Re: flexfax: Meaning of RTN (was: Hylafax question)

[Dmitry Bely <dbely@mail.ru>]

David Woolley <david@djwhome.demon.co.uk> writes:

> > >Can you explain what the RTN problem is?
> > 
> > In fact I'm not too sure ! As far as I understand, after sending a page, modems
> > of both sides will do some hand shake to check the connection (retrain), if the
> 
> RTN is sent *before* any retraining.  At the end of a page, the modems
> switch back from a typical, phase C, 9600 bps (1200 baud) to 300 bps
> (and baud), of phase D, and the sender sends a message saying whether
> there is any more to send (typically MPS - "more pages, same document")
> and the receiver sends one saying what it wants to do next.  The receiver
> can say continue normally, but, if it thinks the analogue signal quality
> has degraded a bit can send "retrain positive" to indicate that the
> sender should send training signals to recalibrate the receiving modem's
> equaliser before continuing.
> 
> If it thinks that, in addition, the image quality was unacceptably poor,
> it can, instead, send "retrain negative" to indicate that the page should
> be resent after the retraining process.  Other possibilities include an
> indication that the operator has interrupted reception, with or without
> successful receipt of the page.
> 
> Many fax machines don't actually monitor the true analogue signal quality

I.e. there are some fax machines that *can* monitor analogue signal
quality??? Then what is the criterion of the error?

In reality the rules are following (don't ask me about the source :-))

[---cut---]
Quality criterion

The receiver adds up the run length identified between 2 recognized line
end code words. If this sum is not equivalent to the set number of scan
elements per line, a line error has occured. A line error may cover
serveral lines if line end code reception is erroneous. 

>From the above mentioned algorithm the receiver has to derive the
following quality criterion:

- If the number of line errors is >15% of all lines received, the receiver
must recognize the received copy as not successful and send the respective
negative message(RTN or PIN) in phase D.
- If the number of lines is < 15% of all lines received, the receiver must
recognize the received copy as successful and send the respective positive
message (MCF, RTP or PIP) in phase D.
- If the number of line errors is >= 5% and <= 15% of all lines received,
the receiver may recognize the recieved copy as successful or not
successful and send the respective message in phase D.

The two-dimensional coding method counts all non-decodeable lines as line
errors.
[---cut---]

> and use protocol violations in the recovered image data to determine
> whether the line quality is degrading.  If there is a systematic error
> in the coding of the image, they will repeatedly treat this as requiring
> retraining, but retraining won't help.  If the receiving system is 
> unreasonably fussy about image quality, it may send RTN when the
> human user of the system cannot see a problem, with the result that 
> the first page is sent multiple times for no apparent reason.
> 
> One such systematic error is caused by incorrect splicing of the tag line 
> into the top of the image.  A patch for this has existed for a long time,

BTW, this tag line error (without tagline patch) cannot lead to RTN -- one
broken scan line in the middle of the image is *not* enough to reject the
page and generate RTN.

> but is not in the released version of the Hylafax, although probably in the
> current beta.  The basis of the recent patch is that the "return to
> control" code at the end of image data is duplicated if the program that
> created the TIFF file also included one.

Not only this one. Hylafax also appends extra RTC itself (but it's
prohibited by Class 2.0 specs). Also there is a problem of "byte-aligned"
first EOL -- according to T.4 it should be exactly 000000000001, and extra
fill (zero) bits are *not* allowed.

> Some fax receivers appear
> to require an immediate carrier drop after the first one for them to
> consider the protocol valid.
> 
> > handshake failed, the RTN will report. Please check the HylaFAX home page
> > for details or ask Dmitry.
> 
> The original subject would cover almost every posting on this list, so I
> have changed it.

Hope to hear from you soon,
Dmitry