Skip to content

0.2.4 release

Choose a tag to compare

@zhubonan zhubonan released this 16 Dec 05:16
· 12 commits to master since this release

Changes

  • Greatly improved iteration speed and get_node / get_group via node and group object as attributes of GroupPathX of the objects created
    • This mode is not enabled by default, since once cache is used, changes on the data may not be reflected. We may enable it by default in the future or add modifiable configuration options.
    • Enable faster iteration using the .fast_iter property, e.g. for path in basepath.fast_iter instead of for path in basepath. The resulting GroupPathX objects will have the cache enabled.
    • The show_tree is now much faster than before.

New feature

  • Added GroupLauncher class which allows keeping N active concurrent workflows in from a pool of jobs. The user need to define a callback function that turns a node and a label into a running job plus a label. The running job will be stored in a pre-defined group path. This can be useful when it is necessary to limit the number of running jobs due to local resource or remote cluster constrains.

Full Changelog: 0.2.3...0.2.4