robertwb commented on a change in pull request #14491:
URL: https://github.com/apache/beam/pull/14491#discussion_r624152633
##########
File path: sdks/python/apache_beam/runners/dataflow/internal/apiclient.py
##########
@@ -588,11 +588,25 @@ def _stage_resources(self, pipeline, options):
'Found duplicated artifact: %s (%s)',
type_payload.path,
type_payload.sha256)
+ staged_name = hashs[type_payload.sha256]
dep.role_payload = beam_runner_api_pb2.ArtifactStagingToRolePayload(
- staged_name=hashs[type_payload.sha256]).SerializeToString()
+ staged_name=staged_name).SerializeToString()
else:
- resources.append((type_payload.path, role_payload.staged_name))
- hashs[type_payload.sha256] = role_payload.staged_name
+ staged_name = role_payload.staged_name
+ resources.append((type_payload.path, staged_name))
+ hashs[type_payload.sha256] = staged_name
+
+ if google_cloud_options.staging_location.startswith('gs://'):
+ dep.type_urn = common_urns.artifact_types.URL.urn
+ dep.type_payload = beam_runner_api_pb2.ArtifactUrlPayload(
+ url=FileSystems.join(
+ google_cloud_options.staging_location,
+ staged_name)).SerializeToString()
+ else:
+ dep.type_payload = beam_runner_api_pb2.ArtifactFilePayload(
Review comment:
Not resolved. I don't think we should be special-casing GCS here and
turning it into a URL artifact (and then, presumably on the other end,
recognising GCS "urls" to download them).
(With regard to Maven/PyPI, etc. those wouldn't have URLs but dependency
specs with version ranges that could be merged.)
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
