Re:I Just Can't Kick Lisp
by Test0129 on Sunday October 02, 2022 @02:30PM
(#33058773)
SICP is a good way to learn about programming from first principles. It is not the best way to learn CL. CL and scheme are extremely different despite having similar roots.
[ Parent
| on HN ]
It's not even the best book to learn actual Scheme. SICP's main purpose is to teach computer science basics using a tiny subset of Scheme (no macros, no error handling, no libraries, ...).
Re:I Just Can't Kick Lisp
by throwaway81523 on Sunday October 02, 2022 @03:02PM
(#33059098)
Scheme didn't have macros when SICP was first written, iirc.
Scheme already had macros in 1975 [1]
SICP was first published in 1984 [2]
[1] https://stackoverflow.com/a/69261645/3414663
[2] https://en.wikipedia.org/wiki/Structure_and_Interpretation_o...
Re:I Just Can't Kick Lisp
by throwaway81523 on Monday October 03, 2022 @03:14AM
(#33064408)
That says (some) implementations had macros from the beginning, though those macros were much different than what was eventually standardized in (I think) r5rs. I guess I can believe that. I don't know when Guile got macros.
Philip Wadler's paper discussed and linked from here is also interesting: https://www.wisdomandwonder.com/link/1055/why-calculating-is...
SICP was largely written before Scheme was standardized and before SRFI existed.
For example SICP uses a so-called special form called cons-stream, which is in no Scheme definition (it was in no Scheme reports) and for which no implementation is given in the book, since it can't be implemented by the Scheme subset used for the book. It assumes that it is somehow provided without giving any details.
Re:I Just Can't Kick Lisp
by throwaway81523 on Monday October 03, 2022 @04:17PM
(#33073070)
Thanks, yeah ok, it hadn't occurred to me that cons-stream has to be a macro. I also remember the book also explaining how force and delay work in terms of wrapping and later evaluating a lambda, but don't remember if it explained that delay has to be a macro too. I guess the Common Lisp approach to macro hygiene is using gensyms for temporaries, but Schemers saw that as too ugly. I wonder how Clojure does it. I'm not that big on Scheme or Lisp these days anyway though.