[Nut-upsdev] Problem with bestfcom and old Ferrups

[Charles Lepple]

On Feb 12, 2008 9:46 AM, Chris Adams <cmadams at hiwaay.net> wrote:
> Once upon a time, Charles Lepple <clepple at gmail.com> said:
>
> > On Feb 7, 2008 10:13 AM, Arjen de Korte <nut+devel at de-korte.org> wrote:
> > > >> 193-                 ser_get_line_alert(upsfd, buf, sizeof(buf), '\012', "",
> > > >> 194-                         POLL_ALERT, alert_handler, 0, 20000);
> > > >> 193+
> > > >> 194+                 ser_flush_io(upsfd);
> > > >> 195
> > > >> 196          ser_send(upsfd, cmd);
> > > >
> > > > I tried this, but it didn't catch the extra data.  I still needed the
> > > > ser_get_line() calls.
> > >
> > > That's probably because in you ser_get_line() function, you specify quite
> > > a long delay (3 seconds), while the ser_flush_io() commmand will not wait
> > > for data. So the reply you're trying to get rid of may not have arrived
> > > completely at the time of flushing the buffers. So we may need both.
> > > Reading the reply through ser_get_line() and ser_flush_io() to get rid of
> > > unexpected messages.
> >
> > What was the final decision regarding the placement and order of the
> > ser_get_line/ser_flush_io calls?
>
> I need the ser_get_line() calls in my patch either way.
>
> While the ser_get_line_alert() call above isn't really doing anything
> (the alert handler just throws away the input) does have a wait that is
> longer than the ser_flush_io() call, so I don't know if that will
> function the same.
>
> I have not yet had a chance to try it on my old UPS (since I have to
> copy updates to my notebook and carry it to the power room).

OK. When you get a chance, let me know the order that works, and we
can commit it.

-- 
- Charles Lepple