Sakari Ikonen

Using `tags` as a Parameter name breaks flow.

1

original discussion here: https://outerbounds-community.slack.com/archives/C02116BBNTU/p1707942631741849 cause is due to the calls to `obj.flow._set_constants(obj.graph, kwargs)` in `cli.py`, as `tags` is an argument for the function, so it is not a part of...

fix: Deterministic foreach task id's for Argo Workflows

4

changes the Task ID generation for tasks in foreach branches to be deterministic. This is so that the following join steps do not require an enormous list of ID's to...

Argo Workflows template exceeds max size with preceding large foreach

8

The task_id's of the preceding foreach steps for a join appear multiple times in the Argo Workflow init containers `ARGO_TEMPLATE`, which bloats up the size significantly. With wide foreaches the...

There are some problematic names for steps due to an open bug with Argo Workflows: https://github.com/argoproj/argo-workflows/issues/12037 This affects even the provided tutorials, where a `join` step causes the flow to...

feature: custom annotations revisited

3

reintroduces the feature first merged in #1442 which had to be reverted in #1516 due to a regression that broke annotations for recording user and project info. This feature keeps...

introduce ability to add new Azure credential providers through Metaflow extensions

feature: limit argo workflow name length

1

first draft for fixing argo workflows workflow name maxLength. Previously 253 characters have been allowed, but this causes launching workflows through Argo UI to fail for names longer than 63...

The currently prominent way of running the checks for a flow (including linting) for a platform specific CLI is to call `obj.check` in the entrypoint command, for example `argo_workflows()`. This...

refactors MetaflowEnvironment init and changes one getter into an instance variable

feature: add argo workflows list-runs command

2

- add command for listing runs (and filtering by status) for argo-workflows addresses part of #1387 Note: due to API limitations, this feature required adding a custom label to workflows...