Re: [boost] Predictable Layout for certain non-POD data types

From: Gabriel Dos Reis (Gabriel.Dos-Reis_at_[hidden])
Date: 2002-03-17 13:40:19

"David Abrahams" <david.abrahams_at_[hidden]> writes:

| Ralf and I have authored this paper for the upcoming C++ committee
| meeting
| 
| 
| http://cci.lbl.gov/~rwgk/tmp/enhanced_pod_proposal.html

Ralf asked me to make some comments about that document.

Dave and Ralf have made good job and I have virtually no comments!

However,

* I'm uncomfortable with the sentence:

  For example, some people interpret the standard as saying a polar
  internal representation might be legal. 

The standard text 26.2.2/1 is clear.  However, that doesn't invalidate
your point about complex<> layout; but I don't think it serves it and
it tends to lower your concern.

* I would replace 

  A predictable data layout or direct access through references is
  also a prerequisite to enabling essential speed optimizations

with

  A predictable data layout and direct access to individual components
  through modifiable lvalues is also a prerequisite to enabling
  essential speed optimizations 

Again, nice job!

-- Gaby