CVS difference for ai05s/ai05-0018-1.txt

Differences between 1.2 and version 1.3
Log of other versions for file ai05s/ai05-0018-1.txt

--- ai05s/ai05-0018-1.txt	2007/04/06 04:37:11	1.2
+++ ai05s/ai05-0018-1.txt	2007/07/26 02:58:05	1.3
@@ -1,6 +1,7 @@
-!standard 4.9(38/2)                                    07-04-05    AI05-0018-1/02
+!standard 4.9(38/2)                                    07-06-16    AI05-0018-1/03
 !standard 12.7(6/2)
-!class binding interpretation 06-11-09
+!class pathology 07-06-02
+!status No Action (8-1-1) 07-06-03
 !status work item 06-11-09
 !status received 06-09-15
 !priority Medium
@@ -792,6 +793,44 @@
 property that named numbers are essentially the same as literals.
 
 (Named numbers are not one of my favorite features, by the way.)
+
+****************************************************************
+
+From minutes of the 32nd ARG meeting (Paris, France)
+
+Steve claims that the RM says that the two values are not equal, and he changed the
+matching rules to make that work. Erhard and Ed wonder why this doesn't occur in
+many other cases? For instance, in static matching of range constraints. Steve
+claims that the problem can only occur when the types are different.
+
+Erhard complains that 3.14 and 3.14000000000000001 would match using this rule. That's ugly.
+
+Jean-Pierre complains that the legality of a program would depend on your target.
+Pascal says that is already the case, for instance if you use a float compare as
+a when branch in a case statement.
+
+Tucker wonders why his proposal (from last time) was rejected (that is, the value is
+always exact until used). Steve says that it would be an earthquake in some
+implementations (his).
+
+Randy agrees with Tucker. Tucker says that a named constant of a type would be
+rounded (because this is effectively an assignment); but other cases would not
+be rounded until use.
+
+Pascal says that they round right away, so it would be an earthquake to fix. Tucker
+thinks this is a single special case, and should be fixed.
+
+Erhard says that he doesn't want a large paragraph in the manual to fix this
+Baird pathology. Pascal says that he is willing to sweep this under the rug.
+
+Tucker suggests that we could just say "prior to rounding or truncation". Pascal
+would prefer to vote it No Action as just a pathology.
+
+Tucker suggests putting these minutes into the !appendix of this AI.
+
+Reclassify this as pathology, and vote it No Action: 8-1-1. Bibb opposes since he
+is not convinced; while this case is a pathology, he worries that there are other
+related cases which are not.
 
 ****************************************************************
 

Questions? Ask the ACAA Technical Agent