Testing¶
Test Environment¶
To create a test environment (called “test” in command below)
conda create --name test python=2
To switch to “test” environment:
conda activate test
To reset “test” environment to initial install:
conda install --name test --revision 0
Savu Training Examples¶
Training examples are documented here: trainingexamples
Simple
This can be run if in the TrainingExample root directory as follows:
savu data/24737.nxs process_lists/simple_tomo_pipeline_cpu.nxs .
Data files have to be opened and saved by savu_config to avoid import error for a module nxtomo_loader; pending a future formal fix to Savu.
Parallel
This can be run if in the TrainingExample root directory as follows assuming a writeable “/tmp2” folder exists:
savu_mpijob_local.sh data/24737.nxs process_lists/simple_tomo_pipeline.nxs . -d /tmp2
This currently fails, but is not a priority as savu-lite was not intended to run in parallel.
Current Error is:
Global Name ‘mpi4py’ is not defined
Existing Savu Tests¶
These can be run successfully from a conda installed savu-lite.
- savu_quick_tests
- savu_full_test
savu_quick_tests
This is run as part of the build. scikit-image is a pre-requisite for it to run and has therefore been included in savu-lite build.
savu_full_test
This can be run after savu-lite has been installed, success is indicated by it completing with no errors reported at its conclusion. Pre-requisites for this to run are: tifffile, pyfai, peakutils, fabio, tomopy, pymca
Savu Issues¶
Refresh.py is awaiting a fix that will be introduced in next release.
Training Example Errors
The Savu 2.4 training examples are out-of-date: trainingexamples
Refresh.py is awaiting a fix that will be introduced in next release. python process_lists/refresh.py
In the meantime this can be resolved by opening data files with savu_config and then re-saving them. Thus avoiding import error for a module nxtomo_loader.