dmj@cunixa.cc.columbia.edu (Douglas M Jaffe) (05/25/91)
i'm trying to evaluate an rs/6000 530 running aix 3.1.5
(at least that's what our ibm rep has been saying). i would
appreciate any comments on its pros/cons, especially on
how well suited it is for use as a compute server.
also, the ibm reps are under the impression that the machine
is near-perfect. either (1) they're high on drugs, or (2) their
standards are low (probably _both_).
if anyone knows of definite bugs concerning 3.1.5, please email
me at the above address (dmj@cunixa.cc.columbia.edu). i'll
compile the list and post it to this newsgroup.
so far, i know of one bug relating to the development system:
BUG: cc or xlc has problems casting a modifiable
lvalue to an lvalue of a different type. i checked
the official ansi reference (ie. not k&r 2ed), and
it specifically allows cast expressions that result
in modifiable lvalues.
below is a session log:
======================================================================
$
$ cat cast.c
#include <stdio.h>
main()
{
char *c;
int buf[100];
((int *)c) = &(buf[0]);
exit(0);
}
$
$ make cast
cc -O cast.c -o cast
8 | ((int *)c) = &(buf[0]);
.........a......................
a - 1506-025: (S) Operand must be a modifiable lvalue.
1254-004 The error code from the last failed command is 1.
Make Quitting.
$
======================================================================
Please note:
1. i'm using the expression "((int *)c) = &(buf[0])"
to specifically show that there are no ambiguities
in the expression.
2. "c = (char *) buf" is better for _this_ toy case, but
for _my_ purposes (too complicated to mention here), i
need to be able to write macros that can cast modifiable
lvalues. for those interested, email me.
as a final word, please email me any bugs that you have encountered
using aix 3.1, (esp. 3.1.5). i will compile and verify the list, send
copies to those who emailed to me, and post it to this newsgroup.
thanks in advance.
----------------------------------------------------------------------
_
,---------'_)
(_/\ O O /
\ v /
\___/
U
"doggie fresh"
----------------------------------------------------------------------
_
,---------'_)
(_/\ O O /jfh@rpp386.cactus.org (John F Haugh II) (05/26/91)
In article <1991May25.005319.29296@cunixf.cc.columbia.edu> dmj@cunixa.cc.columbia.edu (Douglas M Jaffe) writes: > cc -O cast.c -o cast > 8 | ((int *)c) = &(buf[0]); > .........a...................... >a - 1506-025: (S) Operand must be a modifiable lvalue. >1254-004 The error code from the last failed command is 1. The expression ((int *) c) is not a modifiable l-value. It is the value of the variable "c" treated as a pointer to an integer. If you want to modified what "c" points to, insert a "*" before it. That is, *((int *) c) = &(buf[0]); The result of a cast is not an l-value. If you want "c" to contain the value of the address of the 0-th element, you must do something completely different, like the c = (char *) buf; example that you gave, or possibly *((int *) &c) = &buf[0]; which will do something similiar (only different ;-) -- John F. Haugh II | Distribution to | UUCP: ...!cs.utexas.edu!rpp386!jfh Ma Bell: (512) 255-8251 | GEnie PROHIBITED :-) | Domain: jfh@rpp386.cactus.org "If liberals interpreted the 2nd Amendment the same way they interpret the rest of the Constitution, gun ownership would be mandatory."