#1678 Dep: Use protos of new `kubeflow.pytorch` plugin instead of legacy `pytorch` plugin Pull request opened by fg91 TL;DR In flyteorg/flyteidl@5a3a44f the kubeflow training operator plugins' protos where refactored. In this PR I update the task to use the new

flyteidl.plugins.kubeflow.pytorch_pb2.ElasticConfig

instead of the legacy

from flyteidl.plugins.pytorch_pb2.ElasticConfig

(notice the addional

kubeflow

). Without this fix, registration fails with this error:

│ /home/fabiogratz/miniconda3/envs/dev/lib/python3.9/site-packages/flytekitplugins/kfpytorch/task.py:371 in get_custom                               │
│                                                                                                                                                    │
│ ❱ 371 │   │   │   job = pytorch_task.DistributedPyTorchTrainingTask(                                                                               │
╰────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯
TypeError: Parameter to MergeFrom() must be instance of same class: expected <class 'flyteidl.plugins.kubeflow.pytorch_pb2.ElasticConfig'> got <class 
'flyteidl.plugins.pytorch_pb2.ElasticConfig'>.

Here, I used

flytekit==1.6.2

and

flyteidl==1.5.8

and the cluster was running images of version

1.6.0

. I pin

flyteidl

to at least version

1.5.1

which introduced the new

flyteidl.plugins.kubeflow

module. Type ☑︎ Bug Fix ☐ Feature ☐ Plugin Are all requirements met? ☑︎ Code completed ☑︎ Smoke tested ☐ Unit tests added ☐ Code documentation added ☐ Any pending items have an associated Issue Complete description NA Tracking Issue NA Follow-up issue NA flyteorg/flytekit ✅ All checks have passed 30/30 successful checks