Make an RFC for the hole syntax

Holes (unfinished expression to be filled) are natural in programming. Interactive theorem prover users use holes to fill in proofs. HtDP has been using ... x ... to mean a hole where x is available. Etc. Other languages that have holes include (but not limited to): Pyret, which uses ...; Idris, which uses ?id.

The problem with ... x ... in HtDP is that it is not syntactically correct. That is, the program that contains this syntax fragment is not runnable even if the hole will not be evaluated.

I propose that we find a syntax for holes. Here are some design choices:

• The syntax: should it be a compound term (so that we can attach other terms inside it), or atomic, or both?
• If the hole can be a compound term, would stuff inside need to be syntactically correct?
• Where could the hole appear? Only expression positions? What about binding positions?
• What's the semantics of the hole? I think the only answer that makes sense is that it should error when evaluated at run-time, but others might disagree.
• How can DrRacket integrate with this feature?

Note that Racket already has Todo List by @david-christiansen, which is one potential solution to this issue. I hope that if this package is picked as a solution, it will be added to base, to both standardize this functionality and promote it.

Start by reviewing the proposed syntax, placement, runtime semantics, and DrRacket integration questions in this issue. Compare the existing Racket Todo List package as the mentioned potential solution. Done would be a settled RFC specifying the hole syntax, allowed contexts, semantics, and integration direction.