home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!usc!snorkelwacker.mit.edu!ai-lab!zurich.ai.mit.edu!gjr
- From: gjr@zurich.ai.mit.edu (Guillermo J. Rozas)
- Newsgroups: comp.lang.scheme
- Subject: Re: What are good references on learning Scheme
- Date: 23 Nov 92 22:25:46
- Organization: M.I.T. Artificial Intelligence Lab.
- Lines: 33
- Message-ID: <GJR.92Nov23222546@chamarti.ai.mit.edu>
- References: <1992Nov20.001132.28991@pasteur.Berkeley.EDU>
- <1992Nov22.033413.1020@linus.mitre.org>
- <GJR.92Nov23094656@chamarti.ai.mit.edu> <By6MoB.7qH@rice.edu>
- Reply-To: gjr@zurich.ai.mit.edu
- NNTP-Posting-Host: chamartin.ai.mit.edu
- In-reply-to: dorai@cs.rice.edu's message of 23 Nov 92 18:43:23 GMT
-
- In article <By6MoB.7qH@rice.edu> dorai@cs.rice.edu (Dorai Sitaram) writes:
-
- | That said, while I agree that SICP chooses to teach thinking
- | about programs over teaching programming, I don't think the
- | passing over of call/cc in the book can be attributed to
- | that choice. Arguably, if SICP were written today, it would
- | contain call/cc or some variant, without such an act
- | spelling a compromise of any of its lofty goals.
-
- I think that you are wrong. The course has not remained fixed at MIT,
- and each semester different topics have replaced some of the sections
- of the book (which is a polished old snapshot of the course). To my
- knowledge, not _once_ has cwcc been mentioned except possibly as an
- aside. Some of the experiments have introduced continuations (without
- calling them by that name) within the workings of an interpreter, in
- order to implement back-tracking or some such. However, the focus was
- not on the implementation, but on the new language and using it to
- perform speculative computation or whatever. Never has something like
- cwcc been used, rather than implemented directly by modifying an
- interpreter.
-
- Hal and Gerry can speak for themselves, but I am quite confident in
- stating that they think cwcc is an interesting detail for systems
- construction, but it does not add that much to the teaching aspects.
- The more complex systems have typically been based on modifications of
- an interpreter, which can supply its own control mechanisms whether
- cwcc is provided by the system or not.
-
- In other words, CWCC is a tool, and the tool is not nearly as
- interesting as the things that can be built with it. It is even less
- interesting when there are other ways to accomplish it that (in a
- teaching setting) might be more illuminating about what's really going
- on.
-