Summary: | when viewing .dvi with included .eps files, these .eps figures are shown in wrong places | ||
---|---|---|---|
Product: | [Applications] okular | Reporter: | Andrey Grozin <A.G.Grozin> |
Component: | DVI backend | Assignee: | Okular developers <okular-devel> |
Status: | CONFIRMED --- | ||
Severity: | normal | CC: | kensington, luigi.toscano, mark.vanrossum, martin.sandsmark |
Priority: | NOR | ||
Version: | 0.20.2 | ||
Target Milestone: | --- | ||
Platform: | unspecified | ||
OS: | Linux | ||
Latest Commit: | Version Fixed In: | ||
Attachments: |
bug.tex
a.eps bug.dvi in okular bug.ps in okular bug.pdf in okular bug.dvi in xdvi |
Description
Andrey Grozin
2015-01-11 16:11:47 UTC
Created attachment 90351 [details]
bug.tex
Created attachment 90352 [details]
a.eps
Test case checked on current Debian testing-almost-stable 8 Jessie, Okular 0.20.2 with KDE Platform 4.14.2. I checked the test case on Okular 0.14.3 (shipped with KDE SC 4.8.4, on Debian 7.8 Wheezy), and indeed the rendering was the same on the DVI, PS and PDF versions. There haven't been many changes to the DVI backend, quite strange. Anyway, there is a regression somewhere. This bug still exists in the current kde5 version from git (in Gentoo it is called okular-5.9999, don't know exactly to which branch it corresponds). Extremely annoying for everybody, who (like me) wants to insert pictures (e.g., Feynman diagrams) into formulas. Is there any hope that somebody could do bisection and find which commit has introduced this bug? The Frameworks version is unofficiale ("frameworks" branch) and that code was basically not touched. If you have some resources (which I don't right now), you may want to try the versions (kdelibs4-based) between 0.14.3 and 0.20.2, were the issue appeared. What versions of ghostscript do you have? Currently 9.19. When I filed this bug, surely, I had an older version. I tried to do bisection. Compiled okular-0.14.3 (which was in KDE SC 4.8.4) which was mentioned in comment #3. But the bug was already present. I have not compiled the whole KDE SC 4.8.4 together with all libraries and programs on which okular depends. So, it seems that the bug is *not* in okular itself. Behaviour of some external library or program has changed some time ago, and induced the buggy behaviour in okular. I'll try to compile some old version of ghostscript. Does anybody know what version of ghostscript was in Debian 7.8? I'm fairly certain that ghostscript either has a regression or somehow changed behavior. Okular doesn't render the EPS files itself, it just calls ghostscript: https://github.com/KDE/okular/blob/master/generators/dvi/psgs.cpp#L147-L292 Created attachment 157720 [details]
bug.dvi in okular
Created attachment 157721 [details]
bug.ps in okular
Created attachment 157722 [details]
bug.pdf in okular
Created attachment 157723 [details]
bug.dvi in xdvi
The bug still exists in okular-22.12.3 (my current ghostscript is 10.01.0). |