From: Gregory Colvin (gregory.colvin_at_[hidden])
Date: 2003-05-02 10:51:53
On Friday, May 2, 2003, at 09:18 America/Denver, Joel de Guzman wrote:
> Noel Yap wrote:
>> Joel de Guzman wrote:
>>>> But I don't think that a lack of global usage
>>>> guaruntees hinders the potential usefulness of the idea.
>>> Why? If it's not applied globally, it would be like installing some
>>> security alarm system only in some places in your house. How useful
>>> is that?
>>> There would still be uncertainty.
>> IIRC, at $70.00/window and possibly over a couple of dozen windows,
>> most don't use the security system for every possible entrance to
>> house, at least not all at once. How did you go about doing it?
> I don't. No offense meant, but I think such incomplete security systems
> are useless. Wasn't that my point?
Depends on how accessible the unsecured windows are.
As I understand it, if the author of an interface imposes the use of
decorations then the users of that interface must comply, and thus
certain mistakes, which could translate into fewer bug reports to that
Whether this is a net win depends on a lot of factors, but my take is
might be helpful for interfaces like COM that use a lot of out and
parameters. How helpful, I don't know, though when I did COM
don't recall this being a big issue.
So, I don't care much whether this goes in Boost, but would find it more
compelling as part of larger library that takes good advantage of it,
than as a standalone.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk