best approach for parallel sessions

[epifanio]

Hi, I was wondering what would be the best approach to run a list of independent commands in parallel that use input file from the same location.

My approach is to create a mapset for each command, so that for each process i repeat something like:

import uuid
mapset = str(uuid.uuid4())

with Session(gisdb="grassdata", 
             location="location_name"):
    gcore.run_command("g.mapset", flags="c", mapset=mapset)
    
with Session(gisdb="grassdata", 
             location="location_name", mapset=mapset):
    gcore.run_command('r.process.data', 
                  param='value-for-process-i_n,
                  output=prefix_param-value-for-process-i_n)
    gcore.run_command('r.export.output', 
                  input=prefix_param-value-for-process-i_n,
                  output=prefix_param-value-for-process-i_n)

I run this for each process by changing just the value for the input parameters and then I remove such mapset after the session exit with a brute rm -rf /path/to/location_name/mapset_uuid_name

Is there an option to use grass-session with a temporary mapset that gets created when the session starts and removed when it exit?

[zarch]

@epifanio I've added this feature last week in the develop branch, see the commit: f9b4c09da74.

The commit add a new TmpSession class that do exactly what you are looking for. I'm still testing this new class locally, let me know if you have any issue using it.
Once I've properly test it I will add an example in the README and release the new stable.

[zarch]

@epifanio the TmpSession has been released in v0.5 (current stable version). Let me know if you have any feedback.

[thagorx]

Hey Zarch I'm also currently looking into heavily parallelize some code. Is there already an example in how to use TmpSession? Also I observed that for example an command like r.mapcalc(f'mapcalc = if({curr_raster} > 0, layer_1, null())', overwrite = True) creates a mapcalc file probably containing the result of the mapcalc in the base folder of the python script. Is there an option to control where this file is create rather than in the base folder?