Red Hat Bugzilla – Bug 52254
Latex/dvi shifts page down by 1/2 inch vs. RedHat 7.1
Last modified: 2007-04-18 12:36:08 EDT
From Bugzilla Helper: User-Agent: Mozilla/4.77 [en] (X11; U; Linux 2.4.3-12 i686) Description of problem: When running latex on relatively simple latex files, the resulting (Roswell 2) PostScript file resulting from running dvips -Ppdf on the file is shifted down on the page about half an inch vs. RedHat 7.1. I also tried GhostScript 7.0 from the GhostScript site and the problem still occurs. I will provide the latex file I have if it you think it might be unique to my file or do not want to bother with your own latex file. email: dreed@capital.edu On both the 7.1 and roswell, I have changed the file: /usr/share/texmf/xdvi/XDvi so that the one line reads: XDvi*paper: us The problem could be with ghostscript, but the dvi files differ on the two systems so I'm not certain. Version-Release number of selected component (if applicable): How reproducible: Always Steps to Reproduce: 1. latex file.tex 2. dvips -Ppdf file -o 3. view the PostScript file and compare it to the same one on RedHat 7.1 Additional info:
Yes, please attach a small file, so that we know we're looking at the same thing. Could you also attach the output of 'zcat /etc/alchemist/namespace/printconf/local.adl'? Thanks.
Created attachment 28874 [details] /etc/alchemist/namespace/printconf/local.adl
Created attachment 28875 [details] syllabus.sty
Created attachment 28876 [details] syllabus.tex
We (Red Hat) should really try to fix this before next release.
I think the change is introduced by the time the DVI is created. On a 7.1 system, a Roswell-produced DVI is converted to PS containing %%BoundingBox: 0 0 612 792 On a Roswell system, a 7.1-produced DVI is converted to PS containing %%BoundingBox: 0 0 596 842 This is the same as doing the whole lot on a 7.1 system. The main change between 7.1 and Roswell in the tetex package is to use a newer set of LaTeX macros. Perhaps that's where the change has come from. Interestingly enough, when these PS files are loaded into Ghostscript (Roswell), the Roswell-produced PS file is shown as 'Letter', while the 7.1-produced PS file is shown as 'BBox'. Also, gimp (again, on Roswell) shows the Roswell-produced PS to be 11"x8.5", while the 7.1-produced PS is 8.28"x11.7". Could it be that the change is in fact a feature rather than a bug? ;-)
It could be a "new feature"; if so, could you please let me know what changes were made to the Latex macros (or where you got them). We just installed Latex on a Sparc Solaris system at work about a week ago and it works the same as RedHat 7.1. This is going to be really annoying if I have to edit the file everytime I try to print at work vs. home :-(
The updated LaTeX macros came from the Mandrake package (from Guiseppe Ghibr's texmf-gg archive).
Ok, I am willing to admit that the RedHat 7.1 Latex and Solaris Latex are wrong. Page 85 of the "Latex Companion" by Goossens et al. shows the page layout and what the margins, etc. set with \setmarginsrb mean. Even though I requested a top margin of 1 inch, the printout from 7.1 and Solaris have a margin of less than inch. I'll check the roswell page printout tonight to see if it appears correct. Looks like I'll have to get Latex compiled for Solaris from the texmf-gg source archive :-(
Okay, I spoke to Guiseppe about this, and he thinks that it's probably a paper size bug in the dvips configuration. I'll take a closer look at it, and try things like 'dvips -t a4' to see if that makes things better.
After measuring the output of the roswell PostScript file and looking at the Latex Companion, I think the 7.1 output (and the binary Solaris files that were installed at work) are wrong and that roswell is correct. I'd really like to know how to make 7.1 produce the same output as roswell so I can try the same thing on the Solaris machine.
I tried switching from the Latex macro: \setmarginsrb{1.0in}{0.5in}{1.0in}{0.25in}% {12pt}{0.25in}{12pt}{0.5in} to: \setmargins{1.0in}{0.5in}{6.5in}{9.0in} {12pt}{0.25in}{12pt}{0.5in} and using: dvips -t letter Then both the RedHat 7.1 system and the Roswell system produce PostScript that look the same. Unfortunately, trying to print the PostScript file from the 7.1 system hangs the printer. Printing the PostScript file generated on the 7.1 system (or the roswell system) from the roswell system to the HP LaserJet 1100 (connected to the 7.1 system) works fine. This appears to be the 7.1 bug that has not been fixed yet: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=42663 I believe a member of our local Linux User Group (colug) who has contacts at RedHat talked to someone and they tracked it down to manually processing the PostScript file for printing vs. letting the filter do it (it worked when manually processing the filters). Which roswell src.rpm would I try to rebuild on the 7.1 system? I'm guessing: LPRng
I guess, from the sound of it. Marking this bug closed.
Actually, it's not LPRng (since I recompiled the src RPM on my 7.1 system. It's somewhere in the printconf RPMS and it's dependencies. There were too many RPMS that needed updated to install the roswell printconf on a 7.1 system. Yes, I agree that this bug should be closed for roswell. The 7.1 bug https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=42663 still exists, although upgrading to RedHat 7.2 whenever it comes out will solve it.