Raise Exceptions on Timeout about bigquery-python HOT 4 CLOSED

I'm not sure if I like this over the current design. For example, with query, it's perfectly reasonable to set timeout=0. This simply means the query is submitted and the results will be fetched at some point in the future. It wouldn't make sense to force people to catch an exception there. I think the current design encourages people to think asynchronously, which they should.

That's not to say the API can't be improved. Maybe using futures would make more sense?

job_id, results_future = client.query('SELECT * FROM dataset.my_table LIMIT 1000')
try:
    results_future.result(timeout=5)
except TimeoutException:
    print "Timeout"

Just something to think about.

from bigquery-python.

How about setting query timeout default value to 0 so that it will run asynchronously by default or when the user wants to?

My suggestions was because there isn't a straightforward way to know if a query is returning empty list due to timeout or due to empty result.

If users wants to run asynchronously, they can still use

job_id, _results = client.query('SELECT * FROM dataset.my_table LIMIT 1000')
# Do other stuffs
complete, row_count = client.check_job(job_id)
if not complete:
    print "Timeout

If they want to run in synchronous manner with timeout, they can use

try:
    job_id, results = client.query('SELECT * FROM dataset.my_table LIMIT 1000', timeout=5)
except BigQueryTimeoutException:
    print "Timeout"

Futures is another possible way but I'm biased here and I prefer the appoarch above =). For my use case, I need to run queries in a synchronous way with timeouts since the second query depends on the first query. Async would still work here, though not as clean as I wanted it to be.

from bigquery-python.

Yes, I like the synchronous approach you describe and defaulting to async. That makes sense to me.

from bigquery-python.

Great, I'll follow up with a PR

from bigquery-python.