Alex Ganose
Alex Ganose
Ah I missed that. Yes, that should work!
I'm not against using shorter IDs, especially since they're currently encoded as string not bson uuids. Is there a way to generate those short IDs without an additional dependency?
Agreed that given the emmet limitations, having a sortable id is essential. Of the sortable options mentioned - UUID1 is 36 characters - [ULID](https://github.com/ulid/spec) is 26 characters - [Snowflake-id](https://pypi.org/project/snowflake-id/) is...
Ok, happy to go with that. @jmmshn would you be able to submit a PR? I will then release a new version of jobflow.
I would also be happy with ULID. The only drawback is that we would need to either: 1. Add ULID as a dependency 2. Copy the code into jobflow. What...
Jobflow doesn't depend on emmet though (only atomate2 does). From some quick reading, it seems like the sorting should be trivial without needing ULID as a dependency.
I agree that the wording is confusing because of the Fireworks context. That said, the detour does route back to the original workflow but the outputs can't be used. The...
I think the actual detour is still useful in contrast to addition. You might want to run some post processing and update a file in place before the next job...
Hi @computron, the options available are in this comment: https://github.com/materialsproject/jobflow/issues/358#issuecomment-1613102577 There are also options to `stop_children` and `stop_jobflow` which are the equivalent of `defuse_children` and `exit` in fireworks.
How about we change `detour` to `insert`?