Question 1 @ 2025-04-08 19:23

Why are symbols produced by (gensym) guaranteed to be unique?

---

• A

  (gensym) checks all symbols in the program and produces a “fresh” one in the sense that it is not used in it.

• B

  (gensym) increments a number which is part of the produced symbol every time it is called.

• C

  (gensym) produces uninterned symbols, which ensures they are not eq? to any other symbol (regular one, or a result of another (gensym)).

• D

  They are not guaranteed to be unique: (gensym) produces a highly random symbol name using a cryptographically-robust random number (effectively a sha256 random number) so chances of name collisions are negligibly small.

• E

  The produced symbol is based on machine+user identification, thereby producing a symbol which is, like a UUID, completely unique.

Question 2 @ 2025-04-08 19:25

Say that we run the following code in #lang racket with the define-macro tool that we’ve seen:

What would the result be, and is it showing that the macro is implemented properly?

---

• 1a

  The proper result is 1.

• 1b

  The improper result is 1 (incorrect because both of the orelse subforms are evaluated).

• 2a

  The proper result is 2.

• 2b

  The improper result is 2 (incorrect because the first subform is evaluated twice).

• 3a

  The proper result is 3.

• 3b

  The improper result is 3 (incorrect because the first subform is evaluated twice, and the second once).

• C

  The result is a void value (so, no output at all), since all of the forms, including the last one, are used for their side-effects.

• D

  The result would be incorrect regardless of macro details, because we have a case of identity “aliasing” where two values are the same exact box object.

Question 3 @ 2025-04-08 19:31

Focusing on the macro from the last question:

How could we solve the problem of this macro?

(Important: choose the best answer!)

---

• A

  There is no problem with it, so no solution is needed.

• B

  The problem was with box identity (more precisely, with aliasing + mutation), so changing the macro cannot improve this case.

• C

  Use define-syntax + syntax-rules.

• D

  Like (C), but better: use define-syntax-rule.

• E

  Use a temporary gensym-ed binding.

• F

  Like (E), but better: also use quasiquotes and unquotes to make it look more like a code template.

Question 4 @ 2025-04-08 19:32

What is the biggest advantage of define-macro over define-syntax + syntax-rules?

(Important: choose the best answer!)

---

• A

  It is more hygienic.

• B

  It is less hygienic.

• C

  It avoids name captures.

• D

  It can “inject” known identifiers into macro results

• E

  It is using the host language (Racket) for the macro definition, which means that you can do arbitrary computations at the macro level.

• F

  It can be used to implement a whole compiler instead of just macros.

Question 5 @ 2025-04-08 19:34

Your friend Random J. Hacker heard about macros and wants to use one to create a useful debugging tool:

The idea is to print the expression and its result in the ??? line.

Could this be achieved, and how?

---

• A

  There is no way to get it done, because it will inevitably lead to evaluating the expression twice.

• B

  It’s easily done, eg: (printf "result: ~s\n" result), but there’s not much point in doing that as a macro since the same could be done as a simple function too.

• C

  We could try to do it like so: (printf "~s => ~s\n" expr result) but as we clearly see, this will lead to a double evaluation of the input expression.

• D

  We can use (printf "~s => ~s\n" 'expr result) to print the actual (quoted!) expression as well as what it evaluated to.

• E

  We can use (printf "~s => ~s\n" 'expr 'result) to print the actual expression as well as what it evaluated to, using quotes to avoid undesired evaluations.