How can I tell codegen the argument type is double?

I tried to use -args {1.0}, but it generates a lot of real_T stuff, that's not what I want. I was wondering if it is possible to generate a function whose arguments are of type double ? I can do that using Matlab coder, is it also possible to use codegen to achieve it?

Answer 1

[Walter Roberson]

https://www.mathworks.com/help/coder/ug/mapping-matlab-types-to-cc-types.html

real_T is double by default. Is it a problem for you that the code generates real_T that is a typedef to double, instead of directly generating double ?

[JINGYI]

Thanks for your reply. And yes, I would like codegen to directly generate double.

[Walter Roberson]

What difficulties are you encountering arising from codegen emitting a typedef'd or macro name that resolve to double instead of emitting double directly ?

See also https://www.mathworks.com/help/coder/ug/specify-data-types-used-in-generated-code.html

[Walter Roberson]

Oner reason I have for asking about the problem you are encountering, is to try to figure out whether the problem could be resolved by adding a -Dreal_T=double to the command line, which would not change the source code but would cause references to real_T to be compiled directly to double instead of going through a typedef. That could alter some warning messages slightly. (But it could also potentially lead to problems if there is indeed a typedef of that name, as the compiler would then read

typedef double real_T

as if you had written

typedef double double

)

[JINGYI]

The reason is that I would like to integrate the generated code directly into my program without altering it since that would introduce some potential bugs into my program (because I cannot understand the details of those generated code, it is impossible for me to spot the problems if I mistakenly change the code).

And the workaround you mentioned looks acceptable to me. Thanks!