From ... From: Erik Naggum Subject: Re: CLOS: read only slots? Date: 1999/09/11 Message-ID: <3146034386741697@naggum.no>#1/1 X-Deja-AN: 523723789 References: <87ogfe3kdx.fsf@pdm.pvt.net><87k8q2t3qe.fsf@piracy.red-bean.com> <37d6e2ea$0$232@newsreader.alink.net> <3145859967030620@naggum.no> <37d989d8$0$229@newsreader.alink.net> mail-copies-to: never X-Complaints-To: usenet@news.eunet.no X-Trace: oslo-nntp.eunet.no 937045614 20 193.71.66.49 (11 Sep 1999 10:26:54 GMT) Organization: Naggum Software; +47 8800 8879; +1 510 435 8604; http://www.naggum.no NNTP-Posting-Date: 11 Sep 1999 10:26:54 GMT Newsgroups: comp.lang.lisp * Harley Davis | (And I think the only successful strategy these days to promote new | languages is large-scale, loss-leading corporate sponsorship, similar to | Sun's with Java and Microsoft's with C++ and Basic, where first-degree | money-losing promotion of a language is intended to have secondary | benefits for the corporation via platform application availability. So | the next chance for Lisp will be the next generation popular software | platform, probably in another 5 years. Better have something ready to | dust off by then, guys!) two immediate thoughts: (1) is the money spent on languages this way actually recovered, or are they marketing vehicles and thus merely marketing costs for something else? (2) does unlimited source access help recover the money spent or just help promote the language (and depending on question 1, helpful to that something else)? | C is pretty much the epitome of simple, predictable efficiency models. : | Simplicity isn't important to you but it is important to beginning and | intermediate users, I would wager. OK, so if we set the baseline for both simple and predictable at C, I would say that we don't need it _that_ simple. for instance, there are no basic operationgs in C that are worse than O(1). in Common Lisp, = is O(n), but /= is O(n²). in C, (lo <= x && x <= hi) requires computing x twice if it isn't known to be unchanging, and it's fast, because the types are known, but in CL, it may surprise people used to the C mindset that (<= lo x hi) is usually a function call away, but (TYPEP x '(INTEGER lo hi)) is fast and inlined (when lo and hi are constants, obviously). I think the latter is simple, too, but it is clearly predicated on having groked the dynamic typing concept. does this make it less simple, or only simple in non-C terms? let's make it as simple as possible, but no simpler. (incidentally, is this a Feynmanism, or was it Einstein?) #:Erik -- it's election time in Norway. explains everything, doesn't it?