[comp.lang.c] c compiler for ibm

jru@etn-rad.UUCP (John Unekis) (09/15/87)

Does anyone out there know of a C compiler for the IBM 43xx series
running OS/MVS . 
Who sells it?
How much?
Does it support CICS, or TSO, or anything else?

Any information would be greatly appreciated.

ihnp4!wlbr!etn-rad!jru

sullivan@marge.math.binghamton.edu (fred sullivan) (09/15/87)

In article <280@etn-rad.UUCP> jru@etn-rad.UUCP (0000-John Unekis) writes:
>Does anyone out there know of a C compiler for the IBM 43xx series
>running OS/MVS . 
There is a Waterloo C which runs under VM.  Probably there is an MVS
version.
>Who sells it?
University of Waterloo
>How much?
Don't know.
>Does it support CICS, or TSO, or anything else?
Don't know.

Fred Sullivan
Department of Mathematical Sciences
State University of New York at Binghamton
Binghamton, New York  13903
Email: sullivan@marge.math.binghamton.edu

KJH%APLVM.BITNET@wiscvm.wisc.EDU (Kenneth J. Heeres) (09/16/87)

We are using the Lattice C compiler as modified by SAS Institute.  It is availa
ble for both OS/MVS and VM/CMS.

daves@brspyr1.UUCP (09/16/87)

in article <280@etn-rad.UUCP>, jru@etn-rad.UUCP (John Unekis) says:
> 
> Does anyone out there know of a C compiler for the IBM 43xx series
> running OS/MVS . 

There are several C compilers out in the mainframe world.  

The three that I know of are:
	1) Whitesmith's / IBM C 
		(IBM licenses and sells it, Whitesmith develops it, and
		 sells it as well.)
	2) Waterloo C
		(I don't know much about it, except that it's low cost, and
		has quick compilation; used by several universities)
	3) SAS / Lattice C
		(This is the package we use for our product development on
		VM/CMS and MVS/TSO.  A very good package; very robust runtime
		library, similar to UNIX's.)

We've used both the SAS and Whitesmith's compiler in the past, but now we
only use SAS.  By the way, none of the compilers have a CICS interface that
I know of, but with SAS, you can write assembler "glue" routines to get to
the CICS routines, if you want. I think that there are some SAS C users out
there who are doing that.

-- 
|Dave Schubmehl      | UUCP Map:   daves@brspyr1.UUCP                       |
|BRS                 | UUCP Path:  seismo!rochester!steinmetz!brspyr1!daves |
|Latham, NY   12110  |        or:  ihnp4!dartvax!brspyr1!daves              |
|(518) 783-1161      --------------------------------------------------------

ron@topaz.rutgers.edu (Ron Natalie) (09/17/87)

The Lattice C compiler is notoriously non-standard.  There are C
compiler for both VM/CMS and MVS from IBM that we have acquired.
I haven't had a chance to look at it yet.

-Ron

STROBL%DBNGMD21.BITNET@wiscvm.wis (Wolfgang Strobl 49+228303223) (09/17/87)

> Does anyone out there know of a C compiler for the IBM 43xx series
> running OS/MVS .
We tested the Lattice C as modified by SAS Institute (Version 2.10C).
We have Waterloo C (Version 1.3) and the IBM C (Release
1.0).  All three compilers look like ports from VM/CMS to MVS.

> Who sells it?
Lattice C:  SAS Institute Inc. Box 8000, Cary, NC 27511-8000
            or SAS Institute GmbH, Rohrbarcher Str. 22,
                                   D-6900 Heidelberg 1, Germany
Waterloo C: WATCOM Products Inc,   415 Phillip St., Waterloo, Ontario,
                                   Canada N2L 3X2
IBM C:      IBM

> How much?
Lattice C: If I remember correctly, 12.000 DM for the first year and
8.000 DM for the following years. This prices are nearly two years old.
At that time the Lattice C was the only working C Compiler for MVS. I
don't know the actual price.  Waterloo C: 1800 $ per year (commercial)
IBM C: 5000 $ One-time Charge, 313 $ monthly.

> Does it support CICS, or TSO, or anything else?
CICS: don't know. TSO: all three compilers produce load modules
which can be run under TSO - but that does not say very much, there
is nearly no special support for TSO.

>
> Any information would be greatly appreciated.
>
The MVS interface of all three compilers (of the versions we had or
have) is lousy. For example: both Lattice C and Waterloo C can fopen
preallocated files only, via a DDNAME. The Lattice C manual mentioned
that as to be fixed in a later version.  The IBM C's fopen creates
unblocked files by default and the compiler uses unblocked intermediate
workfiles. One of my programs which compiles in 4 minutes on a AT-03
using MSC 4.0 takes 35 minutes execution time on our 3090-150 if the
machine is empty.  If you write unblocked files to 3380 disks, this may
use up to 30 times the space it takes with blocked files.  The rename
function is not implemented in all three C's.

The following list of critera tries to give you an impression of my
experience whith the compilers. If I put a compiler in 3. place that
does not mean it's bad. It only says the other two look better to me.

                              1.Place      2.Place            3.Place
---------------------------------------------------------------------

Language                      IBM          Waterloo           Lattice

   Example: IBM translates ANSI C, Lattice C has no enum and no
   structure assignement.

Installation:                 IBM          Waterloo           Lattice

Speed:                        Waterloo     Lattice            IBM

   All three compilers are  s l o w ! If one had used Dsmet C on a fast
   AT before and moves to a 10 MIPS 3090, one get's really surprised.
   Turnaround times go up, not down.

Library
 (# of functions)             Lattice      Waterloo           IBM
 (design of stdio port)       Lattice      IBM                Waterloo
 (quality of implementation)  IBM          Waterloo           Lattice

Let me explain what I mean with "design of stdio port" by giving an
example. With Waterloo C you must specify the attributes of a data set
in the fopen call, if they do not match the default attributes, which
resemble card image with a fixed record length of 80. Otherwise one gets
an open error.  With Lattice C and IBM C you can specify attributes, but
one is not forced to do that. Which "quality of implementation" I mean
the # and severity of bugs or omissions I found.

>
> ihnp4!wlbr!etn-rad!jru

Wolfgang Strobl, GMD Z1.BN, STROBL@DBNGMD21.BITNET

PS:  I want to get in contact with other people which have
     to maintain and use a C Compiler under MVS, in order to
     exchange bug reports, programs, horror stories :-) ...

PPS: the above stated opinions are my own, but I am not allowed to sell
     them.