Bugzilla Bug 200531

Error is:
epstopdf eg7.eps
Unrecoverable error: undefined in is
Unrecoverable error: undefined in .uninstallpagedevice
during emake pdf
It's a problem with ghostscript - I've got app-text/ghostscript-gpl-8.60-r1
I've seen this error on google about cups, but, as you can see it does not seem
device related.

Not enough information, and I can't reproduce here...please elaborate

Well, there's not much to add, besides the fact that I've got
texlive-recomended installed, but not complete teexlive, but this should not
matter, cause as I said, it fails as it executes `epstopdf eg7.eps`.
If I run this command as `epstopdf -d eg7.eps`, I get
* Input filename: eg7.eps
* BoundingBox comment: %%BoundingBox:
* Output filename: eg7.pdf
* Ghostscript command: gs
* Compression: on
* Ghostscript pipe: gs -q -sDEVICE=pdfwrite  -dAutoRotatePages=/None
-sOutputFile='eg7.pdf' - -c quit
* Scanning header for BoundingBox
* Old BoundingBox: 50 50 410 302
* New BoundingBox: 0 0 360 252
* Offset: -50 -50
Unrecoverable error: undefined in is
Unrecoverable error: undefined in .uninstallpagedevice

Created an attachment (id=137273) [details]
eps it chokes on

This is the eps files, generated during build  from plt, that fails to be
converted.

No problems here....with more information I meant your USE flags for gnuplot
and your emerge  --info.  Plus your exact Ghostscript, and please list
everything you have installed from TeXLive: app-portage/eix can help you here
with $(eix -I dev-texlive/*).

Created an attachment (id=137304) [details]
output of emerge --info

Well, OK, but it will be misleading - it states x86 whereas my
packages.keywords negates that in several places (ditto for my overlay).

equery l dev-texlive/
dev-texlive/texlive-basic-2007
dev-texlive/texlive-documentation-base-2007
dev-texlive/texlive-documentation-chinese-2007
dev-texlive/texlive-documentation-japanese-2007
dev-texlive/texlive-fontsrecommended-2007
dev-texlive/texlive-genericrecommended-2007
dev-texlive/texlive-langcjk-2007-r1
dev-texlive/texlive-langpolish-2007
dev-texlive/texlive-latex-2007
dev-texlive/texlive-latex3-2007
dev-texlive/texlive-latexrecommended-2007
dev-texlive/texlive-mathextra-2007
dev-texlive/texlive-psutils-2007
dev-texlive/texlive-science-2007
dev-texlive/texlive-xetex-2007

emerge -1pv sci-visualization/gnuplot
USE="X doc gd ggi latex%* pdf plotutils readline wxwindows -emacs -svga -xemacs
(-tetex%*)"

Packages only you have:
dev-texlive/texlive-documentation-chinese-2007
dev-texlive/texlive-documentation-japanese-2007
dev-texlive/texlive-langcjk-2007-r1
dev-texlive/texlive-langpolish-2007
dev-texlive/texlive-xetex-2007

Packages only I have:
texlive-bibtexextra
texlive-documentation-german
texlive-fontsextra
texlive-formatsextra
texlive-genericextra
texlive-htmlxml
texlive-humanities
texlive-langgerman
texlive-latexextra
texlive-metapost
texlive-pictures
texlive-plainextra
texlive-pstricks
texlive-texinfo


I cannot reproduce with your set of USE flags, at the moment there are other
more important things to do, so please try to reemerge Ghostscript, try a
different variant of Ghostscript and maybe you find a TeXLive package needed
for it (though I don't believe the problem lies there).  Adding tex team to CC.

Could you post your eg7.ps or is it the same for you but it works for you ?

(In reply to comment #7)
> Could you post your eg7.ps or is it the same for you but it works for you ?

 It is the same file apart from /CreationDate, /Author and %%CreationDate.

In this case I think the printing herd should be CC'd, not tex team.

(In reply to comment #9)
> In this case I think the printing herd should be CC'd, not tex team.

 tex team maintains epstopdf, so I think they should be look here first.  Could
you test with other EPS files, if you have a general problem or if it is a
special cornercase for GS, too.

I don't know if I find any eps files on my drive.
As far as I tested epstopdf is a perl script calling ghostscript with some
arguments and the command line from comment #2 :
gs -q -sDEVICE=pdfwrite  -dAutoRotatePages=/None
-sOutputFile='eg7.pdf' eg7.eps
fails with exactly the same errror, so I'm nearly 100% sure it's a ghostscript
issue. 

OK, consider this bug fixed and not fixed at the same time.
I was right, it was a problem with ghostscript.
When I emerged as of today in portage ghostscript-gpl-8.61,
gnuplot could be successfully emerged.

(In reply to comment #12)
> OK, consider this bug fixed and not fixed at the same time.
> I was right, it was a problem with ghostscript.
> When I emerged as of today in portage ghostscript-gpl-8.61,
> gnuplot could be successfully emerged.

 Would be nice if you would have tried with a remerge of ghostscript-gpl-8.60. 
Problem is not too serious, as 8.61 will be stable soon because of security
reasons.

Nah, reemerge of 8.60-r1 did not work, I've tried it yesterday.
This should probably be advertised on the forums, cause I've seen some
cups-related bugs with the same error, so they may be fixed now (in 861, of
course).

(In reply to comment #14)
> Nah, reemerge of 8.60-r1 did not work, I've tried it yesterday.
> This should probably be advertised on the forums, cause I've seen some
> cups-related bugs with the same error, so they may be fixed now (in 861, of
> course).

 Ok, we will wait a bit longer for 8.61 (the security bug is fixed in 8.60-r1)
and then stabilise...it is on my todo list...so have patience.  Thanks for the
report.

I close this one as arches started to stabilise ghostscript-gpl-8.61 (ppc,
ppc64 and x86 already done).