Automated bioreactor for bacteria growth.

See https://blog.krastanov.org/2016/11/15/bioreactor/ for pictures and description.

• The code assumes the existence of a single reactor, hence numerous global variables (reactor of class Reactor, db as an sqlite3 connection object, calibration as a json dict, etc.) spread over submodules (reactor, database, calibration, etc.). The number of wells is hard coded (maybe it will move to calibration in the distant future).

• Three threads are created: the scheduler from submodule scheduler; the web interface from submodule web; the temperature control in reactor. The web interface talks to the scheduler from a single location. The scheduler does not talk to anybody.

• An sqlite on-disk database is used by most threads. Threads access the database for reading and writing, relying only on sqlite's internal locks. No optimizations of disk access are done (might lead to wear of flash-based drives).

• The templating for the web UI is rudimentary, relying only on str.format. The navigation toolbar is hardcoded.

• Not much attention is paid to encoding url parameters. Non ASCII parameters are not guaranteed to work. Special symbols might explode. More testing necessary.

• Temperature control is done with a PID loop in a separate (third) thread. Some protection and resets through usbdevicesfs is enabled (requires the compilation of usbreset.c) in the case of a hangup. Additional watchdogs are possible in the Arduino, but are not currently enabled.

• Proper database normalization would be to have a single table with data measurements with a column dedicated to measurement type, but the more naive approach with multiple tables (one per measurement type) is good enough for this simple project.

Copyright (C) 2017 Hegarty, Krastanov, Racharaks