[hylafax-devel] Re: config/usr-xon & usr-rts

[Dmitry Bely <dbely@mail.ru>]

Darren Nickerson <darren@dazza.org> writes:

>   Dmitry> Sure. But the question is: should AA be enabled with the *default*
>   Dmitry> setting or not? If yes (although I don't like it) we should do it
>   Dmitry> uniform way. i.e. for *all* modem templates supplied, where it is
>   Dmitry> possible (all Class2/Class2.0 modems).
> 
> I agree. We should probably reach consensus on which way to go on this. 
> Dmitry, it would be useful to us if you could provide any data which supports 
> your suggestion that having this enabled can actually cause fax reception 
> problem.

Just my experience (and also what I have heard from others). Considerable
number of incoming call were lost by by Zyxel 1496 with something like

c00006465:Jun 06 16:19:18.28: [31766]: ANSWER: Ring detected without successful handshake

when Adaptive answer was enabled. IMHO many fax machines just hang up,
having modem response instead of fax one. Also longer connection means that
it may have no time to repeat traiting if the first attempt was
failed. Also there are security issues (see Nico's messages).

> The quote from the HylaFAX.ORG web pages is pretty compelling, but is 
> that information as out of date as the rest of it? ;-)

I think it's up to date: the trouble happens with the remote fax machine
(T1 timeout expires etc), not our modem.

> This feature is potentially pretty important to a lot of installations. If
> we were to lean towards disabling it in a default install, might it be wise
> to have the AA config params present in each template (in a way which WOULD
> actually work with that modem) but to have them commented out?
>
>   Dmitry> Understand me right: I have nothing against auto-answer feature in
>   Dmitry> general (although it makes connect longer and sometimes do not work
>   Dmitry> reliably -- useless waste of time in the fax-only environment). But
>   Dmitry> the purpose of modem templates IMHO is working aroung specific modem
>   Dmitry> quirks and nothing else.
> 
> This is, in itself, an interesting point. And to a large degree I think that's 
> what they're used for. What about when you have an advanced setup feature 
> which is quirky, but which we've decided not to enable by default? It should 
> probably be IN the config file, but commented out. Agreed?

Yes, then it's already generated (nearly all possible parameters should be
there). But don't think, that we should create modem *templates* the same
way. Let me quote myself:

[---cut---]
I also think that faxaddmodem script should parse two files when generating 
modem config: generic ClassX template and specific model template. If some
parameter is present in model template, it will be used, otherwise generic
ClassX value will do. This way we can move all Class-specific settings (
ModemSetupAACmd etc. ) to ClassX templates while model templates will
contain only minimal model specific settings (flow control etc.) Is this
acceptable approach?
[---cut---]

>   Dmitry> So all non modem-specific defaults should comply to
>   Dmitry> config(5f). Regarding ModemSetupAACmd is says that auto-answer is
>   Dmitry> disabled by default.
> 
> Doesn't "default" in this context mean purely what HylaFAX will do in the
> ABSENCE of an entry in the config file? config(5f) does not indicate a
> default for ModemSetupAACmd, at least not the copy I have. It DOES indicate
> a default of "No" for AdaptiveAnswer however, which seems to be supported
> by the following caveat in config(5f):

AdaptiveAnswer parameter has nothing to do with ModemSetupAACmd :-) These
are two different strategies: ModemSetupAACmd setups the modem to decide
which type of call is arrived (possible for Class2/2.0 and small amount of
Class1 modem -- see recent Harald Pollack's message), while AutoAnswer
(together with AnswerRotary and others) enables software-only call type
detection methods.

>               Note that the adaptive answer strategy depends on many factors
>               that limit its usefulness.  When calls are answered first as
>               fax, it typically only works with a Class 1 modem because it is
>               important that the sequence of operations related to answering
>               as fax be completed in a short enough time that a subsequent
>               answer for data be done before the caller times out and hangs up
>               the telephone (for this reason Class1RecvIdentTimer should be set
>               to a value that is shorter than FaxT1Timer.)
> 
> Are a lot of people actually using AA in production?
> 
> Anyway, I'm not sure lines which enable AA in modem config files are at
> all inconsistent with the man pages . . . since the whole purpose of modem
> config files are to OVERRIDE the defaults mentioned therein. 

I believe that config defaults in *templates* should be only changed to
work arong the problems in specific firmware (also see above). Thus all
other defaults should comply with config (5f). It seems to be fundamental
difference between my point of view and yours.

> It's late though, and I joined this thread late so I may be missing the
> point ;-)

Hope to hear from you soon,
Dmitry




____________________ HylaFAX(tm) Developers Mailing List ____________________
 To unsub: mail -s unsubscribe hylafax-devel-request@hylafax.org < /dev/null