Should there be syntactic support for sets and set operations in Groovy? Arguments in favor are:
- They're very useful (see any CS algorithms textbook for examples).
- Syntax and semantics would be unambiguous (if I could figure out how to put curly braces in-line without Confluence thinking they're a macro invocation, I'd give some examples
).
- Java's container classes provide a ready-made implementation.
Arguments against:
- It's mostly enough to focus on Java Collections. Most usage of sets really treats them as generic collections, so just support collections well. (-- jrose)
- Small sets can be specified as lists; this is appropiate for anything you might write as a literal in a program. (Note also that Common Lisp successfully represents sets as lists, for most purposes.) (-- jrose)
- Anything you can do using sets, you can do using a map with don't-care values. But using sets directly is clearer, and eliminates arguments over what do with non-empty values when doing "set union" on dictionaries...
- Why stop at sets? Why not multi-sets, skip lists, ... Because they aren't used nearly as often, just as multi-dimensional arrays aren't used as often as their one-dimensional brethren.
See the Python Sets PEP http://www.python.org/peps/pep-0218.html for more.
- Greg Wilson
Labels