Sync hylafax 2:6.0.4-10 (universe) from Debian unstable (main)

[tildar]

I believe this might be due to recent ghostscript update.

ii hylafax-client 2:6.0.3-5.1ubuntu1 Flexible client/server fax software - client
ii hylafax-server 2:6.0.3-5.1ubuntu1 Flexible client/server fax software - server
ii ghostscript 8.71.dfsg.1-0ubuntu5.1 The GPL Ghostscript PostScript/PDF interpret
ii ghostscript-cups 8.71.dfsg.1-0ubuntu5.1 The GPL Ghostscript PostScript/PDF interpret
ii ghostscript-x 8.71.dfsg.1-0ubuntu5.1 The GPL Ghostscript PostScript/PDF interpret

same or simular problem reported here: https://bugzilla.redhat.com/show_bug.cgi?id=569957

Please advise on any additional debugging that I can provide.

[tildar (tildar)]

ok, this is due to a change in ghostscript.

The default MaxStripSize had changed from 0 in ghostscirpt-8.70 to 8192 in ghostscript-8.71

I'm testing to see if adding -dMaxStripSize=0 to bin/ps2fax.gs and bin/pdf2fax.gs changes anything for me.

[tildar (tildar)]

Adding adding -dMaxStripSize=0 to bin/ps2fax.gs and bin/pdf2fax.gs seems to have corrected this issue for me.

[Moshe Ortov (jim-networksystemssolutions)]

Indeed, this does fix the problem of Ubuntu 10.04 LTS hylafax sending blank pages for me.

I modified bin/pdf2fax.gs as described by adding the -dMaxStripSize=0 onto the GS command at the end of the file and that fixed the problem.

As this fix has been around for around a month and seems to be correct it would be great to see it added to the online updates - it does not seem to have much likelihood of breaking anything else so looks like a nice clean fix to roll out.

Patch file is trivial and attached.

[Bodo (bodo-bodo-m)]

A similar fix must be added to pdf2fax.

[Giuseppe Sacco (eppesuig)]

From what I understand from https://bugzilla.redhat.com/show_bug.cgi?id=571520 it seems the problem is in ghostscript and has been fixed upstream.
I would suggest uploading a new ghostscript package instead of a new hylafax package.

[Bodo (bodo-bodo-m)]

Giuseppe Sacco wrote:
> it seems the problem is in ghostscript and has been fixed upstream.

That's a matter of opinion.

HylaFAX relies on Ghostscript having a default MaxStripSize that is large enough. The default that used to be 0 (unlimited) was changed to a small value that broke HylaFAX. In a recent version the default has been changed again to a value that seems to be big enough. Who knows what might get changed in future versions of Ghostscript.

I think it's a good idea to make HylaFAX more robust by explicitly specifying parameters for Ghostscript and maybe checking the resulting TIFF file.

In the HylaFAX mailing list a message was posted with a link to a version that checks the TIFF file created by Ghostscript and modifies it if necessary. IIRC this patch is included in HylaFAX+. The scripts in HylaFAX+ may be different from HylaFAX, but it should be possible to copy the mechanism to appropriate places in HylaFAX.

Bodo

[Spencer Rhodes (spencer-rhodespa)]

Solved the problem for me, thanks.

[tildar (tildar)]

I would have to agree with Bodo Meissner's point about explicitly specifying parameters, especially if default setting keep changing around thus breaking things, then fixing, then braking.....

I would even go so far as to ask why we do not have HylaFAX+ packaged as an optional replacement? or even just drop the org version and go to hylafax+ instead!

It seems that every time I have a problem with hylafax, it has already been fixed in hylafax+., or was never a problem to begin with.

[Giuseppe Sacco (eppesuig)]

I am uploading hylafax 6.0.4-10 to Debian. This package include the supplied patch and should hopefully fix the problem. Please let me know if anything is still broken.

Bye,
Giuseppe

[Sebastien Bacher (seb128)]

subscribing sponsors to review the sync request

[Hurikhan77+bgo (hurikhan77+bgo)]

This is a bug already discovered in Ubuntu (see URL). Ghostscript changed processing defaults so ps2fax and pdf2fax (etc) cannot generate a valid TIFF. The result is that the fax recipient only gets an empty page. The fix described there (MaxStripSize=0) resolves the issue.

Involved versions:
net-misc/hylafax-6.0.3
app-text/ghostscript-gpl-8.71-r1

I propose to use the MaxStripSize=0 approach in HylaFax to make it more robust to ghostscript upstream changes.

Reproducible: Always

[Giuseppe Sacco (eppesuig)]

You may get Debian patches from http://patch-tracker.debian.org/package/hylafax/2:6.0.4-10 . It is the one called 808_ghostscript_args.patch and keep two corrections: the one you already proposed in this report, and adding '-P' that seems to be required with '-dSAFER=true'

[Stefano Rivera (stefanor)]

Guiseppe: hylafax currently ftbfs http://udd.debian.org/cgi-bin/ubuntu_ftbfs.cgi

I had to modify the configure script to add __THROW.

[Launchpad Janitor (janitor)]

This bug was fixed in the package hylafax - 2:6.0.4-10ubuntu1

---------------
hylafax (2:6.0.4-10ubuntu1) maverick; urgency=low

  * Sync from Debian unstable (LP: #580568)
  * Add 901_fchmod-prototype.patch: glibc now includes an fchmod prototype
    which doesn't match hylafax's.

hylafax (2:6.0.4-10) unstable; urgency=low

  * Check for upstart scripts in /etc/init instead of /etc/event.d.
  * Added -dMaxStripSize=0 to all ghostscript invokation. (See
    https://bugzilla.redhat.com/show_bug.cgi?id=569957)

hylafax (2:6.0.4-9) unstable; urgency=low

  * Fixed a problem with faxadduser/faxdeluser overwriting hosts.hfaxd file.
  * Updated to Debian Policy 3.9.1 (no changes required).
  * Added a NEWS about faxmail behaviour changed since hylafax 4.
  * Added an example on how to make exim4 deliver email to faxmail.
 -- Stefano Rivera <email address hidden> Sun, 05 Sep 2010 23:23:42 +0200

[Giuseppe Sacco (eppesuig)]

Thanks,
I am going to report this upstream.

[Giuseppe Sacco (eppesuig)]

Hi, before reporting it, I better checked all bug reports. This problem has already been reported both against Debian package and against Hylafax upstream developers. The latest statement from upstream is that it is probably a bug in latest gcc. Are you telling me that Ubuntu gcc-4.4_4.4.4-13ubuntu2 include patches from gcc 4.5?

More info:
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=565001
http://bugs.hylafax.org/show_bug.cgi?id=924

[Stefano Rivera (stefanor)]

> I better checked all bug reports.

Thanks, I didn't see that one when I skimmed the reports in Debian, and I didn't look ino what was causing configure to misdetect this prototype.

> Are you telling me that Ubuntu gcc-4.4_4.4.4-13ubuntu2 include patches from gcc 4.5?

Yes, according to the changelog it's "built from VCS".

Looks like this is our culprit: http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15638

[Samuli Suominen (ssuominen)]

Closing as WONTFIX because hylafax was removed from Portage. But adding 361187 to Blocks: -line since this bug should get fixed if introduced back to Portage.