Re: Comments on the thread library

From: williamkempf_at_[hidden]
Date: 2001-03-27 14:59:52

--- In boost_at_y..., Jeremy Siek <jsiek_at_l...> wrote:
> On Mon, 26 Mar 2001 williamkempf_at_h... wrote:
> willia> This illustrates one of the big reasons why I'm not sure I 
care for a 
> willia> generative design for Boost.Threads.  Acquisition 
scheduling is 
> willia> nearly always going to be defined by the underlying 
implementation, 
> willia> so exposing an interface to specify this is misleading at 
best.  
> willia> You're not going to be able to specify this in a portable 
manner.  I 
> willia> designed the semantics of the Boost.Threads mutex types 
very 
> willia> carefully, taking factors like this into account.
> 
> But why punish those platforms that do support acquisition 
scheduling? It
> seems to me that it is acceptable for Boost.Threads to support more 
things
> on some platforms and less on others. We just have to carefully 
document
> what is supported where, so users can know what kind of portability 
to
> expect.

No one's being "punished".  The interface is simple and can be used 
for implementation of platform specific types.  IMHO this is a better 
approach than having a portable library provide non-portable types.

Bill Kempf