Hi Bertrand,

The “busy” behavior does not occur when when I start TeXmacs from the command 
line as opposed to launching from GUI (something in the environment which the 
Mathematica session sees is different in the two cases).

Cheers,
Peter.

> On 01 Sep 2016, at 12:23, BB <[email protected]> wrote:
> 
> Hi everybody,
> 
> I try to get the Mathematica plugin to work on OS X 10.9.5 and 10.9.6   with 
> mathematica 10.2.0  and some versions TeXmacs  ( including  Qt-based 1.99.5 
> and the last Fink distributions )
> 
> For doing that, I used an installer ( written in Mathematica, you can find it 
> here 
> <https://dl.dropboxusercontent.com/u/4023570/TeXmacs_Mathematica_Plugin_installer.zip>
>  )  which gave good results on Mac OS  10.8 with Texmacs 1.0.7 , and I tried  
> to improve it , taking into account the modifications suggested by Peter and  
> Tian-Min.
> Without surprise I got the a similar result ( fig 1: « En cours »= « in 
> progress »,  in French ).
> 
> Using the command ‘top’, I noticed that tm_mathematica and the mathematica 
> kernel are initially launched when we start the session  ( fig 1 )
> but both vanishe soon after an attempt to use mathematica ( fig 2 ).
> 
> 
>    <Capture d’écran 2016-08-30 à 15.58.01.png> ( fig 1 )
> 
>   <Capture d’écran 2016-08-30 à 15.58.30.png> ( fig 2 )
> 
> Best,
> Bertrand
> 
> ———————————————
> 
> 
> 
> Le 28 juin 2016 à 04:14, Peter Rapčan <[email protected] 
> <mailto:[email protected]>> a écrit :
> 
>> Hello everybody, 
>> 
>> 
>> I tried to get the Mathematica plugin to work on OS X 10.11.5, TeXmacs 
>> 1.99.4, with Mathematica 10.0.0.0.
>> 
>> 
>> First of all, I copied the tm_mathematica.c from Tian-Min Yan to the 
>> /Applications/TeXmacs.app/Contents/Resources/share/TeXmacs/plugins/mathematica/src.lazy/
>>  directory.
>> 
>> Then I performed two changes to the file: 
>> - replaced all occurences of MLDisownString by MLReleaseString
>> - replaced all occurences ofMLDisownSymbol by MLReleaseSymbol
>> (According to the Wolfram documentation, as of Mathematica 6.0 the former 
>> functions have been superseded by the latter functions. I am not sure from 
>> which version of Mathematica the old functions cannot be used; in any case I 
>> could not compile the tm_mathematica.c with the libMLi3.a choice without 
>> these changes in the case of  Mathematica 10.0.0.0). I suppose these changes 
>> need to be done also if one wanted to use the stock tm_mathematica.c 
>> provided in a TeXmacs installation.
>> 
>> The adapted tm_mathematica.c  is file is in the attached.
>> 
>> 
>> Next, I used the script provided by Singh Madhusudan and made some changes 
>> to make it work. The changes I had to make are the following:
>> - change the Mathlink_Path variable (Mathematica version 9 or greater 
>> contains the architecture specification [in my case MacOSX-x86-64] in the 
>> path to the directory where Mathlink resides).
>> - change the /usr/lib/ to /usr/local/lib/ since from OS X 10.11 the System 
>> Integrity Protection is enabled, forbidding to write to the /usr/lib/ 
>> directory irrespective of administrator rights.
>> - change the compiler commant to use g++ instead of gcc and use the oprion 
>> -l$lib_option -stdlib=libstdc++ instead of the -l$lib_option -lstdc++
>> 
>> With these changes, running the script compiled the tm_mathematica.c just 
>> fine, with a single warning “tm_mathematica.c:544:46: warning: format 
>> specifies type 'int' but the argument has type 'size_t' (aka 'unsigned long')
>>      [-Wformat]
>>    printf("\2prompt#\\red In[%1d]:= {}\5\5",InNum++);”
>> 
>> I attach the working script as “tmMMAscript".
>> 
>> 
>> Next, as TeXmacs did not recognize Mathematica as available in the list of 
>> supported sessions, I created a wrapper “math” command that runs the 
>> MathKernel  and placed in into /usr/local/bin/. The wrapper is attached. 
>> (note that creating a symlink to MathKernel does not work). After doing 
>> this, TeXmacs offers the Mathematica session in its list of supported 
>> sessions.
>> 
>> 
>> Upon including a Mathematica session, I was prompted whether I allow 
>> tm_mathematica.bin to accept incoming connections, from which I deduce the 
>> MathLink is working… well… almost. Unfortunatelly, the session does not 
>> return results of computations; instead it shows a “busy…” message (see the 
>> attached screenshot).
>> 
>> <Screen Shot 2016-06-28 at 03.43.48.png>
>> 
>> I looked at the tm_mathematica file provided by TeXmacs, and tried to tweak 
>> it (I think the paths are based on some old mathematica version, so I 
>> changed the relevant command to
>> 
>> export MATHLINK_PATH="$MATH3/CompilerAdditions"
>> 
>> or 
>> 
>> export MATHLINK_PATH="$MATH3/CompilerAdditions/AlternativeLibraries"
>> (According to the readme file provided in the AlternativeLibraries 
>> directory: “This AlternativeLibraries directory contains a versions of the 
>> MathLink
>> libraries compiled with -stdlib=libc++ compiler flags.  These libraries 
>> exist to link programs that need compatibility with Apple's newer libc++ C++ 
>> library.” 
>> Since I compiled with the -stdlib=libstdc++ option the second choice should 
>> be the correct one).
>> 
>> In any case, none of the changes I tried to make to the tm_mathematica file 
>> helped to make the plugin work, hence I am stuck with the busy message. Any 
>> help to resolve the issue would be greatly appreciated.
>> 
>> 
>> Cheers,
>> -Peter.
>> 
>> <tmMMAscript><tm_mathematica.c><math>
>> 
>> 
>>> On 31 Oct 2012, at 11:35, Tian-Min Yan <[email protected] 
>>> <mailto:[email protected]>> wrote:
>>> 
>>> Hi all,
>>> 
>>> Now I know how to output PS/EPS graphics. According to the manual
>>> section "Interfering TeXmacs with other programs", I simply put
>>> 
>>>   fputs("\2ps:", stdout);
>>>   fflush(stdout);
>>>   system("cat ~/tmp.eps");
>>>   fputs("\5{}{}\n\n", stdout);
>>> 
>>> into the code (see the attachment), now the EPS figure shows up. Can we
>>> adjust the size and the position of the figure within TeXmacs?
>>> 
>>> Besides, I am wondering if we can display a PNG file, since those
>>> rasterized formats have significantly smaller size and the program runs
>>> faster. We should be able to choose between different formats to meet
>>> different purposes.
>>> 
>>> One more question, can anybody who is familiar with Mathematica figure
>>> out why "Export["./tmp.eps", expr]" sometimes outputs figures in the
>>> wrong position, as is shown in the attachment where the main part is not
>>> at the center? In contrast, PNG file does not have this problem.
>>> 
>>> Best,
>>> Tian-Min
>>> 
>>> On Tue, 2012-10-30 at 19:01 +0100, [email protected] 
>>> <mailto:[email protected]>
>>> wrote: 
>>>> Hi Tian-Min,
>>>> 
>>>> The plugin worked perfectly for graphics in old times when Mathematica 
>>>> produced pure postscript graphics (this was in versions <=5).
>>>> If you are happy with the old-style Mathematica graphics, in your 
>>>> mathematica session simply evaluate : Get["Version5`Graphics`"]
>>>> and then the plugin will recognize the ps graphics and show them in 
>>>> TeXmacs (but only for old -or unchanged- graphics functions, old options 
>>>> etc.)
>>>> 
>>>> It would be nice to benefit from the new mathematica graphics, but that 
>>>> would require someone to volunteer for improving the plugin in that area.
>>>> 
>>>> Best,
>>>> Philippe
>>>> 
>>>> 
>>>> Tian-Min Yan - [email protected] <mailto:[email protected]> 
>>>> wrote:
>>>>> Hi all,
>>>>> 
>>>>> I am using Mathematica 8. The Mathematica plugin in TeXmacs works
>>>>> perfectly with a single expression if it does not contain graphical
>>>>> elements. But if we use functions involving plots, the output is
>>>>> problematic. Has anybody figured out the implementation of embedded
>>>>> figures with the Mathematica plugin?
>>>>> 
>>>>> Best,
>>>>> Tian-Min
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>> 
>>> 
>>> 
>>> <texmacs_wrong_position.png><tm_mathematica.c>
>> 
>> 
> 

_______________________________________________
Texmacs-dev mailing list
[email protected]
https://lists.gnu.org/mailman/listinfo/texmacs-dev

Reply via email to