2011-08-04 Emmanuel Briot <briot@adacore.com>
* makeutl.adb (Complete_Mains): if the user has passed a relative file
name on the command line (for instance "../dummy"), never assume it can
be a unit name.
2011-08-04 Tristan Gingold <gingold@adacore.com>
* bindgen.adb: Remove code the reference or create the
Break_Start/__gnat_break_start procedure.
* s-stalib.adb (Break_Start): Removed.
From-SVN: r177374
2011-08-04 Yannick Moy <moy@adacore.com>
* par-ch4.adb (P_Primary): preferentially issue an error message about
a missing parenthesis arount a conditional or case expression in Ada
2012 mode, if we detect that the alignment is not correct for a
statement.
2011-08-04 Hristian Kirtchev <kirtchev@adacore.com>
* exp_ch7.adb (Process_Declarations): Do not consider the result of a
tag-to-class-wide conversion as needing finalization actions.
* exp_util.adb (Has_Controlled_Objects): Do not consider the result of
a tag-to-class-wide conversion as needing finalization actions.
(Is_Finalizable_Transient): The result of a tag-to-class-wide
conversion does not need finalization actions.
(Is_Tag_To_CW_Conversion): New routine.
* exp_util.ads (Is_Tag_To_CW_Conversion): New routine. Determines
whether an object is the result of a tag-to-class-wide conversion.
2011-08-04 Yannick Moy <moy@adacore.com>
* sem_ch13.adb (Analyze_Aspect_Specifications): correct order in which
the left-hand-side and right-hand-side of a conjunct are inserted when
translating a pre- or postcondition
* sem_ch6.adb: Correct typo in comment
2011-08-04 Ed Schonberg <schonberg@adacore.com>
* gnat_rm.texi: Ravenscar does not prohibit dependence on
Unchecked_Conversion and Unchecked_Deallocation.
2011-08-04 Thomas Quinot <quinot@adacore.com>
* make.adb: Minor reformatting.
From-SVN: r177371
2011-08-04 Emmanuel Briot <briot@adacore.com>
* prj.adb, prj.ads, makeutl.adb (Complete_Mains): search for the
sources in the project and its imported projects, but not in the whole
tree.
From-SVN: r177370
2011-08-04 Emmanuel Briot <briot@adacore.com>
* make.adb: Remove debug trace.
2011-08-04 Vincent Celier <celier@adacore.com>
* makeutl.adb (Do_Complete): Get the absolute path name of a relative
path in canonical form, to be able to compare path names on platforms
such as Windows or Darwin.
2011-08-04 Thomas Quinot <quinot@adacore.com>
* par_sco.adb (Traverse_Declarations_Or_Statements.Set_Statement_Entry):
For a pragma (statement with C1 = 'P'), record the sloc of the pragma.
* scos.ads: Update documentation accordingly.
* put_scos.adb (Output_SCO_Line): Omit statement SCOs for disabled
pragmas.
From-SVN: r177369
2011-08-04 Richard Guenther <rguenther@suse.de>
PR fortran/49957
* trans-array.c (add_to_offset): New function.
(gfc_conv_array_ref): Build the array index expression in optimally
associated order.
(gfc_walk_variable_expr): Adjust for the backward walk.
* gfortran.dg/vect/O3-pr49957.f: New testcase.
From-SVN: r177368
2011-08-04 Ed Schonberg <schonberg@adacore.com>
* sem_attr.adb (Bad_Attribute_For_Predicate): flag illegal use of
attribute only if prefix type is scalar.
2011-08-04 Emmanuel Briot <briot@adacore.com>
* make.adb, makeutl.adb, prj-env.adb (Check_Mains): put back support
in gnatmake for specifying mains on the command line that do not belong
to the main project. These mains must currently all belong to the same
project, though.
(Ultimate_Extension_Of): removed, since duplicated
Ultimate_Extending_Project.
From-SVN: r177367
2011-08-04 Arnaud Charlet <charlet@adacore.com>
* make.adb (Do_Codepeer_Globalize_Step): Removed. Use CodePeer_Mode
instead.
(CodePeer_Mode_String): New.
(Linking_Phase, Binding_Phase): Call gnatlink with -P switch in
CodePeer mode.
(Scan_Make_Arg): Do not disable binding/linking phase in CodePeer mode.
* bindgen.adb (Gen_Elab_Calls_Ada): Ignore subprograms in CodePeer
mode, since no useful elaboration subprogram is needed by CodePeer.
* gnatlink.adb (Gnatlink): Add support for -P switch (CodePeer mode).
In this mode, compile binder file with -gnatC and do stop after this
step.
2011-08-04 Vincent Celier <celier@adacore.com>
* exp_ch7.adb: Minor comment fix.
From-SVN: r177366
2011-08-04 Emmanuel Briot <briot@adacore.com>
* make.adb, makeutl.adb, makeutl.ads (Make): major refactoring.
Create several new subprograms to move code out of Make. This makes the
code more readable, removes code duplication, and is a preparation work
for adding support for aggregate projects.
2011-08-04 Tristan Gingold <gingold@adacore.com>
* s-po32gl.ads, s-po32gl.adb: New files.
* exp_ch7.ads (Get_Global_Pool_For_Access_Type): New function.
* exp_ch7.adb (Get_Global_Pool_For_Access_Type): New function.
(Build_Finalization_Collection): Use it.
* exp_ch4.adb (Complete_Controlled_Allocation): Ditto.
* rtsfind.ads: System_Pool_32_Global, Re_Global_Pool_32_Object: New
literals.
* gcc-interface/Makefile.in: Use s-po32gl.o on VMS.
* gcc-interface/Make-lang.in: Update dependencies.
From-SVN: r177364
2011-08-04 Emmanuel Briot <briot@adacore.com>
* makeutl.adb (Complete_Mains): add support for specs with no suffix on
the command line.
From-SVN: r177363
2011-08-04 Emmanuel Briot <briot@adacore.com>
* make.adb, makeutl.adb (Complete_Mains): if a main specified on the
command line is not a file name, also check whether it could be a unit
name.
From-SVN: r177362
2011-08-04 Eric Botcazou <ebotcazou@adacore.com>
* bindgen.adb (Gen_Finalize_Library_Ada): Factor out code to generate
the header of the finalization routine.
If the unit has no finalizer but is a body whose spec has one, then
generate the decrement of the elaboration entity only.
If the unit has a finalizer and is a spec, then do not generate the
decrement of the elaboration entity.
(Gen_Finalize_Library_C): Likewise.
2011-08-04 Hristian Kirtchev <kirtchev@adacore.com>
* exp_ch7.adb (Alignment_Of): New subsidiary routine.
(Bounds_Size_Expression): Removed.
(Double_Alignment_Of): New subsidiary routine.
(Make_Finalize_Address_Stmts): New local variable Index_Typ. Account
for a hole in the dope vector of unconstrained arrays due to different
index and element alignments.
2011-08-04 Ed Schonberg <schonberg@adacore.com>
* sem_res.adb (Resolve_Allocator): diagnose task allocator that will
raise program_error because body has not been seen yet.
2011-08-04 Ed Schonberg <schonberg@adacore.com>
* sem_ch10.adb (Analyze_With_Clause): Protect against child unit with
an unresolved name.
2011-08-04 Vincent Celier <celier@adacore.com>
* makeutl.adb (Do_Complete): Check absolute paths in canonical forms
2011-08-04 Yannick Moy <moy@adacore.com>
* alfa.adb, alfa.ads (Unique_Defining_Entity): move function from here
* sem_util.adb, sem_util.ads (Unique_Defining_Entity): ...to here
2011-08-04 Thomas Quinot <quinot@adacore.com>
* sem_ch12.adb (Analyze_Package_Instantiation): Do not omit body for
instantiation in RCI.
2011-08-04 Emmanuel Briot <briot@adacore.com>
* make.adb: Share more code with gprbuild
From-SVN: r177361
2011-08-04 Emmanuel Briot <briot@adacore.com>
* projects.texi: Added documentation for the IDE'Gnat project file
attribute.
2011-08-04 Nicolas Roche <roche@adacore.com>
* gnat_rm.texi: Minor editing.
2011-08-04 Javier Miranda <miranda@adacore.com>
* bindgen.adb (Gen_Adafinal_Ada): Do not differentiate the main case
and the library case for VM targets.
(Gen_Adainit_Ada): Likewise.
2011-08-04 Robert Dewar <dewar@adacore.com>
* g-altive.ads: Minor comment updates.
* prj-nmsc.adb: Minor reformatting.
2011-08-04 Javier Miranda <miranda@adacore.com>
* opt.ads
(Normalize_Scalars_Config): Value of the configuration switch set by
pragma Normalize_Scalars when it appears in the gnat.adc file.
(Normalize_Scalars): New field for record Config_Switches_Type. Used
to save and restore settings of this pragma.
* opt.adb
(Register_Opt_Config_Switches, Save_Opt_Config_Switches,
Restore_Opt_Config_Switches): Add missing support for Normalize_Scalars.
2011-08-04 Vincent Celier <celier@adacore.com>
* gnat_ugn.texi: Document gnatlink options -M and -M=mapfile
From-SVN: r177360
2011-08-04 Arnaud Charlet <charlet@adacore.com>
* makeutl.adb: Minor reformatting.
2011-08-04 Olivier Hainque <hainque@adacore.com>
* g-altive.ads: Update comments to reflect that only the soft
binding is available today.
From-SVN: r177359
2011-08-04 Arnaud Charlet <charlet@adacore.com>
* prj-env.adb: Remove local debug traces.
2011-08-04 Yannick Moy <moy@adacore.com>
* checks.adb (Apply_Float_Conversion_Check): correct a typo where Lo_OK
was used instead of Hi_OK, which could cause a read of an uninitialized
value later on. Detected while working on the new warning.
* exp_ch9.adb (Expand_N_Entry_Declaration): remove useless assignment
to local variable.
* sem_ch5.adb (Analyze_Assignment): set the last assignment component
in more cases, in order to detect more unreferenced values.
* sem_util.adb, sem_util.ads (Get_Enclosing_Object): return enclosing
object for expression, if any.
2011-08-04 Eric Botcazou <ebotcazou@adacore.com>
* exp_intr.adb (Expand_Binary_Operator_Call): Look at the RM size of
the operand types instead of the Esize.
2011-08-04 Steve Baird <baird@adacore.com>
* switch-b.adb (Scan_Binder_Switches): Add -P binder switch, used to
enable CodePeer_Mode.
* bindusg.adb (Display): Add help message describing -P binder switch.
* gnat_ugn.texi: Document -P binder switch.
* bindgen.adb (Gen_Main_Ada): If CodePeer_Mode is set, then call the
user-defined main program directly.
(Gen_Output_File_Ada): If CodePeer_Mode is set, generate a with of the
user-defined main program in the context clause of the package body.
From-SVN: r177355
2011-08-04 Emmanuel Briot <briot@adacore.com>
* makeutl.adb (Complete_Mains): when a multi-unit source file is
specified on the gprbuild command line, we need to compile all units
within that file, not just the first one we find
Fix error message for mains that are not found.
2011-08-04 Thomas Quinot <quinot@adacore.com>
* sem_ch6.adb: Update comment.
* sem_ch12.adb: Minor reformatting.
2011-08-04 Bob Duff <duff@adacore.com>
* s-tasren.adb (Task_Do_Or_Queue): Previous code was reading
Acceptor.Terminate_Alternative without locking Acceptor, which causes a
race condition whose symptom is to fail to lock Parent. That, in turn,
causes Parent.Awake_Count to be accessed without locking Parent, which
causes another race condition whose symptom is that Parent.Awake_Count
can be off by 1 (either too high or too low). The solution is to lock
Parent unconditionally, and then lock Acceptor, before reading
Acceptor.Terminate_Alternative.
From-SVN: r177352
2011-08-04 Arnaud Charlet <charlet@adacore.com>
* debug.adb: Update comment.
2011-08-04 Robert Dewar <dewar@adacore.com>
* par_sco.adb, exp_util.adb, make.adb, a-tags.adb, sem_prag.adb,
sem_util.adb, sem_attr.adb, exp_ch4.adb, g-comlin.adb, g-comlin.ads,
exp_ch6.adb: Minor reformatting.
2011-08-04 Gary Dismukes <dismukes@adacore.com>
* gnat_ugn.texi: Add sentence about the case of compile-time-detected
checks for -gnateE.
2011-08-04 Ed Schonberg <schonberg@adacore.com>
* gnat_rm.texi: update Ravenscar desription to reflect Ada2012 RM.
2011-08-04 Thomas Quinot <quinot@adacore.com>
* sinfo.ads: Minor reformatting.
2011-08-04 Emmanuel Briot <briot@adacore.com>
* makeutl.adb, makeutl.ads, prj-nmsc.adb (Complete_Mains): report error
even if there is no location.
From-SVN: r177351
2011-08-04 Thomas Quinot <quinot@adacore.com>
* sinfo.adb, sinfo.ads, sem_prag.adb, sem_ch12.adb (Pragma_Enabled):
This flag of N_Pragma nodes is not used, remove it as well as all of
the associated circuitry.
2011-08-04 Javier Miranda <miranda@adacore.com>
* exp_disp.adb (Make_DT): Switch -gnatdQ disables the generation of the
runtime check on duplicated externa tags
* debug.adb Document switch -gnatdQ.
2011-08-04 Gary Dismukes <dismukes@adacore.com>
* a-fihema.ads: Minor typo fix.
2011-08-04 Yannick Moy <moy@adacore.com>
* sem_ch10.adb: Minor comment update.
2011-08-04 Hristian Kirtchev <kirtchev@adacore.com>
* einfo.adb: Update the node field usage to reflect the renaming of
Return_Flag to Return_ Flag_Or_Transient_Decl.
(Return_Flag): Renamed to Return_Flag_Or_Transient_Decl.
(Set_Return_Flag): Renamed to Set_Return_Flag_Or_Transient_Decl.
(Write_Field15_Name): Change Return_Flag to
Return_Flag_Or_Transient_Decl.
* einfo.ads: Rename node field Return_Flag to
Return_Flag_Or_Transient_Decl. Update the associated comment and all
occurrences in entities.
(Return_Flag): Renamed to Return_Flag_Or_Transient_Decl. Update
associated Inline pragma.
(Set_Return_Flag): Renamed to Set_Return_Flag_Or_Transient_Decl. Update
associated Inline pragma.
* exp_ch4.ads, exp_ch4.adb (Expand_N_Expression_With_Actions): New
routine.
* exp_ch6.adb (Expand_N_Extended_Return_Statement): Update the calls to
Return_Flag and Set_Return_Flag.
* exp_ch7.adb (Process_Declarations): Add code to recognize hook
objects generated for controlled transients declared inside an
Exception_With_Actions. Update the calls to Return_Flag.
(Process_Object_Declaration): Add code to add a null guard for hook
objects generated for controlled transients declared inside an
Exception_With_Actions. Update related comment.
* exp_util.adb (Has_Controlled_Objects): Add code to recognize hook
objects generated for controlled transients declared inside an
Exception_With_Actions. Update the calls to Return_Flag.
* expander.adb (Expand): Add new case for N_Expression_With_Actions.
2011-08-04 Ed Schonberg <schonberg@adacore.com>
* sem_util.adb:(Wrong_Type): Improve error message on a one-element
positional aggregate.
2011-08-04 Vincent Celier <celier@adacore.com>
* par_sco.adb (Process_Decisions.Output_Header): Check and record pragma
SLOC only for pragmas.
2011-08-04 Emmanuel Briot <briot@adacore.com>
* projects.texi: Minor typo fix.
2011-08-04 Emmanuel Briot <briot@adacore.com>
* prj-nmsc.adb (Check_File): Minor change to traces, to help debugging
on case-sensitive file systems.
From-SVN: r177349
2011-08-04 Thomas Quinot <quinot@adacore.com>
* put_scos.adb (Put_SCOs): Do not emit decision SCO for an X decision
nested in a disabled pragma.
* scos.ads, scos.adb, par_sco.ads, par_sco.adb: Record sloc of
enclosing pragma, if any, for X decisions.
2011-08-04 Thomas Quinot <quinot@adacore.com>
* sem_prag.adb: Minor reformatting.
From-SVN: r177347
2011-08-04 Vincent Celier <celier@adacore.com>
* a-tags.adb (Check_TSD): Avoid concatenation of strings, as it is not
allowed in no run time mode.
2011-08-04 Vincent Celier <celier@adacore.com>
* make.adb (Check): When -gnatc is used, check for the source file
inside the ALI file with a canonical case file name.
From-SVN: r177346
2011-08-04 Emmanuel Briot <briot@adacore.com>
* g-comlin.adb, g-comlin.ads (Add_Switch): Put back support for
overriding the separator.
From-SVN: r177345
2011-08-04 Robert Dewar <dewar@adacore.com>
* sem_aggr.adb, par_sco.adb, sem_type.adb, exp_util.adb, exp_ch9.adb,
prj-nmsc.adb, sem_ch13.adb, exp_strm.adb: Minor reformatting.
2011-08-04 Robert Dewar <dewar@adacore.com>
* gnat_rm.texi: Minor documentation fix for pragma Annotate.
2011-08-04 Yannick Moy <moy@adacore.com>
* sem_attr.adb (Analyze_Attribute): add check during pre-analysis that
'Result only appears in postcondition of function.
2011-08-04 Thomas Quinot <quinot@adacore.com>
* a-tags.adb (Check_TSD): When raising PROGRAM_ERROR for a duplicated
external tag, include the value of the external tag in the exception
message.
From-SVN: r177344
2011-08-04 Yannick Moy <moy@adacore.com>
* sem_attr.adb (Result): modify error message for misplaced 'Result
2011-08-04 Sergey Rybin <rybin@adacore.com>
* gnat_rm.texi (pragma Annotate): Fix syntax description to make it
clear that the second argument must be an identifier.
2011-08-04 Thomas Quinot <quinot@adacore.com>
* exp_ch9.adb (Build_Barrier_Function): When compiling with
-fpreserve-control-flow, insert an IF statement on the barrier
condition to ensure that a conditional branch instruction is generated.
2011-08-04 Emmanuel Briot <briot@adacore.com>
* prj-part.adb, prj.adb, prj.ads, prj-tree.ads
(Processing_Flags.Ignore_Missing_With): new flag.
2011-08-04 Emmanuel Briot <briot@adacore.com>
* prj-nmsc.adb (Find_Sources, Path_Name_Of): Fix handling of
Source_List_File on case-insensitive systems where the file is actually
on a case-sensitive file system (NFS,...).
2011-08-04 Hristian Kirtchev <kirtchev@adacore.com>
* sem_ch6.adb (Analyze_Function_Return): In a rare case where a
function return contains a controlled [extension] aggregate and the
return statement is not part of a handled sequence of statements, wrap
the return in a block. This ensures that all controlled temporaries
generated during aggregate resolution will be picked up by the
finalization machinery.
2011-08-04 Ed Schonberg <schonberg@adacore.com>
* sem_aggr.adb (Resolve_Aggregate): If aggregate has box-initialized
components, freeze type before resolution, to ensure that default
initializations are present for all components.
* sem_res.adb (Resolve_Actuals): the designated object of an
accces-to-constant type is a legal actual in a call to an
initialization procedure.
2011-08-04 Hristian Kirtchev <kirtchev@adacore.com>
* exp_util.adb (Extract_Renamed_Object): Add N_Type_Conversion and
N_Unchecked_Type_Conversion to the possible containers of a renamed
transient variable.
From-SVN: r177343
2011-08-04 Yannick Moy <moy@adacore.com>
* par-ch13.adb (Aspect_Specifications_Present): recognize
"with Identifier'Class =>" as an aspect, so that a meaningful warning
is issued in Strict mode.
* par.adb: Fix typos in comments.
2011-08-04 Yannick Moy <moy@adacore.com>
* sem_attr.adb (Result): modify error message to take into account Post
aspect when compiling Ada 2012 (or newer) code.
2011-08-04 Nicolas Roche <roche@adacore.com>
* env.c (__gnat_clearenv): Avoid use of dynamic size array in order to
remove need for GCC exceptions.
2011-08-04 Vincent Celier <celier@adacore.com>
* makeutl.adb (Do_Complete): Call Debug_Output with the name of the
project, not the source file name.
* prj.adb (Find_Sources.Look_For_Sources): If the source has been
excluded, continue looking. This excluded source will only be returned
if there is no other source with the same base name that is not locally
removed.
2011-08-04 Ed Schonberg <schonberg@adacore.com>
* sem_res.adb (Resolve_Intrinsic_Operator): if the result type is
private and one of the operands is a real literal, use a qualified
expression rather than a conversion which is not meaningful to the
back-end.
From-SVN: r177342
2011-08-04 Yannick Moy <moy@adacore.com>
* sem_ch13.adb (Aspect_Loop): when an aspect X and its classwise
corresponding aspect X'Class are allowed, proceed with analysis of the
aspect instead of skipping it.
2011-08-04 Tristan Gingold <gingold@adacore.com>
* s-flocon.adb, s-flocon.ads: Add a body for s-flocon.ads
This will allow to have an Ada implementation for the subprogram.
From-SVN: r177341
2011-08-04 Ed Schonberg <schonberg@adacore.com>
* sem_ch3.adb (Analyze_Full_Type_Declaration): If the declaration is a
completion and aspects are present, apply them to the entity for the
type which is currently the partial view, but which is the one that
will be frozen.
* sem_ch13.adb (Analyze_Aspect_Specifications): if the predicate
applies to a partial view, indicate that the full view has predicates
and delayed aspects.
(Replace_Type_Reference): Handle properly predicates that apply to the
full view of a private completion.
From-SVN: r177340
2011-08-04 Eric Botcazou <ebotcazou@adacore.com>
* layout.adb (Layout_Type): For composite types, do not set Esize.
* freeze.adb (Set_Small_Size): Remove test on alignment and do not
set Esize.
(Size_Known): Look at the RM size of components instead of the Esize.
(Freeze_Record_Type): Look at the RM size instead of the Esize to
issue warning and activate Implicit_Packing.
(Freeze_Entity): Likewise. Do not issue a warning for alias/atomic
if the Esize is not known.
* sem_ch13.adb (Analyze_Attribute_Definition_Clause) <Size>: Set Esize
for elementary types only.
(Analyze_Record_Representation_Clause): Look at the RM size instead
of the Esize to issue errors.
* gcc-interface/decl.c (gnat_to_gnu_entity): Do not set Esize if it
is not known.
<E_Record_Type>: Look at the RM size instead of the Esize. Remove
obsolete block.
Look at the RM size instead of the Esize for types if the latter is
not known.
(gnat_to_gnu_field): Use Known_Esize instead of Known_Static_Esize.
2011-08-04 Ed Schonberg <schonberg@adacore.com>
* sem_type.adb: proper handling of equality not involving anonymous
access types.
From-SVN: r177339
2011-08-04 Hristian Kirtchev <kirtchev@adacore.com>
* exp_ch7.adb (Create_Finalizer): Remove local variables Spec_Nod and
Vis_Decls. When creating a library-level finalizer for a package spec,
both the declaration and body of the finalizer are inserted either in
the visible or private declarations of the package spec.
2011-08-04 Javier Miranda <miranda@adacore.com>
* sem_ch3.adb (Derive_Subprograms): Complete assertion to request the
use of the full-view of a type when invoking Is_Ancestor.
* sem_type.adb (Is_Ancestor): For consistency, when the traversal of
the full-view of private parents is requested, then use also the
full-view of the parent of the first derivation.
From-SVN: r177338
2011-08-04 Tristan Gingold <gingold@adacore.com>
* s-taprop-vxworks.adb (Enter_Task): Use System.Float_Control.Reset
instead of the locally imported procedure.
* s-taprop-mingw.adb (Enter_Task): Ditto.
* s-valrea.adb (Scan_Real): Ditto.
* s-imgrea.adb (Set_Image_Real): Ditto.
* s-flocon.ads: Make the package pure.
2011-08-04 Thomas Quinot <quinot@adacore.com>
* sinfo.ads, sinfo.adb (Debug_Statement, Set_Debug_Statement): Remove.
* tbuild.ads, tbuild.adb (Make_Pragma): Adjust accordingly.
* sinfo-cn.ads, sinfo-cn.adb (Change_Name_To_Procedure_Call_Statement):
New subprogram, moved here from...
* par.adb, par-ch5.adb (P_Statement_Name): ... here.
* par-prag.adb (Par.Prag, case Pragma_Debug): Do not perform any
rewriting of the last argument into a procedure call statement here...
* sem_prag.adb (Analyze_Pragma, case Pragma_Debug): ...do it there
instead.
2011-08-04 Thomas Quinot <quinot@adacore.com>
* par_sco.adb: Minor reformatting.
From-SVN: r177337
2011-08-04 Javier Miranda <miranda@adacore.com>
* exp_strm.adb (Build_Array_Input_Function): if the array type is
constrained, do not attempt to build a subtype of it.
2011-08-04 Yannick Moy <moy@adacore.com>
* errout.ads Change comments: remove 'R'M as an example where quotes
are needed; add ALFA as another case where quotes are not needed
* erroutc.adb (Set_Msg_Insertion_Reserved_Word): add ALFA as another
case where quotes are not needed.
* sem_ch11.adb, sem_ch13.adb, sem_ch2.adb, sem_ch3.adb, sem_ch4.adb,
sem_ch5.adb, sem_ch6.adb, sem_ch9.adb, sem_prag.adb, sem_res.adb:
Remove all occurrences of 'A'L'F'A
2011-08-04 Ed Schonberg <schonberg@adacore.com>
* sem_ch13.adb: check legality of invariant pragma.
From-SVN: r177335
2011-08-04 Eric Botcazou <ebotcazou@adacore.com>
* bindgen.adb: Add comments.
2011-08-04 Yannick Moy <moy@adacore.com>
* einfo.adb, einfo.ads: Free Flag254 and make Formal_Proof_On a
synthesized flag.
* sem_prag.adb (Analyze_Pragma): record the pragma Annotate
(Formal_Proof, On/Off) in the Rep_Item list of the current subprogram.
2011-08-04 Robert Dewar <dewar@adacore.com>
* exp_ch7.adb, exp_ch6.adb, sem_ch3.adb, layout.adb, sem_ch5.adb,
osint-c.ads, sem_util.ads, gnat1drv.adb, targparm.ads, sem_ch6.adb,
sem_ch13.adb, s-pooloc.adb: Minor reformatting.
From-SVN: r177334
2011-08-04 Thomas Quinot <quinot@adacore.com>
PR ada/47880
* s-pooloc.adb (Deallocate): Fix the case of deallocating the only
allocated object.
From-SVN: r177332
2011-08-04 Ed Schonberg <schonberg@adacore.com>
* exp_ch4.adb (Expand_Composite_Equality): code cleanup: use component
type in all cases to compute list of primitive operations, because full
view may be an itype that is not attached to the list of declarations.
2011-08-04 Eric Botcazou <ebotcazou@adacore.com>
* bindgen.adb (Gen_Adafinal_Ada): Generate an early return if the
library has already been finalized.
(Gen_Adafinal_C): Likewise.
(Gen_Adainit_Ada): Generate an early return if the library has
already been elaborated.
(Gen_Adainit_C): Likewise.
(Gen_Output_File_Ada): Generate an elaboration flag.
(Gen_Output_File_C): Likewise.
From-SVN: r177331
2011-08-04 Hristian Kirtchev <kirtchev@adacore.com>
* exp_ch4.adb (Expand_Allocator_Expression): Disable the generation of
Set_Finalize_Address_Ptr for CodePeer because Finalize_Address was not
built in the first place.
(Expand_N_Allocator): Disable the generation of Set_Finalize_Address_Ptr
for CodePeer because Finalize_Address was not built to begin with.
* exp_ch6.adb (Make_Build_In_Place_Call_In_Allocator): Ditto.
From-SVN: r177330
2011-08-04 Pascal Obry <obry@adacore.com>
* adaint.c (__gnat_tmp_name): Use current process id to create temp
filenames, this ensures unicity of filenames across processes.
2011-08-04 Hristian Kirtchev <kirtchev@adacore.com>
* bindgen.adb (Gen_Finalize_Library_Ada): Rename generated routine
Raise_Controlled to Raise_From_Controlled_Operation. Update the
signature of Raise_From_Controlled_Operation by adding flag From_Abort.
Add False as the actual to From_Abort in the generated call to
Raise_From_Controlled_Operation.
2011-08-04 Jerome Lambourg <lambourg@adacore.com>
* osint-c.ads, osint-c.adb (Set_Library_Info_Name): Move to declaration
so that the ali file name can be retrieved from outside of this package.
* back_end.ads, back_end.adb (Gen_Or_Update_Object_File): New method
doing nothing in the general case, but used to update the object file
timestamp if directly generated by the backend.
* gnat1drv.adb (Gnat1drv): Make sure the object file's timestamp is set
to a later time than the ali file one.
From-SVN: r177329
2011-08-04 Yannick Moy <moy@adacore.com>
* einfo.adb, einfo.ads (Formal_Proof_On): new flag set on subprogram
entities whose body contains an Annotate pragma which forces formal
proof on this body.
* sem_ch11.adb, sem_ch2.adb, sem_ch3.adb, sem_ch4.adb, sem_ch5.adb,
sem_ch6.adb, sem_ch9.adb, sem_res.adb: Adapt call to
Mark_Non_ALFA_Subprogram to pass in a message and node.
* sem_prag.adb (Analyze_Pragma): add treatment of pragma Annotate
(Forma_Proof, On) which sets the flag Formal_Proof_On in the
surrounding subprogram.
* sem_util.adb, sem_util.ads (Mark_Non_ALFA_Subprogram,
Mark_Non_ALFA_Subprogram_Unconditional): if the subprogram being marked
as not in ALFA is annotated with Formal_Proof being On, then an error
is issued based on the additional parameters for message and node.
* snames.ads-tmpl (Name_Formal_Proof): new name for annotation.
* gcc-interface/Make-lang.in: Update dependencies.
2011-08-04 Hristian Kirtchev <kirtchev@adacore.com>
* exp_ch3.adb (Expand_Freeze_Class_Wide_Type): Do not generate
Finalize_Address when CodePeer is enabled.
2011-08-04 Pascal Obry <obry@adacore.com>
* adaint.c (__gnat_tmp_name): Use _tempnam() instead of tempnam() as
the latter returns a pointer to a static buffer which is deallocated
at the end of the routine.
From-SVN: r177328
2011-08-04 Yannick Moy <moy@adacore.com>
* sem_ch3.adb (Array_Type_Declaration): move test for type in ALFA
after index creation; mark unconstrained base array type generated as
being in/not in ALFA as well
(Make_Index): mark subtype created as in/not in ALFA
* sem_ch5.adb (Analyze_Iteration_Scheme): mark entity for iterating
over a loop as in/not in ALFA, depending on its type and form of loop
iteration.
2011-08-04 Hristian Kirtchev <kirtchev@adacore.com>
* exp_ch7.adb (Build_Object_Declarations): Initialize flag Abort
directly to False on .NET and JVM.
From-SVN: r177326
2011-08-04 Yannick Moy <moy@adacore.com>
* sem_ch3.adb (Constrain_Enumeration, Constrain_Integer): remove
constraint that subtype must come from source code to be in ALFA, so
that anonymous subtypes can be in ALFA too.
From-SVN: r177325
2011-08-04 Yannick Moy <moy@adacore.com>
* sem_ch3.adb (Analyze_Object_Declaration): remove obsolete addition of
reference for ALFA mode.
* lib-xref-alfa.adb, lib-xref.ads (Traverse_Compilation_Unit): make
procedure public so as to use it from formal verification back-end.
From-SVN: r177324
2011-08-04 Jerome Lambourg <lambourg@adacore.com>
* g-comlin.adb (For_Each_Simple_Switch.Is_In_Config): Remove the
invalid checks of the Parameter, making Remove_Switch not working
correctly.
2011-08-04 Arnaud Charlet <charlet@adacore.com>
* targparm.ads: Update header: no run-time exception is needed on this
file.
2011-08-04 Vincent Celier <celier@adacore.com>
* a-fihema.adb: Use System.IO instead of GNAT.IO
2011-08-04 Thomas Quinot <quinot@adacore.com>
* sem_ch3.adb (Access_Definition): Anonymous access to subprogram types
are forbidden in Compiler_Unit mode.
From-SVN: r177323
