.3.5/contrib/emacs“ wird verlassen

$ make install

...
make[1]: Verzeichnis „/home/kfp/fricas-1.3.5/contrib/emacs“ wird verlassen

79 FriCAS contrib installation finished.

****

$ fricas -nosman
Checking for foreign routines
AXIOM="C:/msys64/mingw32/lib/fricas/target/i686-w64-mingw32"
spad-lib="C:/msys64/mingw32/lib/fricas/target/i686-w64-mingw32/lib/libspad.so"
foreign routines found

openServer result -2
                       FriCAS Computer Algebra System
                            Version: FriCAS 1.3.5

                    Timestamp: Di, 12. Feb 2019 20:56:11

-----------------------------------------------------------------------------
   Issue )copyright to view copyright notices.
   Issue )summary for a summary of useful system commands.
   Issue )quit to leave FriCAS and return to shell.
-----------------------------------------------------------------------------

(1) -> )lisp  (lisp-implementation-type)

Value = "SBCL"
(1) -> )lisp  (lisp-implementation-version)

Value = "1.4.14"
(1) -> )q

kfp@NUC MINGW32 ~/fricas-1.3.5

[Kurt Pagani]

>
>
> Yes completely faster. As an example :
>
> WSL on left, MinGW64 on the right (same machine) :
>
> https://drive.google.com/file/d/16cq6a_d2jDdGTiIFr7S4nu_eJL-gr-5a/view?usp=sharing
>

> --

Indeed, I got similar results. WSL certainly is the method of choice for Windows
10 users, however, there still seems to be numerous working on Windows 7 ;

May I ask how you managed the MingW 64bit build? I can only build the mingw32
version.

[oldk1331]

On 2/13/19 3:02 AM, Waldek Hebisch wrote:

> oldk1331 wrote:
>
> IIUC "obey" is not needed for buld from release tarball. Concerning

I didn't know that. We should mention it in the notes for Windows users.

Still, we should make it possible to build trunk on windows.

> 'texmacs.lsp' I at loss now, from earlier mails by Greg and Kurt
> I thougt that it builds.
>
>

With the release tarball, I didn't encounter this problem.
I've successfully build fricas-1.3.5 on mingw64 on windows10.

BTW, IIUC, in mingw, "fork" is not available, so sman is not
built.

[oldk1331]

On the website of msys2, there is a 64bit version. Just use that to
install build dependencies (and 64bit sbcl), then you will have a
64bit build of fricas.

[Grégory Vanuxem]

Hello,

>
> May I ask how you managed the MingW 64bit build? I can only build the mingw32
> version.
>

On the website of msys2, there is a 64bit version.  Just use that to
install build dependencies (and 64bit sbcl), then you will have a
64bit build of fricas.

Yes on the https://www.msys2.org/ web site download it and SBCL (x86-64).

After I would suggest to use the package manager 'pacman' :

use
pacman -Syu

to upgrade your installation and close the terminal window and redo
the same things several time until your installation is up to date.

Don't ask why you need to close the terminal window each time I don't know why.

Afer : 'pacman -S' make and the x86-64 gcc to install them

If you want to build FriCAS from scratch i.e. without pre-generated files add the diffutils package,

the FriCAS build process use the 'cmp' command somewhere.

Cheers
__

Greg

 

[Kurt Pagani]

Thank you both, it works!

The problem was an outdated version (pacman -Syu solved the problems)

before: MINGW64_NT-6.1 NUC 2.9.0(0.318/5/3) 2017-09-13 23:16 x86_64 Msys
now: MINGW64_NT-6.1 NUC 2.11.2(0.329/5/3) 2018-11-26 09:22 x86_64 Msys

I cant's say which components were responsible for the failure, unfortunately
... but a "pacman -Syu" might be indicated anyway ;)

> --
> You received this message because you are subscribed to the Google Groups
> "FriCAS - computer algebra system" group.
> To unsubscribe from this group and stop receiving emails from it, send an email
> to fricas-devel...@googlegroups.com
> <mailto:fricas-devel...@googlegroups.com>.
> To post to this group, send email to fricas...@googlegroups.com
> <mailto:fricas...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/fricas-devel/CAHnU2dbpQ0-1wH5Sj_RBHsrkKoi8uSAy%2BgfpovrDOcMaB%2BWCDA%40mail.gmail.com
> <https://groups.google.com/d/msgid/fricas-devel/CAHnU2dbpQ0-1wH5Sj_RBHsrkKoi8uSAy%2BgfpovrDOcMaB%2BWCDA%40mail.gmail.com?utm_medium=email&utm_source=footer>.
> For more options, visit https://groups.google.com/d/optout.

[Grégory Vanuxem]

Hi,

Le mar. 12 févr. 2019 à 20:09, Waldek Hebisch <heb...@math.uni.wroc.pl> a écrit :

> WSL on left, MinGW64 on the right (same machine) :
>
> https://drive.google.com/file/d/16cq6a_d2jDdGTiIFr7S4nu_eJL-gr-5a/view?usp=
> =3Dsharing

Interesting.  90s versus 120s.  I wonder why, do both build use
GMP?

None were built with GMP support.

Shame on me

__

Greg

 

--
                              Waldek Hebisch

--
You received this message because you are subscribed to the Google Groups "FriCAS - computer algebra system" group.

To unsubscribe from this group and stop receiving emails from it, send an email to fricas-devel...@googlegroups.com.
To post to this group, send email to fricas...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/fricas-devel/E1gtdQr-0007Fd-MH%40hera.math.uni.wroc.pl.

[Waldek Hebisch]

Thanks for info. I have put preliminary version
of this at:

http://www.math.uni.wroc.pl/~hebisch/fricas/windows.txt

If you think that this is reasonably accurate and useful
I will add this to documentation pages at Sourceforge.

--
Waldek Hebisch

[Ralf Hemmecke]

Hi Waldek,

Do you have any opinion on writing such things in restructuredText?

I've already asked here
https://www.mail-archive.com/fricas...@googlegroups.com/msg12749.html
and up to now got no reaction.

Ralf

[Kurt Pagani]

Perfect. I think Win$ows users now have a sufficient number of options.
Thanks a lot!
Kurt

[Kurt Pagani]

Hi Ralf
I guess "rst" on SF will not diplay properly.
You could make/select a new branch on github/fricas in the settings, then adding
the suffix .rst to the files you mentioned should do (maybe some cosmetics will
be necessary, but seems peanuts to me;)
Kurt

[Ralf Hemmecke]

On 2/16/19 11:40 PM, Kurt Pagani wrote:
> Hi Ralf
> I guess "rst" on SF will not diplay properly.

Huh?

rst is a text format that is usually nicely readable also as pure text,
i.e., not translated into HTML by some converter.

Look at

https://raw.githubusercontent.com/hemmecke/fricas-ws/master/README.rst

I would still call this properly readable.

> You could make/select a new branch on github/fricas in the settings, then adding
> the suffix .rst to the files you mentioned should do (maybe some cosmetics will
> be necessary, but seems peanuts to me;)

Yeah I could do that on my own clone of FriCAS. I intend to keep

https://github.com/fricas/fricas/

an exact mirror of the SVN repo.

I definitely will not invest time in manually updating the respective
.rst files if the original .txt file changes. SVN should go away. I also
don't think that the sourceforge pages are particularly attractive.

But it's obviously only me who wants a change. :-(

Ralf

[Kurt Pagani]

On 17.02.2019 00:20, Ralf Hemmecke wrote:
> On 2/16/19 11:40 PM, Kurt Pagani wrote:
>> Hi Ralf
>> I guess "rst" on SF will not diplay properly.
>
> Huh?
>
> rst is a text format that is usually nicely readable also as pure text,
> i.e., not translated into HTML by some converter.

I thought you were interested in the HTML rendered form, displayed on SF ... a
misunderstanding.

>
> Look at
>
> https://raw.githubusercontent.com/hemmecke/fricas-ws/master/README.rst
>
> I would still call this properly readable.

Yes, of course.

>
>> You could make/select a new branch on github/fricas in the settings, then adding
>> the suffix .rst to the files you mentioned should do (maybe some cosmetics will
>> be necessary, but seems peanuts to me;)
>
> Yeah I could do that on my own clone of FriCAS. I intend to keep
>
> https://github.com/fricas/fricas/
>
> an exact mirror of the SVN repo.
>
> I definitely will not invest time in manually updating the respective
> .rst files if the original .txt file changes.

Understandable ;)

> SVN should go away. I also
> don't think that the sourceforge pages are particularly attractive.

I've been thinking that the problem will sort out itself someday, just because
of the unattractiveness and the tedious advertisments ... being tenacious.

>
> But it's obviously only me who wants a change. :-(

Not at all. The day will dawn when SF will resign ... principle of least
inherent necessity :)

>
> Ralf
>

[oldk1331]

Hi Kurt, is this line correct:

$ ./configure --with-lisp="sbcl --dynamic-space-size 1048"

I think it should be "sbcl.exe" instead of "sbcl". (One has
to also setup PATH properly.)

Also, "1048" feels strange -- is it 1024 or 2048?

And Waldek, will you add this file to the repo?

[Kurt Pagani]

On 18.02.2019 05:55, oldk1331 wrote:
>
>
> On 2/17/19 12:29 AM, Waldek Hebisch wrote:
>> Thanks for info. I have put preliminary version
>> of this at:
>>
>> http://www.math.uni.wroc.pl/~hebisch/fricas/windows.txt
>>
>> If you think that this is reasonably accurate and useful
>> I will add this to documentation pages at Sourceforge.
>>
>
> Hi Kurt, is this line correct:
>
> $ ./configure --with-lisp="sbcl --dynamic-space-size 1048"
>

$ ls -l $(which sbcl)
-rwxr-xr-x 1 kfp None 3163221 29. Nov 10:00 /usr/bin/sbcl

It worked but you're right it is not correct, at least not state of the art:

> I think it should be "sbcl.exe" instead of "sbcl". (One has
> to also setup PATH properly.)

I probably copied/renamed sbcl.exe to /usr/bin/sbcl and one should mention to
set up the PATH variable properly, indeed.

>
> Also, "1048" feels strange -- is it 1024 or 2048?

I recall that I tried 2048 in the first place (what didn't work) so I change the
"2" to "1", therefore 1048. Better is 1024 of course.

>
> And Waldek, will you add this file to the repo?
>

Do you agree to this one?
$ ./configure --with-lisp="/pathto/sbcl.exe --dynamic-space-size 1024"

[Grégory Vanuxem]

Hi all,

Yes SBCL should be on the PATH or the path to the executable need to be given.

Nevertheless wherever the sbcl executable is SBCL_HOME need to be defined on Windows since sbcl need to find the sbcl.core file.

Cheers

__

Greg

--
You received this message because you are subscribed to the Google Groups "FriCAS - computer algebra system" group.
To unsubscribe from this group and stop receiving emails from it, send an email to fricas-devel...@googlegroups.com.
To post to this group, send email to fricas...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/fricas-devel/f9054b7d-5937-7513-95fc-a250594f0155%40gmail.com.

[Waldek Hebisch]

Kurt Pagani wrote:
>
> On 18.02.2019 05:55, oldk1331 wrote:
> >
> > Hi Kurt, is this line correct:
> >
> > $ ./configure --with-lisp="sbcl --dynamic-space-size 1048"
> >

<snip>

> >
> > Also, "1048" feels strange -- is it 1024 or 2048?
>
> I recall that I tried 2048 in the first place (what didn't work) so I change the
> "2" to "1", therefore 1048. Better is 1024 of course.

Hmm, IIRC old Windows had 2GB, that is 2048 MB max for user programs.
Giving it all to sbcl would means that there are no space for
anything else, so this would not work. If one want to max space
given for FriCAS something like 1800 (+- 100) is probably better.

> >
> > And Waldek, will you add this file to the repo?
> >

Yes, assuming it is good enough.

> Do you agree to this one?
> $ ./configure --with-lisp="/pathto/sbcl.exe --dynamic-space-size 1024"

"/pathto/sbcl.exe" -- yes.

--
Waldek Hebisch

[Grégory Vanuxem]

Hello,

I agree with you. It would be good. And now, sadly, Microsoft owns github. I'm perplex. Not that I'm against github but I don't know what we can think about that.

A part that github is/was a good repository from my point of view.

__

Greg

--
You received this message because you are subscribed to the Google Groups "FriCAS - computer algebra system" group.
To unsubscribe from this group and stop receiving emails from it, send an email to fricas-devel...@googlegroups.com.
To post to this group, send email to fricas...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/fricas-devel/cab614f4-528f-7680-713f-0628a092b3c9%40hemmecke.org.

[Grégory Vanuxem]

Not for polemic of course. Just an internal though 

[Ralf Hemmecke]

The .txt vs. .rst is not so much about github or sourceforge, since I
use sphinx locally before I upload the generated .html files that form
fricas.github.io. In fact, I don't need github features at all.

But still, I think that .rst is the better text format.

Concerning Microsoft owning github. In fact, I don't care so much.
We are doing everything in the open anyway, I cannot see why anything
changed after github got aquired by M$.

All I want is using git instead of SVN and github looks more attractive
than SF. But I don't want to bind too tightly to github anyway. Just use
a few features but otherwise always be prepared if github shuts down one
day. In such a case, git is anyway the better repository format.

Ralf

[Grégory Vanuxem]

Le lun. 18 févr. 2019 à 22:20, Ralf Hemmecke <ra...@hemmecke.org> a écrit :

The .txt vs. .rst is not so much about github or sourceforge, since I
use sphinx locally before I upload the generated .html files that form
fricas.github.io. In fact, I don't need github features at all. 

Ah ok

But still, I think that .rst is the better text format.

Concerning Microsoft owning github. In fact, I don't care so much.
We are doing everything in the open anyway, I cannot see why anything
changed after github got aquired by M$.

All I want is using git instead of SVN and github looks more attractive
than SF. But I don't want to bind too tightly to github anyway. Just use
a few features but otherwise always be prepared if github shuts down one
day. In such a case, git is anyway the better repository format.

Ralf

I think I share your opinion on that stuff

(though I have to pull a git request somewhere and I have not found the time during the day to do this in a precise way :-) 

__

Greg

To view this discussion on the web visit https://groups.google.com/d/msgid/fricas-devel/bb9f7193-b501-a2f1-13c2-a9e4996bc3dd%40hemmecke.org.