Thread: SPARC/Solaris to PC/SuSE problems

Started: 2005-12-28 02:16:10
Last activity: 2006-01-03 00:53:46
Topics: SAC Help
Jim Lawson
2005-12-28 02:16:10
We have changed from SPARC/Solaris to PC/SuSE Linux.
Since this change we have several problems with
SAC. The problems persist for the following
versions of SAC: 00.59.48 100.00
and the following SuSE OS: 8.0pro,
8.2pro 9.3pro.

Problem 1. Whenever a SAC graphics
window is covered totaly or partly,
and is reexposed, the reexposed
parts are blank. Putting the
window "always on top" or frequent
replotting are the only workarounds
we have found.

Perhaps this is an XFree86
config problem. I have not found
any other application which acts
this way.

Problem 2. After "plotpk" the
SAC window will only display a
small "+", not a full crosshairs.

Problem 3. I can't use rollback.
It always gives "out of memory"
and sometimes other statements.
Because most of our Linux boxes
have ca 1GB RAM and ca 1GB swap,
but will not rollback even one
tiny SAC file, I do not believe
we are truly out of either RAM
or swap.

Problem 4. The "color" statement
seems to have little or no effect.
I seem to only be able to plot
black and white.

Problem 5. Help followed by a
SAC command which is mispelled
exits with "segmentation fault".

Help!!!

Jim

Jim Lawson
Oklahoma Geological Survey
Observatory
jim<at>okgeosurvey1.gov
www.okgeosurvey1.gov

  • Kris Walker
    2005-12-27 20:04:26
    On Dec 27, 2005, at 10:16 AM, jim<at>okgeosurvey1.gov wrote:

    We have changed from SPARC/Solaris to PC/SuSE Linux.

    Congrats. As you probably know, you may not experience immediate
    benefits, except maybe for the speed factor, but they'll start to shine
    through as you climb the linux learning curve.

    Since this change we have several problems with
    SAC. The problems persist for the following
    versions of SAC: 00.59.48 100.00
    and the following SuSE OS: 8.0pro,
    8.2pro 9.3pro.

    Problem 1. Whenever a SAC graphics
    window is covered totaly or partly,
    and is reexposed, the reexposed
    parts are blank. Putting the
    window "always on top" or frequent
    replotting are the only workarounds
    we have found.

    I had same experience with redhat versions >=6.2 if I recall. I don't
    recall my experience with redhat v. <6.2. I think this is due to the
    graphics drivers. If there were better graphics drivers used, it
    probably wouldn't be a problem.


    Perhaps this is an XFree86
    config problem. I have not found
    any other application which acts
    this way.
    Problem 2. After "plotpk" the
    SAC window will only display a
    small "+", not a full crosshairs.


    I don't remember what happens on my linux boxes (and can't access them
    currently), but on my mac os X, I get the full crosshairs, but then get
    the small "+" after I type "q" to quick picking.

    Problem 3. I can't use rollback.
    It always gives "out of memory"
    and sometimes other statements.
    Because most of our Linux boxes
    have ca 1GB RAM and ca 1GB swap,
    but will not rollback even one
    tiny SAC file, I do not believe
    we are truly out of either RAM
    or swap.

    Interesting. I've never used this command before, but I have found
    other similar memory-flavor bugs as well.
    http://sail.ucsd.edu/~walker/andmore.htm (look for SAC link--but will
    be down until Jan. 3 for security reasons). I recall the specific
    problem of spectrograms calls and do loops crashing sac2000. It would
    be a good if the developers of SAC could use a good memory leak finder
    like "purify" during testing on each supported platform.


    Problem 4. The "color" statement
    seems to have little or no effect.
    I seem to only be able to plot
    black and white.

    I've never been able to plot color.


    Problem 5. Help followed by a
    SAC command which is mispelled
    exits with "segmentation fault".

    I recall having this problem on some installations but not on others.

    You will no doubt encounter more similar pains and annoyances as you
    further program in SAC. I have found that using shell scripts to do
    most of the data flow/processing organization (piping suites of sac
    commands output with "echo" into sac) is a way to get around most of
    the memory-related problems, which are the most severe problems for me.
    As far as graphics go, I'm not sure there is any workaround.

    Best,
    kris

    ================================================================
    Dr. Kristoffer T. Walker http://sail.ucsd.edu/~walker
    Green Scholar email: walker<at>ucsd.edu
    IGPP, Scripps MC 0225 phone: 858-534-0126
    University of California, San Diego fax: 858-534-9873
    La Jolla, CA 92093-0225 room: 2201 Revelle Lab
    ================================================================


    • Frederik Tilmann
      2006-01-03 00:53:46
      It sounds like the refreshing problem has been fixed in 100.1 but for
      older versions a change in the X configuration can address this problem:

      Frederik

      -------- Original Message --------

      This email describes how to fix a problem with the display of both
      SAC2000 graphics windows on most Linux machines.

      The Problem:

      On some of the newer Linux packages (e.g. SuSe 8.0+, Redhat 8.0+,
      also depends on the chosen WM), the graphics windows of both programs
      are not buffered, ie. when moving another window on top and then moving
      it away again, the part of the display previously covered has been wiped
      out.

      Why does it do that?

      In X11, refreshing the content of windows is really the
      responsibility of the application program (e.g. SAC).
      However, in order to make life easier for the casual programmer, an
      option is provided for letting the display manager (X11) take care of
      the refreshing using the so-called backing store. Both the display
      manager and the window need to be opened specifically requesting the
      automatic refresh for it to work. For performance reasons, the
      automatic refreshing is turned off by default in many newer
      distributions.

      How is it fixed?

      Change the default X configuration to enable the backing store. You
      will need to have root permission to carry out this change. On Linux
      systems, find the file XF86Config or XF86Config-4 (e.g. for Redhat it is
      in /etc/X11 ). Save the original file, e.g.
      cp -p XF86Config XF86Config.orig

      Edit XF86Config:
      Find the Section "Device". As last line of this section(just before
      EndSection), insert the line

      Option "backingstore"

      e.g., if before it looks like:

      Section "Device"
      # no known options
      #BusID
      Identifier "ATI Rage 128"
      Driver "r128"
      VendorName "ATI Rage 128"
      BoardName "ATI Rage 128"
      EndSection

      after it should be:

      Section "Device"
      # no known options
      #BusID
      Identifier "ATI Rage 128"
      Driver "r128"
      VendorName "ATI Rage 128"
      BoardName "ATI Rage 128"
      Option "backingstore"
      EndSection

      If there is more than one Device section, carry out this change for each
      one. Reboot. The backing store should now be switched on. You can
      check with the command "xdpyinfo". The output should contain a line like:

      options: backing-store YES, save-unders YES

      (if it doesn't there is probably something wrong with your change to the
      XF86Config). This is all that is needed to enable refreshing for
      SAC2000.


      Kris Walker wrote:
      On Dec 27, 2005, at 10:16 AM, jim<at>okgeosurvey1.gov wrote:

      ...

      Problem 1. Whenever a SAC graphics
      window is covered totaly or partly,
      and is reexposed, the reexposed
      parts are blank. Putting the
      window "always on top" or frequent
      replotting are the only workarounds
      we have found.


      I had same experience with redhat versions >=6.2 if I recall. I don't
      recall my experience with redhat v. <6.2. I think this is due to the
      graphics drivers. If there were better graphics drivers used, it
      probably wouldn't be a problem.


      ...

      --
      Frederik Tilmann
      Bullard Laboratories Tel. +44 1223 765545
      Department of Earth Sciences Fax. +44 1223 360779
      University of Cambridge email: tilmann<at>esc.cam.ac.uk
      Madingley Road
      Cambridge CB3 0EZ
      UK





  • Arthur Snoke
    2005-12-27 22:28:30
    I have not heard reports from others and have not done testing on a
    Sun/Linux system, so ...

    However, many of these problems existed on other systems and were fixed in
    version 100.1. I suggest trying that version and then report back about
    problems that persist.


    • Kris Walker
      2005-12-27 20:09:40
      Hi,

      I've reported similar such problems in the past to Peter Goldstein, but
      from what I understand, they didn't have the resources to look into
      them. After a while I thought it would be good to write them all down,
      thus the webpage that is the first to come up on google searching for
      "Problems with SAC2000":

      http://sail.ucsd.edu/~walker/sac.html

      Anyway, what is version 100.1 and where is discussed? I don't recall
      hearing about this release, and the website doesn't talk about it:

      http://www.llnl.gov/sac/SAC_Info_Install/New_Feature_Release.html

      Is there a new SAC website for which I missed the announcement?

      kris

      On Dec 27, 2005, at 11:28 AM, Arthur Snoke wrote:

      I have not heard reports from others and have not done testing on a
      Sun/Linux system, so ...

      However, many of these problems existed on other systems and were
      fixed in version 100.1. I suggest trying that version and then report
      back about problems that persist.

      _______________________________________________
      sac-help mailing list
      sac-help<at>iris.washington.edu
      http://www.iris.washington.edu/mailman/listinfo/sac-help


      ================================================================
      Dr. Kristoffer T. Walker http://sail.ucsd.edu/~walker
      Green Scholar email: walker<at>ucsd.edu
      IGPP, Scripps MC 0225 phone: 858-534-0126
      University of California, San Diego fax: 858-534-9873
      La Jolla, CA 92093-0225 room: 2201 Revelle Lab
      ================================================================


Page built 09:31:53 | v.b'ea7a92