source literals shouldn't have their own AST type

[TwitchBronBron]

Source literals should not have a dedicated expression type of SourceLiteralExpression. These work exactly the same as VariableExpression, so we should just use that. And then the list of known source literals can be handled in the validation and transpilation layers.

This is a breaking change, so we should do this before v1.0.0 lands.

[elsassph]

Is there a semantic value?

[TwitchBronBron]

I can't think of much semantic benefit. SourceLiteralExpression can contain have several different names (line_num, source_file_path, function_name, etc...) so we're either doing this (pseudocode)

ast.findChild(isSourceLiteralExpression).token.text.toLowerCase() === 'line_num'

or this

ast.findChild(isVariableExpression).name.text.toLowerCase() === 'line_num'

[markwpearce]

The value of having them as SourceLiteralExpression is that validation is done against VariableExpression -- we'd have to add the specifics to an allow list or something if they were all VariableExpression

I think having these separated makes the transpilation easier to understand too...