2011-08-02 Robert Dewar <dewar@adacore.com>
* exp_ch4.adb: Minor reformatting.
2011-08-02 Ed Schonberg <schonberg@adacore.com>
* sem_ch5.adb (Analyze_Loop_Statement): If the iteration scheme is an
Ada2012 iterator, the loop will be rewritten during expansion into a
while loop with a cursor and an element declaration. Do not analyze the
body in this case, because if the container is for indefinite types the
actual subtype of the elements will only be determined when the cursor
declaration is analyzed.
2011-08-02 Arnaud Charlet <charlet@adacore.com>
* sem_ch13.adb (Analyze_Attribute_Definition_Clause): Ignore
size/alignment related attributes in CodePeer_Mode.
2011-08-02 Gary Dismukes <dismukes@adacore.com>
* sem_ch3.adb (Check_Ops_From_Incomplete_Type): Remove call to
Prepend_Element, since this can result in the operation getting the
wrong slot in the full type's dispatch table if the full type has
inherited operations. The incomplete type's operation will get added
to the proper position in the full type's primitives
list later in Sem_Disp.Check_Operation_From_Incomplete_Type.
(Process_Incomplete_Dependents): Add Is_Primitive test when checking for
dispatching operations, since there are cases where nonprimitive
subprograms can get added to the list of incomplete dependents (such
as subprograms in nested packages).
* sem_ch6.adb (Process_Formals): First, remove test for being in a
private part when determining whether to add a primitive with a
parameter of a tagged incomplete type to the Private_Dependents list.
Such primitives can also occur in the visible part, and should not have
been excluded from being private dependents.
* sem_ch7.adb (Uninstall_Declarations): When checking the rule of
RM05-3.10.1(9.3/2), test that a subprogram in the Private_Dependents
list of a Taft-amendment incomplete type is a primitive before issuing
an error that the full type must appear in the same unit. There are
cases where nonprimitives can be in the list (such as subprograms in
nested packages).
* sem_disp.adb (Derives_From): Use correct condition for checking that
a formal's type is derived from the type of the corresponding formal in
the parent subprogram (the condition was completely wrong). Add
checking that was missing for controlling result types being derived
from the result type of the parent operation.
git-svn-id: svn+ssh://gcc.gnu.org/svn/gcc/trunk@177156
138bc75d-0d04-0410-961f-
82ee72b054a4