| gabriel rosenkoetter on Thu, 4 Jul 2002 12:43:18 -0400 |
|
On Thu, Jul 04, 2002 at 11:37:26AM -0400, Noah Silva wrote: > I am confused how people never saw it when TurboPascal is the best > selling compiler of all time... Oh well ;) Best *selling*... I think many of the folks on this mailing list regularly use a compiler that's not sold at all (or, at least, isn't counted as a sale when it's included with everywhere Linux and BSD distribution). But your point still stands. :^> As far as that goes, though, LISP dialects like Scheme and Haskell are clearly better (especially for prototyping) than the "standards", and the old LISP machines would, arguably, have been a better model for real world than Unix, but got shunted aside by the Berkeley and AT&T folks. On Thu, Jul 04, 2002 at 11:40:35AM -0400, Noah Silva wrote: > On Thu, 2002-07-04 at 11:09, Leonard Rosenthol wrote: > > Don't get me started on Objective C - it's just WAY TOO WEIRD > > and needs to be updated to the 1990's (let alone 2000)... > As I understand it, apple is working on this now, as they are sing it > for the development of OS X stuff. C++ was designed to be a temporary > solution, and imo, it shows. Ummm... way to weird for whom, Len? For you? It certainly seems perfectly sane, what I've used of it under NeXTStep. As for use in Mac OS X, I'd love to know that was true, but I somehow doubt it. Objective C just isn't taken seriously. On Thu, Jul 04, 2002 at 12:06:49PM -0400, Noah Silva wrote: > I have seen C people tell me that it's better to do something like: > > a++ > a++ > a++ > a++ > > than to use > > a=a+4; > > because four inc's is faster than an addition. But it isn't on newer > processors. Examples of this are nearly endless. I want some of what those C people were smoking, because it must be pretty potent stuff. This is clearly the compiler's job, not the programmer's. If your compiler's too dumb to do that, go fix it. (Yes, I've written a minimal C compiler, thanks. As has anyone else who deserves their CS degree.) In any case, INC's being faster than ADD is true on almost all processor architectures. It's definitely no more or less true on newer IA32 processors than it was on older ones (building an adder takes flipflops and then somewhere else to store the result, building an accumulator just requires enough bits, and flipping them is pretty fast), it's just that you're talking about a difference in speed that no human is going to notice. If you compare two loops each doing one of those things compiled with optimization forced off, you *might* start to notice a difference around 2^20 passes. But probably not. And, at any rate, a case like this is the *really* easy stuff for a compiler to optimize, and we've known how to do it since the early '70s. > 4.) A lot of fuzzy-logic type name matching libraries to handle things > where someone is in one system as "bob smith" and in another system as > "william Z. SMYTH, III" and we want to link them. > etc. Guh. That stuff hurts. Names are almost always hard. The folks at work do this in Perl. I wouldn't (and, fortunately, don't have to). > I can understand people fighting commercialization to some degree. On > the other hand, debian isn't going to suddenly disappear, and > commercialization means you can get drivers for odd hardware that we > like. Getting back to where that started, XEmacs is, itself, not a bad development environment (including the debugger in the same interface, yadda yadda). For quite a while, I'd been saying that, when I got back to doing a large coding project, I'd probably use XEmacs in vi editing mode. I'm told that gvim has gotten pretty good at doing the same stuff as of version 6, so maybe I'll just use that. (I'll know soon enough, as long as playing with them doesn't take up too much of my time and I throw up my hands and just go back to xterms ;^>.) > Since there has been a reasonable amount of traffic on this, and it > isn't -strictly- linux related, if anyone wants us to take it off list, > I have no problem with it... I still say that computer-related talk, especially about software that does, in fact, work on Linux (or, imho, Unix-like OSes at large) is probably not off-topic. But whatever, nobody died and left me in charge. -- gabriel rosenkoetter gr@eclipsed.net Attachment:
pgpDSjgYkdwv1.pgp
|
|