Type safety!

[d.healey]

@Christoph-Hart Found another one

In file included from ../../../../../HISE/hi_scripting/hi_scripting_01.cpp:98,
                 from ../../JuceLibraryCode/include_hi_scripting_01.cpp:9:
../../../../../HISE/hi_scripting/scripting/api/ScriptingApi.cpp:3601:64: error: macro "ADD_TYPED_API_METHOD_2" requires 3 arguments, but only 2 given
 3601 |     ADD_TYPED_API_METHOD_2(setAttribute, VarTypeChecker::Number);

[d.healey]

What's going on here?

Too many arguments in API call Server.callWithPOST(). Expected: 1

[oskarsh]

super nice! now I can obsessively type everything :D

[d.healey]

I notice this now returns an error

local myVariable;
myComponent.set("tooltip", myVariable);

In my actual project myVariable may or may not contain a value. So now I have to add an isDefined check.

Would it be possible to add something like:
myComponent.set("tooltip", myVariable || ""); ?

Edit: Oh actually that appears to already work!

[aaronventure]

From the source code, these are the types and the identifiers, right?

static const Identifier Undefined("undefined");
    static const Identifier Integer("int");
    static const Identifier Double("double");
    static const Identifier Number("number");
    static const Identifier String("string");
    DECLARE_ID(Colour);
    DECLARE_ID(Array);
    DECLARE_ID(Buffer);
    DECLARE_ID(ObjectWithLength);
    DECLARE_ID(JSON);
    DECLARE_ID(ScriptObject);
    static const Identifier Object("object");
    DECLARE_ID(Function);
    DECLARE_ID(ComplexType);
    DECLARE_ID(NotUndefined);

@Christoph-Hart Please correct me, then add this to the docs

[Christoph Hart]

In my actual project myVariable may or may not contain a value. So now I have to add an isDefined check.

Yes but this was already the case - you can't call functions with an undefined variable - I've introduced this "limitation" to HiseScript years ago because the times it silently fails is much higher than when you actually want to pass on an undefined value. It was just that the detection logic so far didn't catch these problems when the object that you call it to is dynamic:

// A const object
const var MyEffect = Synth.getEffect("My Effect");

// The same object but not accessed through a const var
const var MyDynamicEffect = [MyEffect];

// This will throw an error
MyEffect.setAttribute(undefined, 12);

// Up until now this went through fine, but should also cause an error just like the line above
MyDynamicEffect[0].setAttribute(undefined, 12);

then add this to the docs

Done, but it's not in the HISE docs website because I need to rebuild the HTML files when I'm back tomorrow.

[d.healey]

I'm finding since this change I am getting some fun errors, but the line numbers given are not correct (or at least are not clear).

For example:

[Christoph Hart]

@d-healey ah good catch, it mistakenly interpreted a function as a JSON object (which is technically correct, but it should identify as Function before). Should be fixed now.

[d.healey]

@Christoph-Hart Thank you

[Dan Korneff]

This safety check is pretty aggressive. I'm trying to track down some API call with undefined parameter 0 errors, but HISE isn't being helpful at the moment.
I get an error at a line that doesn't contain code, and if I comment out the ENTIRE script I still get the error
The only way to make it clear the error is to restart HISE.

It seems to be circling around addComboBox, and that combo box initializes with a value of 0.0.

edit:
it seems to initialize to 0.0 if "saveInPreset": "0" . Is that by design?

[d.healey]

@Dan-Korneff said in Type safety!:

I get an error at a line that doesn't contain code

I'm seeing this a lot too. However if you look above the Console, on the same line as the Compile button you might get a more helpful message.

ComboBox values start at 1, so that might be the issue.

[Christoph Hart]

You guys know you can just change a flag in HISE to restore the old compiler strictness?

HISE/hi_scripting/hi_scripting.h at d91fb47d5cec35358dbf56235fb412bc931a723e · christophhart/HISE

The open source framework for sample based instruments - HISE/hi_scripting/hi_scripting.h at d91fb47d5cec35358dbf56235fb412bc931a723e · christophhart/HISE

GitHub (github.com)

I really stand by this decision from a language design perspective, undefined as function arguments cause more pain than it is useful, but I understand how annoying it is if you just want to pull the latest changes and find your interface completely broken because of a new restriction.

Anyways, I've changed the default value here so that old projects load without issues, and you even can deactivate the warning in the Hise preferences (Development -> Warn If Undefined Parameters) if you want to go full on ostrich on your code quality :)

[Dan Korneff]

@Christoph-Hart nah, I'd rather know that these issues exist and deal with them. Thankful for the feature, annoyed with debugging

[Adam_G]

@Christoph-Hart do we just need to change #define HISE_WARN_UNDEFINED_PARAMETER_CALLS 0 to 1? or is there something else that needs to be changed?

[Christoph Hart]

@Adam_G pull the latest changes, then this flag is set and prints the warning. If you then want to disable the warning, untick the box in the preferences.

[Adam_G]

@Christoph-Hart getting some errors on compiling

[d.healey]

@Adam_G I think that means you are low on RAM. Try closing other programs that you are not using when compiling or reduce the number of threads that VS is using during the compilation.

[Lindon]

@d-healey said in Type safety!:

@Adam_G I think that means you are low on RAM. Try closing other programs that you are not using when compiling or reduce the number of threads that VS is using during the compilation.

• its more likely to mean that the plugin is using too many png's or wav files stored inside the plugin itself - past about 50Mb (if I recall correctly) the compiler runs out of heap space....on Windows only

[Adam_G]

@d-healey doh! you were right. too many other programs open

[Lindon]

@Adam_G said in Type safety!:

@d-healey doh! you were right. too many other programs open

careful --- re-running the compile when th eissue is the one I pointed at can sometimes run successfully...