gwyn@smoke.BRL.MIL (Doug Gwyn) (09/28/89)
If somebody knows an e-mail path to Mike Westerfield of ByteWorks, could you relay this to him? Meanwhile it may be useful to other Apple IIGS C programmers: ORCA.C Version 1.0 BUGS: (0) Many of the standard header files define non-reserved identifiers, and some standard headers #include others, which is a no-no. (1) fread( buffer, 16384, 1, stdin ) returned 14 when a 14-character string was typed on the keyboard followed by EOF (Ctrl-@). It should have returned 0. (2) Executable compiled & linked via "Run/Compile to Disk" is not usable from the shell (traps). (3) (void) cast in front of strcpy(...) and other functions flagged as "illegal type cast" by the compiler. (4) Use of block-scope static char array leads to "Unresolved reference" at link time. (5) strtok() fails to skip all of a multiple run of separator characters; it returns an empty string in such cases. (6) "const" type qualifier not supported. (7) A tentative extern data declaration causes a later non-extern initialized definition to produce a "duplicate symbol" compiler diagnostic. (8) (char *)NULL in char * initializer flagged as "type conflict". (9) getenv() and vfprintf() are not supplied. (10) <signal.h> and signal() are not supplied. (11) Compiler dies when processing sizeof array_name, when the array was declared [] and size set by supplied number of initializers. (12) Compiler produces no object code files when source consists solely of an initialized datum.
samt@pro-europa.cts.com (Sam Theis) (09/29/89)
Comment to message from: adm!smoke!gwyn@nyu.edu (Doug Gwyn) Re: your Orca/C bugs. Give ByteWorks a call. Mike is always interested in hearing about the difficulties that his customers are having. There is an update that should be sent to registered owners real soon now. Mike also has a Industry Connection section up on America Online. Sam UUCP: crash!pro-europa!samt ARPA: crash!pro-europa!samt@nosc.mil INET: samt@pro-europa.cts.com
bird@cbnewsd.ATT.COM (j.l.walters) (09/30/89)
From article <11174@smoke.BRL.MIL>, by gwyn@smoke.BRL.MIL (Doug Gwyn): > If somebody knows an e-mail path to Mike Westerfield of ByteWorks, > could you relay this to him?.... All you have to do is print your bugs out and mail them to ByteWorks. You will get a very nice answer to each and every one of you complaints. The address is: Byte Works Inc. 4700 Irving Blvd. NW, Suite 207 Albuquerque, NM 87114 (505) 8998-8183 If you call, you might ask when the September free upgrade is going to be ready and pass it on to the rest of us. -- Joe Walters att!ihlpf!bird IHP 1F-240 (312) 713-5356
gwyn@smoke.BRL.MIL (Doug Gwyn) (09/30/89)
In article <2088@cbnewsd.ATT.COM> bird@cbnewsd.ATT.COM (j.l.walters) writes: >All you have to do is print your bugs out and mail them to ByteWorks. Gee, I would never have thought of that. As you noted, rumor had been spreading about a September upgrade, so speed was of the essence. As I said, I figured that the information might be useful to others anyway. I know Westerfield is accessible via e-mail, since I once exchanged messages with him, but I misplaced his e-address. In fact another respondent (in private e-mail) promised to forward the list and discussed other useful Orca/C matters, so I think my posting accomplished everything that I intended.
lvirden@pro-tcc.UUCP (Larry Virden) (09/30/89)
Network Comment: to #530 by adm!smoke!gwyn@nyu.edu There was someone else - perhaps Jason Blow... (my list of net personalities is missing :-() had a list of problems as well. Perhaps all the problems should be collected and passed along, so that someone could fix them. I know that is what Doug was wanting to do. I checked on Compuserve and there isnt anyone from Byteworks (a VERY small company) who gets on and does online support. Mike or friends get onto Aline regularly - I might try to download the problem list and pass it along to him there. P.S. Look at how long and how many version of gnu c there have been - and then some folks are distressed that the one or two programmers for BW only got 60% or so of the bugs out of Orca/C...
gwyn@smoke.BRL.MIL (Doug Gwyn) (10/01/89)
In article <8909300944.AA10553@trout.nosc.mil> lvirden@pro-tcc.UUCP (Larry Virden) writes: >then some folks are distressed that the one or two programmers for BW only >got 60% or so of the bugs out of Orca/C... I don't recall anyone indicating distress, upset, or anything else of the sort. One person commented that he had enough trouble with Orca/C V1.0 that he found it "unusable" and reverted to APW C, but that can be taken as a simple description of his experience. I know I've had to work around all the problems that I reported; some people would just give up. To me the Orca development environment is sufficiently superior to the other IIGS alternatives that I prefer to develop work-arounds and hope that future releases will fix the problems. "Getting 60% of the bugs out" is a meaningless software metric, by the way.
martyp@pnet02.gryphon.com (Martin Peckham) (10/01/89)
The MAJOR differences between gnu C and ORCA/C are:
1. ORCA/C is reputed to be finished commercial software! As far as I'm
concerned it never should have been released from BETA testing!
2. gnu C is "free" for the downloading... ORCA/C retails for $150.00
Mike Westerfield blew it by releasing it too early. I have found it totally
unusable.
Marty
UUCP: {ames!elroy, <routing site>}!gryphon!pnet02!martyp
INET: martyp@pnet02.gryphon.comlvirden@pro-tcc.UUCP (Larry Virden) (10/02/89)
Network Comment: to #627 by gem.mps.ohio-state.edu!ginosko!uakari.primate.wisc.edu!uflorida!haven!adm!smoke!gwyn@tut.cis.ohio-state.edu
I never particularly indicated that you HAD indicated distress. I did
however recently read around here someone who was giving up on Orca/C to
go back to APW because of all the bugs - thats all.
Larry W. Virden ProLine: pro-tcc!lvirden
674 Falls Place Work: lvirden@cas.bitnet
Reynoldsburg, OH 43068-1614 Aline: LVIRDEN
CIS: 75046,606mmunz@pro-beagle.cts.com (Mark Munz) (10/02/89)
Network Comment: to #11148 by usc!henry.jpl.nasa.gov!elroy.jpl.nasa.gov!gryphon!pnet02!martyp@bloom-beacon.mit.edu > The MAJOR differences between gnu C and ORCA/C are: > > 1. ORCA/C is reputed to be finished commercial software! As far as I'm > concerned it never should have been released from BETA testing! > > 2. gnu C is "free" for the downloading... ORCA/C retails for $150.00 > > Mike Westerfield blew it by releasing it too early. I have found it tota > unusable. > > Marty > Did you expect it to be completely BUG FREE? Come on. I hope not. Any large program will have all kinds of bugs that will never be found in beta testing (no matter how long it is). I'm sure that languages are even more difficult to test. All the different combinations that a language can have. Being a programmer, I've seen this happen (where beta testing doesn't catch all). I say we force all publishers not to put out anything if there's even the slighest chance of a bug in it. That way, there will be no software available for us to buy and use. We can then go about and just play games, since all the real applications will be in beta testing for 10, maybe 20 years.. As for comparing Orca/C and gnu C, you forgot to mention that gnu C isn't available on the IIGS. And in fact, your only other option with C on a IIGS is APW C (now there's a bug free language!). And remember, in the Mac & IBM world, $150 is absolutely nothing. Most programs start at $200 and work their way up. --Mark Munz
gwyn@smoke.BRL.MIL (Doug Gwyn) (10/03/89)
In article <8910020336.AA13294@trout.nosc.mil> mmunz@pro-beagle.cts.com (Mark Munz) writes: > Did you expect it to be completely BUG FREE? Come on. I hope not. Any > large program will have all kinds of bugs that will never be found in > beta testing (no matter how long it is). The fellow's point was probably that it would have been impossible to miss most of the reported bugs with even a minimal attempt to apply the compiler to existing production code. The ones I reported all turned up quite soon during an attempt to port an application that had been carefully designed for portability and during initial straightforward implementation of a program in portable C to simply read in OMF files and display the contents of their segment headers. > I'm sure that languages are even more difficult to test. All the different > combinations that a language can have. Being a programmer, I've seen this > happen (where beta testing doesn't catch all). C isn't particularly hard to test; there are numerous test suites commercially available. ByteWorks may have felt that they couldn't afford any of them. I'm pretty sure that ORCA/C V1.0 couldn't possibly have passed the Plum Hall Standard C validation suite. > As for comparing Orca/C and gnu C, you forgot to mention that gnu C isn't > available on the IIGS. That is a valid rebuttal. Who cares how good and cheap a product is if it doesn't meet our needs? I assume ByteWorks is working on an improved release of ORCA/C. It would be nice to be able to quit worrying about the compiler and get back to worrying about our programs.
jabernathy@pro-houston.cts.com (Joe Abernathy) (10/04/89)
Network Comment: to #847 by gem.mps.ohio-state.edu!uakari.primate.wisc.edu!aplcen!haven!adm!smoke!gwyn@tut.cis.ohio-state.edu > C isn't particularly hard to test; there are numerous test suites > commercially available. ByteWorks may have felt that they couldn't > afford any of them. Byte Works uses a suite of over 800 individual tests on all of their compilers, and on top of that, send out thousands of dollars worth of every new release to beta testers. They go to great lengths to get all of the bugs out. Then why are there still bugs after this process? With any application several thousand lines in length, there are going to be things that get missed. And with a language compiler, this fact of life is doubled or tripled. Particularly in the case of a C compiler, you're going to run full tilt into trouble. C is a very unkind language on its own; add to this the fact that most IIGS programmers don't have a lot of C experience, and the fact that the IIGS Toolbox doubles the amount of complexity with which you're faced. You end up with programs setting memory pointers to left field, mysterious error messages, and a lot of complaints. Another problem is that some people even fail to read the manual. Some people even fail to BUY the manual. The thinking with ORCA/C among many people seems to be that all you need to do is download a copy of the language somewhere, get a C manual at B. Dalton, and go to town. I've been having a lot of fun with ORCA/C. Yes, there are some problems still, but I certainly don't find it to be an unusuable compiler. I've only had problems with two programs out of about three dozen I've compiled with it. (Yes, I was one of the beta testers, but I'm not affiliated with Byte Works in any fashion.) Think back to the original APW C. Or TML Pascal (still). Or TML BASIC (still). Or GS BASIC (still). Or Pascal 1.3. Or ... UUCP: crash!pro-houston!jabernathy ARPA: crash!pro-houston!jabernathy@nosc.mil INET: jabernathy@pro-houston.cts.com
martyp@pnet02.gryphon.com (Martin Peckham) (10/05/89)
mmunz@pro-beagle.cts.com (Mark Munz) writes: >Network Comment: to #11148 by usc!henry.jpl.nasa.gov!elroy.jpl.nasa.gov!gryphon!pnet02!martyp@bloom-beacon.mit.edu > > >> The MAJOR differences between gnu C and ORCA/C are: >> >> 1. ORCA/C is reputed to be finished commercial software! As far as I'm >> concerned it never should have been released from BETA testing! >> >> 2. gnu C is "free" for the downloading... ORCA/C retails for $150.00 >> >> Mike Westerfield blew it by releasing it too early. I have found it tota >> unusable. >> >> Marty >> > > Did you expect it to be completely BUG FREE? Come on. I hope not. Any > large program will have all kinds of bugs that will never be found in > beta testing (no matter how long it is). > > I'm sure that languages are even more difficult to test. All the different > combinations that a language can have. Being a programmer, I've seen this > happen (where beta testing doesn't catch all). > > I say we force all publishers not to put out anything if there's even the > slighest chance of a bug in it. That way, there will be no software available > for us to buy and use. We can then go about and just play games, since all > the real applications will be in beta testing for 10, maybe 20 years.. > > As for comparing Orca/C and gnu C, you forgot to mention that gnu C isn't > available on the IIGS. And in fact, your only other option with C on a IIGS > is APW C (now there's a bug free language!). And remember, in the Mac & IBM > world, $150 is absolutely nothing. Most programs start at $200 and work their > way up. > >--Mark Munz No I don't expect bug free compilers or any other non-trivial program. I make my living writing software and I know from experience. What I expect is "USABLE" software. I'll not waste bandwidth here reiterating the bugs that someone else has already done. Look at the list and see if anyone can write anything usable considering those. Yes there are bugs in APW C and some of them are documented... unfortunately not by Apple... but none of them that I know of generate bad code. ORCA/C generates bad code. The prime example is: char *someptr = NULL; Disassemble that and you will find that the compiler generates code that would be equivalent of: char *someptr; *(long *)someptr = NULL; That transates into, put 4 zero bytes at an arbitrary place in memory!!!! Now, what other nasties are lurking that I haven't found. BTW, I will not find anymore because I will not use it. I will NOT pay to beta test somebody's software. Beta testing my own costs me enough. Marty UUCP: {ames!elroy, <routing site>}!gryphon!pnet02!martyp INET: martyp@pnet02.gryphon.com