Hi Rodney,
On Mon, 19 May 2008, Rodney Smith wrote:
> We've had a user complaining about failure of ardgen, ardplot within kappa
> under the Humu release. There's lots of things here and I don't know which
> are relevant.....
>
> Under Scientific Linux 4.2, the Starlink Hokulei release (and glibc 2.3.4)
> he used ardgen to define an ard region and then used ardplot to define that
> area on a postscript image by using "epsf_p;filename.ps" as the device name.
> He then used Starlink's psmerge to produce a plot.
>
> Using a SL5.0 machine and the Starlink Humu release, 32-bit architecture he
> found that ardplot would produce a postscript image with no ard region
> defined, i.e. just a header. I corrected the error on his machine as it had
> glibc 2.5 installed but was running glibc2.3.4. However, now ardplot gives
> the error
> !! ERR_/EMS_BEGIN: STATUS set with no error report (improper use of EMS).
> ! ARDPLOT: Error plotting regions described in an ARD file.
> ! Application exit status FACERR__BADARG, error 385101 (not a facility error
> ! code)
> I noticed there was an EMS bug fix, even though the error was not supposed
> to be evident in ardplot. So I installed the fix and I still get the same error.
>
> Using a SL5.1 machine, glibc2.5 and 64-bit everything appears to work OK.
>
> However, using a SL4.5 machine, 32-bit, glibc2.3.4 and running the relevant
> verion of humu gives me a segmentation fault under ardgen(running the
> glibc2.5 version gives a floating exception, so I don't have the 2.3.4
> mislabelled). This suggests a system mismatch somewhere.
>
> Anybody else found similar problems or is it just our set-up?
I have a patch available for you to try. It's downloadable from
http://starlink.jach.hawaii.edu/patches/humu/ard-Linux-glibc2_3_4.tar.gz
When you unzip it, it'll create a lib/ directory. Copy the files in that
directory to your $STARLINK_DIR/lib directory, overwriting the files that
are already there.
Please let us know if this fixes your bug.
Cheers,
Brad.
|