Unnamed objects in list are lost

[noamross]

When a list contains non-list objects without names, they are lost, as in the following:

> a = list(a=1,2, list(c=2))
> b  =as.Node(a)
> b
  levelName
1     Root 
2      °--1
> b$fields
[1] "a"

As you can see, the first object, a, is captured, and the third object, an unnamed list, is also captured, but the unnamed object, 2, disappears. These fields should be captured and named. I'll see what I can do. Should their names be numbers, as with unnamed list objects? Or some other designation, .1, .2, perhaps?

[gluc]

I get your point. Currently, only named list elements are considered, except if mode = "simple", then lists are interpreted to contain children and the as.Node auto-assigns a name if none is provided.

For fields, it's a bit different. Not sure how much sense un-named Node fields make. Nodes are environments, and as such their members are not ordered. If you auto-assign a name (similar to how it's done with children), then the user has a hard time figuring out which value represents what.

Happy to discuss, though!

[noamross]

I see. My use case is dealing with a large, heterogenous tree imported via JSON. Within the JSON there are unnamed arrays, so the elements of those arrays, whether simplified via jsonlite::fromJSON(...,simplifyVector) or not, are unnamed atomic types in list. So when I convert this list to data.tree there are always going to be unnamed values that I need to retrieve, and suspect there is plenty of JSON out there with similar structures.

Giving the unnamed fields numeric numbers helps somewhat: it provides the information that was previously stored in ordering of the array. I also like this because, when going back from data.tree to lists or JSON, I can remove names of both children and fields by looking for numbers.