gibsondan
gibsondan
Hi @psahgal - while we sort this out, this issue seems related - dagster uses the grpc library which seems to have some m1-specific problems: https://github.com/grpc/grpc/issues/25082 Do you know what...
But we could still be executing code in the scheduler in cloud that was generated using pipelines On Thu, Jul 28, 2022 at 7:44 PM Sandy Ryza ***@***.***> wrote: >...
That sounds like a reasonable philosophy but I can't think of any existing tests today that actually do that. There's a test suite that runs a fixed set of code...
I guess the problem i'm describing is still a problem independent from this PR
(the problem being that we wouldn't have a great way of knowing if we introduced a change in dagit that broke old users using pipeline and solid)
Current dependencies on/for this PR: * master * **PR #9678** 👈 * **PR #9762** This comment was auto-generated by [Graphite](https://app.graphite.dev/github/pr/dagster-io/dagster/9678?utm_source=stack-comment).
Actually maybe I’m on board, let me run down that option a bit more On Thu, Sep 29, 2022 at 7:36 AM Daniel Gibson ***@***.***> wrote: > Joe can the...
@jmsanders and @Ramshackle-Jamathon latest rev responds to your feedback about task definitions by changing the plan - instead of having a path where the run launcher constructs a new task...
Well this has been a journey! @jmsanders i think this is ready for perusal again
Joe can the image be overridden? If not I think that’s the big blocker here - although maybe the gRPC servers could expose a task definition to use rather than...