flexfax: Solved: squashed/stretched pages & format_failed

[Jacco de Leeuw <jacco2@dds.nl>]

Hi,

We have been using the hylafax-v4.0pl2 RPM (recompiled with minor
changes) on Linux RedHat 5.2. We had a couple of problems which
we seem to have been solved now, and I would like to share them with
you.

- We are using Multitech internal ISA 33K6 modems (MT2834ZPXI-022,
  if I'm correct; not at the office right now). This model is not
  specifically mentioned in the list so we used the Multitech
  configuration file included with Hylafax. The problem was that
  we sometimes received "squashed" faxes. I added some lines from the
  generic Class 2 configuration file and this fixed most of the
  squashing but not all of it.

- I reduced the serial port speed from 57600 bps to 38400 bps and
  this solved the rest of the squashed pages problem.

- At irregular times, users who used WHFC to fax documents got the
  following message:

      Your facsimile job to xxxxxxxxx was not sent because
      document conversion to facsimile failed.  The output from the
      converter program was:
     
      Check any PostScript documents for non-standard fonts and
      invalid constructs.

  At first we thought that MS Office might be generating lousy
  Postscript, but that could not be the cause: exactly the
  same document *could* be faxed some time later. Examining the
  queue directories showed that the Postscript files actually had been
  converted to facsimile (G3 TIFF) alright. But Hylafax's 3 second timer
  had kicked in immediately so it thought the conversion had failed.
  
  You have to know that we used an "at" job to stop and start Hylafax,
  to have it reread its configuration files. Somehow this "at" job
screws
  up Hylafax alarms/signals/timers. I'm not a hard-core Unix systems
  programmer so I don't know the exact cause, but at least I was able
  to find out where and why the error occurred. Hurray for
  source code... ;-).

  Upon further examination of the source code we noticed that hfaxd
  does a stat() on its configuration files so we didn't have to do a
  /etc/rc.d/init.d/hylafax stop and start after all. So we removed
  those. This seemed to solve the problem. Still, I don't understand
  why the "at" job confuses Hylafax's timer, while a cron job doesn't.

- We use $FAX2PS -S $FILE | lpr to print out received faxes.
  Note the -S switch which scales the fax to fill the page. When
  large (i.e. mostly too wide) faxes were received, they were scaled
  down to A4 format, which is great. However, this had a side effect:
  small faxes were blown out of proportion and stretched to fill the
  whole page. John Williams' patch
  (http://www.hylafax.org/patches/tiff2ps.patch) seemed to fix this,
  but we were using fax2ps and not tiff2ps (AFAIK both programs do
  similar things but fax2ps generates more efficient Postscript output
  for fax purposes). So I wrote a new patch for fax2ps (libtiff) which
  only scales *down*, not up. And this separately for height resp.
  width. It works for us, but perhaps one of Hylafax experts should
  review it. If anyone wants the patch, just mail me.

Jacco
-- 
Jacco de Leeuw            mailto:jacco2@dds.nl
van Wessemstr. 54         http://huizen.dds.nl/~jacco2/
1501 VM Zaandam, Holland  tel:+31(0)756352068
In Windows, the power switch is an integral part of the user interface.