Bug 176044 - ports: print/ghostview (1.5_3) segfault/coredump
ports: print/ghostview (1.5_3) segfault/coredump
Status: Closed Overcome By Events
Product: Ports & Packages
Classification: Unclassified
Component: Individual Port(s)
Latest
Any Any
: Normal Affects Only Me
Assigned To: FreeBSD ports mailing list
:
Depends on:
Blocks:
  Show dependency treegraph
 
Reported: 2013-02-12 03:20 UTC by rfg
Modified: 2014-06-19 10:09 UTC (History)
2 users (show)

See Also:


Attachments
output.ps (338.97 KB, text/plain)
2013-02-12 03:20 UTC, rfg
no flags Details

Note You need to log in before you can comment on or make changes to this bug.
Description rfg 2013-02-12 03:20:00 UTC
The attached modest .ps file causes ghostview-1.5_3 to immediately segfault
and dump core.

Fix: I haven't looked at the code, so I don't know offhand what the fix is.

I will work on trying to debug it, if the current FreeBSD maintainer is not
able to invest the time to get to the bottom of this.

How-To-Repeat: 
portinstall print/ghostview
ghostview output.ps
Comment 1 Martin Wilke freebsd_committer 2013-02-28 17:04:57 UTC
State Changed
From-To: open->feedback

Hi, unforgently this port don't have a maintainer.
Comment 2 Stefan Walter freebsd_committer 2013-04-28 23:37:37 UTC
Responsible Changed
From-To: freebsd-ports-bugs->stefan

Take this, since I have also taken ports/162045.
Comment 3 Stefan Walter freebsd_committer 2013-04-29 10:56:06 UTC
Hello Ronald,

after trying a bit, I can add the following facts:

- Ghostview dumps core for me, too, when I try to open the PostScript file
  you attached on my (amd64) FreeBSD 9.1 system.
- Gv (from port print/gv) does not dump core when I open the file on my
  FreeBSD 9.1 system.
- Ghostview does not dump core when I open the file on a (amd64) Debian
  system.

So it seems to be a FreeBSD-specific issue.

I haven't investigated the actual cause of the segmentation fault, and
thus do not have a fix, either. Your best bet to make any progress would
probably be sending a mail to the freebsd-ports mailing list. (To be
honest, though, I wonder if there's a good reason for using an ancient,
unmaintained software when there's an actively maintained and working
replacement like print/gv.)

Best regards,
Stefan

p.s.: I'm going to close the older PR about this (ports/162045) to avoid
redundancy.
Comment 4 rfg 2013-04-29 21:25:25 UTC
In message <20130429095605.GA1720@birne.dunkelkammer.void>, you wrote:

>after trying a bit, I can add the following facts:
>
>- Ghostview dumps core for me, too, when I try to open the PostScript file
>  you attached on my (amd64) FreeBSD 9.1 system.

OK.  Thank you very much for verifying.

>- Gv (from port print/gv) does not dump core when I open the file on my
>  FreeBSD 9.1 system.

That is interesting.

>- Ghostview does not dump core when I open the file on a (amd64) Debian
>  system.

I see.

>So it seems to be a FreeBSD-specific issue.

That seems like an entirely reasonable conclusion.

>I haven't investigated the actual cause of the segmentation fault, and
>thus do not have a fix, either. Your best bet to make any progress would
>probably be sending a mail to the freebsd-ports mailing list.

I'm sorry.  I don't follow you.

I filed a formal PR.  In an ideal Universe, that alone would cause the
port maintainer to at least make some modest effort at finding a fix,
don't you think?

>(To be
>honest, though, I wonder if there's a good reason for using an ancient,
>unmaintained software

I'm sorry.  Again I am not following you.

If the package in question is no longer maintained (e.g. because it no
longer has a living breathing formally designated maintainer who is also,
conveniently, resident in the same dimensional time-space continuum as
you and I) then why is said port still present in the ports tree?

(The presence in the ports tree of ports with no active maintainer would
seem to me to be counterproductive.)


Regards,
rfg
Comment 5 Stefan Walter freebsd_committer 2013-04-29 23:39:45 UTC
Ronald F. Guilmette, 29.04.13, 22:25h CEST:

> >So it seems to be a FreeBSD-specific issue.
> 
> That seems like an entirely reasonable conclusion.
> 
> >I haven't investigated the actual cause of the segmentation fault, and
> >thus do not have a fix, either. Your best bet to make any progress would
> >probably be sending a mail to the freebsd-ports mailing list.
> 
> I'm sorry.  I don't follow you.
> 
> I filed a formal PR.  In an ideal Universe, that alone would cause the
> port maintainer to at least make some modest effort at finding a fix,
> don't you think?

I agree that would be ideal from the point of view of someone who merely
wants to use a port.

Yet: I am not the port's maintainer. The port does not currently have a
maintainer, i.e. a single individual who has enough interest in it to step
forward and continuously take care of it. And apparently, nobody else has
shown any interest in trying to resolve this issue since your first PR on
it - and that was in 2011.

The reason I mentioned the mailing list was that discussing it there would
expose it to a larger audience and raise chances of finding someone who is
able to fix it AND is interested enough to actually do it. Something that
the PR hasn't achieved so far...

> >(To be
> >honest, though, I wonder if there's a good reason for using an ancient,
> >unmaintained software
> 
> I'm sorry.  Again I am not following you.
> 
> If the package in question is no longer maintained (e.g. because it no
> longer has a living breathing formally designated maintainer who is also,
> conveniently, resident in the same dimensional time-space continuum as
> you and I) then why is said port still present in the ports tree?

Because it works. It is not uncommon for ports to be maintained "by the
community". There is more than one port in the tree that receives updates
and bugfixes now and then without anybody wanting to be THE maintainer.

As for print/ghostview: You might just be the first to notice the issue
you reported. Or others noticed it and didn't care enough about it. (Given
that, you would be welcome to take over maintainership of the port, of
course.)

> (The presence in the ports tree of ports with no active maintainer would
> seem to me to be counterproductive.)

I understand that opinion to a degree, and I disagree, but discussing that
here would serve no purpose and would not change anything, either. The
freebsd-ports mailing list would be a better place for such a discussion.

Best regards,
Stefan
Comment 6 rfg 2013-04-30 21:30:59 UTC
In message <20130429223943.GN1720@birne.dunkelkammer.void>, you wrote:

>Ronald F. Guilmette, 29.04.13, 22:25h CEST:
>> I filed a formal PR.  In an ideal Universe, that alone would cause the
>> port maintainer to at least make some modest effort at finding a fix,
>> don't you think?
>
>I agree that would be ideal from the point of view of someone who merely
>wants to use a port.
>
>Yet: I am not the port's maintainer. The port does not currently have a
>maintainer, i.e. a single individual who has enough interest in it to step
>forward and continuously take care of it. And apparently, nobody else has
>shown any interest in trying to resolve this issue since your first PR on
>it - and that was in 2011.

That would seem to be the case, yes.

>The reason I mentioned the mailing list was that discussing it there would
>expose it to a larger audience and raise chances of finding someone who is
>able to fix it AND is interested enough to actually do it. Something that
>the PR hasn't achieved so far...

Well, if I can find some time, I'll try to fix it myself.  I suspect that
doing that may be not substantially more labor intensive than trying to
find a person on the ports mailing list to gives a darn about this
specific package.

>> If the package in question is no longer maintained (e.g. because it no
>> longer has a living breathing formally designated maintainer who is also,
>> conveniently, resident in the same dimensional time-space continuum as
>> you and I) then why is said port still present in the ports tree?
>
>Because it works. It is not uncommon for ports to be maintained "by the
>community". There is more than one port in the tree that receives updates
>and bugfixes now and then without anybody wanting to be THE maintainer.

In all ernesty, I did not know that.  Thank you for educating me.

I am just a lowly user, so I have, at most, only minimal say in community
policy, but if it were up to me I would banish any and all ports for which
no specific individual would step forward and take final responsibility.
The absence of such clearly identified persons is, I think, what causes
perfectly valid PRs (by the thousands?) to languish with no action for
years on end.

>As for print/ghostview: You might just be the first to notice the issue
>you reported. Or others noticed it and didn't care enough about it.

Yes.  I try diligently to file formal PRs on every bug I find because I
feel that other people... other users... may see the same issues but not
have time to do so.

>(Given that, you would be welcome to take over maintainership of the port, of
>course.)

Thank you, but even if I were so inclined, that would have to go onto my
already overflowing TO-DO list.  I probably wouldn't get to actually doing
anything until, like, 2017.

>> (The presence in the ports tree of ports with no active maintainer would
>> seem to me to be counterproductive.)
>
>I understand that opinion to a degree, and I disagree, but discussing that
>here would serve no purpose and would not change anything, either.

On this, we agree.

I thank you for having looked into (and for trying to reproduce) my PR.
That is certainly more  than anyone else has done in about 2 years.

Just curious... Are you single-handedly tyring to clear out the massive
PR backlog?

If so, you have my undying admiration.


Regards,
rfg
Comment 7 Stefan Walter freebsd_committer 2013-05-05 11:54:15 UTC
For the record: Discussion taken off GNATS, as it doesn't have anything to
do with the problem reported.
Comment 8 Stefan Walter freebsd_committer 2013-05-26 13:25:19 UTC
Responsible Changed
From-To: stefan->freebsd-ports-bugs

Return to the pool for someone to investigate the actual cause of the 
segmentation fault.
Comment 9 John Marino freebsd_committer 2014-06-19 09:22:35 UTC
rfg,
I suggest that you provide a patch to update ghostview to version 3.5.8:
http://pages.cs.wisc.edu/~ghost/gv/gv.htm

or rather, it seems gv is a new project deriving from dead ghostview, so you could create a new port that would replace this one.

Try it.
If it solves the segfault, then submit the new port to this PR.  I'll personally handle it if you do that.

John
Comment 10 John Marino freebsd_committer 2014-06-19 09:23:54 UTC
oh, I spoke too soon, it already exists:
http://www.freshports.org/print/gv/

So why use ghostview over gv?
Comment 11 John Marino freebsd_committer 2014-06-19 09:27:07 UTC
okay, I just read the backlog.  It seems this discussion has already been had.

Why shouldn't I just deprecate this port in favor of gv?  it's not even staged.
Comment 12 commit-hook freebsd_committer 2014-06-19 10:07:13 UTC
A commit references this bug:

Author: marino
Date: Thu Jun 19 10:06:15 UTC 2014
New revision: 358374
URL: http://svnweb.freebsd.org/changeset/ports/358374

Log:
  Deprecate print/ghostview for removal at end of July 2014

  Gentoo removed ghostview in favor of print/gv eight years ago.  The
  latter is maintained both in ports and upstream, and it's a continuation
  of the original ghostview 1.5.

  This port is unstaged, unmaintained, and segfaults.  It's time to pull
  the plug I think.

  PR: 		176044
  Approved by:	portmgr (implicit)

Changes:
  head/print/ghostview/Makefile
Comment 13 John Marino freebsd_committer 2014-06-19 10:09:11 UTC
Gentee removed ghostview in favor of gv eight years ago.  Time to follow suit, I think.

Deprecated, to be removed on 1 Aug 2014