"Minutes" of today's meeting

[Jim Dehnert]

The meeting today had low attendance, so we did not do anything
momentous.  I apologize for the rather late reminder.  To make up
for it, I'm giving you a reminder now that we will meet in two weeks,
on 6 July.  I expect the primary subject of discussion to be the
mangling suggestions Alex sent today -- since both he and HP, at least,
are well into implementation, we should be able to resolve the issues
with some substantial implementation experience to guide us.

Today, we discussed Mark's example of a virtual function in a primary
virtual base class A, which therefore does not need a thunk, where A
then becomes non-primary in a more-derived class D, requiring the thunk,
even though D does not override the function.  We concluded that the
thunk must be associated with the D definition (the need is not known
with the A definition), and since that may occur in multiple files, as
well as occurring for multiple possible D's, it must have vague linkage
(COMDAT).  I will update the draft accordingly.

Otherwise, we went through the draft ABI changes from last time.  We
identified no significant problems, so if you haven't looked it over,
please do so and comment if necessary.  The red highlighting will
disappear early next week (to be replaced by the next set, of course).

Jim

-		Jim Dehnert  x3-4272