feat(bigquery/v2): query client based on bigquery/v2 package #12512
Conversation
product-auto-label
bot
added
the
api: bigquery
|
|
||
| // Client is a client for running queries in BigQuery. | ||
| type Client struct { | ||
| c *apiv2_client.Client |
There was a problem hiding this comment.
nit: just JobClient so you can choose a single client or the monoclient.
There was a problem hiding this comment.
when using the Storage integration, the table service is needed. It's used for fetching table schema, which in some cases is not cached, like if the user uses the AttachJob method
| rc *storagepb.BigQueryReadClient | ||
| projectID string | ||
| billingProjectID string | ||
| defaultJobCreationMode bigquerypb.QueryRequest_JobCreationMode |
There was a problem hiding this comment.
To consider: defaultPostRequest rather than a single field, and use proto merge semantics to pick up defaults.
There was a problem hiding this comment.
good point, still need to work on this one.
bigquery/v2/query/query.go
Outdated
| } | ||
|
|
||
| // state is one of a sequence of states that a Job progresses through as it is processed. | ||
| type state = string |
There was a problem hiding this comment.
I really wish we had a v2 enum to use for this. Since we don't, do we need it at all?
There was a problem hiding this comment.
I removed that enum, is not needed
bigquery/v2/query/query.go
Outdated
| Done state = "DONE" | ||
| ) | ||
|
|
||
| func (q *Query) checkStatus(ctx context.Context) error { |
There was a problem hiding this comment.
this fn name confuses me a bit. The implementation is a bit wonky since it explicitly asks for no rows but appears to have some row cache logic embedded via consumeQueryResponse.
There was a problem hiding this comment.
I was going to reuse the consumeQueryResponse method on the newQueryJobFromQueryResponse method.
| } | ||
|
|
||
| // JobReference returns the job reference. | ||
| func (q *Query) JobReference() *bigquerypb.JobReference { |
There was a problem hiding this comment.
Why isn't JobReference just a member of the object? What about stateless queries?
There was a problem hiding this comment.
good point, I can make it a member, but what happens if uses fiddle with it and set it to nil somehow ? And for stateless queries, the job reference is build using the QueryID. I need to double check if is working as intended in that part, like trying to call AttachJob to a stateless query.
bigquery/v2/query/reader.go
Outdated
| "cloud.google.com/go/bigquery/v2/apiv2/bigquerypb" | ||
| ) | ||
|
|
||
| // Reader is used to read the results of a query. |
There was a problem hiding this comment.
we have a Query, a Reader, and a RowIterator? This feels like an overlap at a high level, but we can dig in when we talk next.
There was a problem hiding this comment.
just Query and RowIterators are going to be public, pushed some changes to outline that. Reader is an interface for reading data using jobs.query or using Storage Read API.
Draft on new query experience using the new bigquery/v2 client. Will break it down into smaller PR down the line, just testing the interface across languages to settle on a good interface.