[asterisk-users] FAX reliability
Lee Howard
faxguy at howardsilvan.com
Mon Apr 13 09:48:35 CDT 2009
Steve Underwood wrote:
> Hi Lee,
>
> Lee Howard wrote:
>> Hi Steve,
>>
>> Steve Underwood wrote:
>>> In chan_dahdi.c there is now code that extends the buffering inside
>>> dadhi when a FAX is detected, and puts the buffering back to normal
>>> at the end. This isn't really a cure - its more of a bandaid.
>>> However, I expect it has the desired effect if they have put it into
>>> the trunk code.
>>>
>>> You need to enable this feature in chan_dahdi.conf.
>>>
>>
>>> Very interestingly, many people who have problems sending from
>>> app_txfax or app_fax have no problem sending from iaxmodem +
>>> HylaFAX. It seems Asterisk can feed a zaptel/dahdi channel much more
>>> reliably when passing a signal through, than when generating on in
>>> an app.
>>
>> So are the chan_dahdi.c developments mentioned of any practical value
>> to iaxmodem users?
> That's a good question, and I have no idea about the answer. Some
> people who have had problems sending from app_txfax say iaxmodem +
> HylaFAX works OK on the same machine. This seems strange, as you might
> expect a problem in scheduling I/O would affect passthrough as well as
> applications. What I don't know is whether they just have a lot less
> trouble with iaxmodem, or they have no trouble at all.
>
> Do you get reports from people who say receive is stable, but transmit
> is not?
There have been a few reports of that situation, but all of them that I
recall had to do with people who were trying to use VoIP for fax
(HylaFAX and iaxmodem tend to be much more tolerant of the audio
cut-outs caused by jitter than are other receivers).
Otherwise, no, I don't have any of those situations that I can point out.
I suppose that I'd need to make some recordings to say for certain
whether audio cut-outs were occurring. However, what would really be
nice would be to see some comments come from whatever developer at
Digium made those chan_dahdi.c modifications.
Thanks,
Lee.
More information about the asterisk-users
mailing list