Your usage looks right to me. There are a couple of things you can check here:
1) bdroot may refer to a different model after the call to compile. This can happen if the top model contains referenced model or if there are any model callbacks which interact with another model. Or if you simply click on another model between the call to compile and terminate. I suggest using a more explicit syntax to avoid this:
>> mdl = 'myModel'
>> feval(mdl, [], [], [], 'compile')
2) if the original code snippet is more than what you've shown, this may be due to the queued behvaior of compile and terminate commands: if compile command was issued more than once on the same model, without any calls to term in between, you would need to call term the same number of times to actually termiante the compilation. You should see warnings in this case though. For example:
>> feval(‘myModel’,[],[],[],’compile’);
>> feval(‘myModel’,[],[],[],’compile’); % Warning: myModel is already compiled
>> feval(‘myModel’,[],[],[],’term’); % Warning: Termination of 'myModel' deferred
>> feval(‘myModel’,[],[],[],’term’); %no warning; model compilation is undone.
Btw, I tried this in R2022a, and i see the compiled state labeled as 'compiled' on the UI, after the compilation call.
