henry@utzoo.uucp (Henry Spencer) (04/09/88)
> Future language standardizations should have more representation by > users, and this should be required by ANSI... How do you propose that they should require this? Forbid standardization without adequate user representation? In practice this would lead to very few standards being written. There is NO LAW against more users getting involved in ANSI standardization work!! The problem is that few of them bother. Without users who actively care and get involved, the situation is not going to get better; with them, the problem will cure itself without changes to the rules. ANSI standards committees are quite explicitly open to anyone who wants to join. If you care so much, why weren't you in X3J11? (I don't see your name in the membership list that came with the second-public-comment draft.) Remember that participation in something like X3J11 takes non-trivial amounts of time and money, particularly time. How many users are going to get assigned to such things as part of their job? Without that, it really is quite difficult to find the time. I am *not* an X3J11 member partly because I can't find the time to do a proper job of it; it's not easy to find the time for the peripheral involvement that I do have. -- "Noalias must go. This is | Henry Spencer @ U of Toronto Zoology non-negotiable." --DMR | {allegra,ihnp4,decvax,utai}!utzoo!henry
purtill@faline.bellcore.com (Mark Purtill) (04/11/88)
In article <> henry@utzoo.UUCP writes: >> Future language standardizations should have more representation by >> users, and this should be required by ANSI... > >How do you propose that they should require this? Forbid standardization >without adequate user representation? In practice this would lead to >very few standards being written. There is NO LAW against more users >getting involved in ANSI standardization work!! The problem is that >few of them bother. One of the reasons that few people bother is that ANSI charges large sums of money for copies of the standard; last I heard, it was on the order of $50. I think it costs even more to actually join X3J11 (altho I'm not sure). If ANSI would allow net distribution, they probably get a lot more response. (Actually, early on someone was sending out free copies of the standard; I've got one, but its on the order of two years old, and I've been told X3J11 will ignore comments not based on the latest standard. It didn't have such stuff as noalias in it anyway...) ^.-.^ Mark Purtill ((")) purtill@math.mit.edu
henry@utzoo.uucp (Henry Spencer) (04/12/88)
> One of the reasons that few people bother is that ANSI charges large > sums of money for copies of the standard; last I heard, it was on the > order of $50. I think it costs even more to actually join X3J11 (altho > I'm not sure)... X3J11 committee membership costs some nominal amount, $100 or something like that, after which things like drafts are free. In fact, being a member means getting vast mounds of paper for free. The problem is that you are supposed to read it all. The actual cost of being a committee member is mostly (a) expenses for attending meetings, and (b) the time spent reading (and thinking about) endless documents. As I recall, you have the right to vote only if you attend the (quarterly) meetings fairly regularly. And you will put your foot in your mouth with some frequency if you don't make some effort to read the endless reams of proposals, comments, and drafts. Nobody who hasn't tried it can possibly imagine what a grind it is to do a careful, thorough, line-by-line reading of the latest draft of a complex technical document, especially when you have already seen 57 earlier drafts and are thoroughly sick of it. People who read comp.lang.c will, however, have some grasp of what a grind it is to have to read, comment on, and shoot down the same old dumb ideas for the 57th time. Standards committees have to do a lot of that too. > If ANSI would allow net distribution, they probably get a lot more response. There is a real and legitimate debate about (non)availability of machine- readable copies of the document, but I would observe that ANSI has some cause for restricting distribution to those who are motivated enough to make an effort to obtain the document. It has never been terribly hard to participate in X3J11 (or whatever) if you really want to. -- "Noalias must go. This is | Henry Spencer @ U of Toronto Zoology non-negotiable." --DMR | {allegra,ihnp4,decvax,utai}!utzoo!henry
lvc@tut.cis.ohio-state.edu (Lawrence V. Cipriani) (04/12/88)
Henry Spencer writes: >> I (Larry Cipriani - that's pronounced sip ree ah knee) write: >> Future language standardizations should have more representation by >> users, and this should be required by ANSI... >How do you propose that they should require this? Forbid standardization >without adequate user representation? Yes. >In practice this would lead to very few standards being written. Maybe so, then again, maybe not. I don't think predictions like this can be made with certainty. If so, I can accept it. >There is NO LAW against more users >getting involved in ANSI standardization work!! The problem is that >few of them bother. As another writer said in response, there are many perfectly good reasons why users don't get involved, cost is a big one, lack of time another. >Without users who actively care and get involved, >the situation is not going to get better; with them, the problem will cure >itself without changes to the rules. So far language standards have resulted in such ugly languages. How is that any better? I'd rather not have portability than use ANSI-C. Actually, as long as noalias was removed I would be content (but not overjoyed) to use ANSI-C. Consider the recent FORTRAN-8X standards, by all accounts I've read the vendors are really screwing up FORTRAN. >ANSI standards committees are quite >explicitly open to anyone who wants to join. Glad to hear it. Part of the problem though is that people don't even know that the standardization of something or other has started. Maybe an alternative to requiring user participation, is that the standardization effort should be advertised in a way that reaches most of the users. At least they will know about it. I bet only 20% of the C users ever heard of ANSI-C. >If you care so much, why weren't you in X3J11? (I don't see your name >in the membership list that came with the second-public-comment draft.) AT&T was already represented, and it would be inappropriate for me to do so independently. -- Larry Cipriani, AT&T Network Systems and Ohio State University Domain: lvc@tut.cis.ohio-state.edu Path: ...!cbosgd!osu-cis!tut.cis.ohio-state.edu!lvc (weird but right)
barmar@think.COM (Barry Margolin) (04/12/88)
In article <10314@tut.cis.ohio-state.edu> lvc@tut.cis.ohio-state.edu (Lawrence V. Cipriani) writes: >As another writer said in response, there are many perfectly good >reasons why users don't get involved, cost is a big one, lack of >time another. Why is this? Why are companies that produce compilers richer than companies that write everything else? If a company has a vested interest in producing portable C programs then it should be worth $100/year and some person-time to make sure that its needs are represented in the standard. If by "users" you mean individuals, rather than companies, I think this effect is by design; one purpose of the membership fee is to prevent complete randoms from joining, making sure that committees are made up of those whose livelihood is significantly impacted by the lack of a standard (in the case of languages, this generally includes vendors of compilers and vendors of programs written in that language). >>ANSI standards committees are quite >>explicitly open to anyone who wants to join. > >Glad to hear it. Part of the problem though is that people don't >even know that the standardization of something or other has started. >Maybe an alternative to requiring user participation, is that the >standardization effort should be advertised in a way that reaches >most of the users. At least they will know about it. Communications of the ACM has a regular column that reports on computer-related standards activities. I believe that IEEE Computer and ComputerWorld may also have similar columns. What more can be done, a mass mailing? > I bet only >20% of the C users ever heard of ANSI-C. I'm sure that most books on C in the last few years mention the standardization. For example, Harbison & Steele mentions it throughout the book. There have been articles in magazines such as Byte and Computer Languages. Unfortunately, personal phone calls to everyone who has purchased Turbo-C are not feasible :-) Barry Margolin Thinking Machines Corp. barmar@think.com uunet!think!barmar
gwyn@brl-smoke.ARPA (Doug Gwyn ) (04/12/88)
In article <10314@tut.cis.ohio-state.edu> lvc@tut.cis.ohio-state.edu (Lawrence V. Cipriani) writes: >As another writer said in response, there are many perfectly good >reasons why users don't get involved, cost is a big one, lack of >time another. Now, hold on. If someone doesn't care enough to get involved, why should any attention be paid to their desires? I've been known to spend my own money to attend meetings and conferences, and I'm not exactly rolling in money myself (and I certainly don't have any spare time!). At the very least, concerned individuals could communicate with the X3J11 committee or its individual members, and many have indeed done so, thereby helping shape the proposed Standard. I don't have much sympathy for those who think X3J11 should have made a special effort to seek out their opinions (except possibly for Dennis Ritchie, who appears to have been kept pretty much apprised of X3J11 developments). >I bet only 20% of the C users ever heard of ANSI-C. Only the 20% who care enough about the language to have opinions worth listening to, I bet. The ANSI C standardization effort has been publicized for years in columns in trade journals and elsewhere. It has hardly been a secret. Do you think X3J11 should have bought space for ads in the comics pages of major newspapers in order to reach the rest of the C programmers?
rns@se-sd.sandiego.NCR.COM (Rick Schubert) (04/13/88)
In article <10314@tut.cis.ohio-state.edu> lvc@tut.cis.ohio-state.edu (Lawrence V. Cipriani) writes: > ... >Actually, as long as noalias was removed I would be content (but not >overjoyed) to use ANSI-C. When and if there is an ANSI C Standard that contains "noalias", feel free to write programs that do not contain the token "noalias". -- Rick Schubert
daveb@laidbak.UUCP (Dave Burton) (04/13/88)
In article <7666@brl-smoke.ARPA> gwyn@brl.arpa (Doug Gwyn (VLD/VMB) <gwyn>) writes: |The ANSI C standardization effort has |been publicized for years in columns in trade journals and elsewhere. |It has hardly been a secret. Do you think X3J11 should have bought |space for ads in the comics pages of major newspapers in order to |reach the rest of the C programmers? That certainly would have been appropriate for noalias, et al. ;-) -- --------------------"Well, it looked good when I wrote it"--------------------- Verbal: Dave Burton Net: ...!ihnp4!laidbak!daveb V-MAIL: (312) 505-9100 x325 USSnail: 1901 N. Naper Blvd. #include <disclaimer.h> Naperville, IL 60540
lvc@tut.cis.ohio-state.edu (Lawrence V. Cipriani) (04/14/88)
In article <1509@se-sd.sandiego.NCR.COM> rns@se-sd.sandiego.NCR.COM (Rick Schubert) writes: >In article <10314@tut.cis.ohio-state.edu> lvc@tut.cis.ohio-state.edu (Lawrence V. Cipriani) writes: >> ... >>Actually, as long as noalias was removed I would be content (but not >>overjoyed) to use ANSI-C. > >When and if there is an ANSI C Standard that contains "noalias", feel >free to write programs that do not contain the token "noalias". Certainly, but I will have to deal with code that I did not write that contains "noalias". After all, most of a programmers work is in maintenance not development. -- Larry Cipriani, AT&T Network Systems and Ohio State University Domain: lvc@tut.cis.ohio-state.edu Path: ...!cbosgd!osu-cis!tut.cis.ohio-state.edu!lvc (weird but right)
jwhitnel@csi.UUCP (Jerry Whitnell) (04/14/88)
In article <1509@se-sd.sandiego.NCR.COM> rns@se-sd.sandiego.NCR.COM (Rick Schubert) writes: >In article <10314@tut.cis.ohio-state.edu> lvc@tut.cis.ohio-state.edu (Lawrence V. Cipriani) writes: >> ... >>Actually, as long as noalias was removed I would be content (but not >>overjoyed) to use ANSI-C. > >When and if there is an ANSI C Standard that contains "noalias", feel >free to write programs that do not contain the token "noalias". Yes, but what about the programmers who write the code that I'll be maintaining? How can I make sure they don't use noalias as well? Shoot them? > >-- Rick Schubert Jerry Whitnell Been through Hell? Communication Solutions, Inc. What did you bring back for me? - A. Brilliant
dsill@NSWC-OAS.arpa (Dave Sill) (04/14/88)
Larry Cipriani writes: > Future language standardizations should have more representation by > users, and this should be required by ANSI... Henry Spencer replies: >How do you propose that they should require this? Forbid standardization >without adequate user representation? Does the phrase "No taxation without representation" ring a bell? Oh that's right, Henry's Canadian. >There is NO LAW against more users getting involved in ANSI >standardization work!! The problem is that few of them bother. It's not that they don't bother. Compiler-marketing companies obviously have more at stake in the standardization than the typical company that uses their compilers. Hence, they are more willing to support an employee on a standards committee. It's not quite as easy to be on a committee as some have suggested. Yes, the membership fee is nominal; and yes, everyone is eligable. But the cost of attending meetings all over the country is more than most individuals can afford. Then there's the time that must be spent, probably 10-20 hours/week if you want to do it right, maybe even more. I'm not ready to be a martyr for an ANSI standard. There are much more worthwhile things one can do. (Support the Free Software Foundation, Amnesty International, WHO, et cetera.) Are comments the only form of input a non-ANSI member has to an ANSI committee/standard? The comments are a good idea, but X3J11 is not bound use them. It seems like a public ballot would be reasonable. Isn't that what IEEE does? ========= The opinions expressed above are mine. "A point comes when enough money has been invested in a certain paradigm that something has to be truly revolutionary to throw it over." -- Bill Joy
rns@se-sd.sandiego.NCR.COM (Rick Schubert) (04/15/88)
In article <10511@tut.cis.ohio-state.edu> lvc@tut.cis.ohio-state.edu (Lawrence V. Cipriani) writes: +In article <1509@se-sd.sandiego.NCR.COM> rns@se-sd.sandiego.NCR.COM (Rick Schubert) writes: ++In article <10314@tut.cis.ohio-state.edu> lvc@tut.cis.ohio-state.edu (Lawrence V. Cipriani) writes: +++ ... +++Actually, as long as noalias was removed I would be content (but not +++overjoyed) to use ANSI-C. ++ ++When and if there is an ANSI C Standard that contains "noalias", feel ++free to write programs that do not contain the token "noalias". + +Certainly, but I will have to deal with code that I did not +write that contains "noalias". After all, most of a programmers +work is in maintenance not development. I was responding to your claim that you would use ANSI C if it didn't contain "noalias" but you would not use it if it did; I did and do maintain that if the presence of "noalias" was the determining factor for you, that you could program in ANSI C--, which would be ANSI C - "noalias". I think that dealing with other programmers' code containing "noalias" confuses the issue. When you say that you will refuse to use ANSI C if it contains "noalias", what do you plan to do? Use existing C compilers? or use another language? That wouldn't address the issue of what you do with other people's ANSI C programs. Do you avoid them altogether? I guess so, if you do not plan on using ANSI C. But if you're going to avoid them altogether, you still have an independent choice to make for your own programs. And for this choice I say: "feel free to write programs that do not contain the token 'noalias'." +Larry Cipriani, AT&T Network Systems and Ohio State University Rick Schubert (rns@se-sd.sandiego.NCR.COM)
gwyn@brl-smoke.ARPA (Doug Gwyn ) (04/15/88)
In article <12960@brl-adm.ARPA> dsill@NSWC-OAS.arpa (Dave Sill) writes: >Are comments the only form of input a non-ANSI member has to an ANSI >committee/standard? The comments are a good idea, but X3J11 is not >bound use them. It seems like a public ballot would be reasonable. >Isn't that what IEEE does? Restricting balloting to actively participating members is a good idea for the same reasons as those that led our (USA) founding fathers to set up a representative republic instead of a democracy. (Let's not discuss why the original plan was subverted in this newsgroup, please!)
friedl@vsi.UUCP (Stephen J. Friedl) (04/15/88)
In article <1510@se-sd.sandiego.NCR.COM>, rns@se-sd.sandiego.NCR.COM (Rick Schubert) writes: > When and if there is an ANSI C Standard that contains "noalias", feel > free to write programs that do not contain the token "noalias". I am certainly not a |noalias| wizard, but it strikes me that Rick might be premature here; those who understand |noalias| are well encouraged to correct me. While I may be free to not use |noalias| myself, I am required to interface with the library, most of whose routines have arguments declared with some permutation of |const| and/or |noalias|. Because of this I think I think I have to have some nominal awareness of what these keywords do; the violent flamage against |noalias| leads me to believe that just pretending they are not there will get me into trouble. Anybody? -- Steve Friedl V-Systems, Inc. "Yes, I'm jeff@unh's brother" friedl@vsi.com {backbones}!vsi.com!friedl attmail!vsi!friedl
throopw@xyzzy.UUCP (Wayne A. Throop) (04/15/88)
> dsill@NSWC-OAS.arpa (Dave Sill) >> (Henry Spencer) >>There is NO LAW against more users getting involved in ANSI >>standardization work!! The problem is that few of them bother. > It's not that they don't bother. Compiler-marketing companies > obviously have more at stake in the standardization than the typical > company that uses their compilers. Hence, they are more willing to > support an employee on a standards committee. Dave's position doesn't make sense to me. Don't companies that *use* those compilers have a stake in the future portability of their code, and thus have a very convincing motive to support employees on the standards comittee? The psychology of such consumer-oriented participation (or rather: lack thereof) seems to me to be based on the "We'll live with whatever the comittee comes up with." fallacy. This fallacy seems to fit neatly into the category of "not bothering". (And by the way, many compiler-vendor representatives have much more reason to be conservative about feeping creaturism than do compiler users. After all, they have to spend money to develop the feeping creatures that folks come up with.) -- If the argument to .TH contains any blanks and is not enclosed by double quotes, there will be dird-dropping-like things on the output. --- Unix User's Manual, MAN(7) entry, BUGS -- Wayne Throop <the-known-world>!mcnc!rti!xyzzy!throopw
mike@arizona.edu (Mike Coffin) (04/15/88)
From article <1510@se-sd.sandiego.NCR.COM>, by rns@se-sd.sandiego.NCR.COM (Rick Schubert): > ... And for > this choice I say: "feel free to write programs that do not contain the > token 'noalias'." > > Rick Schubert (rns@se-sd.sandiego.NCR.COM) Two points: First, you can't ignore noalias --- the interfaces to the standard library functions are liberally sprinkled with with "noalias" and "const noalias". Unless you want to rewrite <stdio>, you had better understand noalias. Second, the reason that C is a beautiful example of language design is that the ratio of power to complexity is so high. Noalias adds nothing to the expressive power of the language, while hugely increasing its complexity. -- Mike Coffin mike@arizona.edu Univ. of Ariz. Dept. of Comp. Sci. {allegra,cmcl2,ihnp4}!arizona!mike Tucson, AZ 85721 (602)621-4252
meissner@xyzzy.UUCP (Michael Meissner) (04/16/88)
In article <12960@brl-adm.ARPA> dsill@NSWC-OAS.arpa (Dave Sill) writes: | Are comments the only form of input a non-ANSI member has to an ANSI | committee/standard? The comments are a good idea, but X3J11 is not | bound use them. It seems like a public ballot would be reasonable. | Isn't that what IEEE does? ANSI makes the rules, the individual technical commitee (X3J11 in this case) doesn't. The rules go something as follows: 1) A technical committee is formed for a specific purpose 2) The technical committee works until it has a draft for public review. 3) Said draft is sent out for public review, and published by ANSI. The first review period is ~4 months, and additional review periods are then 2 months. 4) Anybody interested writes their comments and sends them to the committee. 5) All letters received must be answered. The answer either yes or no, but it must be answered. If in doing so, any substantive changes are made to the document, go back to step 2 (editorial changes such as spelling mistakes, etc. don't count). Additionally there is a review by the parent committee (X3). I'm not sure whether this is in parallel with the public review, or after the public review. To be on the X3 review, the cost is several thousand dollars, and you have to promise to review all documents within two weeks of getting it (and you have to review ALL ANSI X3 documents, on things like tape formats, character sets, etc.). Note that for a technical committee and X3, each organization (ie, a company, government agency, etc.) can only have one voting member at any one time, as well as an alternate, who can only vote officially when the prinicipal member is not present. The highest level of review is the ISO level, where each country gets one vote (it's standardization body). C is going through the standardization for ANSI (U.S.A standard) and ISO at the same time, which is fairly common these days. IEEE is a different standards body, and has quite different rules. The public balloting procedures are different, as well as the assumption that each person votes as an individual, not as an organization representative. Also, IEEE tends to stress reaching concensus, rather than 2/3 votes like ANSI does. (I'm not as familar with IEEE voting rules, as with ANSI, I do know the Pascal standard was delayed by at least a year because it tried to be both an ANSI and an IEEE committee at the same time). Each system has it's pluses and minuses, but like anything else, once you are in the system, you pretty much have to abide by the system's rules. -- Michael Meissner, Data General. Uucp: ...!mcnc!rti!xyzzy!meissner Arpa/Csnet: meissner@dg-rtp.DG.COM
gwyn@brl-smoke.ARPA (Doug Gwyn ) (04/16/88)
In article <543@vsi.UUCP> friedl@vsi.UUCP (Stephen J. Friedl) writes: >While I may be free to not use |noalias| myself, I am required >to interface with the library, most of whose routines have arguments >declared with some permutation of |const| and/or |noalias|. Because >of this I think I think I have to have some nominal awareness of >what these keywords do; the violent flamage against |noalias| leads >me to believe that just pretending they are not there will get me >into trouble. As the January 1988 draft was worded, I believe you're correct. The intent was that "normal" programmers could ignore the type qualifiers in the library interface specs and continue to use the functions as they're accustomed to. This is something we'll have to fix next week.
lvc@tut.cis.ohio-state.edu (Lawrence V. Cipriani) (04/17/88)
>> = Larry Cipriani > = Doug Gwyn >>As another writer said in response, there are many perfectly good >>reasons why users don't get involved, cost is a big one, lack of >>time another. >Now, hold on. If someone doesn't care enough to get involved, why >should any attention be paid to their desires? Now, hold on. Don't put words in my keyboard, I'm only citing reasons why they don't get involved. I recognize that their desires probably won't get any attention unless they get involved. If they should be considered or not is another matter. If they aren't considered, and X3J11 has done what it thinks is a fine job, but the users hate it what will happen? ANSI-C will fail. I don't think X3J11 wants that, neither do I. And why does the converse necessarily hold. If someone cares enough to get involved why should their desires get any attention. What if their ideas are so bizarre that they should not participate. How does X3J11 tell someone to bug off? Who decides what is bizarre? Me! Really, I don't know. Maybe X3J11 should just close up shop and let Dennis Ritchie take over the job if he wants it. >The ANSI C standardization effort has >been publicized for years in columns in trade journals and elsewhere. >It has hardly been a secret. Do you think X3J11 should have bought >space for ads in the comics pages of major newspapers in order to >reach the rest of the C programmers? Well the first thing I read in the newspaper are the comics! :-) The world of C programmers is not made up only of computer jocks. Not all the excellent C programmers are programmers by profession. Many of them are scientists in other fields, they read their own journals and magazines not computing/computer trade journals. I do think X3J11 should have made a special effort to publicize at the start, not now though. Don't assume that the people that aren't involved don't have worthwhile opinions, and don't assume that the people that are involved do have worthwhile opinions. It is the latter group that concerns me the most. By standardizing C I expect that C will be standardized, I don't expect that a new language will be invented which is what is happening. At this point ANSI-C is so different than the C I know that I wouldn't call it C, maybe D, but not C. -- Larry Cipriani, AT&T Network Systems and Ohio State University Domain: lvc@tut.cis.ohio-state.edu Path: ...!cbosgd!osu-cis!tut.cis.ohio-state.edu!lvc (weird but right)
dsill@NSWC-OAS.arpa (Dave Sill) (04/19/88)
> Wayne Throop >> dsill@NSWC-OAS.arpa (Dave Sill) >>> (Henry Spencer) >>>There is NO LAW against more users getting involved in ANSI >>>standardization work!! The problem is that few of them bother. >> It's not that they don't bother. Compiler-marketing companies >> obviously have more at stake in the standardization than the typical >> company that uses their compilers. Hence, they are more willing to >> support an employee on a standards committee. > >Dave's position doesn't make sense to me. Don't companies that *use* >those compilers have a stake in the future portability of their code, >and thus have a very convincing motive to support employees on the >standards comittee? Yes, of course they have a motive. They just don't have as strong a motive. Let's look at the vendor side. The success of, say, Lattice's C compiler probably has a direct influence on every employee of the company. Even Microsoft, with hundreds of other products, would feel the pinch if their C compiler sales dropped significantly. The PC compiler market is very competitive; and the vendors have much at stake and much to gain by being able to claim that their compiler is the best. They want things like `volatile' and `noalias' so they can write mega-optimizing compilers to enter in The Big Benchmark Contest. Now let's look at the user side. The typical company (excluding C compiler vendors, of course) has a much smaller stake in the compiler wars. They aren't going to have to lay off 50% of their employees because they can't buy a C compiler that handles `volatile'. Even the `support the standard for portability reasons' argument doesn't hold water: a standard will be created whether Company X supports a committee member or not. >The psychology of such consumer-oriented participation (or rather: lack >thereof) seems to me to be based on the "We'll live with whatever the >comittee comes up with." fallacy. This fallacy seems to fit neatly into >the category of "not bothering". To an extent, this is true. As I pointed out above, they just don't have enough reason to support a committee position. >(And by the way, many compiler-vendor representatives have much more > reason to be conservative about feeping creaturism than do compiler > users. After all, they have to spend money to develop the feeping > creatures that folks come up with.) Yeah, right. Just like automakers curse air-conditioning, FM radios, power steering, et cetera. A product is the sum of its features. ========= The opinions expressed above are mine. "Faith is believing what you know ain't true." -- Anonymous
franka@mmintl.UUCP (Frank Adams) (04/19/88)
In article <7666@brl-smoke.ARPA> gwyn@brl.arpa (Doug Gwyn (VLD/VMB) <gwyn>) writes: >Now, hold on. If someone doesn't care enough to get involved, why >should any attention be paid to their desires? The following is an oversimplification, but I think it is basically accurate. There are perhaps 100 organizations involved in writing C compilers. Each of these has considerable interest in the standard -- let us say 100 units worth. This gives us a total of 10,000 units of interest by compiler development organizations. There are perhaps 10,000 organizations using C compilers. Each of these has much less interest in the standard than the individual compiler writers -- let us say 10 units worth. This still adds up to a total of 100,000 units of interest by users -- ten times as much as the compiler folks. But, it costs the equivalent of maybe 20 units of interest for an organization to involve itself in the process. So the applications developers don't bother, and the process is dominated by the compiler writers. A further complication is that many of the larger applications development organizations, who might indeed find it worthwhile to send a representative from the applications side, are also compiler vendors; and the rules only allow one representative per organization. I don't have any answers here. But there is a very real problem. -- Frank Adams ihnp4!philabs!pwa-b!mmintl!franka Ashton-Tate 52 Oakland Ave North E. Hartford, CT 06108
throopw@xyzzy.UUCP (Wayne A. Throop) (04/20/88)
> dsill@NSWC-OAS.arpa (Dave Sill) >> throopw@dg-rtp.UUCP (Wayne Throop) >>Don't companies that *use* >>those compilers have a stake in the future portability of their code, >>and thus have a very convincing motive to support employees on the >>standards comittee? > Yes, of course they have a motive. They just don't have as strong a > motive. Well, I agree that many people perceive this motive as "weak", but I am convinced that they are mistaken. > Even the > `support the standard for portability reasons' argument doesn't hold > water: a standard will be created whether Company X supports a > committee member or not. Quite right. But, the standard may not represent anything that Company X can use. For example, it may invalidate too much code that X has already written in K&R C. Or it may mandate a lanaguage too large to be implemented on the equipment that X uses. Or it may mandate BCD arithmetic that X's compiler vendors can only supply in criplingly slow form. Or it may incorporate features (dare I say "noalias"?) (I dare, I dare) which make it harder to understand the standard library interfaces and hence make writing standard-conforming code far too difficult. All of these things are, I submit, potentially threatening to the profitibility and even survivability of Company X. The fact that X's CEO doesn't see it that way simply means to me that X's CEO is wrong. >>(And by the way, many compiler-vendor representatives have much more >> reason to be conservative about feeping creaturism than do compiler >> users. After all, they have to spend money to develop the feeping >> creatures that folks come up with.) > Yeah, right. Just like automakers curse air-conditioning, FM radios, > power steering, et cetera. A product is the sum of its features. But even auto vendors argue against expensive features that they think (for whatever reason) that customers don't want to buy. Air bags, for example. Emissions controls. 5-mph bumpers. These features exist (when they do) by the demand of users, not vendors. And the fact that US automakers can compete at all against import vendors primarily because they can offer to leave features OFF to save money or customize. But to clarify: I wasn't trying to say vendors will always take the KISS side of things. Just that they sometimes do, and often have motive to. As I understand it, "noalias" in particular was not proposed by vendors, but by users who wanted a feeping creature. I may be wrong about that. -- And you may ask yourself "Am I right? ... Am I wrong?" And you may say to yourself "MY GOD! ... WHAT HAVE I DONE?" --- "Once in a Lifetime", Talking Heads -- Wayne Throop <the-known-world>!mcnc!rti!xyzzy!throopw
gwyn@brl-smoke.ARPA (Doug Gwyn ) (04/24/88)
In article <10811@tut.cis.ohio-state.edu> lvc@tut.cis.ohio-state.edu (Lawrence V. Cipriani) writes: >At this point ANSI-C is so different than the C I know ... How so? It includes more than you are accustomed to, undoubtedly (e.g. prototypes and type qualifiers), but it is substantially the same language. Most existing more-or-less portable C code should continue to work under an ANSI C implementation with no change. I expect that C compiler vendors will make every effort to supply the same non-ANSI C extensions that their customers are already using; for example, although <stdarg.h> is the new form for variadic argument handling macros, most current implementations of <varargs.h> will continue to be provided as extensions to the ANSI C environment. Experiments wherein existing UNIX PCC-developed code has been recompiled in an ANSI C environment have shown that most of it works no worse than before. Sounds like the same language to me.