"Christophe Meessen" <christophe@meessen.net> wrote in message news:4087C462.5080504@meessen.net...

Hello,

I have axamined the thread_dev header file. Unfortunately the cpp file doesn't seem to correspond.

In what way don't they correspond?

I must admit that the proposed change is a great improvement and removes many objections I had regarding boost thread.

I'm glad it looks good to you.

Changing the thread object into a handle allows now to copy it. It is then possible to add a thread member variable to a class. A start method can then assign the started thread instance to it. So the gap between boost threads and start&run model gets narrow. The source is one year old. What is the time plan to release it ?

I'm merging the thread_dev branch into the main branch a piece at a time, after first examining each piece, finishing unfinished bits, and requesting comments on this list. I've already merged some of the more independent pieces (barrier class, updated thread_specific_ptr). I'm currently working on the mutex class, doing something about adding back the option to build a statically linked version of Boost.Thread, and updating the docs. I hope to have all of these things in the next release, which is planned for sometime around the end of June if I remember correctly. Unfortunately, I'm afraid I won't have time to get to the thread class and other changes until the release after that, however. Mike