VarLenExtend pattern matching front end

[acquamarin]

We should allow the user to omit the lower bound/ upper bound of the variable length extend query.
If the user omit the lower bound of the query:
eg. (a)-[*..5]->(b), the front end should put 1 as the lower bound.
If the user omit the upper bound of the query:
eg. (a)-[*3..]->(b), the front end should put MAX_VAR_LENGTH_EXTEND_PATH as the upper bound.

Neo4j documentation for varLenExtend pattern:
https://neo4j.com/docs/cypher-manual/current/syntax/patterns/

[andyfengHKU]

I think we should just ask user to always give us lowerBound and upperBound.

If lowerBound is omitted, it's unclear whether our default is 0 or 1.
If upperBound is omitted, it usually means Kleene star which we cannot support with walk semantic.

So I think we won't solve this issue for a very very long time. Let me know if we should just close it.

[jkanche]

I would love to omit the upper bound on the queries to traverse the subgraph and collect all descendants with a specific relationship.

If the user omit the upper bound of the query:
eg. (a)-[*3..]->(b), the front end should put MAX_VAR_LENGTH_EXTEND_PATH as the upper bound.

[andyfengHKU]

I would love to omit the upper bound on the queries to traverse the subgraph and collect all descendants with a specific relationship.

If the user omit the upper bound of the query:
eg. (a)-[*3..]->(b), the front end should put MAX_VAR_LENGTH_EXTEND_PATH as the upper bound.

In such case we need a different semantic where either no node or no edge can be repeatedly visited on the path. Otherwise the computation will not be able to stop if a cyclic pattern is found. We are looking into this but it will take a bit of time for it to be available in the release.

[andyfengHKU]

Solved in PR #2041