"Constant" Error when loading in model using loadLearnerForCoder()

On a project at work, we are trying to convert a ML model to C++ using the MATLAB coder. We have done it sucessfully before but now something has gone wrong. In MATLAB Coder, when it runs the error check, it complains the function call for loadLearnerForCoder fails b/c "The Model must be a compiler-time constant". For the sake of a testing, I regenerated a smaller model (b/c our last one took a whole day to generate). When I load it in the same way, it gives no errors. When comparing the two models in MATLAB I can't really see an issue. Does anyone know how to debug this better or see what MATLAB is erroring out on? Also I cannot provide any code for this due to work. I've tried using coder.constant() and it doesn't work. I also used coder.screener() and it reported no problems.

[Walter Roberson]

Roughly speaking, that error would be generated if you had something of the form

model_name = "YourModelNameGoesHere";
do_the_compile_stuff(model_name)

when instead you need

do_the_compile_stuff("YourModelNameGoesHere")

[Joshua]

The code is currently written roughly like this:

function [x,y] = predictFuntion(input) %#codegen

mdlPath = "path\to\model";

mdl = loadLearnerForCoder(mdlPath)

[x,y] = predict(mdl, input);

end

This works for one model but not the other. Would what you stated still be the case and I should just try it as one line instead of two?

[Walter Roberson]

Try it as one line.

[Joshua]

Unfortunately that did not work.