Added forced casting of int values ​​to float for float-typed fields

[EsipovPA]

Added casting int to floats for float-typed fields in messages
Added test for new feature

Closes this issue

[clalancette]

I have to say that personally, I'm not a fan. int and float are not the same thing, and thus I think any conversion like this should be up to the user to explicitly do. I'd like to hear opinions from @ros2/team, though.

[EsipovPA]

Hello @clalancette!

Thanks for the quick reply.

It seems to me that this is a case of a widening conversion, that should be an option for these types of fields. In case of C++ code, the int values would be casted to float automatically, when passed to = operator. The idea behind this code is to make possible something like this for python code. Also it just makes python code of message initialization simpler and easier to read and write, as I see it.

For example, instead of something like this

msg = Msg(check_fields=True, value=float(1))

or

msg = Msg(check_fields=True, value=1.)

user should be able to write just

msg = Msg(check_fields=True, value=1)

and get the float value without raised exceptions.

Still confused about the failed test though. Checking

[clalancette]

It seems to me that this is a case of a widening conversion, that should be an option for these types of fields. In case of C++ code, the int values would be casted to float automatically, when passed to = operator. The idea behind this code is to make possible something like this for python code. Also it just makes python code of message initialization simpler and easier to read and write, as I see it.

Yeah, I understand what you are trying to do. I still don't think it is a good idea, even if C++ allows it (there is nothing we can do about that, it is part of the language).