summaryrefslogtreecommitdiff
path: root/gcc/ada/sem_prag.adb
diff options
context:
space:
mode:
authorcharlet <charlet@138bc75d-0d04-0410-961f-82ee72b054a4>2004-03-29 12:03:27 +0000
committercharlet <charlet@138bc75d-0d04-0410-961f-82ee72b054a4>2004-03-29 12:03:27 +0000
commitfa7497e853a70dd5d253a1313d0dfa7ddbc02eec (patch)
treeb41024c63720bbb6fd3edef3c99d72097d058a4e /gcc/ada/sem_prag.adb
parent611a00932367c28e71b5682cb244fb548eeac5e7 (diff)
downloadgcc-fa7497e853a70dd5d253a1313d0dfa7ddbc02eec.tar.gz
2004-03-29 Javier Miranda <miranda@gnat.com>
* checks.adb (Null_Exclusion_Static_Checks): New subprogram (Install_Null_Excluding_Check): Local subprogram that determines whether an access node requires a runtime access check and if so inserts the appropriate run-time check. (Apply_Access_Check): Call Install_Null_Excluding check if required (Apply_Constraint_Check): Call Install_Null_Excluding check if required * checks.ads: (Null_Exclusion_Static_Checks): New subprogram * einfo.ads: Fix typo in comment * exp_ch3.adb (Build_Assignment): Generate conversion to the null-excluding type to force the corresponding run-time check. (Expand_N_Object_Declaration): Generate conversion to the null-excluding type to force the corresponding run-time check. * exp_ch5.adb (Expand_N_Assignment_Statement): Generate conversion to the null-excluding type to force the corresponding run-time check. * exp_ch6.adb (Expand_Call): Do not generate the run-time check in case of access types unless they have the null-excluding attribute. * sprint.adb (Sprint_Node_Actual): Give support to the null-exclusing part. * exp_util.ads: Fix typo in comment * par.adb (P_Null_Exclusion): New subprogram (P_Subtype_Indication): New formal that indicates if the null-excluding part has been scanned-out and it was present * par-ch3.adb, par-ch4.adb, par-ch6.adb: Give support to AI-231 * sem_aggr.adb: (Check_Can_Never_Be_Null): New subprogram (Aggregate_Constraint_Checks): Generate conversion to the null-excluding type to force the corresponding run-time check (Resolve_Aggregate): Propagate the null-excluding attribute to the array components (Resolve_Array_Aggregate): Carry out some static checks (Resolve_Record_Aggregate.Get_Value): Carry out some static check * sem_ch3.adb (Access_Definition): In Ada 0Y the Can_Never_Be_Null attribute must be set only if specified by means of the null-excluding part. In addition, we must also propagate the access-constant attribute if present. (Access_Subprogram_Declaration, Access_Type_Declaration, Analyze_Component_Declaration, Analyze_Object_Declaration, Array_Type_Declaration, Process_Discriminants, Analyze_Subtype_Declaration): Propagate the null-excluding attribute and carry out some static checks. (Build_Derived_Access_Type): Set the null-excluding attribute (Derived_Type_Declaration, Process_Subtype): Carry out some static checks. * sem_ch4.adb (Analyze_Allocator): Carry out some static checks * sem_ch5.adb (Analyze_Assignment): Carry out some static checks * sem_ch6.adb (Process_Formals): Carry out some static checks. (Set_Actual_Subtypes): Generate null-excluding subtype if the null-excluding part was present; it is not required to be done here in case of anonymous access types. (Set_Formal_Mode): Ada 0Y allows anonymous access to have the null value. * sem_res.adb (Resolve_Actuals): Carry out some static check (Resolve_Null): Allow null in anonymous access * sinfo.adb: New subprogram Null_Exclusion_Present All_Present and Constant_Present available on access_definition nodes * sinfo.ads: New flag Null_Exclusion_Present on subtype_declaration, object_declaration, derived_type_definition, component_definition, discriminant_specification, access_to_object_definition, access_function_definition, allocator, access_procedure_definition, access_definition, parameter_specification, All_Present and Constant_Present flags available on access_definition nodes. 2004-03-29 Robert Dewar <dewar@gnat.com> * fname.adb, fname.ads, fname-sf.adb, fname-uf.adb, fname-uf.ads, gnat1drv.adb, lib.adb, lib.ads, lib-load.adb, lib-writ.adb, opt.ads, osint.adb, osint.ads, osint-c.adb, par.adb, par-ch10.adb, par-load.adb, par-prag.adb, sfn_scan.adb, sfn_scan.ads, sinput-l.adb, sinput-l.ads, switch-c.adb, sem_prag.adb: Updates to handle multiple units/file * par.adb: Change test for s-rpc to s-rp for detecting rpc and children * par.adb, memtrack.adb, prj-makr.adb, prj-part.adb, sem_util.adb: Minor reformatting * sem_ch12.adb: Add comment for previous change 2004-03-29 Laurent Pautet <pautet@act-europe.fr> * osint.adb (Executable_Prefix): Set Exec_Name to the current executable name when not initialized. Otherwise, use its current value. * osint.ads (Exec_Name): Move Exec_Name from body to spec in order to initialize it to another executable name than the current one. This allows to configure paths for an executable name (gnatmake) different from the current one (gnatdist). 2004-03-29 Ed Schonberg <schonberg@gnat.com> * exp_ch6.adb (Expand_Call): A call to a function declared in the current unit cannot be inlined if it appears in the body of a withed unit, to avoid order of elaboration problems in gigi. * exp_ch9.adb (Build_Protected_Sub_Specification): Generate debugging information for protected (wrapper) operation as well, to simplify gdb use. * sem_ch6.adb (Analyze_Subprogram_Body): For a private operation in a protected body, indicate that the entity for the generated spec comes from source, to ensure that references are properly generated for it. (Build_Body_To_Inline): Do not inline a function that returns a controlled type. * sem_prag.adb (Process_Convention): If subprogram is overloaded, only apply convention to homonyms that are declared explicitly. * sem_res.adb (Make_Call_Into_Operator): If the operation is a function that renames an equality operator and the operands are overloaded, resolve them with the declared formal types, before rewriting as an operator. 2004-03-29 GNAT Script <nobody@gnat.com> * Make-lang.in: Makefile automatically updated git-svn-id: svn+ssh://gcc.gnu.org/svn/gcc/trunk@80055 138bc75d-0d04-0410-961f-82ee72b054a4
Diffstat (limited to 'gcc/ada/sem_prag.adb')
-rw-r--r--gcc/ada/sem_prag.adb50
1 files changed, 39 insertions, 11 deletions
diff --git a/gcc/ada/sem_prag.adb b/gcc/ada/sem_prag.adb
index 62eb47a5c0a..3c8ca3df41b 100644
--- a/gcc/ada/sem_prag.adb
+++ b/gcc/ada/sem_prag.adb
@@ -1692,7 +1692,6 @@ package body Sem_Prag is
is
Id : Node_Id;
E1 : Entity_Id;
- Comp_Unit : Unit_Number_Type;
Cname : Name_Id;
procedure Set_Convention_From_Pragma (E : Entity_Id);
@@ -1908,12 +1907,9 @@ package body Sem_Prag is
end if;
-- For the subprogram case, set proper convention for all homonyms
- -- in same compilation unit.
- -- Is the test of compilation unit really necessary ???
- -- What about subprogram renamings here???
+ -- in same scope.
else
- Comp_Unit := Get_Source_Unit (E);
Set_Convention_From_Pragma (E);
-- Treat a pragma Import as an implicit body, for GPS use.
@@ -1931,7 +1927,10 @@ package body Sem_Prag is
-- That is deliberate, we cannot chain the rep item on more
-- than one Rep_Item chain, to be fixed later ???
- if Comp_Unit = Get_Source_Unit (E1) then
+ if Comes_From_Source (E1)
+ and then Nkind (Original_Node (Parent (E1))) /=
+ N_Full_Type_Declaration
+ then
Set_Convention_From_Pragma (E1);
if Prag_Id = Pragma_Import then
@@ -8561,9 +8560,39 @@ package body Sem_Prag is
-- Source_File_Name --
----------------------
+ -- There are five forms for this pragma:
+
+ -- pragma Source_File_Name (
+ -- [UNIT_NAME =>] unit_NAME,
+ -- BODY_FILE_NAME => STRING_LITERAL
+ -- [, [INDEX =>] INTEGER_LITERAL]);
+
-- pragma Source_File_Name (
- -- [UNIT_NAME =>] unit_NAME,
- -- [BODY_FILE_NAME | SPEC_FILE_NAME] => STRING_LITERAL);
+ -- [UNIT_NAME =>] unit_NAME,
+ -- SPEC_FILE_NAME => STRING_LITERAL
+ -- [, [INDEX =>] INTEGER_LITERAL]);
+
+ -- pragma Source_File_Name (
+ -- BODY_FILE_NAME => STRING_LITERAL
+ -- [, DOT_REPLACEMENT => STRING_LITERAL]
+ -- [, CASING => CASING_SPEC]);
+
+ -- pragma Source_File_Name (
+ -- SPEC_FILE_NAME => STRING_LITERAL
+ -- [, DOT_REPLACEMENT => STRING_LITERAL]
+ -- [, CASING => CASING_SPEC]);
+
+ -- pragma Source_File_Name (
+ -- SUBUNIT_FILE_NAME => STRING_LITERAL
+ -- [, DOT_REPLACEMENT => STRING_LITERAL]
+ -- [, CASING => CASING_SPEC]);
+
+ -- CASING_SPEC ::= Uppercase | Lowercase | Mixedcase
+
+ -- Pragma Source_File_Name_Project (SFNP) is equivalent to pragma
+ -- Source_File_Name (SFN), however their usage is exclusive:
+ -- SFN can only be used when no project file is used, while
+ -- SFNP can only be used when a project file is used.
-- No processing here. Processing was completed during parsing,
-- since we need to have file names set as early as possible.
@@ -8580,9 +8609,7 @@ package body Sem_Prag is
-- Source_File_Name_Project --
------------------------------
- -- pragma Source_File_Name_Project (
- -- [UNIT_NAME =>] unit_NAME,
- -- [BODY_FILE_NAME | SPEC_FILE_NAME] => STRING_LITERAL);
+ -- See Source_File_Name for syntax
-- No processing here. Processing was completed during parsing,
-- since we need to have file names set as early as possible.
@@ -8597,6 +8624,7 @@ package body Sem_Prag is
-- Check that a pragma Source_File_Name_Project is used only
-- in a configuration pragmas file.
+
-- Pragmas Source_File_Name_Project should only be generated
-- by the Project Manager in configuration pragmas files.