OpenE2K/gcc
2
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

[multiple changes]

Browse Source 
2010-10-04  Vincent Celier  <celier@adacore.com>

	* sinput-l.adb (Load_File): Do not fail when switch -gnateG is
	specified and the processed file cannot be written. Just issue a
	warning and continue.

2010-10-04  Thomas Quinot  <quinot@adacore.com>

	* sem_res.adb: Minor reformatting.

2010-10-04  Ed Schonberg  <schonberg@adacore.com>

	* sem_ch8.adb (Analyze_Subprogram_Renaming): If the renamed operation
	is an overridden inherited operation, the desired operation is the
	overriding one, which is the alias of the visible one.

2010-10-04  Ed Schonberg  <schonberg@adacore.com>

	* sem_ch6.adb (Find_Corresponding_Spec): Check that the wrapper body is
	present before deleting from the tree, when an inherited function with
	a controlling result that returns a null extension is overridden by a
	later declaration or body.

2010-10-04  Gary Dismukes  <dismukes@adacore.com>

	* checks.adb: Update comment.

From-SVN: r164933
This commit is contained in:
Arnaud Charlet 2010-10-04 15:16:56 +02:00
parent d4d24ba450
commit 1366997bf2
6 changed files with 72 additions and 37 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

27
gcc/ada/ChangeLog
View File

@ -1,3 +1,30 @@
2010-10-04 Vincent Celier <celier@adacore.com>
* sinput-l.adb (Load_File): Do not fail when switch -gnateG is
specified and the processed file cannot be written. Just issue a
warning and continue.
2010-10-04 Thomas Quinot <quinot@adacore.com>
* sem_res.adb: Minor reformatting.
2010-10-04 Ed Schonberg <schonberg@adacore.com>
* sem_ch8.adb (Analyze_Subprogram_Renaming): If the renamed operation
is an overridden inherited operation, the desired operation is the
overriding one, which is the alias of the visible one.
2010-10-04 Ed Schonberg <schonberg@adacore.com>
* sem_ch6.adb (Find_Corresponding_Spec): Check that the wrapper body is
present before deleting from the tree, when an inherited function with
a controlling result that returns a null extension is overridden by a
later declaration or body.
2010-10-04 Gary Dismukes <dismukes@adacore.com>
* checks.adb: Update comment.
2010-09-30 Joseph Myers <joseph@codesourcery.com>
* gcc-interface/misc.c (optimize, optimize_size): Undefine as macros

9
gcc/ada/checks.adb
View File

@ -819,15 +819,6 @@ package body Checks is
-- node is retained, in order to avoid the warning for
-- redundant conversions in Resolve_Type_Conversion.
-- The above comment is uncomfortable. This seems like
-- an awkward covert channel, since there isno general
-- requirement in sinfo.ads or einfo.ads that requires
-- this rewrite. Instead, the issue seems to be that in
-- the old code, some node was incorrectly marked as
-- coming from source when it should not have been and/or
-- the warning code did not properly test the appropriate
-- Comes_From_Soure flag. ???
Rewrite (N, Relocate_Node (N));
Set_Etype (N, Target_Type);

19
gcc/ada/sem_ch6.adb
View File

@ -5910,8 +5910,8 @@ package body Sem_Ch6 is
-- that was created for an operation inherited by a null
-- extension, it may be overridden by a body without a previous
-- spec (one more reason why these should be shunned). In that
-- case remove the generated body, because the current one is
-- the explicit overriding.
-- case remove the generated body if present, because the
-- current one is the explicit overriding.
elsif Ekind (E) = E_Function
and then Ada_Version >= Ada_05
@ -5922,15 +5922,20 @@ package body Sem_Ch6 is
then
Set_Has_Completion (E, False);
if Expander_Active then
if Expander_Active
and then Nkind (Parent (E)) = N_Function_Specification
then
Remove
(Unit_Declaration_Node
(Corresponding_Body (Unit_Declaration_Node (E))));
(Corresponding_Body (Unit_Declaration_Node (E))));
return E;
-- If expansion is disabled, the wrapper function has not
-- been generated, and this is the standard case of a late
-- body overriding an inherited operation.
-- If expansion is disabled, or if the wrapper function has
-- not been generated yet, this a late body overriding an
-- inherited operation, or it is an overriding by some other
-- declaration before the controlling result is frozen. In
-- either case this is a declaration of a new entity.
else
return Empty;

15
gcc/ada/sem_ch8.adb
View File

@ -2100,6 +2100,21 @@ package body Sem_Ch8 is
if No (Old_S) then
Old_S := Find_Renamed_Entity (N, Name (N), New_S, Is_Actual);
-- The visible operation may be an inherited abstract operation that
-- was overridden in the private part, in which case a call will
-- dispatch to the overriding operation. Use the overriding one in
-- the renaming declaration, to prevent spurious errors below.
if Is_Overloadable (Old_S)
and then Is_Abstract_Subprogram (Old_S)
and then No (DTC_Entity (Old_S))
and then Present (Alias (Old_S))
and then not Is_Abstract_Subprogram (Alias (Old_S))
and then Is_Overriding_Operation (Alias (Old_S))
then
Old_S := Alias (Old_S);
end if;
-- When the renamed subprogram is overloaded and used as an actual
-- of a generic, its entity is set to the first available homonym.
-- We must first disambiguate the name, then set the proper entity.

36
gcc/ada/sem_res.adb
View File

@ -8847,9 +8847,8 @@ package body Sem_Res is
-- own expression is a possibly overloaded function call. The
-- qualified expression is needed to be disambiguate the call,
-- but it appears in a context in which a name is needed, forcing
-- the use of a conversion.
-- In Ada2012 a qualified expression is a name, and this idiom
-- is not needed any longer.
-- the use of a conversion. In Ada 2012, a qualified expression is
-- a name, and this idiom is no longer needed.
elsif Nkind (Orig_N) = N_Qualified_Expression
and then Nkind (Expression (Orig_N)) = N_Function_Call
@ -9255,9 +9254,9 @@ package body Sem_Res is
Rewrite (N, Op_Node);
-- If the context type is private, add the appropriate conversions
-- so that the operator is applied to the full view. This is done
-- in the routines that resolve intrinsic operators,
-- If the context type is private, add the appropriate conversions so
-- that the operator is applied to the full view. This is done in the
-- routines that resolve intrinsic operators.
if Is_Intrinsic_Subprogram (Op)
and then Is_Private_Type (Typ)
@ -9277,9 +9276,8 @@ package body Sem_Res is
elsif Ekind (Op) = E_Function and then Is_Intrinsic_Subprogram (Op) then
-- Operator renames a user-defined operator of the same name. Use
-- the original operator in the node, which is the one that Gigi
-- knows about.
-- Operator renames a user-defined operator of the same name. Use the
-- original operator in the node, which is the one Gigi knows about.
Set_Entity (N, Op);
Set_Is_Overloaded (N, False);
@ -9290,12 +9288,12 @@ package body Sem_Res is
-- Set_Slice_Subtype --
-----------------------
-- Build an implicit subtype declaration to represent the type delivered
-- by the slice. This is an abbreviated version of an array subtype. We
-- define an index subtype for the slice, using either the subtype name
-- or the discrete range of the slice. To be consistent with index usage
-- elsewhere, we create a list header to hold the single index. This list
-- is not otherwise attached to the syntax tree.
-- Build an implicit subtype declaration to represent the type delivered by
-- the slice. This is an abbreviated version of an array subtype. We define
-- an index subtype for the slice, using either the subtype name or the
-- discrete range of the slice. To be consistent with index usage elsewhere
-- we create a list header to hold the single index. This list is not
-- otherwise attached to the syntax tree.
procedure Set_Slice_Subtype (N : Node_Id) is
Loc : constant Source_Ptr := Sloc (N);
@ -9401,10 +9399,10 @@ package body Sem_Res is
if Is_OK_Static_Expression (Low_Bound) then
-- The low bound is set from the low bound of the corresponding
-- index type. Note that we do not store the high bound in the
-- string literal subtype, but it can be deduced if necessary
-- from the length and the low bound.
-- The low bound is set from the low bound of the corresponding index
-- type. Note that we do not store the high bound in the string literal
-- subtype, but it can be deduced if necessary from the length and the
-- low bound.
Set_String_Literal_Low_Bound (Subtype_Id, Low_Bound);

3
gcc/ada/sinput-l.adb
View File

@ -591,10 +591,9 @@ package body Sinput.L is
if not Status then
Errout.Error_Msg
("could not write processed file """ &
("?could not write processed file """ &
Name_Buffer (1 .. Name_Len) & '"',
Lo);
return No_Source_File;
end if;
end;
end if;