Re: Ada and robots

[Huy Vo <vo@CTRON.COM>]

aebrain@dynamite.com.au (Alan Brain) wrote
>
>Huy Vo wrote:
>
>> I don't know a C compiler that doesn't have __LINE__. Do you?
>
>DEC LSI-11/02 C for RTS-11SJ. The first time I did an embedded system in
>C, back in 1978.
>
>Turbo C++ 2.0 has problems with __LINE__ too, and I'll check with MVC++
>5 and CW 10, CW 12 to see if they give identical results.
>
>> Do you know of 2 compilers that will not agree on the value of __LINE__
>> in the  following program:
>
>> #include <stdio.h> /* since you insist on having this */
>>
>> main() {
>>         printf("the value of __LINE__ is independent of the compiler %d\n",
>>  __LINE__);
>>         return 0;
>> }
>
>Yes, the LSI-11/02 and Turbo C++ 2.
>


DEC has a compiler that accepts this line
include stdio.h      /* no brackets */
given the kind of liberty they allow the compiler to have,
I wouldn't worry much what a DEC compiler has to say.

I don't know where you digged out your copy of Turbo C;
mine (Turbo C 4.5.1 ) compiles __LINE__ without a blink; if your compiler had
 problem with __LINE__ then it's a bug, and not a portability issue.

The following compilers have  __LINE__:
Metaware Hi-C
Sun cc and Sun acc
gcc
AIX xlc
Microsoft 7.1
Mix Power C
That's 7 modern compilers againts your ancient DEC compiler.
I am sure the list is not exhaustive. If only 5 out of 10 Ada compilers
provided an mechanism equivalent to __LINE__, would you forget the portability
crap and use it, or are you  still saying "Gee, I can't use this functionality
because only half the compilers have it. I just have to hard code in the line
 number
to make my code portable." IMHO, if there is a worse thing than goto then it has
 to
be hard coded constants. You brought up the portability
issue, whereas I intended to use __LINE__ solely as a  method of debugging.