Org-babel-dev: PROPOSED Create simple vectors in R?

PROPOSED Create simple vectors in R?

State "PROPOSED" from "" 2010-09-27 Mon 13:38

Currently all variables are received in R as data frames: we cannot create simple vectors and we cannot create matrices (i.e. simple vectors with a dimension attribute).

Proposals


	Current	A	B
Create data frames	Yes	Yes	No
Create R vectors?	No	Yes	No
Pass arbitrarily nested lists?	No	No	Yes

NA = no native representation

Current behaviour


Org	elisp	R	python
string	string	length 1 character vector	string
number	number	length 1 numeric vector	number
NA	list	data frame	list
table	list of lists	data frame	list of lists
NA	other nested list	ERROR	other nested list

Rows omitted from the proposals below are the same as in the current behaviour above.

Advantages

Disadvantages

Cannot create vectors

Cannot pass arbitrarily nested lisp lists

Proposal A: My initial patch in this thread


Org	elisp	R
NA	list	vector

Advantages

Simple way to create vectors in R

Disadvantages

Cannot pass arbitrarily nested lisp lists

Proposal B: Using R lists


Org	elisp	R
NA	list	list
table	list of lists	list of lists
NA	other nested list	other nested list

Advantages

Conceptually simple

lisp lists == R lists

Appropriate

These are the analogous data structures in the two languages (mutable, mixed-type, arbitrarily nested lists). It solves the problem of sending nested lisp structures to R.

Disadvantages

Unnatural conversion to data frame

a b

1 2

is '((a b) (1 2))

which becomes list(list("a", "b"), list(1,2))

It would be natural to use data.frame(lapply(x, unlist)) but this would give us the transpose of what we want.

a	b
1	2

Cases

integer

print(str(x))

1D vector

print(str(x))

Column table


1
2

print(str(x))

Row table


1	2

print(str(x))

2D table


1	3
2	4

print(str(x))