Boost logo

Boost :

From: William Kempf (sirwillard_at_[hidden])
Date: 2000-10-17 09:24:44


--- In boost_at_[hidden], "Moore, Paul" <paul.moore_at_u...> wrote:
> From: Peter Dimov [mailto:pdimov_at_m...]
> I see a fundamental distinction between graph and regex, and py_cpp
and
> (say) threading or GUI libraries. It is that the former two have no
> dependency on the platform (OS, other language libraries, etc etc).
Whether
> this pushes graph and regex down a layer, or adds a fourth layer, I
don't
> know.

I understand the sentiment, and it's consistent with historical
precedent, but IMHO threads are not something that any language can
ignore any more. This includes C++. Concurrent processing has
become mandatory for a very large range of problems, and desirable
for an even larger range. Only on embedded systems would inclusion
of thread primitives be considered "too much"... and for them even
iostreams goes to far.

Lack of standardized language support for threads has lead to a huge
base of code being unportable and even unstable. Advances in thread
safety is hindered by each researcher "speaking a different
language", even if they are programming using the same language, such
as C++.

From what I've heard from members of the C++ comittee, they are more
than open to inclusion of threads in the C++ language, but are
concerned with portability, efficiency and safety. So I don't think
that threads fall in a category of "beyond the scope of future C++
standards", and strongly feel that in fact, the C++ language _must_
consider the concept if it is to survive as a language.

Bill Kempf


Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk