# Operations and Conditions¶

This chapter introduces the two fundamental concepts for the implementation of workflows with the signac-flow package: Data Space Operations and Conditions.

## Data Space Operations¶

### Concept¶

It is highly recommended to divide individual modifications of your project’s data space into distinct functions. In this context, a data space operation is defined as a unary function with an instance of Job as its only argument.

We will demonstrate this concept with a simple example. Let’s initialize a project with a few jobs, by executing the following script within a ~/my_project directory:

# init.py
import signac

project = signac.init_project('MyProject')
for i in range(10):
project.open_job({'a': i}).init()


A very simple operation, which creates a file called hello.txt within a job’s workspace directory could then be implemented like this:

# operations.py

def hello(job):
print('hello', job)
with job:
with open('hello.txt', 'w') as file:
file.write('world!\n')


We could execute this operation for the complete data space, for example in the following manner:

>>> import signac
>>> from operations import hello
>>> project = signac.get_project()
>>> for job in project:
...     hello(job)
...
hello 0d32543f785d3459f27b8746f2053824
hello 14fb5d016557165019abaac200785048
>>>


### The flow.run()-interface¶

However, we can do better. The flow package provides a command line interface for modules that contain operations. We can access this interface by calling the run() function:

# operations.py

def hello(job):
print('hello', job)
with job:
with open('hello.txt', 'w') as f:
file.write('world!\n')

if __name__ == '__main__':
import flow
flow.run()


Since the hello() function is a public, top-level function within the module with only one argument, it is interpreted as a dataspace-operation. That means we can execute it directly from the command line:

## Conditions¶

In the context of signac-flow, a workflow is defined by the ordered execution of operations. The execution order is determined by specific conditions.

That means in order to implement a workflow, we need to determine two things:

1. What is the current state of the data space?
2. What needs to happen next?

We answer the first question by evaluating unary condition functions for each job. Based on those conditions, we can then determine what should happen next.

Following the example from above, we define a greeted condition that determines whether the hello() operation was executed, e.g. the hello.txt file exists:

def greeted(job):
return job.isfile('hello.txt')


Executing this workflow in an ad-hoc manner could be accomplished like this:

for job in project:
if not greeted(job):
hello(job)


This approach is fine for simple workflows, but would become very cumbersome for even slightly more complex workflows and is not very flexible. In the next chapter, we will demonstrate how to integrate operations and conditions into a well-defined workflow using the FlowProject class.