From: Arnaud Charlet <charlet@adacore.com>
To: gcc-patches@gcc.gnu.org
Cc: Ed Schonberg <schonberg@adacore.com>
Subject: [Ada] Attributes of arrays with an index type derived from a formal type
Date: Wed, 23 Jun 2010 07:27:00 -0000 [thread overview]
Message-ID: <20100623062644.GA29656@adacore.com> (raw)
[-- Attachment #1: Type: text/plain, Size: 1266 bytes --]
The attributes of formal discrete types, and of arrays whose indices are
formal types, are not static and cannot be constant-folded. The same is true
if an index type is derived from a formal type. The check for types derived
from formal types was missing, leading to improper static evaluation..
The following must compile and execute quietly:
with G1;
procedure Q is
type Indx is range 1 .. 10;
package Inst is new G1 (Indx);
use Inst;
Obj : R := ((others => 0), 10);
begin
if not Is_Full (Obj) then
raise Program_Error;
end if;
end;
---
generic
type Ix is range <>;
package G1 is
type Iterator is new Ix;
type Arr is array (Iterator) of integer;
type R is record
Nodes : Arr;
Count : Positive;
end record;
function Is_Full (It : R) return Boolean;
end G1;
---
package body G1 is
function Is_Full (It : R) return Boolean is
begin
return It.Count = It.Nodes'Length;
end Is_Full;
end G1;
Tested on x86_64-pc-linux-gnu, committed on trunk
2010-06-23 Ed Schonberg <schonberg@adacore.com>
* sem_attr.adb (Eval_Attribute): If the prefix is an array, the
attribute cannot be constant-folded if an index type is a formal type,
or is derived from one.
* checks.adb (Determine_Range): ditto.
[-- Attachment #2: difs --]
[-- Type: text/plain, Size: 2239 bytes --]
Index: checks.adb
===================================================================
--- checks.adb (revision 161078)
+++ checks.adb (working copy)
@@ -3351,6 +3351,14 @@ package body Checks is
Indx := Next_Index (Indx);
end loop;
+ -- if The index type is a formal type, or derived from
+ -- one, the bounds are not static.
+
+ if Is_Generic_Type (Root_Type (Etype (Indx))) then
+ OK := False;
+ return;
+ end if;
+
Determine_Range
(Type_Low_Bound (Etype (Indx)), OK1, LL, LU,
Assume_Valid);
Index: sem_attr.adb
===================================================================
--- sem_attr.adb (revision 161200)
+++ sem_attr.adb (working copy)
@@ -5633,10 +5633,10 @@ package body Sem_Attr is
while Present (N) loop
Static := Static and then Is_Static_Subtype (Etype (N));
- -- If however the index type is generic, attributes cannot
- -- be folded.
+ -- If however the index type is generic, or derived from
+ -- one, attributes cannot be folded.
- if Is_Generic_Type (Etype (N))
+ if Is_Generic_Type (Root_Type (Etype (N)))
and then Id /= Attribute_Component_Size
then
return;
@@ -6205,13 +6205,13 @@ package body Sem_Attr is
Ind : Node_Id;
begin
- -- In the case of a generic index type, the bounds may appear static
- -- but the computation is not meaningful in this case, and may
- -- generate a spurious warning.
+ -- If any index type is a formal type, or derived from one, the
+ -- bounds are not static. Treating them as static can produce
+ -- spurious warnings or improper constant folding.
Ind := First_Index (P_Type);
while Present (Ind) loop
- if Is_Generic_Type (Etype (Ind)) then
+ if Is_Generic_Type (Root_Type (Etype (Ind))) then
return;
end if;
reply other threads:[~2010-06-23 6:26 UTC|newest]
Thread overview: [no followups] expand[flat|nested] mbox.gz Atom feed
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20100623062644.GA29656@adacore.com \
--to=charlet@adacore.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=schonberg@adacore.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).