BISMG:SarahS/automated testing


 * Set up a Cron job to send and email - holocene talk to the outside world
 * export EDITOR=/usr/bin/emacs to get editor for crontab
 * crontab -e or (-l to list what is in crontab)

* EISMINT-2 (some subset of tests A through H)
 * Create bash script (/home/automat/co_and_build.sh) to
 * 1. check out a clean copy of the code
 * 2. build
 * Bill Wilscombe suggested these tests: 

* ISMIP-HOM (a subset of tests A through F, for higher-order models only)


 * Ross ice shelf (for shelf and/or HO models)


 * the "hump" test on the parallel branch


 * one or more analytical tests, as in the Bueler papers

spinup tests for Greenland and Antarctica. Here it may be less important to get a particular answer than to demonstrate that the code can run for a long time without crashing (though not so long that the test can't run overnight).
 * a glint-example test case for Greenland (I think it's important to test Glint, since it's a key part of coupled climate experiments)


 * exact restart tests--very important for climate models

We'll have to think about which dynamical schemes (e.g., SIA and HO) and which options (e.g., evolution solvers and basal physics) to specify with each case. Hopefully we can design a set of tests that give most of the code a good workout.

In addition to looking at netCDF files, the test scripts could compare diagnostic output in the log files, if that's useful. The diagnostic subroutines are in the module glide_diagnostics.F90, and it's easy to add new diagnostics.

Developers who are familiar with the various test cases can supply appropriate config files (many of which already exist). For instance, I volunteer to set up config files to test exact restart.