Re: [boost] RFC: Multithreading design constraints

From: David Abrahams (abrahams_at_[hidden])
Date: 2000-06-09 20:20:25

I think that sort of thing is normally done with separate executable/lib
objects, each of which is built with a _compile_time_ decision about SMP
support. The appropriate object is dynamically selected at runtime. If it's
all compiled together you need to choose different link names for
SMP/non-SMP and the whole system becomes a lot more complicated.
-Dave

----- Original Message -----
From: "Peter Dimov" <pdimov_at_[hidden]>
To: <boost_at_[hidden]>
Sent: Friday, June 09, 2000 5:53 PM
Subject: Re: [boost] RFC: Multithreading design constraints

> > > But SMP vs non-SMP is not a compile time decision, is it?
>
> > It could be. You could dynamically load a different version of the code
on
> > SMP systems, for example.
>
> The key word here being "dynamically." :)
>
> I meant that the SMP vs non-SMP decision can be deferred until runtime,
with
> a single library containing both code paths.
>
> I admit that there are (rare?) cases where you know beforehand that the
> target platform is not SMP.
>
> --
> Peter Dimov
> Multi Media Ltd.
>
>
> ------------------------------------------------------------------------
> Take your development to new heights. Work with clients like Dell and
> pcOrder. Submit your resume to jobs_at_liaison.com. Visit us at
> http://click.egroups.com/1/4358/3/_/9351/_/960589885/
> ------------------------------------------------------------------------
>
>