CVS difference for ai05s/ai05-0139-2.txt

--- ai05s/ai05-0139-2.txt	2011/06/10 05:58:44	1.18
+++ ai05s/ai05-0139-2.txt	2011/06/20 04:55:17	1.19
@@ -840,14 +840,14 @@
 !corrigendum 4.1(2)
 
 @drepl
-@xcode<@fa<name> ::= 
+@xcode<@fa<name> ::=
      @fa<direct_name> | @fa<explicit_dereference>
    | @fa<indexed_component> | @fa<slice>
    | @fa<selected_component> | @fa<attribute_reference>
    | @fa<type_conversion> | @fa<function_call>
    | @fa<character_literal>>
 @dby
-@xcode<@fa<name> ::= 
+@xcode<@fa<name> ::=
      @fa<direct_name> | @fa<explicit_dereference>
    | @fa<indexed_component> | @fa<slice>
    | @fa<selected_component> | @fa<attribute_reference>
@@ -1055,13 +1055,13 @@
 
 @xcode<   @b<type> Forward_Iterator @b<is limited interface>;
    @b<function> First (Object : Forward_Iterator) @b<return> Cursor @b<is abstract>;
-   @b<function> Next (Object : Forward_Iterator; Position : Cursor) @b<return> Cursor
-      @b<is abstract>;>
+   @b<function> Next (Object : Forward_Iterator; Position : Cursor)
+      @b<return> Cursor @b<is abstract>;>
 
 @xcode<   type Reversible_Iterator @b<is limited interface and> Forward_Iterator;
    @b<function> Last (Object : Reversible_Iterator) @b<return> Cursor @b<is abstract>;
-   @b<function> Previous (Object : Reversible_Iterator; Position : Cursor) @b<return> Cursor
-      @b<is abstract>;>
+   @b<function> Previous (Object : Reversible_Iterator; Position : Cursor)
+      @b<return> Cursor @b<is abstract>;>
 
 @xcode<@b<end> Ada.Iterator_Interfaces;>
 
@@ -4531,6 +4531,28 @@
 
 ****************************************************************
 
+From: Bob Duff
+Sent: Friday, June 10, 2011  6:09 AM
+
+> I note that 4.1.5 (User-defined References) doesn't say anything about
+> "user-defined" in the similar text. So I conclude this is just a mistake.
+> Right??
+
+Right.
+
+****************************************************************
+
+From: Tucker Taft
+Sent: Friday, June 10, 2011  7:52 AM
+
+Your fix sounds fine.  I think "user-defined" is as opposed to "predefined," not
+"language-defined."  Even language-defined packages can have "user-defined
+primitive subprograms" overriding "predefined" operators.  But if you think it
+is creating confusion, or simply that it doesn't add anything, then feel free to
+get rid of it.
+
+****************************************************************
+
 From: Randy Brukardt
 Sent: Thursday, June 9, 2011 11:13 PM
 
@@ -4559,5 +4581,49 @@
 close together.
 
 Perhaps this can be worded more elegantly.
+
+****************************************************************
+
+From: Tucker Taft
+Sent: Friday, June 10, 2011  7:53 AM
+
+Your fix sounds fine, though I have a suspicion that if you include the larger
+context, it may be redundant.
+
+****************************************************************
+
+From: Bob Duff
+Sent: Friday, June 10, 2011  8:16 AM
+
+> So I replaced this paragraph by the following:
+>
+>    In a container element iterator whose /iterable_/name has type T,
+>    if the /iterable_/name denotes a constant or the Variable_Indexing
+>    aspect is not specified for T, then the Constant_Indexing aspect
+>    shall be specified for T.
+
+How about this:
+
+   In an iterator whose /iterable_/name has type T, the Variable_Indexing
+   or Constant_Indexing aspect (or both) shall be specified for T.
+   If the /iterable_/name denotes a constant, the Constant_Indexing aspect
+   shall be specified for T.
+
+****************************************************************
+
+From: Tucker Taft
+Sent: Friday, June 10, 2011  8:35 AM
+
+To be more precise, the expected type for a container element iterator is "any
+iterable type" and from 5.5.1:
+
+   An /iterable type/ is an indexable type ...
+
+and from 5.5:
+
+   An /indexable type/ is (a view of) a user-defined tagged type with at least
+   one of the aspects Constant_Indexing or Variable_Indexing specified.
+
+So I believe your change is redundant.
 
 ****************************************************************