Re: Moving sp_thread_pause, sp_thread_sleep to Core

From: Peter Dimov (pdimov_at_[hidden])
Date: 2023-06-01 14:47:30

Andrey Semashev wrote:
> On 6/1/23 15:28, Peter Dimov via Boost wrote:
> > The motivation here is that Unordered has now acquired a dependency
> > on SmartPtr, just for those primitives, which is seen as undesirable.
> >
> > Putting them somewhere else, such as Thread, solves nothing, as a
> > dependency on Thread for those is much worse.
> >
> > The alternative to moving them to Core is Unordered containing its own
> > local copy.
> 
> That everyone depends on Core doesn't mean we should pile everything in it.

Moving an existing implementation detail of library A into Core because
library B wants to use it doesn't feel wrong to me.

We don't have a better place for things like "a portable implementation of
__builtin_pause() if such a thing existed".