@@ -9,12 +9,17 @@ Glossary
9
9
of the task's parameters, so that the same task with the same parameters can be
10
10
reused.
11
11
12
+ Combiner
13
+ A combiner is used to combine :ref: `State-array ` values created by a split operation
14
+ defined by a :ref: `Splitter ` on the current node, upstream workflow nodes or
15
+ stand-alone tasks.
16
+
12
17
Container-ndim
13
- The number of dimensions of the container object to be iterated over when splitting
14
- a iterable value. For example, a list-of-lists or a 2D array with ` container_ndim=2 `
15
- would be split over the elements of the inner lists into a single 1-D state array.
16
- However, if ` container_ndim=1 `, the outer list/2D would be split into a 1-D state array
17
- of lists/1D arrays.
18
+ The number of dimensions of the container object to be iterated over when using
19
+ a :ref: ` Splitter ` to split over an iterable value. For example, a list-of-lists
20
+ or a 2D array with ` container_ndim=2 ` would be split over the elements of the
21
+ inner lists into a single 1-D state array. However, if ` container_ndim=1 `,
22
+ the outer list/2D would be split into a 1-D state array of lists/1D arrays.
18
23
19
24
Environment
20
25
An environment refers to a specific software encapsulation, such as a Docker
@@ -31,10 +36,10 @@ Glossary
31
36
or send notifications
32
37
33
38
Job
34
- A :ref: `Task ` that has been instantiated with resolved with concrete inputs.
35
- (i.e. not lazy-values or state-arrays) and assigned to a worker. Whereas a
36
- a task describes "what" is to be done and a submitter object describes
37
- "how" it is to be done, a job combines them both to describe a concrete unit
39
+ A job is a discrete unit of work, a :ref: `Task `, with all inputs resolved
40
+ (i.e. not lazy-values or state-arrays) that has been assigned to a worker.
41
+ A task describes "what" is to be done and a submitter object describes
42
+ "how" it is to be done, a job combines both objects to describe a concrete unit
38
43
of processing.
39
44
40
45
Lazy-fields
@@ -61,13 +66,18 @@ Glossary
61
66
by a split operation either at the current or upstream node of a workflow. The
62
67
size of the array is determined by the :ref: `State ` of the workflow node.
63
68
69
+ Splitter
70
+ Defines how a task's inputs are to be split into multiple jobs. For example if
71
+ a task's input takes an integer, a list of integers can be passed to it split
72
+ over to create a ref:`State-array ` of jobs. Different combinations of
73
+
64
74
Submitter
65
- A submitter object parameterises how a task is to be executed, by specifying
66
- an worker, environment, cache-root directory and other key execution parameters.
67
- The submitter
75
+ A submitter object parameterises how a task is to be executed, by defining the
76
+ worker, environment, cache-root directory and other key execution parameters to
77
+ be used when executing a task.
68
78
69
79
Task
70
- A task describeas a unit of work within the system. It represents a unit of processing , either
80
+ A task describes a unit of work to be done (but not how it will be) , either
71
81
standalone or as one step in a larger workflow. Tasks can be of various types,
72
82
including Python functions, shell commands, and nested workflows. Tasks are
73
83
parameterised, meaning they can accept inputs and produce
@@ -79,5 +89,5 @@ Glossary
79
89
80
90
Workflow
81
91
A Directed-Acyclic-Graph (DAG) of parameterised tasks, to be executed in order.
82
- Note that a Workflow object is created by a WorkflowTask, by the WorkflowTask 's
83
- `construct() ` method at runtime.
92
+ Note that a Workflow object is created by a :class: ` WorkflowTask ` 's
93
+ `construct() ` method at runtime and is not directly created by the end user .
0 commit comments