|
Boost : |
From: Beman Dawes (bdawes_at_[hidden])
Date: 2001-06-08 14:45:48
At 10:51 AM 6/8/2001, williamkempf_at_[hidden] wrote:
>The thread class in particular needs very careful scrutiny. There
>are several tricky issues that need to be considered and accounted
>for in the interface for a full thread concept. I'll list the main
>ones that I can think of:
>
> ...
Some of the issues come down to "Design and implement now, or wait until
another revision?"
But others are tougher because it isn't known how to design the feature
such that it would be widely portable. You mention thread priority, for
example.
For some features (thread priority?), it may be possible to define them as
"hints" or otherwise just stub out the implementation. But that only works
for the few features that don't in any way impact program correctness.
But in general Boost.Threads early releases should avoid non-portable or
optional features, IMO. The standard for measuring portability should be
Win32 and Pthreads. If it can't be ported to those environments, avoid the
feature for now.
--Beman
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk