CVS difference for ai12s/ai12-0262-1.txt

Differences between 1.3 and version 1.4
Log of other versions for file ai12s/ai12-0262-1.txt

--- ai12s/ai12-0262-1.txt	2018/10/19 02:34:09	1.3
+++ ai12s/ai12-0262-1.txt	2018/10/19 05:13:38	1.4
@@ -434,7 +434,7 @@
 
 ****************************************************************
 
-From: Brad Moore
+From: Randy Brukardt
 Sent: Thursday, October 18, 2018   9:35 PM
 
 One minor issue:
@@ -455,5 +455,67 @@
 that. (Hope I got it right.) It should be easier to change it back (replace all
 of the square brackets with parens) if we decide that than it would be to change
 it on the fly to square brackets.
+
+****************************************************************
+
+From: Brad Moore
+Sent: Thursday, October 18, 2018  10:16 PM
+
+> This doesn't make any sense (3 curly brackets and one square
+> bracket??); the previous paragraph says
+>
+>>For a reduction_expression where the {identifier of the
+> attribute_designator} [attribute_reference] is Reduce,
+>
+> which is what I expected here; so I changed it accordingly.
+
+Thanks for doing what I meant, and not what I said. :-)
+
+> The other issue is that given the change to AI12-0212-1 to use square
+> brackets, this ought to use those as well. So I changed the syntax and
+> examples to do that. (Hope I got it right.) It should be easier to
+> change it back (replace all of the square brackets with parens) if we
+> decide that than it would be change it on the fly.
+
+I saw the square brackets in Tuckers latest version of his AI, which I think
+makes perfect sense in his AI. But I intentionally did not incorporate them in
+this AI. To me, the square brackets suggest an aggregate value for an array or
+container object is being created, and it is beneficial to use them for that
+purpose. In this AI, there is no aggregate, just an iterator/generator, so it
+made sense to me to not have them, since semantically, the construct is very
+different.
+
+Maybe I need to have my view expanded, but at this point, having square brackets
+in the value_generator syntax doesn't make sense to me. Basically, I dont see
+the benefit for it here. Maybe someone could enlighten me about why this would
+be a good thing?
+
+****************************************************************
+
+From: Tucker Taft
+Sent: Thursday, October 18, 2018  10:33 PM
+
+In my mind, it is definitely a good thing to use [...] here, if we believe we
+are generally going to be using [...] for constructs representing a homogeneous
+collection, as opposed to (...) which is retained for a heterogeneous collection
+like a record, where each element potentially has a different type.
+
+Or more fundamentally, this notation is producing a sequence, just like an array
+or a vector, but then it is immediately consuming the sequence.  It would be
+weird in my view to shift the notation just because the sequence only exists
+temporarily.
+
+****************************************************************
+
+From: Randy Brukardt
+Sent: Friday, October 19, 2018  12:13 PM
+
+What he said.
+
+Right now, the square brackets look weird to us 'cause we haven't been using
+them for 40 years. But once we start using them, they'll quickly become natural,
+and it would be bizarre to use () in one place and [] in another for what is
+fundamentally the same thing. (The "oddity" is the language-defined arrays, they
+should look like just another container. Little late for that, of course.)
 
 ****************************************************************

Questions? Ask the ACAA Technical Agent