In some cases nodes are enumerated/visited multiple times about esprima-dotnet HOT 3 CLOSED

So is your suggestion maybe to have then nullable and nulll when they are missing from the AST?

from esprima-dotnet.

I think the model is fine (AST explorer generates the same AST), so I wouldn't change that. It just feels off that the enumeration includes the same node instance multiple times.

What we could possibly do is to check for these cases and make sure these nodes are enumerated/visited only once.

However, there's the case of shorthand properties of object patterns. Consider var { v } = { } first. In this case the parser doesn't use the same Identifier instance for Key and Value but creates two separate instances with the same content (as opposed to the case of object expression shorthand properties, where only a single instance is created).

Take another example: var { v = 0 } = { }. Now Value is an AssignmentExpression whose Left property is the same identifier as Key. Also two different Identifer instances with the same content are created but in this case they don't even have the same parent node...

How to deal with this? The only correct way I can think of is to change the parser (to be precise, ParsePropertyPattern) to reuse the Key instance in the case of shorthand properties. Then change the enumeration/visitation to skip the Key node in such cases.

from esprima-dotnet.

I don't have a strong opinion here. I always have my Jint glasses on so basically for Jint approach doesn't matter, Jint handles the content of AST without worrying of identity.

Would "back to javascript" be a good use case to prove if the current model is problematic? Like does double-visiting make things hard if not impossible.

I don't se much harm in tweaking the ParsePropertyPattern logic.

from esprima-dotnet.