Defects Index  | Documentation Table of Contents 


97/01/27 09:00 DEFECT in GPSCOM


SYMPTOM: Gpscom failed in CORS processing
PROBLEM: Not enough sites allowed for
CORRECTION: Increase dimensions
FOUND BY: Richard Snay
FIXED BY: Bill Dillinger
SYMPTOM: Error printing routine did not specify the nature of the problem.
PROBLEM: Incorrect version of the error printing subroutine was being used by Gpscom.
CORRECTION: Use a correct version of the error printing subroutine.
FOUND BY: Richard Snay
FIXED BY: Bill Dillinger
SYMPTOM: Associated information not printed in the gpscom.snx file for sites, antennas and recievers.
PROBLEM: Additional Site information files were not being opened when the optimized compiler was used due to a mispelled and thus uninitialized variable.
CORRECTION: Correct the spelling of the variable.
FOUND BY: Richard Snay & Bill Dillinger
FIXED BY: Bill Dillinger

VERSION: Jan. 25,97
SOURCE: sizes.par, gfxnm.f, errtr.f, begin.f, agcnst.f /g1/HPUX.10/Src/Gpscom /g1/HPUX.09/Src/Gpscom
EXECUTABLE: /g1/HPUX.10/Bin/gpscom /g1/HPUX.09/Bin/gpscom
NOTES: The above changes takes care of all of the problems found when we implemented the recent version of the Gpscom program, with the exception of one. That is the question of the number of sites allowed to be solved for. Which is to some degree determined by certian system parameters such as the maximum stack size and how memory is allocated on the stack. Bill Dillinger is currently, with Ed Timmerman's help, looking into how to further increase the available memory and the number of sites. With the assistance of Ed Timmerman, I have pursued the question as to why we seemed to have a limit on the total number of stations that could be solved for in the program GPSCOM. It turned out to be another system parameter that needs to be increased once something over 135 sites is needed. The current setting in Richards version of GPSCOM is 110 sites but he had originally asked for 150 sites which failed. To achieve a limit of more than 135 sites the maximum size of the data segment must be increased. This is the system parameter "maxdsiz". The default value seems to be something slightly over 67,100,000. On "gypsy" Ed doubled the size of the parameter and the new limit in GPSCOM is now somewhere betweed 225 sites which works and 250 sites which doesn't. So the CORS machine will eventually also need to be changed as well. When we first noticed this problem on "neutrino", we also noticed, that after the error message about the memory failure, the program would appear to hang up for a long period of time. We were not able to "kill" the program at this point either. This turned out to be due to writing a very large core file. If one runs into this, it can be avoided by the command, ulimit -c 0 which will cause core files not to be written out. - Bill D



970127.gpscom
June 25, 1999
Steve Hilla