Re: Review: Boost.Threads

From: williamkempf_at_[hidden]
Date: 2001-09-01 14:53:38

--- In boost_at_y..., "Gustavo Guerra" <gustavobt_at_m...> wrote:
> In general, I think the library is well designed and should be 
accepted into
> boost.
> Some comments:
> 
> Have you considered the inclusion of a read/write lock? I think it 
has very
> common use. It's implemented in POSIX, and it's not hard to 
implement on
> other systems on top of a semaphore.

Yes, this has been discussed.  The plan is to add this as a later 
extension.  There are some issues with the design that will take some 
doing and I didn't want this to hold up submission of the library.  
In fact, if you read the introduction you'll see that I've documented 
the plan to add this concept in the next phase.

> I think the semaphore docs need to be clarified. Those "As If" 
parts aren't
> very clear.

I'll look at this, but it's sounding like semaphore may go away.  At 
least two people have suggested removing it in this review process.
 
> There's no mention of thread_resource_error in the docs. Why 
doesn't both
> the exception classes have a 'what' member function?

I goofed with the docs for thread_resource_error and will add it this 
weekend, hopefully.  Both exception classes do have 'what' member 
functions, documented indirectly in the synopsis which shows they 
both derive from std::runtime_error.  I probably should explicitly 
document this, though.
 
> In mutex_concept.html the [Schmidt 00] link is not right. It should 
be
> #Schmidt-00 instead of #Schmidt 00

Thanks, I'll correct this.
 
Bill Kempf