updates

git-svn-id: https://svn.r-project.org/R-dev-web/trunk@2146 c52295ea-58df-0310-926a-d16021944841

Author: ripley

BDR_S4issues.txt

@@ -22,6 +22,9 @@ especially of primitives.
    group generic but log2 is not.  This is probably all history, but
    not at all obvious to end users.
 
+   [Update 2007-07-24: the groups have been harmonized and all of
+   Math is now primitive.  The help pages have been corrected.]
+
 
 2) Members of the Summary group have documented (and S3) arglist (...,
    na.rm = FALSE) and S4 arglist (x, ..., na.rm = FALSE).
@@ -31,6 +34,7 @@ especially of primitives.
 
    Is this necessary?
 
+   [This is being addressed via implicit generics.]
 
 3) trunc() is documented as a member of the S4 Math group (which has
    only one argument), but has been implemented with an implicit
@@ -51,6 +55,9 @@ especially of primitives.
    In R-devel the current definition is trunc(x, ...) and S3 dispatch
    is on the first argument.
 
+   [Now trunc() is a member of the S4 Math group and S4 methods can be
+   set on trunc(x, ...).  The S4 Math group cannot dispatch methods
+   for the extra arguments of log() and trunc().]
 
 4) S3 method dispatch on primitives tries hard to emulate a call to a
    closure, by setting up promises for the unevaluated and evaluated
@@ -63,6 +70,8 @@ especially of primitives.
 
     Why is this thought not to be needed for S4 dispatch?
 
+    [It was necessary: I found an example in the wild where
+    duplication did not happen, and have added code to do this.]
 
 5) Does an S4 object have a class and is the object bit set?
 
@@ -200,4 +209,7 @@ especially of primitives.
      namespace.
 
 
+     [exp and log are now both primitive and so do now behave the same.]
+
+
 BDR 2007-04-30