TST: Add failing test for GH 1423 #1424
Conversation
Add failing test and mark with xfail.
| s = odo(result, pd.DataFrame) | ||
| assert len(s) == 1 | ||
| result2 = compute(t.sample(frac=0.5), sql) | ||
| assert len(s) == 14 |
There was a problem hiding this comment.
should we check that the sorted results are the same ?
There was a problem hiding this comment.
Not sure I understand - could you explain a bit more please.
The sample() method will by default do a random sampling of the data in the SQL query by invoking ORDER BY random().
In both cases we're doing a random sampling; however when using n=100 we have LIMIT defined as 100 while for frac=0.5, the LIMIT is computed as COUNT(*) * 0.5
There was a problem hiding this comment.
sorry, maybe I incorrectly assumed that we were doing a sample of the whole table which should just shuffle it. If this is a subset then I think that we should test that all the records exist in the base table as a sanity check that we are actually sampling correcectly.
|
Thanks for adding an xfail test. This is great way to codify tracking this issue! |
|
Closing as this is subsumed in #1452. |
Add failing test and mark with xfail for GH 1423
To run the test using docker provisioning.sh script, after odo PR#428 is merged; run the ./provisioning.sh and add the blaze/examples/data as a mount point for docker so it can load data into postgres.
Example:
provisioning.sh -t ~/tmp/docker/ -m ${BLAZE_PROJ_PATH}/blaze/blaze/examples/data restart