CVS difference for ais/ai-00229.txt

Differences between 1.13 and version 1.14
Log of other versions for file ais/ai-00229.txt

--- ais/ai-00229.txt	2002/01/24 04:54:11	1.13
+++ ais/ai-00229.txt	2002/03/13 00:48:27	1.14
@@ -1104,3 +1104,94 @@
 want to cover).
 
 *************************************************************
+
+From: Dan Eilers
+Sent: Thursday, February 28, 2002  1:02 AM
+
+One of the source files from gtkada appears to demonstrate the
+compatibility problem created by AI95-00229.
+
+GtkAda-1.2.12/src/gtk-marshallers.adb, line 79:
+    return (Func  => To_General_Handler (Cb), Proxy => Call'Access);
+
+where Call'Access is of type Handler_Proxy, which is declared
+outside the current generic unit (in an enclosing generic unit).
+
+*************************************************************
+
+From: Tucker Taft
+Sent: Friday, March 1, 2002  8:01 AM
+
+Could the expression "Call'Access" be moved up to the private part
+of the generic?  That would hopefully provide an adequate workaround.
+
+*************************************************************
+
+From: Tucker Taft
+Sent: Monday, March 4, 2002  11:20 PM
+
+> I'm not sure.
+> Here's the source code.
+
+Yes, it appears that it would be possible to declare
+
+    Call_Access : constant Handler_Proxy := Call'Access;
+
+immediately after the declaration of each "Call" function
+in each private part, and then use that instead of "Call'Access"
+in the corresponding package body.
+
+*************************************************************
+
+From: Tucker Taft
+Sent: Monday, March 4, 2002  11:40 PM
+
+One thing that seems surprising when I look at the
+wording of AI-00229 is that the following
+is now *legal* whereas before it was illegal:
+
+   generic
+   package GP is
+     pragma Elaborate_Body;
+     type S is access procedure;
+
+     X : S;
+   end GP;
+
+   package body GP is
+     procedure In_Body is begin
+         null;
+     end;
+   begin
+     X := In_Body'Access;
+   end GP;
+
+Was that intentional?  Is there any implementation
+problem for shared generics?
+
+*************************************************************
+
+From: Pascal Leroy
+Sent: Tuesday, March 5, 2002  3:52 AM
+
+This change was suggested at the Leuven meeting by one Tucker Taft.  At the
+time there was concern that it would cause trouble with shared generics, so the
+idea was not explored further during the meeting.  A few weeks later there was
+a flurry of email to fix problems with the wording of the AI, and Randy raised
+the issue of whether we needed to restrict S to be in the generic body; his
+conclusion was that this was not necessary.  The same Tucker Taft came up with
+a wording that nicely unified the last two sentences of the paragraph, and
+resulted in the change above.  This new wording was then approved at the
+Bloomington meeting.
+
+Look at the appendix of the AI for email dated 1-5 June 2001.
+
+*************************************************************
+
+From: Tucker Taft
+Sent: Tuesday, March 5, 2002  4:13 AM
+
+Good to know that someone on this mailing list is awake ;-).
+
+*************************************************************
+

Questions? Ask the ACAA Technical Agent