CVS difference for ai05s/ai05-0125-1.txt

Differences between 1.11 and version 1.12
Log of other versions for file ai05s/ai05-0125-1.txt

--- ai05s/ai05-0125-1.txt	2011/04/01 04:07:22	1.11
+++ ai05s/ai05-0125-1.txt	2011/04/02 07:23:27	1.12
@@ -2,6 +2,7 @@
 !standard  8.3(12.3/2)
 !standard  8.3.1(5/2)
 !class Amendment 08-10-20
+!status No Action (7-0-1) 11-03-17
 !status work item 08-10-20
 !status received 08-10-03
 !priority Low
@@ -966,5 +967,33 @@
 "model" changes.  Typically "observer" is an interface with just a couple of
 operations.  You "register" yourself as an "observer" and wait to be called
+[Excerpt from minutes of March 17, 2011 phone meeting.]
+Tucker explains his position. This is the case of giving a declaration where the
+"grandparent" operation is not inherited, but you have visibility on it.
+The current situation is that if you give such an operation with no overriding
+indicator, you get a new operation. If you say "overriding", you get an error.
+The change if nothing is said, you get an error; if you say "overriding" it
+works.The problem with interfaces is that you get a preference rule; and makes
+interfaces funny.
+We thought about making these illegal, because they are effectively privacy
+breaking. But that causes a ripple effect - anything declared in the private
+part could break existing code.
+Tucker thinks this one is rapidly approaching the "brick wall of complexity".
+Randy says that his opinion on this one flops around like a fish caught in a
+gill net.
+So long as users actually use overriding indicators, they will at least be aware
+of the problem. (That was the main driving force for adding indicators in the
+first place.) They'll get an error if they expect overriding but don't get it.
+It might require heavy restructuring (which is bad), but there will not be
+surprises. So this is not worth crashing into a wall of complexity.

Questions? Ask the ACAA Technical Agent