Re: [boost] Re: Re: Threads & Exceptions

From: William E. Kempf (williamkempf_at_[hidden])
Date: 2002-08-13 17:09:42

From: "Victor A. Wagner, Jr." <vawjr_at_[hidden]>

> What I want to _continue_ is an agreement that _some_ mechanism for
> allowing exceptions to be passed (via join() if that's how one retrieves
> status about a thread) in addition to (instead of?) some "return value".
> It would disappoint me if I had to alter a function in order to implement
a
> "delayed call" or whatever term we're going to use to describe this
feature.
> It's also obvious (to me) that we canNOT possibly pass all possible
> exceptions, I suggested a set, and volunteered to assist in writing the
code.

OK, here's the requirement.  The list of exception types that are to be
propogated should result in exceptions of the type specified being passed in
a complete form.  Slices are not allowed.  Anything less has limited
utility, and is likely to cause at least complaints from users, if not bugs
in corner cases.  If you can implement something with these restrictions I'd
be happy for the assist.  Otherwise, I'm not going to propogate anything
more than a thread_terminated exception, possibly with a copied what()
result from the original exception if it was a std::exception type.

Bill Kempf