[Bug ada/11090] New: Variant Record Initialization.

[Bug ada/11090] New: Variant Record Initialization.

[bdavis9659@comcast.net]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11090

           Summary: Variant Record Initialization.
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: critical
          Priority: P1
         Component: ada
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: bdavis9659@comcast.net
                CC: gcc-bugs@gcc.gnu.org
  GCC host triplet: i386/gnu/linux

Functions used in variant record initialization are called multiple times.


FYI, the VADS ada compiler will only call the function once.




------------------------------------------------------------------------------
Here is a test case that demonstrates the problem, followed by the output:

with text_io;

procedure varrec is

  type my_enum_type1 is (A1, A2, A3, A4, A5, A6);
  type my_enum_type2 is (B1, B2);

  type my_integer is new integer range 0..190;

  type var_rec_a (sel1 : my_enum_type1 := A4; sel2 : my_enum_type2 := B2) is
   record
     case sel1 is
       when A4 | A3 | A1 | A2 =>
         case sel1 is
           when A3 | A1 | A2 =>
             case sel1 is
               when A3 =>
                 case sel2 is
                   when B1 =>
                     vary : integer;
                   when others =>
                     null;
                 end case;
               when others =>
                 case sel2 is
                   when B2 =>
                      var3 : my_integer;
                   when others =>
                      null;
                 end case;
             end case;
           when others =>
             case sel2 is
                when B2 =>
                   varyy : integer;
                when B1 =>
                   varzz : integer;
             end case;
         end case;
       when others =>
         case sel2 is
            when B2 =>
               varyyy : integer;
            when B1 =>
               varzzz : integer;
         end case;
     end case;
   end record;

  type var_rec_2 (sel1 : my_enum_type1 := A4) is
   record
     var1 : integer;
     var2 : integer;
     case sel1 is
        when A4 | A3 | A1 | A2 | A6 =>
           var_rec_1 : var_rec_a;
        when A5 =>
           varx : integer;
        when others =>
           null;
     end case;
   end record;

  rec1 : var_rec_2(A2);
  rec2 : var_rec_2(A2);
  int_cnt : my_integer := my_integer'first;

  function inc_int return my_integer is
  begin
     text_io.put("   int_cnt was : " & my_integer'image(int_cnt));
     int_cnt := int_cnt + 1;
     text_io.put_line(" : is : " & my_integer'image(int_cnt) & " :");
     return int_cnt;
  end inc_int;

begin


  text_io.new_line;
  text_io.new_line;
  text_io.put_line("Initializing rec1");
  text_io.put_line("  Note: inc_int is unexpectedly called more than once");
  text_io.put_line("        when initializing the var3 field");
  rec1 := (sel1  => A2,
            var1  => 123,
            var_rec_1 => (sel1 => A2,
                          sel2 => B2,
                          var3 => inc_int),
            var2  => 456);
  text_io.put_line("Done initializing rec1");

  text_io.new_line;
  text_io.put_line("Initializing rec2");
  text_io.put_line("  Note: inc_int is unexpectedly called more than once");
  text_io.put_line("        when initializing the var3 field");
  rec2 := (sel1  => A2,
            var1  => 123,
            var_rec_1 => (sel1 => A2,
                          sel2 => B2,
                          var3 => inc_int),
            var2  => 456);
  text_io.put_line("Done initializing rec2");

  text_io.new_line;
  text_io.put_line("Printing var3 field in rec1 and then rec2.");
  text_io.put_line("   Note: This is not what was expected.");
  text_io.put_line("         Only expected inc_int to be called once,");
  text_io.put_line("         with returned values of 1 and 2.");
  text_io.put_line("rec1.var_rec_1.var3: " &
my_integer'image(rec1.var_rec_1.var3));
  text_io.put_line("rec2.var_rec_1.var3: " &
my_integer'image(rec2.var_rec_1.var3));

end varrec;


--------------------------------------------------------------------------
Here is the current output:


Initializing rec1
  Note: inc_int is unexpectedly called more than once
        when initializing the var3 field
   int_cnt was :  0 : is :  1 :
   int_cnt was :  1 : is :  2 :
   int_cnt was :  2 : is :  3 :
   int_cnt was :  3 : is :  4 :
   int_cnt was :  4 : is :  5 :
   int_cnt was :  5 : is :  6 :
   int_cnt was :  6 : is :  7 :
   int_cnt was :  7 : is :  8 :
   int_cnt was :  8 : is :  9 :
Done initializing rec1

Initializing rec2
  Note: inc_int is unexpectedly called more than once
        when initializing the var3 field
   int_cnt was :  9 : is :  10 :
   int_cnt was :  10 : is :  11 :
   int_cnt was :  11 : is :  12 :
   int_cnt was :  12 : is :  13 :
   int_cnt was :  13 : is :  14 :
   int_cnt was :  14 : is :  15 :
   int_cnt was :  15 : is :  16 :
   int_cnt was :  16 : is :  17 :
   int_cnt was :  17 : is :  18 :
Done initializing rec2

Printing var3 field in rec1 and then rec2.
   Note: This is not what was expected.
         Only expected inc_int to be called once,
         with returned values of 1 and 2.
rec1.var_rec_1.var3:  1
rec2.var_rec_1.var3:  10



------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug ada/11090] Variant Record Initialization.

[pinskia@physics.uc.edu]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11090


pinskia@physics.uc.edu changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
  GCC build triplet|                            |pc-linux-gnu
   GCC host triplet|i386/gnu/linux              |i386-pc-linux-gnu
 GCC target triplet|                            |pc-linux-gnu
           Priority|P1                          |P2

[Bug ada/11090] Variant Record Initialization.

[dhazeghi at yahoo dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11090


dhazeghi at yahoo dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|critical                    |normal
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
  GCC build triplet|pc-linux-gnu                |i386-pc-linux-gnu
 GCC target triplet|pc-linux-gnu                |i386-pc-linux-gnu
           Keywords|                            |wrong-code
   Last reconfirmed|0000-00-00 00:00:00         |2003-07-06 04:03:04
               date|                            |
   Target Milestone|3.4                         |---
            Version|unknown                     |3.3


------- Additional Comments From dhazeghi at yahoo dot com  2003-07-06 04:03 -------
Confirmed with gcc 3.3 branch and mainline (20030620).

[Bug ada/11090] Variant Record Initialization.

[bdavis9659 at comcast dot net]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11090



------- Additional Comments From bdavis9659 at comcast dot net  2003-09-13 16:51 -------
Below is a patch that resolves this problem:





Index: gcc/gcc/ada/trans.c
===================================================================
RCS file: /cvsroot/gcc/gcc/gcc/ada/trans.c,v
retrieving revision 1.30
diff -c -3 -p -r1.30 trans.c
*** gcc/gcc/ada/trans.c 7 Jul 2003 17:57:27 -0000       1.30
--- gcc/gcc/ada/trans.c 13 Sep 2003 16:24:28 -0000
*************** assoc_to_constructor (gnat_assoc, gnu_ty
*** 4946,4952 ****
        if (Do_Range_Check (Expression (gnat_assoc)))
        gnu_expr = emit_range_check (gnu_expr, Etype (gnat_field));
   
!       gnu_expr = convert (TREE_TYPE (gnu_field), gnu_expr);
   
        /* Add the field and expression to the list.  */
        gnu_list = tree_cons (gnu_field, gnu_expr, gnu_list);
--- 4946,4952 ----
        if (Do_Range_Check (Expression (gnat_assoc)))
        gnu_expr = emit_range_check (gnu_expr, Etype (gnat_field));
   
!       gnu_expr = convert (TREE_TYPE (gnu_field), protect_multiple_eval
(gnu_expr));
   
        /* Add the field and expression to the list.  */
        gnu_list = tree_cons (gnu_field, gnu_expr, gnu_list);

[Bug ada/11090] Variant Record Initialization.

[charlet at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11090


charlet at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED


------- Additional Comments From charlet at gcc dot gnu dot org  2003-10-21 15:48 -------
Fixed with last Ada merge.

Arno

[Bug ada/11090] Variant Record Initialization.

[pinskia at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11090


pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |3.4