Re: [libc-alpha] Fwd: Re: C++ and pthread_create()

[Kaz Kylheku <kaz at ashi dot footprints dot net>]

On Thu, 14 Feb 2002, Alexander Terekhov wrote:

> Date: Thu, 14 Feb 2002 18:05:45 +0100
> From: Alexander Terekhov <terekhov@web.de>
> To: libc-alpha@sources.redhat.com
> Subject: [libc-alpha] Fwd: Re: C++ and pthread_create()
> 
> 
> Thanks a LOT for sharing your opinions/plans/whatever! ;-)
> 
> Perhaps you should read this:
> 
> http://groups.google.com/groups?as_umsgid=DEta8.55%241h5.204%40news.cpqcorp.net
> 
> "...I think any implementation that doesn't build
>  cancelation (and thread exit) on top of a pervasive 
>  common exception infrastructure is broken to the point 
>  of being unusable. Nevertheless, the standard doesn't 
>  (and can't) require a correct implementation, so those 
>  who do it wrong can get away with it as long as their
>  users allow them to. ..."

I seem to remember Dave writing something similar before.  I agree.
In a given programming environment, there should be a single, unified
mechanism for handling non-local transfers of control with unwinding.
Everything else should map to that: thread cleanup handlers in C,
setjmp/longjmp, C++ try/catch and similar features in other languages.