Fix imcompatible between presto and velox when hash double/float -0.0

[shenh062326]

For double/float -0.0 and 0.0, folly::hasher will return the same value.
but in presto/mysql, the return hash values are not equal. We should firstly
cast double/float to longBits, then call folly::hasher to get hash value.

[netlify]

✅ Deploy Preview for meta-velox ready!

Name	Link
🔨 Latest commit	925fb3b
🔍 Latest deploy log	https://app.netlify.com/sites/meta-velox/deploys/662081f49262eb00083e32c7
😎 Deploy Preview	https://deploy-preview-9511--meta-velox.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.

---

To edit notification comments on pull requests, go to your Netlify site configuration.

[Yuhta]

@bikramSingh91 @rschlussel Similar to NaN, how does Presto treat 0.0 and -0.0? It seems 0.0 = -0.0 returns true, but during join/groupby they are treated differently? Should we add this to our plan of NaN unification?

[spershin]

Agree that we need 1st to check what we want it to be in Presto.
Also, just curious, since bit representation of 0.0 and -0.0 look different, does folly::hasher() performa an extra check to return the same hash for 0.0 and -0.0?

[bikramSingh91]

does folly::hasher() performa an extra check to return the same hash for 0.0 and -0.0?

@spershin thats probably true based on this comment in folly

https://github.com/facebook/folly/blob/427a1ef2c1641e0cfaec5c9300487d084175b00e/folly/Range.h#L1664-L1665

[rschlussel]

@bikramSingh91 @rschlussel Similar to NaN, how does Presto treat 0.0 and -0.0? It seems 0.0 = -0.0 returns true, but during join/groupby they are treated differently? Should we add this to our plan of NaN unification?

do you have an example query to reproduce this? Earlier I had the same suspicion about group by and tried to test it out, but didn't see the behavior you are describing.

[Yuhta]

@rschlussel

> select 0.0 = -0.0;
 _col0
-------
 true

> WITH t as (SELECT * FROM unnest(array[0.0, -0.0], array[1, 2]) as _t(k, v)) select * from t join t as t2 on t.k = t2.k;
  k   | v |  k   | v
------+---+------+---
  0.0 | 1 |  0.0 | 1
 -0.0 | 2 | -0.0 | 2

[rui-mo]

With this change, do we need to normalize -0.0 and 0.0 in #9272 for Spark? @zhli1142015

[zhli1142015]

With this change, do we need to normalize -0.0 and 0.0 in #9272 for Spark? @zhli1142015

Yes, we need to also normalize -0.0 and 0.0. If this is correct behavior for Presto, then Presto and spark expect very different behaviors when handling Nans and -0.0 / 0.0. Should we continue the #9272, @mbasmanova? thanks.

[shenh062326]

This is the behavior of mysql, like presto, it also won't aggregate -0 and 0.

mysql> create table double_basic_test_2(id int,col1 double,col2 double) ;
Query OK, 0 rows affected (0.04 sec)

mysql> insert into double_basic_test_2 values(6,'0','-0');
Query OK, 1 row affected (0.04 sec)

mysql> insert into double_basic_test_2 values(5,0,0);
Query OK, 1 row affected (0.04 sec)

mysql> select sum(id),col2 from double_basic_test_2 group by col2;
+---------+------+
| sum(id) | col2 |
+---------+------+
|       6 |   -0 |
|       5 |    0 |
+---------+------+
2 rows in set (0.03 sec)

[aditi-pandit]

@czentgr, @karteekmurthys

[stale]

This pull request has been automatically marked as stale because it has not had recent activity. If you'd still like this PR merged, please comment on the PR, make sure you've addressed reviewer comments, and rebase on the latest main. Thank you for your contributions!