Writing a Backend

What is a backend

A backend is simply a store for objects. It needs to support four operations:

  1. Saving Python objects associated with a key (a key is a string)

  2. Loading Python objects by their key

  3. Create a lock (identified by a key)

  4. Release a lock (identified by a key)

There are a few other operations, like deleting an entry, that are useful, but not strictly necessary.

Backends are identified by a URI type string, generically called jugdir. For example, to connect to a redis backend, use redis:host:port/database. Your backend should support a similar scheme.

How to write a backend

You can start with the file jug/backends/base.py which provides a template with documentation. Implement the functions in there.

This module details all the operations that are necessary to implement a jug backend.

It can be used as a starting point (template) for writing new backends.

class jug.backends.base.base_lock
abstract fail()

Mark a task as failed. Should have no effect if the task isn’t locked

Returns:
failedbool

Whether the task was marked as failed

abstract get()

Try to atomically create a lock

Parameters:
None
Returns:
lockedbool

Whether the lock was created

abstract is_failed()

Returns whether this task is marked as failed.

This code is not race-condition free. It may happen that by the time this function returns, the failed lock has been released.

Returns:
failedboolean
abstract is_locked()

Returns whether a lock exists for name. Note that the answer can be invalid by the time this function returns. Only by trying to acquire the lock can you avoid race-conditions. See the get() function.

This function is provided only because it might be possible to have a fast check before calling the expensive locking operation.

Returns:
lockedboolean
abstract release()

Releases lock