[M3devel] Array indexing

Thu Dec 30 22:31:04 CET 2010

Hi all:
it seems the same to me, perhaps this uncovered bug has been there for years or a systematic problem, we would need to apply verification for the compiler intermediate code generation, though might be a problem in the code generation, we don't know, it should be provable that for a given input the Modula-3 compiler will gave the same intermediate code, this would open also a door for optimization, like in HP, was built a compiler like that, it will take some time, but eventually, before we can believe it, we may have the compiler specified front end (still if it were 100000 lines verifying would take 60 years/man, if its 10000, 6 years/man), like the Hoare utopia was to built itself.
So between more correct more efficient!

--- El jue, 30/12/10, Rodney M. Bates <rodney_bates at lcwb.coop> escribió:

> De: Rodney M. Bates <rodney_bates at lcwb.coop>
> Asunto: Re: [M3devel] Array indexing
> Para: "Dariusz Knociński" <dknoto at next.com.pl>
> CC: m3devel at elegosoft.com
> Fecha: jueves, 30 de diciembre, 2010 15:12
> I reproduced this immediately on
> LINUXLIBC6 and AMD64_LINUX.  It's a compiler bug.
> A version of cm3 several years old, and an old pm3, both
> with the (much older)
> gcc backend and the x86 internal backend also all show the
> same problem.
> 
> The memory area for each row of tab is getting initialized
> with dope for an
> open array, instead of the actual values.  Moreover,
> the dope is not right
> either, though it's recognizable.  Since the open
> array row constructor is
> assignable to the fixed array row of tab, the code is
> correct and the compiler
> should take care of the representation change.
> 
> Making tab a VAR instead of CONST shows the same bad
> behavior.
> 
> As a workaround, you can change the row constructors to
> fixed array constructors:
> 
>  CONST    
>     tab = ARRAY [0..2], [0..15] OF CARDINAL {
>     ARRAY [0..15] OF CARDINAL
>         { 17, 17, 17, 17, 17, 17,
> 17, 17, 17, 17, 17, 17, 17, 17, 17, 17 },
>     ARRAY [0..15] OF CARDINAL
>         {  1,  2, 
> 3,  4,  5,  6,  7,  8,  9, 10,
> 17, 17, 17, 17, 17, 17 },
>     ARRAY [0..15] OF CARDINAL
>         { 17, 11, 12, 13, 14, 15,
> 16, 17, 17, 17, 17, 17, 17, 17, 17, 17 }
>     };
> 
> and this works as expected:
> 
> (m3gdb) run
> Starting program:
> /home/rodney/proj/m3/exp/dknoto/AMD64_LINUX/prog
> Can't disable VM GC.
> Signal        Stop     
> Print   Pass to program Description
> SIG64         No   
>     No      Yes     
>        Real-time event 64
> [Thread debugging using libthread_db enabled]
> [New LWP 9108]
> k = 1, l = 9
> k = tab[k][l] => 0x0a
> k = tab[1][9]   => 0x0a
>  17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17
>  01 02 03 04 05 06 07 08 09 10 17 17 17 17 17 17
>  17 11 12 13 14 15 16 17 17 17 17 17 17 17 17 17
> 
> Program exited normally.
> (m3gdb)
> 
> Amazing that in all these years, nobody has coded a 2-D
> array constructor this
> way before, or at least didn't complain.
> 
> 
> Rodney Bates
> 
> 
> Dariusz Knociński wrote:
> > Hi all,
> > 
> > my name is Dariusz Knociński, I am new in this list,
> it is my first letter.
> > Best wishes for New Year to all.
> > 
> > I have a problem with two-dimensional arrays. In my
> simple source code I can't
> > get value from array indexed by two variables. But, I
> can get these values  by
> > indexing by constants. The program compile
> successfully but generates idiotic
> > output.
> > 
> > Source:
> > 
> > MODULE aib EXPORTS Main;
> > 
> > IMPORT IO, Fmt;
> > 
> > CONST    
> >     tab = ARRAY [0..2], [0..15] OF
> CARDINAL {
> >     ARRAY OF CARDINAL
> >         { 17, 17, 17, 17, 17,
> 17, 17, 17, 17, 17, 17, 17, 17, 17, 17, 17 },
> >     ARRAY OF CARDINAL
>         {  1,  2, 
> 3,  4,  5,  6,  7,  8,  9, 10,
> 17, 17, 17, 17, 17, 17 },
> >     ARRAY OF CARDINAL
>         { 17, 11, 12, 13, 14, 15,
> 16, 17, 17, 17, 17, 17, 17, 17, 17, 17 }
> >     };
> > 
> > VAR
> >     k, l : CARDINAL;
> >     
> > BEGIN
> >     
> >     k := 1;
> >     l := 9;
> >     
> >     IO.Put( "k = " &
> Fmt.Int(k) & ", " );
> >     IO.Put( "l = " &
> Fmt.Int(l) & "\n" );
> >     k :=
> tab[k][l];    
> >     IO.Put( "k = tab[k][l] =>
> 0x" &     Fmt.F( "%02s", Fmt.Unsigned(k))
> & "\n" );
> >     
> >     k := tab[1][9];
> >     IO.Put( "k =
> tab[1][9]   => 0x" &
>     Fmt.F( "%02s", Fmt.Unsigned(k)) &
> "\n" );
> >         
> >     FOR i := 0 TO 2 DO
> >     FOR j := 0 TO 15 DO
> >         IO.Put( Fmt.F( "
> %02s", Fmt.Int( tab[i][j] ) ) );
> >     END;
> >     IO.Put( "\n" );
> >     END;
> >     
> > END aib.
> > 
> > Output:
> > 
> > k = 1, l = 9
> > k = tab[k][l] => 0x00
> > k = tab[1][9]   => 0x0a
> >  6296864 16 00 00 00 00 00 00 00 00 00 00 00 00
> 00 00
> >  6296864 16 00 00 00 00 00 00 00 00 00 00 00 00
> 00 00
> >  6296864 16 00 00 00 00 00 00 00 00 00 00 00 00
> 00 00
> > 
> > My system and compiler version is:
> > 
> > $ uname -a
> > 
> > Linux wenus.next.com.pl 2.6.35.10-74.fc14.x86_64 #1
> SMP Thu Dec 23 16:04:50\
> >     UTC 2010 x86_64 x86_64 x86_64
> GNU/Linux
> >     $ cm3 -version
> > 
> > Critical Mass Modula-3 version 5.8.6
> >   last updated: 2010-04-11
> >   compiled: 2010-07-12 20:10:34
> >   configuration:
> /usr/local/cm3/bin/cm3.cfg
> >   host: AMD64_LINUX
> >   target: AMD64_LINUX
> > 
> > or
> > 
> > Critical Mass Modula-3 version d5.9.0
> >   last updated: 2010-07-21
> >   compiled: 2010-12-16 03:15:26
> >   configuration:
> /usr/local/cm3/bin/cm3.cfg
> >   host: AMD64_LINUX
> > quake runtime error: "/usr/local/cm3/bin/cm3.cfg",
> line 2: quake runtime error:\
> >  undefined variable: SL
> > 
> > --procedure--  -line-  -file---
> >               
>      2  /usr/local/cm3/bin/cm3.cfg
> >   target: 
> > Both compilers generate same results.
> > 
> > Best regards
> > DKnoto.
> > 
> 