From: William Kempf (williamkempf_at_[hidden])
Date: 2002-08-11 20:09:08
>From: "David Bergman" <davidb_at_[hidden]>
>That was very flexible of you: "I refuse to ..." ;-)
Sorry, but this one isn't really negotiable for so many reasons.
>I hope you see Eric's point that the "single thread" rules, (inherited
>from the C main entry convention) regarding C++ and main, are not
>automatically applicable to the MT C++ we are dealing with.
Not really, and more importantly I don't see how it would be relevant to
using a main() wrapper any way.
>You surely must appreciate the fact that we leave the safe world of C++
>when we add this, OS-specific, thread dimension, i.e., we are not
>dealing with the original semantics of C++, as defined in those 700+
Yes, I understand that. Thus I *have* to abide by the rules of the
underlying thread library when I can (and as C libraries they are mute on
this subject) and by the C++ standard when I can't.
>This does not mean that we should not follow that intention as closely
>as possible, but we should have the right to question certain implicit
>mappings of terms made here, or?
You can question it when you are defining the standard, not when building a
library such as Boost.Threads. There all that you can do is, again, to
follow the rules of the underlying thread libraries when they are defined,
and the standard when they are not.
>I truly hope that this forum will be open for diverse interpretations of
>how the C++ standard should be applied to MT C++, at least till such a
>standard emerges (in, say, six years).
Only when you can do so, which means the underlying libraries must define
the behavior. If they do not, then I must obey the defined behavior of the
Send and receive Hotmail on your mobile device: http://mobile.msn.com
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk