dietrich@cernvax.UUCP (Wiegandt) (03/08/85)
We recently got a request from the Central Research Institute for Physics in Budapest, Hungary, for a pointer to any vendor of a C compiler to run under IBM 360/OS and/or IBM 370 VS1 and VM/CMS. I believe that they would not want or would not be able to run UTS or IX/370 under VM on their machine just to get a C compiler. If anyone can help, please send mail to {seismo,decvax}!mcvax!cernvax!dietrich. I shall forward your reply to Hungary. Dietrich Wiegandt CERN-DD-Division CH-1211 GENEVA Switzerland
rhbartels@watcgl.UUCP (Richard Bartels) (03/11/85)
In article <138@cernvax.UUCP> dietrich@cernvax.UUCP (Dietrich WIEGANDT) writes: > >We recently got a request from the >Central Research Institute for Physics in Budapest, Hungary, >for a pointer to any vendor of a C compiler to run under IBM 360/OS >and/or >IBM 370 VS1 and VM/CMS. > Note that the U.S. Govt. restricts the export of technology to certain countries. Has anyone checked whether software systems fall under that restriction? If so, what systems? What countries?
msb@lsuc.UUCP (Mark Brader) (03/12/85)
Dietrich Wiegandt (dietrich@cernvax.UUCP) writes: > >We recently got a request from ... Budapest, Hungary, > >for a pointer to ... a C compiler ... Richard Bartels (rhbartels@watcgl.UUCP) adds: > Note that the U.S. Govt. restricts the export of technology to > certain countries. The interesting thing is that the second posting is from Canada. Mark Brader, Toronto, Canada Serious followups should perhaps go to net.politics.
dave@lsuc.UUCP (David Sherman) (03/12/85)
Sorry, Mark, but U.S. government restrictions do indeed find their way into Canada. Our UNIX source license, for example, contains an explicit undertaking by the Law Society not to export any of the licensed software to certain countries "without the prior written consent of the Office of Export Adminstration of the U.S. Department of Commerce". So while U.S. law has no force in Canada, it causes our contract with a U.S. organization to incorporate some of its requirements. Dave Sherman The Law Society of Upper Canada Toronto -- {utzoo pesnta nrcaero utcs hcr}!lsuc!dave {allegra decvax ihnp4 linus}!utcsri!lsuc!dave
root@bu-cs.UUCP (Barry Shein) (03/15/85)
I know of two C compilers for the IBM/370: 1. A PCC from Bell Labs which we run here at B.U. Call Bell for availability. My suspicion is you would have to have e or essentially be eligible for a UNIX source license (the fact that it is a 370 does *not* exclude this, speak to bell.) 2. A company called RIS (they make a data base system called ORACLE) in California. They sell a C compiler, it isn't cheap but may have advantages that make it worthwhile (support, they use it for their DB sys so it probably is pretty good.) The Bell PCC is 'good', not great. For example, array declarations are limited to 4096 bytes (cuz of base/displacement) although I just malloc() larger things at start-up and it works fine. There are a few small bugs that require similar 'hacks' but nothing intolerable. I remember some problem with returning a float so we had to put defs in <math.h> that went something like: float *_sin(x) ; #define sin(x) (*_sin(x)) no big problem, just a nuisance (more a documentation problem as you explain these workarounds over and over again.) On the other hand, if you like C it really is C, we use it a *lot*! -Barry Shein, Boston University
ron@brl-tgr.ARPA (Ron Natalie <ron>) (03/16/85)
> 2. A company called RIS (they make a data base system > called ORACLE) in California. They sell a C compiler, > it isn't cheap but may have advantages that make it > worthwhile (support, they use it for their DB sys so > it probably is pretty good.) BULLSHIT. The company is called RSI and it sucks. We have compiled a list of over one hundred serious and not so serious bugs. They include not being able to ever reuse a local symbol, the fact that if you declare a variable that has the same name as some assembler symbol (like R12) the compiler does not detect the problem. Performance is terrible as well. The support is a joke. They are not interested in supporting the compiler nor fixing any of these bugs. The compiler implementor was tired of hearing from us. It was never intended that this be used as a product. They wrote it so they could port the database (which is no bargain either) to the IBM. They can live with the idiosyncracies of the compiler long enough to get Oracle compiled, but you can't do any real work with it. It was an entire waste of two thousand dollars. -Ron
root@bu-cs.UUCP (Barry Shein) (03/17/85)
[Reference to my reference to RSI 'C' compiler for IBM/370 and warning that it is lousy] Just to clarify. I have never used it, I was only guessing when I assumed it might be good to look at as they used it. We use the AT&T compiler and are quite happy with it. Thanks to those who used this RSI 'C' and were able to offer more experienced opinions than my own. It sounds like it is a loser...call AT&T. -Barry Shein, Boston University [There was also a summary of data-bases in one of the groups and that seemed to indicate that RSI's ORACLE DB sys is lousy also]
josh@v1.UUCP (Josh Knight) (03/17/85)
> I know of two C compilers for the IBM/370: There are at least two more: (1. There is a C compiler from Waterloo for IBM 370's. We're still in the process of getting it, but I've seen favorable commentary on VMSHARE (an electronic conference about VM/CMS on TYMSHARE connected with the IBM users group, SHARE). Karl Geiger (karl@uscvm.BITNET) provided the following information in <6142@brl-tgr.ARPA> in this group a while ago, the address is: WATCOM Computer Systems Group University of Waterloo 158 University Avenue Ontario, Canada N2L 3E9 It costs $1800/year with a 50% educational discount available. (2. Whitesmiths apparently also sells a 370 C compiler for about $5K. The reviews on this beast have been mixed (I've never seen it). > 1. A PCC from Bell Labs which we run here at B.U. > Call Bell for availability. My suspicion is you would > have to have e or essentially be eligible for a UNIX > source license (the fact that it is a 370 does *not* > exclude this, speak to bell.) I believe the 370 C compiler license is separate from any Unix license. Don't remember the price. > float *_sin(x) ; > #define sin(x) (*_sin(x)) > > no big problem, just a nuisance (more a documentation problem > as you explain these workarounds over and over again.) I would call this a bug in the compiler. I suspect anyone who tried to write scientific/engineering code would as well. The opinions expressed (or implied) are my own, not those of my employer. Josh Knight, IBM T.J. Watson Research josh at YKTVMX on BITNET, josh.yktvmx.ibm on CSnet, ...!philabs!v1!josh