CVS difference for ai12s/ai12-0112-1.txt

--- ai12s/ai12-0112-1.txt	2018/07/28 00:35:54	1.7
+++ ai12s/ai12-0112-1.txt	2018/12/07 07:01:27	1.8
@@ -1995,3 +1995,43 @@
 
 ****************************************************************
 
+From: Randy Brukardt
+Sent: Thursday, December 6, 2018  8:44 PM
+
+I've been thinking about the job of putting these contracts into the Standard.
+I've concluded that writing a full AI for this would just double the work, as
+I'd have to create an AI version of the changes, and then put the same changes
+into RM. While this might be good for my bank account (and that assumes an 
+unlimited willingness of AdaCore to put in more money -- something that does 
+not likely exist), it seems silly.
+
+My thinking is that simply putting the changes directly into the RM, with 
+proper change marks, would be likely more useful for human reviewers than any
+amount of time spent on the actual AI. Moreover, for machine testing, it is
+relatively easy to extract the required specifications from the RM (it would 
+be easy to write a tool to process the HTML to do that, including removing 
+the paragraph numbers which would prevent directly cut-and-pasting). Machine 
+testing seems mandated, although having a compiler that could ignore 
+Nonblocking and Global aspects would seem very useful (else having a tool to 
+strip those would be required - doing it by hand would take ages).
+
+However, in order to follow this approach, I'll need essentially a "blank 
+check" to make such changes as described in AI12-0112-1. Once these changes
+are inserted in the RM (at the same time as changes for AIs 212, 111, and
+266) would pretty much mean that the work to back them out would be equivalent
+to the original work. (Summary: not gonna happen.)
+
+Is this a reasonable approach? Should I put this on the agenda for discussion 
+on Monday???
+
+****************************************************************
+
+From: Tucker Taft
+Sent: Thursday, December 6, 2018  9:35 PM
+
+Makes sense to me.  It would be nice to do this during a "lull" in making 
+other changes to the RM, so we could review the final product before signing 
+off on it.
+
+****************************************************************
+