Re: Boost.Threads draft library submission

From: williamkempf_at_[hidden]
Date: 2001-03-16 10:23:53

--- In boost_at_y..., Doug Gregor <gregod_at_r...> wrote:
> On Thursday 15 March 2001 05:36, you wrote:
> > --- In boost_at_y..., williamkempf_at_h... wrote:
> > > --- In boost_at_y..., <boost_at_y...> wrote:
> > > > Hello,
> > > >
> > > > This email message is a notification to let you know that
> > > > a file has been uploaded to the Files area of the boost
> > > > group.
> > > >
> > > >   File        : /threads/threads.zip
> > > >   Uploaded by : williamkempf_at_h...
> > > >   Description : Boost.Threads draft library submission
> > > >
> > > > You can access this file at the URL
> > > >
> > > > http://groups.yahoo.com/group/boost/files/threads/threads.zip
> > > >
> > > > To learn more about file sharing for your group, please visit
> > > >
> > > > http://help.yahoo.com/help/us/groups/files
> >
> > I've uploaded a newer version of this file.  It addresses most of 
the
> > problems found by others up to this point.  I'd appreciate someone
> > testing this with Comeau (and other compilers) to see if I've 
fixed
> > the syntax problems.  VC++ is too forgiving on many of these 
things,
> > making it difficult on me here.
> 
> A few more minor issues Comeau found:
> - boost/thread/thread.hpp: #include <string> so that implicit 
conversion from 
> (const) char* to std::string can occur in the call to 
std::runtime_error's 
> constructor (not sure if Comeau is acting correctly here or not, 
but it can't 
> hurt).

Reading the standard I don't see how they could be considered to be 
acting "incorrectly", but that's a pain in the neck.  I'd prefer not 
including <string> in the header here.  If Comeau deemed not to in 
<stdexcept> I'm not going to in <boost/thread/thread.hpp> either.  
So, I've moved the constructor implementation into the source file 
where <string> is included.  This should fix the problem and retain 
the intent of the standard library implementation.

> - boost/thread/condition.hpp: timed_wait needs a return value 
if "pred" 
> returns true.

Nice catch.  I wonder why VC++ didn't complain.  Oh well, my kingdom 
for a real compiler!

> Also: it isn't clear to me what the intention of "pred" is: 
> should it be a function object? a volatile reference to something 
convertible 
> to bool?

It should have been a function object.  I see your point here.  The 
code will be corrected later today.  I also obviously need another 
test in the test harness.  I'll work on that later as well.
 
> - src/mutex.cpp: a "-1" that should be a 
> std::numeric_limits<unsigned>::max(). 

One slipped through, huh.  Ok, fixed.

> - src/mutex.cpp:  #include <limits>.

Fixed.

> - src/semaphore.cpp: in the up member function, there is a 
conditional:
> 	if (count < 0 || available + count > max)
>   where "count" is unsigned.

Fixed.

Bill Kempf