dwatola@NEXTASY2.EECS.WSU.EDU (David Watola) (03/31/91)
J. Michael Ashley writes: >It's bad, because GNU cc is *not* the last word in C compilers. I am >positive that several or even lots of people out there are capable of >writing a C compiler that will smoke gcc, but they're never going to write >it, because they can't compete against a free product. Why should the sure they can. good economic decisions aren't based exclusively on price. they are based on all factors that are important to the consumer--including documentation and support. there will always be some amount of people to these factors are most important, and they will spend extra money for them (even for inferior goods). if a company cannot compete, it is not simply because their competitor is free. it is because the performance/price ratio dictates that the costly product just isn't worth that incremental cost. how MUCH better is the nonfree software? certainly, an expensive functioning compiler is better than an intermittenly buggy free compiler. >unsupported software. Yeah, I know there are businesses that exist to >support GNU software, but that doesn't invalidate my previous point, and >at any rate, they only support GNU software. ah, i know how you feel. just the other day i called microsoft and asked for tech support on lotus 123. the laughter was humiliating... :) >NeXT will never have strong software as long as this deluge of free software >exists. Has a situation similar to this existed in another economic market? next already has strong software. gcc is certainly a bad product to criticize on this point. can anyone name any other compiler that is a) freely distributed, with source code and documentation b) (at least) as bug-free as most other compilers on the market c) capable of handling (and enforcing strict rules for) strict-bsd, ansi, or 'old-style' C? d) easily ported to practically any other system? we run gcc on decstations, hp9000s, and nexts here. i personally got gcc running on our decstations in about 30 minutes--a few minutes to ftp the software and read the instructions, about 30 minutes to compile. e) fast if anyone can think of one, please please please name it. i would certainly like to play around with it for a while. no, i think gcc is pretty much the 'last word' in c compilers. certainly on workstation-priced computers. have you taken at look at the assembly code it produces? tough to beat when the optimizer is on. gcc is state-of-the-art technology. it isn't always true that 'you get what you pay for.' there is plenty of exprensive but crappy software out there. and still more expensive and mediocre software. just look through the catalog that next ships-- i see a couple packages in there that cost over $10k which represent just about the amount of work a graduate student would end up doing on a project for a semester... earning mere peanuts as a university lackey. the folks at fsf idealists, and i applaud their efforts. for those who do feel that they are somehow getting inferior software simply because it is free, i suggest the following: fsf can offer two versions of gcc--one which is distributed free, and one that you CAN'T get unless you pay for it. of course they would have to be different. the one that costs can give a different version number when you invoke 'gcc -v'. gosh, it feels good to ramble on and on... "pardon me, but isn't that pin loose in your grenade?" -#86
usenet@lectroid.sw.stratus.com (Mr. Usenet News Manager) (04/01/91)
In article <9103310821.AA09756@nextasy2.eecs.wsu.edu> dwatola@NEXTASY2.EECS.WSU.EDU (David Watola) writes: > ah, i know how you feel. just the other day i called microsoft and asked for > tech support on lotus 123. the laughter was humiliating... :) But if you call Microsoft and say "I'm trying to use Word for Windows with the new Window's based Lotus, and something isn't working, they will help you. If you ask the GNU folks how to use gcc with someone else's product, you generally get a response on the order of "We don't like propriatary software, so please don't use our products with XXXXXX. Use YYYYY instead." Jim Mann jmann@vineland.pubs.stratus.com Stratus Computer
scott@texnext.gac.edu (Scott Hess) (04/02/91)
In article <12573@pt.cs.cmu.edu> ddj@zardoz.club.cc.cmu.edu (Doug DeJulio) writes: In article <4754@lectroid.sw.stratus.com> usenet@lectroid.sw.stratus.com (Mr. Usenet News Manager) writes: > If you ask the GNU folks how to use gcc with someone else's product, > you generally get a response on the order of "We don't like > propriatary software, so please don't use our products with XXXXXX. > Use YYYYY instead." Sure. Now have your company get a support contract with Cygnus or any of the other entities that provide support for a fee for FSF software, and see what *they* say. Or ask NeXT. After all, it's their compiler (no matter _where_ it came from originally). They've made many modifications to it, including mods to generate Mach-O and the like. There's no reason they can't support it, and they do. Most people, when they can't get something to compile on the NeXT, ask about that on comp.sys.next, or they ask NeXT tech support. Which is as it should be. Later, -- scott hess scott@gac.edu Independent NeXT Developer GAC Undergrad <I still speak for nobody> "Simply press Control-right-Shift while click-dragging the mouse . . ." "I smoke the nose Lucifer . . . Banana, banana."
scott@texnext.gac.edu (Scott Hess) (04/02/91)
In article <9103310821.AA09756@nextasy2.eecs.wsu.edu> dwatola@NEXTASY2.EECS.WSU.EDU (David Watola) writes: J. Michael Ashley writes: >NeXT will never have strong software as long as this deluge of free software >exists. Has a situation similar to this existed in another economic market? next already has strong software. gcc is certainly a bad product to criticize on this point. can anyone name any other compiler that is a) freely distributed, with source code and documentation b) (at least) as bug-free as most other compilers on the market c) capable of handling (and enforcing strict rules for) strict-bsd, ansi, or 'old-style' C? d) easily ported to practically any other system? we run gcc on decstations, hp9000s, and nexts here. i personally got gcc running on our decstations in about 30 minutes--a few minutes to ftp the software and read the instructions, about 30 minutes to compile. e) fast I have to quibble on that last point. gcc is not fast. Well, yes, it is fast compared to most Unix compilers, but I mean fast compared to TurboC running on PCs. I used to have shorter compile times on old PS/2 (8086 at 10Mhz) than on the NeXT. This isn't all that bad (if it were, I'd have already begun work on the faster version :-), but it is true. BTW: If anyone out there wants to write a fast C compiler/development environment for the NeXT, give me a call. I'd help, or at the least give a list of what I want. Compilation speed is what I'm looking for - code space/speed is not such a big deal when gcc can whip that. Alternately, drop me a line so I can see what the interest in the community is . . . and what acceptable pricing for such a beast is. A note to Mr. Ashley - if you want to spend money on software, send me some . . . :-) Seriously, though, I've found that of all the software availiable on the NeXT, the cheap or free stuff seems without fail to have the better interface, and to more closely follow NeXT's UI guidelines, and just overall to be nicer. Then again, I don't really feel the need for spreadsheets or word processing software, so maybe I'm biased. But the blatant blunders I've seen seem to indicate that not only are large companies lacking in competitive ability, but they might also be lacking in gray matter - or at least common sense. Just an opinion. Later, -- scott hess scott@gac.edu Independent NeXT Developer GAC Undergrad <I still speak for nobody> "Simply press Control-right-Shift while click-dragging the mouse . . ." "I smoke the nose Lucifer . . . Banana, banana."
ddj@zardoz.club.cc.cmu.edu (Doug DeJulio) (04/02/91)
In article <4754@lectroid.sw.stratus.com> usenet@lectroid.sw.stratus.com (Mr. Usenet News Manager) writes: > If you ask the GNU folks how to use gcc with someone else's product, > you generally get a response on the order of "We don't like > propriatary software, so please don't use our products with XXXXXX. > Use YYYYY instead." Sure. Now have your company get a support contract with Cygnus or any of the other entities that provide support for a fee for FSF software, and see what *they* say. -- DdJ
melling@cs.psu.edu (Michael D Mellinger) (04/03/91)
In article <SCOTT.91Apr2114111@texnext.gac.edu> scott@texnext.gac.edu (Scott Hess) writes:
I have to quibble on that last point. gcc is not fast. Well, yes, it
is fast compared to most Unix compilers, but I mean fast compared to
TurboC running on PCs. I used to have shorter compile times on old
PS/2 (8086 at 10Mhz) than on the NeXT. This isn't all that bad (if
it were, I'd have already begun work on the faster version :-),
but it is true.
Using Turbo C? How many lines of code are we talking? How well does
Turbo C do once you can't keep everything in memory?
BTW: If anyone out there wants to write a fast C compiler/development
environment for the NeXT, give me a call. I'd help, or at the least
give a list of what I want. Compilation speed is what I'm looking for -
code space/speed is not such a big deal when gcc can whip that.
Why not just hack GCC? Most of the compile time is spent parsing the
source. I'm sure something can be done to speed up compilations.
Perhaps someone could write an optimized recursive decent parser to
replace the YACC routines?
Alternately, drop me a line so I can see what the interest in the
community is . . . and what acceptable pricing for such a beast is.
I would love to see an incremental compiler for the NeXT.
-Mike