kwagrs for macro calls

[mcmcgrath13]

I started working on supporting rendering of the vega and vegalite mime types (nominally working!), but I ran into the below issue when trying to have two plots:

I ran @macroexpand on the expressions and got:

:(dataset("cars") |> (VegaLite.VegaLite).VLSpec(VegaLite.convert_frag_tree_to_dict(VegaLite.fix_shortcut_level_spec((VegaLite.VLFrag)(VegaLite.Any[:point], VegaLite.OrderedDict{VegaLite.String, VegaLite.Any}("x" => :Horsepower, "y" => :Miles_per_Gallon))))))

I'm not sure why it's registering the x and y as variables in the main scope. Any idea what's happening here?

[fonsp]

Thanks!

Pluto works by figuring out the set of assigned variables and referenced variables for each cell. If one cell references a symbol x, and and another assigns to x, then that is a reactive link between them.

In this case, a bug in the analysis treats the keyword argument x=:asdf as an assignment. I wrote a special case for function calls, but not for macros.

A more general problem is that Pluto doesn't understand macros (#196), but luckily this macro can be parsed as if it is a function call - it does not need to be expanded for the syntax analysis to work.

[fonsp]

Should be fixed now, can you try again?>

[fonsp]

You need to run the master branch of Pluto

[mcmcgrath13]

I'm using the master branch of Pluto, but still seeing the same issue:

[fonsp]

This can't be fixed without expanding the macro (#196):
@f(x=y)
and
@f x = y
produce the same Expr, and something that looks like the second expression should be recognised as an assignment to x. (@benchmark x = y for example)

Try using a semicolon:

@vlplot(
	:a;
	x = :b,
	...
)

[fonsp]

This has been fixed by #670