@LourensVeen with these recent commits, this will build and run, i,.e. if one adds a line

    print(f"{code.get_nj_max() = }, ")

to Erwan's script after the line

    channel_a.copy()

it will print 131072, which means that the user gets to know the maximum number of stars that Sapporo Light allows for and adjust his/her Python script accordingly.

Closes #1208

Closes #1208

In the sense that the user has access to nj_max.

But when running a simulation with > nj_max stars, you still do not get a meaningful error message.

But when running a simulation with > nj_max stars, you still do not get a meaningful error message.

This has now been fixed, i.e. if the Python script has a line

    code = Ph4(converter, redirection="file", mode="gpu")

then there will be a line

ERROR: nj = 131073 exceeds GPU capacity nj_max = 131072

in the code.out output file.

Codacy Static Code Analysis is complaining about an unused function, i.e. the abort in line 71 of send_fetch_data.cpp

    if (nj > nj_max) {
      fprintf(stdout, "ERROR: nj = %d exceeds GPU capacity nj_max = %d\n", nj, nj_max);
      abort;
    }

but that complaint does not make sense to me, nor does the other complaint - Error prone Method 'get_nj_max' has no argument.

I think the unused function warning Codacy givesmay be about the nj variable, and that it wants you to write if (address_j.size() > nj_max) and avoid the extra variable.

I'm not sure though, and this is reported so spectacularly poorly that I suggest we leave it as is if only out of spite.

I think the unused function warning Codacy givesmay be about the nj variable, and that it wants you to write if (address_j.size() > nj_max) and avoid the extra variable.

Okay, but

    int nj = address_j.size();

is not part of this MR and nj is needed below.

Hey Hanno! I appreciate you working on this, but I am not sure if it is completely fixed.
I recently rebuilt the package and when I went above the particle limit, the code crashed giving the same error message as #1208. I went on to manually increase the particle number, and still the same outcome happened.

After several trials, I built from commit 8dd292c and applied your original fix manually by changing nj_max. With that version, the code worked.

So it looks like something introduced after that commit may have broken or interfered with the fix.

I recently rebuilt the package and when I went above the particle limit, the code crashed giving the same error message as #1208.

And it should, but you should now be able to find a meaningful error message in your output file if you set redirection="file".

After several trials, I built from commit 8dd292c and applied your original fix manually by changing nj_max. With that version, the code worked.

I guess by original fix you mean manually increasing nj_max = 131072;?
It is always an option to set nj_max to the number of stars you need and recompile, but as far as I recall, it was not the idea of this MR to set nj_max to a number much larger than $2^{17}$, since that would just shift the problem.
However, please let me know if you want a different default value, that is obviously super easy to implement.

Ideally, one would like to have nj_max set through the controlling Python script, but that would require a major code overhaul; currently nj_max is set at compile time, if you would want to set that at run time, this would involve quite some effort.

Oh, I probably should have read this more carefully. My bad. Thanks!