Translating JSFX to C++

[shades13]

Hi,

So I’m somewhat of a hobby programmer but have been creating my own JSFX for quite a while now. Over the past few weeks, I’ve been starting to port some of my JSFX scripts into WDL-OL so that I can compile them on my Mac and use them in Logic, since there’s no Audio Unit version of ReaJS.

I’ve had great success with a few of them, but others don’t work anything like has intended and massively overload/clip/produce generally weird results. I think the core of the issue is I don’t understand how EEL translates over into the different math types available in C++.

In particular, I have one plugin, which is a heavily-modified version of the Liteon Deesser, which doesn’t work at all.

I understand that EEL uses doubles for all numbers, but how does this translate into C++?
With C++ doubles, do I need to put decimal points after every number, so e.g. 1 become 1.0 to make sure it is calculated as a double, even when the variable type is specified as a double? And what other potential snags are there to get caught on that could be affecting my results?

[Xenakios]

Originally Posted by shades13

so e.g. 1 become 1.0 to make sure it is calculated as a double, even when the variable type is specified as a double?

Yes, this can be a potential issue.

For example :

int x = 10;
int y = 11;
double z = x/y; // this is *not* going be 0.909090 or so, it will be 0

C++ calculates the expressions on the right of the = first, using the types those variables have. It doesn't first look what is the type of the variable you are assigning the result to. In the above example, the division is done as integers, resulting in 0 and then that is assigned to the double.

You can use a cast like this to force the calculation to happen as floating point :

int x = 10;
int y = 11;
double z = (double)x/y;

Or of course, by just using floating point variables to begin with. (If you are using numeric literals, then yes, you should use 1.0 instead of 1 and so on.)

Other possible issue when converting EEl/JesuSonic code to C++ could be that errors might be handled differently when calling the numerical functions or doing arithmetic. Eel might for example return 0.0 for calculations that in C++ will cause the result to be inf(inity) or nan (not-a-number). Justin could probably explain if there are any possible issues like that. (But if such cases exist, then the original JesuSonic plugin code was kind of wrong to begin with...)