Conference clt::dec_pascal_bugs

835.0. "%descr passes wrong datatype in descriptor" by TLE::REAGAN (All of this chaos makes perfect sense) Fri Mar 21 1997 11:03

          <<< TURRIS::DISK$NOTES_PACK:[NOTES$LIBRARY]PASCAL.NOTE;2 >>>
                             -< DEC Pascal Notes >-
================================================================================
Note 2665.0        %descr passes wrong datatype in descriptor            1 reply
LNZALI::BACHNER "Mouse not found. Click OK to conti" 58 lines  21-MAR-1997 09:27
--------------------------------------------------------------------------------
DEC Pascal V5.5-54
OpenVMS VAX V7.0

When a program passes [byte] -128..127 or [word] -32768..32767 variables by
descriptor, the datatype in the descriptor is always set to longword integer.

The behaviour was correct with V5.1 of the compiler (I could not check other
versions in between yet).

Following are two sample source files which illustrate the problem.

Is it already a known problem or should I IPMT it ?

Thanks for your help,
Hans.

--------------------- sample main program -----------------------
[inherit("sys$library:starlet.pen")]
program siemens_main (output);

TYPE
    $byte = [byte] -128..127;
    $word = [word] -32768..32767;

var
  element : $byte;
  scaling : $word;

PROCEDURE  WRITEATTR ( DSC: DSC1$TYPE ); external;

    begin 
         writeattr ( %descr element );
         writeattr ( %descr scaling );

end.
--------------------- sample subprogram -----------------------
[inherit("sys$library:starlet.pen")]
module siemens_test (output);

TYPE
    $byte = [byte] -128..127;
    $word = [word] -32768..32767;

[GLOBAL]
PROCEDURE  WRITEATTR ( DSC: DSC1$TYPE );

BEGIN
   WITH DSC  DO
           BEGIN
	    writeln (dsc$b_dtype);

           { DSC$B_DTYPE is always 8 (longword integer), but should be
	     6 for the $byte argument and 7 for the $word argument)
           }

	  end{ with };
end;
end.
    

    It changed sometime between V5.4-46 and V5.5-50.  That narrows it down
    to a couple of months.  I can see the difference in the intermediate
    code...
    
    				-John

    OK, I found it.  It was a simple error added last February for another
    fix.  This will be in the next kit.  I'd guess the next network kit
    might be in a couple of weeks.  Haven't picked the exact dates for
    the next SSB/cdrom kit yet...
    
    				-John