Waveform Parameters not writeable?

[edyounis]

I have the following python code:

            if not isinstance(instruction, Instruction.Pulse):
                continue

            waveform = instruction._0.waveform
            waveform.parameters["scale"] = quil.expression.Expression.Number(1.0 + 0.0j)
            # cache this waveform

The goal here is to cache all the waveform shapes regardless of the scale. If you throw away the scale, there is a lot of compression potential in my use case. However, the waveform.parameters don't change after the write. Additionally, this doesn't show as an error, the code executes fine.

Edit:

Trying to explore workarounds, I found the following code errors:

            waveform2 = quil.waveform.Waveform.from_quil(waveform.to_quil())

with:

TypeError: argument 'invocation': 'str' object cannot be cast as 'WaveformInvocation'

[edyounis]

I understand now that in the above example I have a WaveformInvocation, not a Waveform. The problem here is that either 1) the parameters should be writable or 2) they should report an error that it is not writable. Without knowing if this would effect the larger software infrastructure, I would vote for 1 because that would make my use case easier :)

Nevertheless the issue is because the WaveformInvocation struct uses an IndexMap for its parameters. When pyo3 converts this to python, the IndexMap gets copied into a python dict rather than referenced. This can be solved by having WaveformParameters wrap the IndexMap rather than alias it:

pub type WaveformParameters = IndexMap<String, Expression>;

becomes

#[cfg_attr(feature = "python", pyclass)]
pub struct WaveformParameters {
    pub map: IndexMap<String, Expression>,
}

impl Deref for WaveformParameters {...}
impl DerefMut for WaveformParameters {...}

#[cfg(feature = "python")]
#[pymethods]
impl WaveformParameters {
    #[new]
    fn new() -> Self {
        Self { map: IndexMap::new() }
    }

    fn __getitem__(&self, key: String) -> PyResult<Expression> {
        self.get(&key)
            .cloned() // maybe, not quite sure how Expression should be handled.
            .ok_or_else(|| PyKeyError::new_err(key))
    }

    fn __setitem__(&mut self, key: String, value: Expression) {
        self.insert(key, value);
    }

    ...
}

Edit:

So, that still doesn't work fully because the WaveformInvocation pyclass is frozen :( But, it still enabled me to do what I want by constructing a new object :)

            waveform = instruction._0.waveform
            parameters = waveform.parameters
            parameters["scale"] = quil.expression.Expression.Number(1.0 + 0.0j)
            rescaled_waveform = quil.instructions.WaveformInvocation(waveform.name, parameters)