[Info-vax] "bad select 38" (OpenSSL on VMS)

[John Reagan]

On Tuesday, September 13, 2016 at 7:17:40 PM UTC-4, John E. Malmberg wrote:
> On 9/13/2016 5:55 PM, Stephen Hoffman wrote:
> > On 2016-09-13 22:31:07 +0000, John Reagan said:
> >
> >> There is a CRTL feature logical to ask select() to ignore file
> >> descriptors instead of causing an error.  I don't have it on-hand, I'm
> >> at the hospital visiting my father.
> >
> > That likely won't work here, as the s_client code uses standard input as
> > one of the sources for the select().
> >
> > Was going to use the work-around listed, but that's not necessary as
> > upgrading to SSL1 1.0.1h has resolved this bug.
> >
> > Still trying to sort out the underlying TLS problem that led to testing
> > connections with the s_client in the SSL1 kit, but that's fodder for
> > another discussion and another time, and more debugging.
> >
> > Haven't seen whether the changes were fed back, but didn't look too
> > carefully at the github master for the file, and I know the changes
> > aren't in the s_client from 1.0.1 head.
> >
> > I'd expect that enhancing select() to better implement the expected
> > behavior from other systems — removing the present OpenVMS limitations —
> > is on the list somewhere, but probably not near the top.
> 
> Hidden in the source code for many of the updated GNV utilities, 
> including Bash and coreutils is a wrapper for select() and poll() that 
> handles terminals, sockets, pipes, and if you prepare things right with 
> dup2() to reserve the descriptor number, X11 (implemented in VMS as 
> event flags).
> 
> Regards,
> -John
> wb8tyw at qsl.net_work

What's the license on that code?