jkp@cs.HUT.FI (Jyrki Kuoppala) (04/08/91)
In article <966@boing.UUCP>, dale@boing (Dale Luck) writes: >Up until 5.04 there were too many bugs in SAS/C to compiler X source code, >however after 5.04 became available we finally were able to ram the >client source code through SAS/C. >But still Some of these exist in the 5.10a compiler. All we can do as >companies with products on the market that depend on compilers is to help the >companies find the bugs and hope they have timely fixes. SAS has proven to >me that they support their product and that is one of the reasons we recommend >SAS/C for people doing X11 programming on the Amiga. I find it silly that after that much bad experience with a broken compiler you still stick with it, especially concerning that there's an excellent compiler like gcc available with source. I don't know how well SAS fixes bugs, but if the compiler as buggy as you tell and it still takes several revisions to get the bugs fixed, sounds like the FSF makes a lot better job at it. Also, with gcc you have the alternative of fixing the possible bugs yourself; you are not the mercy of the compiler vendor with the only remedy to just 'hope' that they will fix the bugs. Using gcc on the amiga of course might require writing some libraries which are usually provided by the Amiga compiler vendors; some of this has been done (freely distributable, also) - but I'd suspect that doing that wouldn't be as hard as struggling to get perfectly working code thru a broken compiler. //Jyrki
dale@boing.UUCP (Dale Luck) (04/10/91)
In article <1991Apr7.172821.13328@santra.uucp> jkp@cs.HUT.FI (Jyrki Kuoppala) writes: > >I find it silly that after that much bad experience with a broken >compiler you still stick with it, especially concerning that there's >an excellent compiler like gcc available with source. All compilers on the Amiga at that time were seriously deficient. Thats why I used the Greenhills C compiler on my Sun to generate my Amiga X11 code. Also as far as I know, gcc comes with zip for library support. > >I don't know how well SAS fixes bugs, but if the compiler as buggy as >you tell and it still takes several revisions to get the bugs fixed, >sounds like the FSF makes a lot better job at it. >Also, with gcc you >have the alternative of fixing the possible bugs yourself; you are not >the mercy of the compiler vendor with the only remedy to just 'hope' >that they will fix the bugs. We don't wish to be in the business of fixing compiler bugs, that's why we pay for commercial software; so we have someone we can call up on the phone and complain to or send email. Usually there is a way to code around the problem until the bug is actually fixed. SAS has shown an immense improvement in support in the last 2 years. >Using gcc on the amiga of course might require writing some libraries >which are usually provided by the Amiga compiler vendors; some of this >has been done (freely distributable, also) - but I'd suspect that >doing that wouldn't be as hard as struggling to get perfectly working >code thru a broken compiler. a) The compilers are not as broken as they used to be. b) We make use of lot's of functions in the library. Again we are not in the business of writing a C support library when there exists good enough libraries out there. We feel that SAS/c for the Amiga is being support properly by SAS. The Amiga has several unique capabilities that need to be supported directly by the Compiler. Overlays, chip mem, base register addressing, inline shared library calls, parameter passage in registers. etc. We are not aware that gcc is able to compete with SAS/c when it comes to those Amiga capabilities. And the final reason we use lattice and not gcc for client development: Our customers typically do not have gcc as their native development compiler. Most are graphics programmers want to work on graphics problems not compiler problems. It is better for them to buy a commercial compiler so they will get real support. Also the gcc compiler will not run on many of our customer's Amigas because of the ram required. -- Dale Luck GfxBase/Boing, Inc. {uunet!cbmvax|pyramid}!amiga!boing!dale
ben@epmooch.UUCP (Rev. Ben A. Mesander) (04/11/91)
>In article <968@boing.UUCP> dale@boing.UUCP (Dale Luck) writes: >In article <1991Apr7.172821.13328@santra.uucp> jkp@cs.HUT.FI (Jyrki Kuoppala) writes: >> >>I find it silly that after that much bad experience with a broken >>compiler you still stick with it, especially concerning that there's >>an excellent compiler like gcc available with source. > >All compilers on the Amiga at that time were seriously deficient. Thats >why I used the Greenhills C compiler on my Sun to generate my Amiga X11 code. > >Also as far as I know, gcc comes with zip for library support. GCC comes with a decent library. It even has many of the functions that SAS/C is missing. A lot of it is from the PDC and BSD sources. I regularly pull functions out of it to link with the SAS code I develop, because the SAS library is often missing some routines. >We don't wish to be in the business of fixing compiler bugs, that's why >we pay for commercial software; so we have someone we can call up on the >phone and complain to or send email. Usually there is a way to code >around the problem until the bug is actually fixed. SAS has shown an >immense improvement in support in the last 2 years. I don't think that GCC is for everyone, or for every situation, but you can hardly claim that RMS has not supported GCC well. In fact, I've contacted FSF with regards to one of thier other products (BASH), and recieved prompt, courteous, and correct help in a single day. I've written SAS a letter with my bug list and some questions that I could not answer after extensive experimentation and the perusal of the manual. The only reply I ever got was an email letter from an Evil IBM Mainframe (TM) :-) with the single line "Are you Ben Mesander?" as the message. The IBM appeared to be named SAS.COM or some such. I'm not sure the two events are connected (my letter and the mysterious email). This occurred several months ago. Now, why do I use SAS? Well, until there's an Amiga linker that can handle COMMON hunks, GCC's object file generation is a rather cumbersome process. SAS/C also uses less memory. CPR is great, although deficient in FP support. I really REALLY REALLY like CPR!! Whether or not I paid for it has little to do with it. I regularly use GCC on a Sun at work to generate production code. In fact, when I get my new Aviion workstation, I understand Data General will be shipping GCC as the compiler. >Dale Luck GfxBase/Boing, Inc. >{uunet!cbmvax|pyramid}!amiga!boing!dale -- | ben@epmooch.UUCP (Ben Mesander) | "Cash is more important than | | ben%servalan.UUCP@uokmax.ecn.uoknor.edu | your mother." - Al Shugart, | | !chinet!uokmax!servalan!epmooch!ben | CEO, Seagate Technologies |
mwm@pa.dec.com (Mike (My Watch Has Windows) Meyer) (04/12/91)
>In article <968@boing.UUCP> dale@boing.UUCP (Dale Luck) writes: >We don't wish to be in the business of fixing compiler bugs, that's why >we pay for commercial software; so we have someone we can call up on the >phone and complain to or send email. Usually there is a way to code >around the problem until the bug is actually fixed. SAS has shown an >immense improvement in support in the last 2 years. Actually, that's not an excuse for GCC. There's a couple of companies on the SF Peninsula that sell support for GCC. I haven't tried asking them about the Amiga version, but I wouldn't be surprised if they'd do bug fixing. <mike -- It's been a hard day's night, Mike Meyer And I been working like a dog. mwm@pa.dec.com It's been a hard day's night, decwrl!mwm I should be sleeping like a log.