Port improvements from ParallelProcessingTools#19
Conversation
Some use cases need to call elastic_worker, so it should be part of the API. Also rename stdout_to_master to forward_stdout, it's clearer. Non-breaking since stdout_to_master was never documented.
|
Tbh I'm not terribly familiar with the |
The name of the field is confusing. It's much cleaner for the users to call `get_connect_cmd` directly. Workers may be added in different ways to the same ElasticManager and require a different startup. Also, `get_connect_cmd` was not using printing_kwargs for it's kwargs by default, which was inconsistent.
1b9de83 to
1851da5
Compare
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## master #19 +/- ##
==========================================
- Coverage 93.33% 91.75% -1.59%
==========================================
Files 2 2
Lines 90 97 +7
==========================================
+ Hits 84 89 +5
- Misses 6 8 +2 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
|
Test are fixed now. |
|
Still tracking down an issue with persistent tasks. |
Still trying to figure out what's going on there ... |
These are owned by Distributed, there's no reason to export them from ElasticClusterManager.
Removing ElasticManager.printing_kwargs may be considered breaking, it only affects show(::ElasticManager) but it was documented in the README.
1851da5 to
c204136
Compare
It looks like that was just a local artifact of Aqua testing on ParallelProcessingTools. |
|
For future reference, |
No description provided.