[GRASS5] grass testsuite

Andreas Lange Andreas.Lange at Rhein-Main.de
Sun Oct 15 10:22:47 EDT 2000


Hi,

i evaluated how a testsuite or an automated test for grass could be
implemented. I think that DejaGnu would provide a framework for some
tests. I just started working on it, so i can not promise any timeline
for it. 

DejaGnu uses expect and tcl-scripts to run preconfigured scripts that
can run interactive and batch test sessions. The tests can be invoked by
a Makefile target (runtests) and the output is human-readable. 
The whole testing suite conforms to POSIX 1003.3, so we can decorate us
with another buzzword. 

The testing suite can be easily extended by simply adding new test
scripts in a directory "testsuite/grass.test" or creating a new
directory with a script file. 
I could implement Markus' test script with it and i have written a test
for the coordinate conversion library. 

Another idea would be to automatically generate a list of the modules
and to invoke every module in a grass test enviroment to see if it
exists and runs with "help" command line. This would at least detect a
core dump in this condition. Module authors could provide a test script
for their particular module, but i think that is not very realistic. 

The prerequisites for this are:
DejaGnu installation (no binaries, only scripts, only needed on hosts
that run the tests)
Expect (should be installed on most Linux systems)
Tcl/Tk (already needed for tcltkgrass)
GNU make, GNU automake, GNU autoconf, other GNU tools (already needed
for building grass, at least make and tools). 

One drawback seems to be that DejaGNU is not maintained activly, i could
not find a homepage, the cvs server does not respond. But there is a
version from 2000/07/14 floating around the net that seems to work. 

Are there any objections? Any further ideas?
How should test data be included into this setup?

My main goal is to reduce the time spent with testing. With the scripts
grass can be recompiled automatically and the output of the tests mailed
to someone who investigates the fail-cases further. 

If there are no objections or better ideas, i'll check in a basic setup
in a directory "testsuite" in the next time.

cu,

Andreas
-- 
Andreas Lange, 65187 Wiesbaden, Germany, Tel. +49 611 807850
Andreas.Lange at Rhein-Main.de - A.C.Lange at GMX.net


---------------------------------------- 
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo at geog.uni-hannover.de with
subject 'unsubscribe grass5'



More information about the grass-dev mailing list