Fix the comparisions involving EMPTY.

[xtcyclist]

What type of PR is this?

• bug
• feature
• enhancement

What problem(s) does this PR solve?

Issue(s) number:

Fix #5432

Description:

EMPTY == EMPTY shall not return true.

How do you solve it?

EMPTY == (x), return value: 
==      EMPTY		NULL		Others
EMPTY	NULL		FALSE		FALSE

EMPTY < (x), return value:
<	EMPTY		NULL		Others
EMPTY	NULL		NULL		NULL

Tried to return null in all cases if an EMPTY value is involved in comparisions. But some cases have relied on it to return FALSE. Thus, I applied the above truth table for == and <.

Special notes for your reviewer, ex. impact of this fix, design document, etc:

I think we currently haven't explained any truth table involving EMTPY or EMTPY-like values in our documents. Right? Not sure.

Simlar to NULL (https://docs.nebula-graph.com.cn/3.4.0/3.ngql-guide/3.data-types/5.null/), we may need to explain the behaviours involving EMTPY or EMTPY-like values.

For example, queries like

MATCH (v:player)-[]-(v2) WHERE v.player.nnnnnnnnnnnnnn == v2.player.nnnnnnnnnnnnnn RETURN *

would return nothing, as the property nnnnnnnnnnnnnn does not exist (EMPTY) and EMTPY == EMPTY = NULL.

Checklist:

Tests:

• Unit test(positive and negative cases)
• Function test
• Performance test
• TCK

Affects:

• Documentation affected (Please add the label if documentation needs to be modified.)
• Incompatibility (If it breaks the compatibility, please describe it and add the label.）
• If it's needed to cherry-pick (If cherry-pick to some branches is required, please label the destination version(s).)
• Performance impacted: Consumes more CPU/Memory

Release notes:

Please confirm whether to be reflected in release notes and how to describe:

ex. Fixed the bug .....

[jievince]

This writing style is a bit strange. Could you replace it with the following?

if (empty() && v.empty()) return Value::kNullValue;
if (empty() || v.empty()) return false;

[xtcyclist]

That would leave two extra conditions for normal non-empty cases. Changed to another form instead.

[yixinglu]

it seems that u should assign false to the value. maybe you need not to check the result of value, just delete the following DCHECK.

[xtcyclist]

Removed.

[xtcyclist]

It won't pass the tck. Reverted for now.

It is an expression containing at least one EMPTY here. This place is expecting a TRUE value, if nothing more is introduced to change this part.

[abby-cyber]

vesoft-inc/nebula-docs-cn#2716
After discussion, it has been decided not to add the relevant description to the documentation, as the comparison logic for "empty" still requires further consideration.