adnanhemani commented on code in PR #12584:
URL: https://github.com/apache/iceberg/pull/12584#discussion_r2113256999
##########
open-api/rest-catalog-open-api.py:
##########
@@ -1389,6 +1558,101 @@ class CommitTableResponse(BaseModel):
metadata: TableMetadata
+class EventsResponse(BaseModel):
+ next_page_token: Optional[PageToken] = Field(None, alias='next-page-token')
+ highest_processed_timestamp_ms: int = Field(
+ ...,
+ alias='highest-processed-timestamp-ms',
+ description='The highest timestamp processed by the server when
generating this response. This may not necessarily appear in the returned
changes if it was filtered out.\nClients can use this value as the
`after-timestamp-ms` parameter in subsequent requests to continue retrieving
changes after this point.\n',
+ )
+ events: List[Event]
+
+
+class Event(BaseModel):
+ event_id: str = Field(
+ ...,
+ alias='event-id',
+ description='Unique ID of this event. Clients should perform
deduplication based on this ID.',
+ )
+ request_id: str = Field(
+ ..., alias='request-id', description='ID of the request this change
belongs to.'
+ )
+ event_count: int = Field(
+ ...,
+ alias='event-count',
+ description='Number of events in the request / batch of events',
+ )
+ timestamp_ms: int = Field(
+ ...,
+ alias='timestamp-ms',
+ description='Timestamp when this transaction occurred (epoch
milliseconds). Timestamps are not guaranteed to be unique. Typically all events
in a transaction will have the same timestamp.\n',
+ )
+ actor_chain: Optional[List[Actor]] = Field(
+ None,
+ alias='actor-chain',
+ description='An ordered list of actors involved in the operation, with
the most direct actor (the one who actually performed the operation) first,
followed by delegating actors in order. For example, if a service account
(actor[0]) performed an operation on behalf of a role (actor[1]) assumed by a
user (actor[2]), the chain represents this delegation path.\n',
+ )
+ operation: Union[
+ CreateTableOperation,
+ RegisterTableOperation,
+ DropTableOperation,
+ UpdateTableOperation,
+ RenameTableOperation,
+ CreateViewOperation,
+ DropViewOperation,
+ ReplaceViewOperation,
+ CreateNamespaceOperation,
+ UpdateNamespacePropertiesOperation,
+ DropNamespaceOperation,
+ CustomOperation,
+ ] = Field(..., discriminator='operation_type')
+
+
+class CreateTableOperation(BaseModel):
+ operation_type: Literal['create-table'] = Field(
+ ..., alias='operation-type', const=True
+ )
+ identifier: TableIdentifier
+ table_uuid: UUID = Field(..., alias='table-uuid')
+ metadata: TableMetadata
+
+
+class RegisterTableOperation(BaseModel):
+ operation_type: Literal['register-table'] = Field(
+ ..., alias='operation-type', const=True
+ )
+ identifier: TableIdentifier
+ table_uuid: UUID = Field(..., alias='table-uuid')
+ metadata: TableMetadata
Review Comment:
I've commented this under the RegisterTableOperation. I understand your
point for CreateTableOperation, and while I'm still not sure I see the use case
for it, I think it is certainly still possible to store. I'm in agreement that
it should be of a small size. However, when registering an existing Iceberg
table, it is possible that the table has already has a large table metadata
size - and as a result, I don't think that is okay to store in a persistence
for a catalog, personally. WDYT?
I re-read the Federation and Workflow subsections of the use-case but it
does not answer the question I posed above clearly - can you expand more on why
having the `TableMetadata` on the `CreateTableOperation` and
`RegisterTableOperation`, specifically, helps the Federation and Workflow
use-cases?
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
