🚀 go-pugleaf

In article <26AUG199217273382@srzts204>, roth@srzts204 (G.Roth, 2958,H267) writes:
> A recent message asked about bugs in the Terra-Datentechnik (former Logitech)
> Modula-2 compiler.
> Hereafter the list of bugs we currently know:
> 
[some bug descriptions deleted]
> MODULE bug5;
> 
> FROM SYSTEM IMPORT
>   SIZE;
> 
> FROM InOut IMPORT
>   WriteString, WriteInt, WriteLn;
> 
> VAR
[some declarations deleted]
>   d : RECORD x:ARRAY [0..4] OF CHAR; END;
[..deleted...]
> 
> BEGIN
[..deleted...]
>   WriteString( "  SIZE(d)   (* =5 *)" ); WriteInt( SIZE(d),   3 ); WriteLn;
>   WriteString( "  SIZE(d.x) (* =5 *)" ); WriteInt( SIZE(d.x), 3 ); WriteLn;
[..deleted...]
> END bug5.
> 
> 
> Comment:
> 	SIZE() of a record type whose true length is odd and larger than 1
> 	returns a value that is 1 too large:
> 
> $ modula/nolog bug5.mod
> $ link bug5
> $ run bug5
[...deleted...]
>   SIZE(d)   (* =5 *)  6
>   SIZE(d.x) (* =5 *)  5
[...deleted...]

That's not necessarily a bug. An implementation is free to allocate
one or more bytes for padding of compositional types.
It would be a bug if the compiler doesn't allocate 6 bytes for the
entire RECORD type.

-- 
_______________________________________________________________________________

Andreas Borchert, University of Ulm, SAI, D-W-7900 Ulm, Germany
Internet: borchert@mathematik.uni-ulm.de