[Ada] Handling of anonymous access to subprograms in assignments and calls.
Arnaud Charlet
charlet@adacore.com
Mon Aug 4 09:39:00 GMT 2008
These patches fix a number of small problems in the semantic analysis of
expressions involving both regular and protected anonymous access to
subprograms in various contexts. Examples can be found in ACATS test C3A0026.
Tested on i686-pc-linux-gnu, committed on trunk
2008-08-04 Ed Schonberg <schonberg@adacore.com>
* sem_ch3.adb (Access_Definition): A formal object declaration is a
legal context for an anonymous access to subprogram.
* sem_ch4.adb (Analyze_One_Call): If the call can be interpreted as an
indirect call, report success to the caller to include possible
interpretation.
* sem_ch6.adb (Check_Return_Type_Indication): Apply proper conformance
check when the type
of the extended return is an anonymous access_to_subprogram type.
* sem_res.adb:
(Resolve_Call): Insert a dereference if the type of the subprogram is an
access_to_subprogram and the context requires its return type, and a
dereference has not been introduced previously.
-------------- next part --------------
Index: sem_ch3.adb
===================================================================
--- sem_ch3.adb (revision 138583)
+++ sem_ch3.adb (working copy)
@@ -1054,6 +1054,7 @@ package body Sem_Ch3 is
or else
Nkind_In (D_Ityp, N_Object_Declaration,
N_Object_Renaming_Declaration,
+ N_Formal_Object_Declaration,
N_Formal_Type_Declaration,
N_Task_Type_Declaration,
N_Protected_Type_Declaration))
Index: sem_res.adb
===================================================================
--- sem_res.adb (revision 138518)
+++ sem_res.adb (working copy)
@@ -4692,6 +4692,25 @@ package body Sem_Res is
end loop;
end if;
+ if Ekind (Etype (Nam)) = E_Access_Subprogram_Type
+ and then Ekind (Typ) /= E_Access_Subprogram_Type
+ and then Nkind (Subp) /= N_Explicit_Dereference
+ and then Present (Parameter_Associations (N))
+ then
+ -- The prefix is a parameterless function call that returns an
+ -- access to subprogram. If parameters are present in the current
+ -- call add an explicit dereference.
+
+ -- The dereference is added either in Analyze_Call or here. Should
+ -- be consolidated ???
+
+ Set_Is_Overloaded (Subp, False);
+ Set_Etype (Subp, Etype (Nam));
+ Insert_Explicit_Dereference (Subp);
+ Nam := Designated_Type (Etype (Nam));
+ Resolve (Subp, Nam);
+ end if;
+
-- Check that a call to Current_Task does not occur in an entry body
if Is_RTE (Nam, RE_Current_Task) then
@@ -9487,7 +9506,10 @@ package body Sem_Res is
-- Access to subprogram types. If the operand is an access parameter,
-- the type has a deeper accessibility that any master, and cannot
- -- be assigned.
+ -- be assigned. We must make an exception if the conversion is part
+ -- of an assignment and the target is the return object of an extended
+ -- return statement, because in that case the accessibility check
+ -- takes place after the return.
elsif (Ekind (Target_Type) = E_Access_Subprogram_Type
or else
@@ -9497,6 +9519,10 @@ package body Sem_Res is
if Ekind (Base_Type (Opnd_Type)) = E_Anonymous_Access_Subprogram_Type
and then Is_Entity_Name (Operand)
and then Ekind (Entity (Operand)) = E_In_Parameter
+ and then
+ (Nkind (Parent (N)) /= N_Assignment_Statement
+ or else not Is_Entity_Name (Name (Parent (N)))
+ or else not Is_Return_Object (Entity (Name (Parent (N)))))
then
Error_Msg_N
("illegal attempt to store anonymous access to subprogram",
Index: sem_ch4.adb
===================================================================
--- sem_ch4.adb (revision 138518)
+++ sem_ch4.adb (working copy)
@@ -2127,11 +2127,12 @@ package body Sem_Ch4 is
-- is already known to be compatible, and because this may be an
-- indexing of a call with default parameters.
- Formal : Entity_Id;
- Actual : Node_Id;
- Is_Indexed : Boolean := False;
- Subp_Type : constant Entity_Id := Etype (Nam);
- Norm_OK : Boolean;
+ Formal : Entity_Id;
+ Actual : Node_Id;
+ Is_Indexed : Boolean := False;
+ Is_Indirect : Boolean := False;
+ Subp_Type : constant Entity_Id := Etype (Nam);
+ Norm_OK : Boolean;
function Operator_Hidden_By (Fun : Entity_Id) return Boolean;
-- There may be a user-defined operator that hides the current
@@ -2240,6 +2241,13 @@ package body Sem_Ch4 is
-- in prefix notation, so that the rebuilt parameter list has more than
-- one actual.
+ if not Is_Overloadable (Nam)
+ and then Ekind (Nam) /= E_Subprogram_Type
+ and then Ekind (Nam) /= E_Entry_Family
+ then
+ return;
+ end if;
+
if Present (Actuals)
and then
(Needs_No_Actuals (Nam)
@@ -2259,11 +2267,13 @@ package body Sem_Ch4 is
-- The prefix can also be a parameterless function that returns an
-- access to subprogram, in which case this is an indirect call.
+ -- If this succeeds, an explicit dereference is added later on,
+ -- in Analyze_Call or Resolve_Call.
elsif Is_Access_Type (Subp_Type)
and then Ekind (Designated_Type (Subp_Type)) = E_Subprogram_Type
then
- Is_Indexed := Try_Indirect_Call (N, Nam, Subp_Type);
+ Is_Indirect := Try_Indirect_Call (N, Nam, Subp_Type);
end if;
end if;
@@ -2278,13 +2288,21 @@ package body Sem_Ch4 is
return;
end if;
- Normalize_Actuals (N, Nam, (Report and not Is_Indexed), Norm_OK);
+ Normalize_Actuals
+ (N, Nam, (Report and not Is_Indexed and not Is_Indirect), Norm_OK);
if not Norm_OK then
+ -- If an indirect call is a possible interpretation, indicate
+ -- success to the caller.
+
+ if Is_Indirect then
+ Success := True;
+ return;
+
-- Mismatch in number or names of parameters
- if Debug_Flag_E then
+ elsif Debug_Flag_E then
Write_Str (" normalization fails in call ");
Write_Int (Int (N));
Write_Str (" with subprogram ");
@@ -2410,7 +2428,7 @@ package body Sem_Ch4 is
Write_Eol;
end if;
- if Report and not Is_Indexed then
+ if Report and not Is_Indexed and not Is_Indirect then
-- Ada 2005 (AI-251): Complete the error notification
-- to help new Ada 2005 users
Index: sem_ch6.adb
===================================================================
--- sem_ch6.adb (revision 138518)
+++ sem_ch6.adb (working copy)
@@ -542,16 +542,33 @@ package body Sem_Ch6 is
-- "return access T" case; check that the return statement also has
-- "access T", and that the subtypes statically match:
+ -- if this is an access to subprogram the signatures must match.
if R_Type_Is_Anon_Access then
if R_Stm_Type_Is_Anon_Access then
- if Base_Type (Designated_Type (R_Stm_Type)) /=
- Base_Type (Designated_Type (R_Type))
- or else not Subtypes_Statically_Match (R_Stm_Type, R_Type)
+ if
+ Ekind (Designated_Type (R_Stm_Type)) /= E_Subprogram_Type
then
- Error_Msg_N
- ("subtype must statically match function result subtype",
- Subtype_Mark (Subtype_Ind));
+ if Base_Type (Designated_Type (R_Stm_Type)) /=
+ Base_Type (Designated_Type (R_Type))
+ or else not Subtypes_Statically_Match (R_Stm_Type, R_Type)
+ then
+ Error_Msg_N
+ ("subtype must statically match function result subtype",
+ Subtype_Mark (Subtype_Ind));
+ end if;
+
+ else
+ -- For two anonymous access to subprogram types, the
+ -- types themselves must be type conformant.
+
+ if not Conforming_Types
+ (R_Stm_Type, R_Type, Fully_Conformant)
+ then
+ Error_Msg_N
+ ("subtype must statically match function result subtype",
+ Subtype_Ind);
+ end if;
end if;
else
More information about the Gcc-patches
mailing list