Add implicit partial for multiple indexing

[cgarciae]

I am trying to run this:

        return (
            inputs.s
            |> tf.layers.dense$(?, 32, activation=tf.nn.relu, name='relu_layer', **ops)
            |> tf.nn.dropout$(?, inputs.keep_prob)
            |> tf.layers.dense$(?, 1, name='linear_layer', **ops)
            |> .[:, 0]
        )

but I am getting

CoconutParseError: parsing failed (line 147)

This other form is not working either

        return (
            inputs.s
            |> tf.layers.dense$(?, 32, activation=tf.nn.relu, name='relu_layer', **ops)
            |> tf.nn.dropout$(?, inputs.keep_prob)
            |> tf.layers.dense$(?, 1, name='linear_layer', **ops)
            |> lambda t: t[:, 0]
        )

I had to wrap the lambda inside parenthesis for the compiler to work properly:

        return (
            inputs.s
            |> tf.layers.dense$(?, 32, activation=tf.nn.relu, name='relu_layer', **ops)
            |> tf.nn.dropout$(?, inputs.keep_prob)
            |> tf.layers.dense$(?, 1, name='linear_layer', **ops)
            |> (lambda t: t[:, 0])
        )

[evhub]

@cgarciae The problem is that .[:, 0] is not a valid implicit partial, since as currently implemented it doesn't support comma syntax. That's definitely something worth implementing, though. Another workaround until then is (-> _[:, 0]) using implicit lambda syntax.

[evhub]

@cgarciae Fixed on coconut-develop! Just pip install coconut-develop to get the new behaviour.

[cgarciae]

Thanks!