On Mon, 19 Jun 2023 at 00:40, Jason McKesson via Std-Proposals
<std-proposals_at_[hidden]> wrote:
>
> On Sun, Jun 18, 2023 at 5:34 PM Ville Voutilainen
> <ville.voutilainen_at_[hidden]> wrote:
> >
> > On Mon, 19 Jun 2023 at 00:17, Jason McKesson via Std-Proposals
> > <std-proposals_at_[hidden]> wrote:
> > >
> > > On Sun, Jun 18, 2023 at 4:26 PM Smith, Jim via Std-Proposals
> > > <std-proposals_at_[hidden]> wrote:
> > > >
> > > > Hi All,
> > > >
> > > > It would be convenient if std::thread would automatically become not joinable after the thread function ends.
> > >
> > > This just creates a race condition:
> > >
> > > ```
> > > if(thread.joinable())
> > > thread.join();
> > > ```
> > >
> > > Is this functioning code? It looks like it *should* be but it isn't.
> > > The answer to the `joinable` question is out-of-date the moment after
> > > you call the function. If the thread function exits *after* the
> > > joining thread asks if it is joinable, then the call to `join` will
> > > fail.
> >
> > Why would the call to join() fail? It's perfectly okay to call join()
> > after the thread
> > function has exited.
>
> We're assuming the OP's idea were adopted. If a thread is not
> "joinable" after the thread function ends, that *must* mean that
> `join` will fail in that circumstance. Otherwise, what does "joinable"
> even mean? It's even part of the definition of `join` that you get an
> exception if `joinable` is false.

Right, got it. That's all the more reason not to adopt the suggestion.
It makes no sense.