Would fetching the first page of result() not make an API call ? If so, why not reading as per #6117. _query_results of the query job does get updated by result()

@yiga2 Calling result() makes an API request, but not to tabledata.list, where we get the total_rows from. As you point out in #6117, the request to wait for the job to complete actually contains a total_rows value as well, but we don't pass that through to the RowIterator.

I agree that the way I show in this sample is a bit awkward, and I was thinking about fetching the first page automatically in #4152, but I changed my mind about prefetching, since it could mean an extra unnecessary API request if someone just cares that a query completes and not the actual result rows.

Perhaps we could find a way to pass the total_rows through to the RowIterator after it is constructed in result() to avoid the extra API request.

@tswast any further consideration on getting total_rows without an add'l API call ?

@yiga2 I still think it's a good idea. I just haven't gotten around to. We're open to PRs. The change would likely be to the result method of QueryJob in job.py.

@tswast Feel free to bundle with other enhancements as PR would be very light otherwise.

Suggested code change, after

@tswast any feedback on this ?

@yiga2 I've prepared #7622 which addresses this issue, but in a slightly more complicated way than we propose here because I wanted to also handle more cases where Client.list_rows is called directly. My PR has had one pass at review, but I'm waiting on a follow-up now that I've addressed the requested changes.

Cool - thanks Tim !