Test cleanup and accidental overahul
This does a bunch of stuff, maybe too much:
- Add an associative array for the configuration scripts that each test mode uses. Before we were defining new functions for each configuration, which won't really scale well.
- Consolidate the functions to run tests a bit. Some of the functions were nearly identical, but breaking off the configuration helped here
- Set the
disable_calib
flag totrue
in the physlite configuration: there was a whole extra function to run physlite tests with that option passed as a command line flag.
- Set the
- Make the help and options for
easyjet-test
a bit more friendly:- You can set whatever log level with
-l
now. - The
usage
output is a bit more terse, but the-h
output will list the modes and which configuration file they use. - Added info on the environment variables that control the script defaults.
- You can set whatever log level with
- Add a (bash) tab completion script for the test script. I also added a hook to the standard CMake-generated environment setup that sources this. It should be ignored in zsh automatically.
- Try to bring a bit of sanity to the naming conventions for the CI tests. I also made the names shorter so they don't get truncated when you look at the dashboard. This should make it a bit easier to glance at the dashboard and then run the test locally to debug.
- Define a few environment variables to control the test script. These start with
EASYJET_*
so you can also just set them in your.bashrc
or whatever if you don't like the default behavior. - Skip the
config_only
step in the "fast" unit tests if there's a localmetadata.json
file for the input already. - Minor cleanup for flag declarations of the form
addFlag(name, lambda prev: function(prev))
. There's no point in defining a lambda that just calls a callable with one argument, it should just beaddFlag(name, function)
.
Edited by Dan Guest