pcdshub / beams

Beamline Engineering Automation Management Systems

Beamline Engineering Automation Management Systems

Name pending.

What is it

Framework for automating actions done by operators in a modular, procedural way such that we can conretize the intended set of actions we take to permutate beamline state and ensure our systems adhere to explicitly state sequencing logic.

More to come.

High level documentation

Link

To setup

• clone to directory
• mkdir venv && python -m venv venv make venv
• source venv/bin/activate activate venv
• pip3 install requirements.txt

Makefile interface

• make test - runs tests in test directory
• make test_verbose - runs tests in test directory and prints stdout
• make update_beams - updates the locally install beams package
• make flake8 - flake check source files
• make gen_grpc - generates all needed GRPC files
• make docker_image - builds rhel7 docker image with slac-epics base and BEAMS directory
• make run_sequener - launches sequencer engine

Hanging TODOs:

• make a decorator for work functions in beams/sequencer/Worker.py that logs PID
• handle keyboard interrupt kill signal more elegantly
• can package Queue and Event in single object

beams's Issues

JSON generated trees

The best structure of the actual configuration file is slightly eluding me.

For real, interesting, multilayered trees we probably will just need to explicitly have a list of node objects within the json document that specify their parent child relationships within the tree.

This likely looks like adding:
parent and child fields to the: CheckAndDoNodeEntry class.

BEAMS/beams/tree_generator/TreeSerializer.py

Lines 81 to 85 in 0460058

	class CheckAndDoNodeEntry(_NodeEntry):
	check_and_do_type: CheckAndDoNodeType
	check_entry: CheckEntry
	do_entry: DoEntry
	n_type : NodeType = NodeType.CHECKandDO

Then making the notion of config document able to hold multiple of these classes. My unfamiliarity with apischema is throwing me off the scent here.

replumb up with GRPC

the ideal would be that GRPC is just a transport layer and we can now invoke any sequence if the corresponding config file exists.

That requires this function be properly implemented with requisite safeguards:

BEAMS/beams/sequencer/Sequencer.py

Line 34 in 0460058

job = GenerateTreeFromRequest(request)

PV Condition Nodes | Store Value In Memory for Instanteous Return

We can subclass condition nodes such that the boolean they are returning is stored in memory and we do not have to wait for a caget or other non zero return time process.

@ZLLentz has cool ideas about this with camonitor calls backs.