From alexpascarella96 at gmail.com  Fri Apr  1 12:39:05 2022
From: alexpascarella96 at gmail.com (Alex Pascarella)
Date: Fri, 1 Apr 2022 14:39:05 -0500
Subject: [Spice_discussion] Query CSPICE within Matlab MEX dynamical model
Message-ID: <CAPQwiFUsPxhV8o4QQjd_+C_GrmAbStaFfa1DxbP76VErtds6Kg@mail.gmail.com>

Hi all,
For my research, I have a dynamical model implemented as a Matlab file that
I compile into .mex file through the codegen command.
In the dynamical model file, I need to query the position of the planets
using spkezr.
Since mice comes as a precompiled MEX application, I can't use it within my
matlab file, because that is not compatible with code generation.

My current workaround is as follows:

function [...] = dynamics(..)

    [...]

    % Include C files required for code generation
    coder.updateBuildInfo('addIncludePaths', '[...]\mice\include')
    coder.updateBuildInfo('addIncludePaths', '[...]\mice\src\cspice')
    libPriority = '';
    libPreCompiled = true;
    libLinkOnly = true;
    libName = 'cspice.lib';
    libPath = '[...]\mice\lib';
    coder.updateBuildInfo('addLinkObjects', libName, libPath, libPriority,
libPreCompiled, libLinkOnly);
    coder.cinclude('[...]\mice\src\cspice\SpiceUsr.h')

    % Preallocate output
    starg = zeros(6, 1);
    lt = 0;

    % Query cspice
    coder.ceval('spkezr_c', [targ 0], et, [ref 0], [abcorr 0], [obs 0],
coder.wref(starg), coder.wref(lt));

    [...]
end

The code above compiles successfully when using codegen and the .mex files
produces the expected results, however there is a glaring issue: if an
error is encountered within the spkezr_c call (e.g. kernels are not loaded,
strings are incorrect, etc.) the entire Matlab application crashes without
errors.
Is there any workaround that would allow me to fix this problem and keep
using codegen to generate the .mex file?
Thanks!
Alex
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://naif.jpl.nasa.gov/pipermail/spice_discussion/attachments/20220401/f2d1f9c5/attachment-0001.html>

From alexpascarella96 at gmail.com  Tue Apr  5 13:57:17 2022
From: alexpascarella96 at gmail.com (Alex Pascarella)
Date: Tue, 5 Apr 2022 15:57:17 -0500
Subject: [Spice_discussion] Query CSPICE within Matlab MEX dynamical model
Message-ID: <CAPQwiFXoJW8F=eB1A9V2fOHP0aLodYcSRHF1Htvisdc26Q2MAw@mail.gmail.com>

Hi all,
After receiving very useful feedback from the SPICE team, I was able to
solve the issues I was running into and integrate cspice within my MEX
model.
A full example can be found here:
https://urldefense.us/v3/__https://github.com/EllissoideRotondo/mex_cspice_wrapper_example__;!!PvBDto6Hs4WbVuu7!YrD03VH5oZcfwDu-TMtyN2C3ZTI-iZIw_jWjZnUSv0K6j0h3xqxFRbw8yOyQR4AIjG0WRFikIx4dOQ$ .

Some important takeaways:
- To avoid crashes, it is necessary to call the cspice function erract_c
and set it to "RETURN" before any other cspice calls, which prevents cspice
from throwing an exception. This only needs to be called once per program
run.

- It is a good idea to call errdev_c and set the output device to NULL,
which prevents cspice from trying to dump errors to stdout. This also needs
to be run only once.

- Any exceptions encountered by cspice can be captured using the cspice
functions failed_c (check if any error occured), getmsg_c (get the error
message), and reset_c (reset the error status). Note that reset_c must be
called any time an exception occurs, otherwise cspice will keep returning
an error on subsequent calls.
Once captured, a Matlab exception can be thrown using the error command (in
MEX, you will need to define coder.exstrisinc('error'), as the command is
not natively compatible with code generation).

- As a final note, I also recommend upgrading to the latest version of
Matlab (R2022a) if possible. There is a bug in earlier versions that
occasionally causes Matlab to crash when running certain MEX files in debug
mode.

Thanks,
Alex Pascarella
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://naif.jpl.nasa.gov/pipermail/spice_discussion/attachments/20220405/32d35c77/attachment.html>