What's the issue?

This is a collection of issues with Dagster that I personally find important. cc @cmpadden

Blocking development

• Allow calling IOManager dynamically -- big issue in the era of Components, you basically can't load user-defined dependencies via IOManager at all. Implemented in 1.11.3
• PartitionedConfig with Pythonic (pydantic) config

Really annoying, causing sub-optimal patterns

• Support generic type hints -- that's kinda a big deal since generics are very common in modern Python, but there is a work around (omitting them and lying to your type checker)
• Partitioned asset checks -- the most serious issue right now is blocking behavior. It blocks all downstream partitions, which is just plain wrong.
• dg dev uses hard-coded venv and is not compatible with uv workspaces
• Allow mapping unpartitioned assets into partitions
  • should support mapping different partitioned assets into a single partition asset while combining partitions Mapping unpartitioned assets into partitions:

flowchart LR
up1("A") --> down("[A, B]")
up2("B") --> down("[A, B]")

Currently each of the downstream partitions would depend on every upstream asset. This means that if one of them changes, all the downstream partitions will be recomputed (by automation conditions).

Partitions union:

flowchart LR
up1("[A, B]") --> down("[A, B, C]")
up2("[C]") --> down("[A, B, C]")

This is currently not possible at all.

Would kill for this

• Add "re-execute with config" button to the Run page -- the current manual flow is very time-consuming
• Configurables backfills -- it should be easy to implement at this point!
• Support ForwardRef type hints -- also a big deal, they are very common and unlock the ability to instantly derive configs from third party packages which provide them (e.g. obstore)

What did you expect to happen?

No response

How to reproduce?

No response

Dagster version

1.10.19

Deployment type

Dagster Cloud

Deployment details

No response

Additional information

No response

Message from the maintainers

Impacted by this issue? Give it a 👍! We factor engagement into prioritization.