Bug 416321

Created attachment 281821 [details]
Postscript file which spooks ghotscript-8.60

Created attachment 281831 [details]
errors from an attempt to view this PS with evince on F8

A check with rawhide ghostscript-8.61-5.fc9 and feeding a sample file to
'gs -q -dBATCH -dPARANOIDSAFER -dNOPAUSE -sDEVICE=lj4dith -sOutputFile=- -'
command, like printing could do, brings the same kind of errors.

Interestingly enough 'ps2ascii' nearly works.  "Nearly" in that sense that
it does produce a text from the attached PS file but it also has some
additions; like "Wa rnig: ThsfomtupedwcvAbaR." and "Reader. Ul".

The current workaround is to back off ghostscript to a version from F7
repositories and then viewing and printing does work.  No recompilation
required.  Although "ps2ascii additions" are present too.

In my testing ghostscript 8.61 (I have few more patches than Fedora)
works fine so does the new gsview 4.9. Failing programs are: evince,
kghostview, gv (just shows a blank page).

My impression is that these programs make a mistake in invoking gs 8.61
in this particular case otherwise gsview 4.9 would not work either and
version 4.9 was just announced to support gs 8.61. There must be something
different.

> I have few more patches than Fedora
Quite possibly they make this difference.  When I am trying,
with ghostscript-8.61-5.fc9, a command from comment #3 there is
a long pause and after that "Error: /undefined in eartomark" and
all that followup jazz.

ghostscript-8.15.4-3.fc7 produced 19213 bytes of output in
real    0m0.546s
user    0m0.490s
sys     0m0.042s

Does gs itself show the file correctly?

I do get the same error if I run comment #3 but not with gs alone and
not with gv (EXCEPT turn DSC off under gv options) and gsview.

Also, if I run:
 
gs -q -dBATCH -dPARANOIDSAFER -dNOPAUSE -sDEVICE=lj4dith -sOutputFile=- file.ps

seems to work! Clearly the error is coming from the way each application is
feeding the file to gs. This is why some are working and some not. There may
have been a change in gs or it is a bug. Investigation continues.....

OK. I found this under ghostscript bugzilla:

      http://bugs.ghostscript.com/show_bug.cgi?id=689577

The workaround is somewhat involved. Someone needs to look into this.

I filed bugs with cups and foomatic for the needed changes there. In the
mean time the line:

gs -q -dBATCH -dPARANOIDSAFER -dNOPAUSE -sDEVICE=lj4dith -sOutputFile=- -_

works fine (note the underscore after -).

Since you are an evince user could you file a bug with them and direct them
to this bug.
Thanks

Yes, commands from comment #6 and comment #8 do work; at least for
ghostscript-8.61-5.fc9. gs and gsx (that what I able to find; I guess
that gsx is a gsview equivalent) show, on rawhide, my sample file.
Thanks for bugs.ghostscript.com reference.  I guess that it explains
what is happening here.

Contrary to some suggestions on ghostscript bugzilla turning off DSC
in gv does not seem to have any effects; always a blank page.

> Since you are an evince user ...
Actually you can hardly accuse me of that. :-)  I just found that
evince provided a convenient way to demonstrate the problem.  This
is really not an evince bug but one in ghostscript.

The fix for this was committed to ghostscript trunk. I am attaching the patch.
However, this may come back at some point in the future. The fixes that add
an ""_" after "-" are still good and may prevent a future problem but this is
up to each packager to decide.

Created attachment 282271 [details]
Patch for fixing pipe problem

I do not know much about ghotscript internals but that fix smells
awfully like a hack.  Something else will drop into a file another
eexec block instead of "the following eleven lines" from Adobe and
the "magic constant" will not work in such case.  Don't you think
that this is a distinct possibility?

OTOH a comment 17 from a discussion on ghostscript bugzilla does
suggest that foomatic should use '-_' instead of '-' for stdin
anyway (and the same applies for other non-interactive uses).

I agree and urge foomatic to modify foomatic-gswrapper script as suggested
as well as cups to modify pstoraster. Ghostscript has already modified
their pstoraster and pstopxl scripts to use "-_" for keeps!

Could we get some response for this bug?
Thanks

New package built, should be in updates-testing soon.

ghostscript-8.61-4.fc8 has been pushed to the Fedora 8 testing repository.  If problems still persist, please make note of it in this bug report.
 If you want to test the update, you can install it with 
 su -c 'yum --enablerepo=updates-testing update ghostscript'

I am afraid that with ghostscript-8.61-4.fc8 I still see on my test file
while trying
gs -q -dBATCH -dPARANOIDSAFER -dNOPAUSE -sDEVICE=lj4dith -sOutputFile=- -
the same errors as before.  It does work if I will use '-_' for input but
that we knew.

Also evince spills familiar errors and both evince and gv show a blank
page.  OTOH 'gs test_file.ps' does produce an image of a page with
an expected text displayed.  So this did change.  Not much yet ...

I think he meant ghostscript-8.61-5.fc8

> I think he meant ghostscript-8.61-5.fc8

Well, comment #15 says ghostscript-8.61-4.fc8 and that is what I found
in updates-testing.  Granted, looking closer that package is a bit old
but this is also what is listed in "Fedora updates-testing report" from
today.  Sigh!

Indeed, my mistake.  I've pushed ghostscript-8.61-5.fc8 to the updates system
now, and it should get into the next batch of updates.  Sorry!

ghostscript-8.61-5.fc8 has been pushed to the Fedora 8 testing repository.  If problems still persist, please make note of it in this bug report.
 If you want to test the update, you can install it with 
 su -c 'yum --enablerepo=updates-testing update ghostscript'

> ghostscript-8.61-5.fc8 has been pushed ....
WORKSFORME on various tests with samples which gave me troubles
before.  Although "testing" repos seem to lag quite a bit behind,
despite of annoucements, so this was on a copy from koji. Thanks!

ghostscript-8.61-5.fc8 has been pushed to the Fedora 8 stable repository.  If problems still persist, please make note of it in this bug report.