Incorrect result on edge list join query

[kikimo]

Please check the FAQ documentation before raising an issue

Describe the bug (required)

Look at the query below:

(root@nebula) [gdlancer]> match (v1)-[e*1..2]->(v2) where id(v1) in [6] match (v2)-[e*1..2]->(v1) return count(*)
+----------+
| count(*) |
+----------+
| 0        |
+----------+
Got 1 rows (time spent 2.736ms/16.574125ms)

Thu, 29 Dec 2022 14:03:49 CST

We can see that Nebula return no result, if we change the return clause into return size(e), it shows that there are 6 rows in the result:

(root@nebula) [gdlancer]> match (v1)-[e*1..2]->(v2) where id(v1) in [6] match (v2)-[e*1..2]->(v1) return size(e)
+---------+
| size(e) |
+---------+
| 2       |
| 2       |
| 2       |
| 2       |
| 2       |
| 2       |
+---------+
Got 6 rows (time spent 2.583ms/18.1315ms)

Thu, 29 Dec 2022 14:05:48 CST

in contrast the same query on the same dataset in Neo4j return 6 rows:

$ match (v1)-[e*1..2]->(v2) where v1.id in [6] match (v2)-[e*1..2]->(v1) return count(*)

╒══════════╕
│"count(*)"│
╞══════════╡
│6         │
└──────────┘

Your Environments (required)

• OS: uname -a
• Compiler: g++ --version or clang++ --version
• CPU: lscpu
• Commit id (e.g. a3ffc7d8) 967a8c9 (community edition)

How To Reproduce(required)

Steps to reproduce the behavior:

1. Step 1
2. Step 2
3. Step 3

Expected behavior

Additional context

[xtcyclist]

A bug from hash join.

[xtcyclist]

There may be a bug from RemoveAppendVerticesBelowJoinRule.

[xtcyclist]

(root@nebula) [test]> match (v1)-[e*1..2]->(v2) where id(v1) in [6] match (v2)-[e*1..2]->(v1) return size(e)
+---------+
| size(e) |
+---------+
| 1       |
| 1       |
| 1       |
| 1       |
| 2       |
| 2       |
| 2       |
| 2       |
| 2       |
| 2       |
| 2       |
| 2       |
| 2       |
| 2       |
| 2       |
| 2       |
+---------+
Got 16 rows (time spent 357346/357603 us)

Thu, 29 Dec 2022 20:57:58 CST

(root@nebula) [test]> match (v1)-[e*1..2]->(v2) where id(v1) in [6] match (v2)-[e*1..2]->(v1) return count(*)
+----------+
| count(*) |
+----------+
| 16       |
+----------+
Got 1 rows (time spent 297718/297931 us)

I'm not sure what happened. But the results are ok now. Probabaly fixed by some other PR.

[xtcyclist]

Fixed by this PR: 373a847