re PR fortran/15326 ([4.0 only] ICE with assumed length character strings)
PR fortran/15326
* trans-array.c (gfc_add_loop_ss_code): Set ss->string_length in
the GFC_SS_FUNCTION case too.
* trans-expr.c (gfc_conv_function_val): Allow symbols to be bound
to function pointers as well as function decls.
(gfc_interface_sym_mapping, gfc_interface_mapping): New structures.
(gfc_init_interface_mapping, gfc_free_interface_mapping)
(gfc_get_interface_mapping_charlen, gfc_get_interface_mapping_array)
(gfc_set_interface_mapping_bounds, gfc_add_interface_mapping)
(gfc_finish_interface_mapping, gfc_apply_interface_mapping_to_cons)
(gfc_apply_interface_mapping_to_ref)
(gfc_apply_interface_mapping_to_expr)
(gfc_apply_interface_mapping): New functions.
(gfc_conv_function_call): Evaluate the arguments before working
out where the result should go. Make the null pointer case provide
the string length in parmse.string_length. Cope with non-constant
string lengths, using the above functions to evaluate such lengths.
Use a temporary typespec; don't assign to sym->cl->backend_decl.
Don't assign to se->string_length when returning a cached array
descriptor.
re PR fortran/19928 (Reference of constant derived type component causes failure)
PR fortran/19928
* trans-array.c (gfc_conv_array_ref): Call gfc_advance_se_ss_chain
after handling scalarized references. Make "indexse" inherit from
"se" when handling AR_ELEMENTs.
(gfc_walk_variable_expr): Add GFC_SS_SCALAR entries for each
substring or scalar reference that follows an array section.
* trans-expr.c (gfc_conv_variable): When called from within a
scalarization loop, start out with "ref" pointing to the scalarized
part of the reference. Don't call gfc_advance_se_ss_chain here.
Andreas Krebbel [Thu, 8 Sep 2005 13:55:30 +0000 (13:55 +0000)]
s390.c (s390_sr_alias_set): Variable removed.
2005-09-08 Andreas Krebbel <krebbel1@de.ibm.com>
* config/s390/s390.c (s390_sr_alias_set): Variable removed.
(override_options): Setting s390_sr_alias_set removed.
(save_fpr, save_gprs): Set alias set to vararg or frame.
(restore_fpr, restore_gprs, s390_emit_prologue): Replace
s390_sr_alias_set with get_frame_alias_set ().
(s390_gimplify_va_arg): Replace s390_sr_alias_set with
get_varargs_alias_set ().
Jerry DeLisle [Thu, 8 Sep 2005 06:52:04 +0000 (06:52 +0000)]
re PR libfortran/23760 (gfortran incorrectly succeeds on record overflow)
2005-09-07 Jerry DeLisle <jvdelisle@verizon.net>
PR libfortran/23760
* gfortran.dg/g77/1832.f: Remove long string in write statement
to allow the test to pass on correct list directed output with
prepended space.
re PR libfortran/23262 ([mingw32] rewind truncates file)
PR libfortran/23262
* acinclude.m4 (LIBGFOR_CHECK_CRLF): New check.
* configure.ac: Use new check.
* configure.in: Regenerate.
* config.h.in: Regenerate.
* configure: Regenerate.
* io/transfer.c (next_record_w): Add case for CRLF as line
terminator.
* io/unix.c (tempfile, regular_file): Open files with
O_BINARY on systems with CRLF.
* reload1.c (fixup_eh_region_note): Remove assertion.
(fixup_abnormal_edges): Reverted removal of call to
find_many_sub_basic_blocks made on 2005-08-31.
Jakub Jelinek [Tue, 6 Sep 2005 20:07:13 +0000 (22:07 +0200)]
re PR c/23075 (Redundant / bogus warning)
PR c/23075
* c-typeck.c (c_finish_return): Set TREE_NO_WARNING on RETURN_EXPR
if "return with no value, in function returning non-void" warning
has been issued.
* tree-cfg.c (execute_warn_function_return): Don't look at
RETURN_EXPRs with TREE_NO_WARNING set.
* typeck.c (check_return_expr): Add no_warning argument. Set
*no_warning to true if "return-statement with no value, in function
returning" warning has been issued.
* cp-tree.h (check_return_expr): Adjust prototype.
* semantics.c (finish_return_stmt): Set TREE_NO_WARNING if
check_return_expr set *no_warning to true.
* gcc.dg/pr23075.c: New test.
* g++.dg/warn/pr23075.C: New test.
Jakub Jelinek [Tue, 6 Sep 2005 19:57:57 +0000 (21:57 +0200)]
re PR target/22362 (static function calls and global register variables)
PR target/22362
* config/i386/i386.c (ix86_function_regparm): Make sure automatic regparm
for internal functions doesn't use registers used by global registers
variables. Use fewer register parameters if there are global register
variables.
re PR middle-end/14997 (ncurses build fails with Ada)
PR middle-end/14997
* expr.c (expand_expr_real) <normal_inner_ref>: Force op0 to mem
when we would be extracting outside its bit span (bitpos+bitsize
larger than its mode), possible with some VIEW_CONVERT_EXPRs from
Ada unchecked conversions.
Co-Authored-By: Eric Botcazou <ebotcazou@adacore.com>
From-SVN: r103963
Mark Mitchell [Tue, 6 Sep 2005 14:55:06 +0000 (14:55 +0000)]
re PR c++/9782 (constructor not called on higher-dimensional arrays of template types)
* cp-tree.h (rvalue): New function.
* call.c (build_conditional_expr): Use it.
* init.c (build_new_1): Likewise.
* rtti.c (build_dynamic_cast_1): Likewise.
* tree.c (rvalue): New function.
* typeck.c (build_unary_op): Use it.
(build_static_cast_1): Likewise.
* g++.dg/expr/cast6.C: New test.
PR c++/9782
* init.c (build_new_1): Make sure the entire array type is
complete, not just its element types.
Keith Besaw [Tue, 6 Sep 2005 10:59:58 +0000 (10:59 +0000)]
common.opt: Add option ftree-vect-loop-version.
* common.opt: Add option ftree-vect-loop-version.
* params.def: Add --param vect-max-version-checks.
* doc/invoke.texi: Document ftree-vect-loop-version and
--param vect-max-version-checks.
* tree-vectorizer.h (_loop_vec_info): Add ptr_mask and
may_misalign_stmts and defines for accessors.
* tree-vectorizer.c : (new_loop_vec_info): VEC_alloc for
LOOP_VINFO_MAY_MISALIGN_STMTS.
(destroy_loop_vec_info): VEC_free for
LOOP_VINFO_MAY_MISALIGN_STMTS.
* tree-vect-analyze.c (vect_compute_data_ref_alignment):
Update documentation.
(vect_update_misalignment_for_peel): New.
(vect_enhance_data_refs_alignment): Update to choose loop
peeling or loop versioning if appropriate for the (potentially)
unaligned data references in the loop.
(vect_analyze_data_refs_alignment): Remove call to
vect_enhance_data_refs_alignment so the checks can be done
earlier.
(vect_analyze_loop): Add call to vect_enhance_data_refs_alignment
and move up call to vect_analyze_data_refs_alignment.
* tree-vect-transform.c (vect_create_cond_for_align_checks): New.
(vect_transform_loop): Add call to loop_version.
Mark Mitchell [Mon, 5 Sep 2005 15:59:31 +0000 (15:59 +0000)]
re PR c++/21440 (ICE with statement-as-expression)
PR c++/21440
* semantics.c (finish_stmt_expr_expr): Add an explicit
initialization to the last statement in the statement-expression.
* (finish_stmt_expr): Adjust accordingly.
PR c++/21440
* g++.dg/ext/stmtexpr5.C: New test.
* g++.dg/ext/stmtexpr6.C: Likewise.
re PR java/23431 (gcj allows overriding with more restrictive access)
PR java/23431
* typeck.c (lookup_do): Look up interfaces for the original class,
not the base class.
* parse.y (java_check_regular_methods): Fix diagnostic message for
more restrictive overriding of a method from an interface.
re PR target/23683 (review all uses of gen_rtx_MEM in sh port)
PR target/23683
* sh.c (sh_reorg, emit_load_ptr): Use gen_const_mem.
(output_stack_adjust): Use gen_tmp_stack_mem.
(sh_expand_prologue, sh_expand_epilogue): Use gen_frame_mem.
(sh_set_return_address, sh_allocate_initial_value): Likewise.
(sh_get_pr_initial_val): Likewise.
(sh_builtin_saveregs): Use gen_frame_mem and change_address.
(sh_initialize_trampoline): Likewise. Also use adjust_address.
* sh.md (divsi_inv_m0): Use gen_const_mem.
(push_fpscr, pop_fpscr, load_ra): Use gen_frame_mem.
(movdf_i4+1): Use gen_tmp_stack_mem.
(reload_outdf+3, reload_outdf+4, fpu_switch+1): Use change_address.
(fpu_switch+2): Likewise.
(movv4sf_i, movv16sf_i): Use adjust_address.
(symGOT_load): Set MEM_NOTRAP_P bit.
(Install_Null_Excluding_Check): Moved to the package specification
to use it from other packages to generate the run-time check
associated with the null-exclusion attribute.
Makefile.in: Adjust the libgnat target pairs for Xscale to ARM.
2005-09-01 Arnaud Charlet <charlet@adacore.com>
* Makefile.in: Adjust the libgnat target pairs for Xscale to ARM.
Note that the platform-specific version of g-soccon.ads for VMS is now
named g-soccon-vms.ads (it was previously g-soccon-vms.adb, although it
really is a package spec).
Replace s-taspri-linux.ads by s-taspri-posix.ads
Remove references to s-mastop-x86.adb
* system-vxworks-xscale.ads: Removed, no longer used.
* s-vxwork-xscale.ads: Removed, no longer used.
Robert Dewar [Mon, 5 Sep 2005 08:04:55 +0000 (10:04 +0200)]
g-dirope.ads: Minor reformatting Document that bounds of result of Base_Name match the input...
2005-09-01 Robert Dewar <dewar@adacore.com>
Arnaud Charlet <charlet@adacore.com>
* g-dirope.ads: Minor reformatting
Document that bounds of result of Base_Name match the input index
positions.
Add documentation on environment variable syntax for Expand_Path
* gnat_ugn.texi: Update documentation to include mention of -m switches
Document new treatment of wide characters in max line length
style check.
Remove -gnatL/-gnatZ switches, no longer used.
Add note on pragmas Assertion_Policy and Debug_Policy in discussion
of -gnata switch.
* gnat_rm.texi: Add doc for two argument form of pragma
Float_Representation.
Add documentation for pragma No_Strict_Aliasing
Add note that explicit component clause overrides pragma Pack.
Add documentation of pragma Debug_Policy
Ed Schonberg [Mon, 5 Sep 2005 08:03:48 +0000 (10:03 +0200)]
sem_util.ads, [...] (Gather_Components): Omit interface tags from the list of required components.
2005-09-01 Ed Schonberg <schonberg@adacore.com>
* sem_util.ads, sem_util.adb (Gather_Components): Omit interface tags
from the list of required components.
(Is_Controlling_Limited_Procedure): Determine whether an entity is a
primitive procedure of a limited interface with a controlling first
parameter.
(Is_Renamed_Entry): Determine whether an entry is a procedure renaming
of an entry.
(Safe_To_Capture_Value): A value (such as non_null) is not safe to
capture if it is generated in the second operand of a short-circuit
operation.
Do not capture values for variables with address clauses.
(Is_Object_Reference): Treat a function call as an object reference only
if its type is not Standard_Void_Type.
Ed Schonberg [Mon, 5 Sep 2005 08:03:33 +0000 (10:03 +0200)]
sem_type.adb (Add_One_Interp): If a candidate operation is an inherited interface operation that has an...
2005-09-01 Ed Schonberg <schonberg@adacore.com>
Javier Miranda <miranda@adacore.com>
* sem_type.adb (Add_One_Interp): If a candidate operation is an
inherited interface operation that has an implementation, use the
implementation to avoid spurious ambiguities.
(Interface_Present_In_Ancestor): In case of concurrent types we can't
use the Corresponding_Record_Typ attribute to look for the interface
because it is set by the expander (and hence it is not always
available). For this reason we traverse the list of interfaces
(available in the parent of the concurrent type).
(Interface_Present_In_Ancestor): Handle entities from the limited view
Javier Miranda [Mon, 5 Sep 2005 08:03:17 +0000 (10:03 +0200)]
sem_res.adb (Resolve_Membership_Op): In case of the membership test "Iface_CW_Typ in T'Class" we have nothing else...
2005-09-01 Javier Miranda <miranda@adacore.com>
Ed Schonberg <schonberg@adacore.com>
Gary Dismukes <dismukes@adacore.com>
* sem_res.adb (Resolve_Membership_Op): In case of the membership test
"Iface_CW_Typ in T'Class" we have nothing else to do in the frontend;
the expander will generate the corresponding run-time check to evaluate
the expression.
(Resolve_Call): Check for legal type of procedure name or prefix that
appears as a trigger in a triggering alternative.
(Valid_Conversion): If expression is ambiguous and the context involves
an extension of System, remove System.Address interpretations.
(Resolve_Qualified_Expression): Reject the case of a specific-type
qualification applied to a class-wide argument. Enhance comment
to explain checking of Original_Node.
(Resolve_Type_Conversion): The location of the error message was not
general enough to handle the general case and hence it has been removed.
In addition, this patch improves the text of the message.
(Resolve_Type_Conversion): Add missing support for access to interface
types.
(Resolve_Type_Conversion): If the target is a class-wide interface type,
do not expand if the expression is the actual in a call, because proper
expansion will take place when the call itself is expanded.
(Resolve_Allocator): If the context is an unchecked conversion, the
allocator inherits its storage pool, if any, from the target type of
the conversion.
Javier Miranda [Mon, 5 Sep 2005 08:03:00 +0000 (10:03 +0200)]
sem_disp.adb (Check_Controlling_Formals): Anonymous access types used in controlling parameters exclude null because it...
2005-09-01 Javier Miranda <miranda@adacore.com>
Ed Schonberg <schonberg@adacore.com>
* sem_disp.adb (Check_Controlling_Formals): Anonymous access types
used in controlling parameters exclude null because it is necessary to
read the tag to dispatch, and null has no tag.
(Override_Dispatching_Operation): If the previous operation is inherited
from an interface, it becomes hidden and does not participate in later
name resolution.
Ed Schonberg [Mon, 5 Sep 2005 08:02:37 +0000 (10:02 +0200)]
sem_ch8.adb (Analyze_Renamed_Entry): For a renaming_as_declaration...
2005-09-01 Ed Schonberg <schonberg@adacore.com>
* sem_ch8.adb (Analyze_Renamed_Entry): For a renaming_as_declaration,
verify that the procedure and the entry are mode conformant.
(Analyze_Subprogram_Renaming): Emit a warning if an operator is renamed
as a different operator, which is often a cut-and-paste error.
Javier Miranda [Mon, 5 Sep 2005 08:02:21 +0000 (10:02 +0200)]
sem_ch6.ads, [...] (Check_Conformance): In case of anonymous access types the null-exclusion and access-to-constant...
2005-09-01 Javier Miranda <miranda@adacore.com>
Gary Dismukes <dismukes@adacore.com>
* sem_ch6.ads, sem_ch6.adb (Check_Conformance): In case of anonymous
access types the null-exclusion and access-to-constant attributes must
also match.
(Analyze_Return_Statement): When the result type is an anonymous access
type, apply a conversion of the return expression to the access type
to ensure that appropriate accessibility checks are performed.
(Analyze_Return_Type): For the case of an anonymous access result type,
generate the Itype and set Is_Local_Anonymous_Access on the type.
Add ??? placeholder for check to disallow returning a limited object
in Ada 2005 unless it's an aggregate or a result of a function call.
Change calls from Subtype_Mark to Result_Definition.
(Analyze_Subprogram_Body): Change formal Subtype_Mark to
Result_Definition in call to Make_Function_Specification.
(Build_Body_To_Inline): Change Set_Subtype_Mark to
Set_Result_Definition.
(Make_Inequality_Operator): Change formal Subtype_Mark to
Result_Definition in call to Make_Function_Specification.
(Process_Formals): Create the new null-excluding itype if required.
(New_Overloaded_Entity): For an entity overriding an interface primitive
check if the entity also covers other abstract subprograms in the same
scope. This is required to handle the general case, that is, overriding
other interface primitives and overriding abstract subprograms inherited
from some abstract ancestor type.
(New_Overloaded_Entity): For an overriding entity that comes from
source, note the operation that it overrides.
(Check_Conformance, Type_Conformant): Addition of one new formal
to skip controlling formals in the analysis. This is used to
handle overloading of abstract interfaces.
(Base_Types_Match): Add missing case for types imported from
limited-with clauses
(New_Overloaded_Entity): Add barrier to protect the use of
the "alias" attribute.
Ed Schonberg [Mon, 5 Sep 2005 08:01:04 +0000 (10:01 +0200)]
sem_ch4.adb (Transform_Object_Operation): In a context off the form V (Obj.F)...
2005-09-01 Ed Schonberg <schonberg@adacore.com>
Javier Miranda <miranda@adacore.com>
* sem_ch4.adb (Transform_Object_Operation): In a context off the form
V (Obj.F), the rewriting does not involve the indexed component, but
only the selected component itself.
Do not apply the transformation if the analyzed node is an actual of a
call to another subprogram.
(Complete_Object_Operation): Retain the entity of the
dispatching operation in the selector of the rewritten node. The
entity will be used in the expansion of dispatching selects.
(Analyze_One_Call): Improve location of the error message associated
with interface.
(Analyze_Selected_Component): No need to resolve prefix when it is a
function call, resolution is done when parent node is resolved, as
usual.
(Analyze_One_Call): Add a flag to suppress analysis of the first actual,
when attempting to resolve a call transformed from its object notation.
(Try_Object_Operation, Transform_Object_Operastion): Avoid makind copies
of the argument list for each interpretation of the operation.
(Try_Object_Operation): The designated type of an access parameter may
be an incomplete type obtained through a limited_with clause, in which
case the primitive operations of the type are retrieved from its full
view.
(Analyze_Call): If this is an indirect call, and the return type of the
access_to_subprogram is incomplete, use its full view if available.
sem_ch3.ads, [...] (Analyze_Object_Declaration): Go to the underlying type to check if a type is Constrained in cases related to...
2005-09-01 Cyrille Comar <comar@adacore.com>
Gary Dismukes <dismukes@adacore.com>
Ed Schonberg <schonberg@adacore.com>
Javier Miranda <miranda@adacore.com>
* sem_ch3.ads, sem_ch3.adb (Analyze_Object_Declaration): Go to the
underlying type
to check if a type is Constrained in cases related to code generation
(rather than semantic checking) since otherwise we do not generate
similar code for mutable private types depending if their
discriminants are visible or not.
(Check_Abstract_Overriding): Do not complain about failure to override
the primitive operations used in dispatching selects since they will
always be overriden at the freeze point of the type.
(Access_Definition): Separate out handling for resetting the scope
of an anonymous access function result type. Retrieve the scope
of the associated function rather than using Current_Scope, which
does not have a consistent value (depends on whether we're in the
middle of analyzing formal parameters). Add ??? comment about
finding a cleaner way to handle the special cases of scope setting.
(Process_Incomplete_Dependents): A protected operation is never a
dispatching operation (only its wrapper may be).
(Build_Derived_Record_Type): In case of tagged private types that
implement interfaces add derivation of predefined primitive
operations.
(Derive_Subprograms): Replace the Is_Interface_Derivation parameter
by two parameters that are used in case of derivation from abstract
interface types: No_Predefined_Prims is used to avoid the derivation
of predefined primitives from the interface, and Predefined
Prims_Only is used to complete the derivation predefined primitives
in case of private tagged types implementing interfaces.
Fix typo in comments
(Find_Interface_In_Descendant): Protect the frontend against
wrong code with large circularity chains.
(Is_Private_Overriding): Add support for entities overriding interface
subprograms. The test failed because Entities associated with overriden
interface subprograms are always marked as hidden (and used to build
the secondary dispatch table); in this case the overriden entity is
available through the field abstract_interface_alias (cf. override_
dispatching_operation)
(Access_Definition): Set the scope of the type to Current_Scope for the
case of a function with an anonymous access result type.
(Access_Subprogram_Declaration): Handle creation of the type entity for
an access-to-function type with an anonymous access result.
(Check_Anonymous_Access_Types): Change Subtype_Mark to Result_Definition
in handling for N_Access_Function_Definition.
(Analyze_Subtype_Declaration): Modify the text of error message.
(Derived_Type_Declaration): Modify the text of error message.
(Process_Subtype): Modify the text of error message plus cleanup
of one redundant error message.
(Analyze_Component_Declaration): Code cleanup.
(Analyze_Object_Declaration): Code cleanup.
(Analyze_Subtype_Declaration): Propagate the null-exclusion
attribute in case of access types. Code cleanup.
(Array_Type_Declaration): Code cleanup.
(Process_Discriminants): Create the new null-excluding itype
if required. Code cleanup.
(Process_Subtype): Create the new null-excluding itype if
required. Code cleanup.
(Build_Derived_Record_Type): Code cleanup to avoid calling
twice the subprogram derive_subprograms in case of private
types that implement interfaces. In this particular case the
subprogram Complete_Subprograms_Derivation already does the
job associated with the second call.
* exp_strm.adb (Build_Elementary_Input_Call): Add an explicit
conversion to the full view when generating an operation for a
discriminant whose type may currently be private.
Ed Schonberg [Mon, 5 Sep 2005 07:59:10 +0000 (09:59 +0200)]
sem_ch12.adb (Instantiate_Subprogram_Body): When creating the defining entity for the instance body...
2005-09-01 Ed Schonberg <schonberg@adacore.com>
Javier Miranda <miranda@adacore.com>
Gary Dismukes <dismukes@adacore.com>
* sem_ch12.adb (Instantiate_Subprogram_Body): When creating the
defining entity for the instance body, make a new defining identifier
rather than copying the entity of the spec, to prevent accidental
sharing of the entity list.
(Check_Private_View): When exchanging views of private types, build the
list of exchanged views as a stack, to ensure that on exit the exchanges
are undone in the proper order.
(Analyze_Package_Instantiation, Analyze_Subprogram_Instantiation):
Restore the compilation environment in case of instantiation_error.
(Analyze_Generic_Subprogram_Declaration): Handle creation of type entity
for an anonymous access result.
(Instantiate_Generic_Subprogram): Subtype_Mark => Result_Definition
(Formal_Entity): Handle properly the case of a formal package that
denotes a generic package renaming.
Javier Miranda [Mon, 5 Sep 2005 07:58:52 +0000 (09:58 +0200)]
sem_ch10.adb (In_Chain): Moved from the scope of a subprogram to become local to the whole package.
2005-09-01 Javier Miranda <miranda@adacore.com>
Ed Schonberg <schonberg@adacore.com>
* sem_ch10.adb (In_Chain): Moved from the scope of a subprogram to
become local to the whole package.
(Install_Limited_Withed_Unit): Instead of unchaining real entities if
the package was already analyzed the new algorithm "replaces" the
real entities by the shadow ones. This is required to ensure that
the order of these entities in the homonym chains does not change;
otherwise we can have undefined references at linking time because
in case of conflicts the external name of the entities will have
a suffix that depends on the order of the entities in the chain.
(Remove_Limited_With_Clause): Complementary code that completes the
new algorithm and replaces the shadow entities by the real ones.
(Install_Limited_Withed_Unit): When unchaining entities before the
installation of the shadow entities, only regular entities of the
public part must be taken into account. This is required to
keep this routine in synch with the work done by Remove_Limited_
With_Clause
(Install_Limited_With_Clause): Introduce implicit limited_with_clause
even if unit is analyzed, because the analysis of the unit is
idempotent in any case, and the limited view of the unit may have to
be installed for proper visibility.
(Expand_Limited_With_Clause): Even if the unit in the implicit
with_clause has been analyzed already, a limited view of the package
must be built for the current context, if it does not exist yet.
Robert Dewar [Mon, 5 Sep 2005 07:58:38 +0000 (09:58 +0200)]
sem_cat.adb (Check_Categorization_Dependencies): Add more detail to error msgs for most common cases.
2005-09-01 Robert Dewar <dewar@adacore.com>
* sem_cat.adb (Check_Categorization_Dependencies): Add more detail to
error msgs for most common cases.
Use new errout insertion char < (conditional warning)
Thomas Quinot [Mon, 5 Sep 2005 07:58:05 +0000 (09:58 +0200)]
re PR ada/9087 (Component_Size for packed array returns wrong values)
2005-09-01 Thomas Quinot <quinot@adacore.com>
Ed Schonberg <schonberg@adacore.com>
Robert Dewar <dewar@adacore.com>
* sem_attr.adb (Resolve_Attribute, case 'Address): For an illegal
'Address attribute reference with an overloaded prefix, use the
location of the prefix (not the location of the attribute reference) as
the error location.
(Analyze_Attribute, case 'Size): The name of an enumeration literal, or
a function renaming thereof, is a valid prefix for 'Size (where it is
intepreted as a function call).
(Statically_Denotes_Entity): New predicate to determine whether the
prefix of an array attribute can be considered static.
PR ada/9087
(Eval_Attr): Fix failure to evaluate Component_Size for
unconstrained arrays (resulted in wrong value in packed case, since
back end cannot handle this case)
Robert Dewar [Mon, 5 Sep 2005 07:56:02 +0000 (09:56 +0200)]
opt.ads, opt.adb: Add new switches Debug_Pragmas_Enabled[_Config]
2005-09-01 Robert Dewar <dewar@adacore.com>
* opt.ads, opt.adb: Add new switches Debug_Pragmas_Enabled[_Config]
* par-prag.adb: Implement new pragma Debug_Policy
* sem_prag.adb Implement new pragma Debug_Policy
(Analyze_Pragma, case Pack): do not let pragma Pack override an explicit
Component_Size attribute specification. Give warning for ignored pragma
Pack.
* snames.h, snames.ads, snames.adb: Introduce entries in
Preset_Names for Name_Disp_Asynchronous_Select,
Name_Disp_Conditional_Select, Name_Disp_Get_Prim_Op_Kind,
Name_Disp_Timed_Select.
New pragma Debug_Policy
* switch-c.adb (Scan_Front_End_Switches): Set Ada 2005 mode
explicitly.
Switch -gnata also sets Debug_Pragmas_Enabled
* sem.adb, par.adb (Set_Opt_Config_Switch): Add parameter Main_Unit to
handle an explicit -gnata when compiling predefined files.
Vincent Celier [Mon, 5 Sep 2005 07:55:17 +0000 (09:55 +0200)]
lib-writ.adb: Update Copyright notice
2005-09-01 Vincent Celier <celier@adacore.com>
* lib-writ.adb: Update Copyright notice
(Write_With_Lines): On platforms where file names are case-insensitive,
record the file names in lower case.
(Write_ALI): For D lines, on platforms where file names are
case-insensitive, record the file names in lower case.