COVID Airborne Risk Assessment tool
f63e1d3760
This rather large change adds a layer between the underlying (immutable) model and the application. In doing so we can avoid the use of a global state (useful for the purposes of configuring multiple models in the same application later on) and it also unlocks the ability to implement an MVC-like separation of concerns - again, the intention is that when it comes to comparisons, we will just be able to re-use our application views. I was hoping that ``cara.state`` could have been avoided in lieu of using traitlets, but unfortunately I found a number of limitations which were prohibitive for its use here. Foremost of which was the lack of first-class dataclass support and the difficulty in needing either to use instances of the model (immutable) or duplicate the model and its structure in a mutable form and use the ``traitlets.Instance`` type. Instead I opted for doing it myself - the ``cara.state`` module would make a very good standalone project in the future. |
||
|---|---|---|
| app | ||
| cara | ||
| .gitlab-ci.yml | ||
| app.sh | ||
| README.md | ||
| requirements.txt | ||
| setup.py | ||
CARA - COVID Airborne Risk Assessment
Credits
Development guide
Setting up the application
The https://cern.ch/cara application is running on CERN's OpenShift platform. In order to set it up for the first time, we followed the documentation at https://cern.service-now.com/service-portal?id=kb_article&n=KB0004498. In particular we:
- Added the OpenShift application deploy key to the GitLab repository
- Created a Python 3.6 (the highest possible at the time of writing) application in OpenShift
- Configured a generic webhook on OpenShift, and call that from the CI of the GitLab repository