GRASS Concurrence

[Michael Shapiro]

What is at issue is that GRASS communicates information among the
various modules thru files. The information about what database a
user is using is contained in the file  .grassrc  in  the  user's
home directory.  If a user were to run GRASS twice using the same
home address, both session would share the .grassrc file. If  the
contents  of  this  file  were  changed  by one session, it would
automatically and instantly be changed for the  other.  This  can
happen  if  the user logs on to the same machine twice or logs in
for two different machines  that  have  the  home  directory  NFS
mounted  and  shared.  The difference is that GRASS can`t tell if
you do it  from  2  machines  (it  uses  process  ids  to  detect
concurrent usage).

However, it IS possible to get GRASS  to  use  a  different  file
other  than  $HOME/.grassrc.  In fact GRASS decides which file to
use based on the setting of the unix environment variable  GISRC.
Of  course,  the  grass4.0 command sets GISRC so if you wanted to
have two sessions each using a different  GISRC  file  you  would
need to write your own grass4.0 command and add code to set GISRC
to a unique file name.  (One  consequence  is  that  grass  would
"forget"  which  database  you were using from session to session
unless you implemented a session save/recall database).

But this still isn't quite enough. Even if you managed  to  allow
the same user to run GRASS with two distinct GISRC files, if both
sessions chose the same database location and mapset, then  there
would be trouble.  The region and mask are stored in files in the
mapset and changing them  in  one  session  would  instantly  and
automatically change them for the other session.

|
|I put out an email asking about GRASS concurrence and have received  
|no response other than one phone call telling me this problem has  
|been discussed in the past. I was also told that the problem could  
|only be overcome by creating complete multiple copies of GRASS (an  
|unsubstantiated rumor?). OK. If that's true, how many files actually  
|need to be copied? How many can just be links? If you already have  
|one running copy of GRASS on a computer (a 2-3 day process), surely  
|creating the second copy can be done using work already accomplished  
|on the first. Did anyone ever define this procedure in technical  
|detail? If so, please refer me to the proper document or former email  
|(I'm new to this exploding list).
|
|Greg Colello
|Carnegie Institution, Department of Plant Biology
|290 Panama St
|Stanford, CA  94305
|ph: 415-325-1521   fax: 415-325-6857
|gcolello at biosphere.stanford.edu
|
|


-----------------------------------------------------------------
Michael Shapiro                        U.S. Army CERL                  
email:   shapiro at zorro.cecer.army.mil  Environmental Division          
phone:   (217) 352-6511  ext 526       P.O. Box 9005                   
fax:     (217) 373-7222                Champaign, Ill. 61826-9005
-----------------------------------------------------------------