Big update to prettier-extension-monkeyc

flocsy said:

Is there a way I can know if the build is with or without Prettier Optimizer?

I think so, but it's a little awkward. You'd have to add "when_with_prettier" to your base.excludeAnnotations in your monkey.jungle. Then we'd need an option in the optimizer to add and remove excludeAnnotations; in this case you'd want to add "when_without_prettier" and remove "when_with_prettier".

Or we could drop the option and just say that those are the names, and the optimizer always removes "when_with_prettier" and adds "when_without_prettier" (if we go that route, I'd probably pick different names).

But I think the option could be generally useful. We could have an "extraExcludes" option, which could be overridden in build and launch configs, and you would set something like "extraExcludes: foo, -bar", which would add foo and remove bar. Then in addition to making this feature possible, you could have multiple build configs that build different variants of your app.

Let's start with the "built-in" that doesn't need configuration (you can add the more complex things later :) 

I agree a better name would be welcome.... Maybe one f these pairs:

optimizer, no_optimizer

optimized, not_optimized

flocsy said:

Application.loadResource(Rez.JsonData.label_data_1);

Yes, sorry, my fix was pretty much entirely wrong.

It was supposed to check whether Toybox.Lang.ResourceId existed, and use that if so, or Toybox.Lang.Symbol otherwise. But the check was incorrect and always said it existed, so it always tried to use ResourceId, which wasn't defined under 6.x, and got interpreted as Any. But my tests were (accidentally) set to TypeCheckLevel=Gradual, which meant that "Any" was acceptable as a parameter, and no error/warning was produced.

Anyway, a new fix should be going out soon.

Fix is out: v2.0.77 (may take a few minutes to show up)

v2.0.78 is out, and adds an extraExcludes option, and more Toybox documentation on hover.

I noticed something that in contra intuitive. Do you know how to explain it?

function is_numeric(str as String) as Boolean {
    var chars = str.toCharArray(); // api 1.3.0
    var result = true;
    for (var i = chars.size() - 1; i >= 0; i--) {
        // if (NUMERIC_CHARS.find(chars[i].toString()) == null) {
        if (NUMERIC_CHARS.find("" + chars[i]) == null) { // -3 bytes
            result = false;
            break; // +3
        }
    }
    return result;
}

How come that c.toString() is 3 bytes longer than "" + c where c is a char?
I would think that "" + c probably implicitly calls toString anywhere and it also includes an concatenation so I would expect it to be longer code than to just call toString implicitly.

pmc-analysis bug:

function foo(c as Char) {
    return '0' <= c && c <= '9';
}

Gives a warning: "Unexpected types for operator '<=': [Char vs Char]"

I'll take a look...

I think it's because the conversion really is implicit (I think you meant that the compiler would insert a call to toString, but it doesn't). Assuming the char is in a local "c", the byte code is just

    spush "" // 5 bytes, plus storage for string
    lgetv c  // 2 bytes
    addv     // 1 byte

While a call is pretty complex:

    lgetv c         // 2 bytes
    spush toString  // 5 bytes
    getv            // 1 byte
    frpush          // 1 byte
    invokem 1       // 2 bytes

Note that identical strings are deduped, so if you mention "" anywhere in your program, there's no extra overhead for mentioning it here. I've not really looked into how much space a string takes up, but I would guess its 4 bytes plus the string data.