Boost logo

Boost :

From: Moore, Dave (dmoore_at_[hidden])
Date: 2003-12-17 07:58:56


Just to chime in, from back when I had time to contribute to the thread_dev
branch (rw_mutex, thread_pool, and barrier), I recall emailing with Bill
about this. The dynamic linking requirement exists for TLS resource
cleanup.

There are surely extensive discussions of this in the mailing list archives,
but IIRC, one -could- come up with a cleanup callback type scheme which
would work to deallocate the OS TLS resources in the case of a simple
application statically linking to the thread library, but a more complex
application architecture like an app using a DLL which statically linked to
Boost.Threads could easily bypass this cleanup (inadvertently).

A static library just for mutex, thread, and condition should be fine. It's
TLS and call_once that need the dynamic (un)loading hooks for cleanup. Some
of the experimental thread identification code may also have used TLS, but
that could be backed out. The details should be in the archives.

Regards,
Dave

> -----Original Message-----
> From: Russell Hind [mailto:rhind_at_[hidden]]
> Sent: Wednesday, December 17, 2003 7:21 AM
> To: boost_at_[hidden]
> Subject: [boost] Re: Moving Boost.Thread forward
>
>
> Michael Glassford wrote:
> >
> > I did get them and will look at them more closely when I
> get the chance.
> > Along with you and several others, I too would be interested in a
> > statically linked version of Boost.Thread (even if that requires
> > eliminating some features in the statically linked
> version), so I am
> > quite interested in looking at your suggestion about this.
> >
>
> I would be worth searching the archives for this, as it was discussed
> numerous times, at length IIRC, by Bill and others as to why this
> shouldn't be done.
>
> If it is just to do with TLS, then, as you suggest, these
> features could
> be removed from the static version (I build it statically myself, and
> only use thread, condition and mutex) but I would be careful
> about this
> move because there have obviously been very good reasons for
> not doing
> it so far.
>
> Thanks
>
> Russell
>
> _______________________________________________
> Unsubscribe & other changes:
http://lists.boost.org/mailman/listinfo.cgi/boost

The information contained and transmitted in this electronic message may
contain confidential proprietary or legally priviledged information. It is
intended only for the individual or entity to which it is addressed. No
confidentiality or privilege is waived or lost by mistransmission to an
unintended recipient. If the reader of this message is not the intended
recipient or the employee or agent responsible for delivering this message
to the intended recipient, you are hereby notified that any use,
dissemination, distribution, copying or other retention of this message is
unauthorized and prohibited. If you receive this message in error please
notify the sender immediately by telephone or electronic message.

Vie Financial Group, Inc. (Vie) and its affiliates reserve the right to
monitor all electronic communications through its network. Any views
expressed in this message are those of the individual sender, except where
the message states otherwise and the sender is authorized to represent them
to be the views of Vie or Vie affiliate.


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