Re: first sight

From: bill_kempf (williamkempf_at_[hidden])
Date: 2002-01-16 18:43:27

--- In boost_at_y..., "David Abrahams" <david.abrahams_at_r...> wrote:
> 
> ----- Original Message ----- 
> From: "bill_kempf" <williamkempf_at_h...>
> 
> > > Also, with respect to auto_rethrow_exception:
> > > 
> > >  scoped_lock lock( systemWideMutex );
> > > 
> > >  beginUpdates(); // modify something; nothrow
> > > 
> > >  try
> > >  {
> > >     // ... plug-in oper ;-) strong
> > >  }
> > >  // ...
> > >  catch(...)
> > >  {
> > >     // ...
> > >  }
> > > 
> > >  finalizeUpdates(); // final modifications; nothrow
> > > 
> > >  return;
> > 
> > Such code isn't exception safe any way.  I don't find that to be 
a 
> > compelling argument.
> > 
> > In other words, "finalizeUpdate" needs to be either in a RAII 
> > mechanism or placed in the catch(...).
> 
> Uh, why do you say that?

Ooops, sorry, my mistake. (Man, it's been a bad day.)  I should have 
thought about that one a bit more.  Not sure I've ever seen code 
structured this way, but it's a valid point.

So, all this discussion leads us to:

1) Use deferred cancellation (and allow the user to turn cancellation 
off).
2) Use a C++ exception for cancellation.
3) Allow the user to catch the exception.
4) Allow the user to "eat" the exception.
5) Make the cancellation "flag" a sticky flag, so if the user "eats" 
the exception it will eventually be re-thrown when they call another 
cancellation point.
6) Ensure cancellation points don't throw when uncaught_exception() 
returns true.
7) Document that in general one should always rethrow a caught 
cancelation exception.

Bill Kempf