Support edge CAS in arbitrary rank

[wey-gu]

Initial question: to update edge w/o knowing its rank.

insert edge follow(degree) values "player126"->"player116"@1:(13);
insert edge follow(degree) values "player126"->"player116"@2:(13);

UPDATE EDGE ON follow "player126"->"player116" SET degree = 1000 WHEN degree < 1000;

(root@nebula) [basketballplayer]> UPDATE EDGE ON follow "player126"->"player116" SET degree = 1000 WHEN degree < 1000;
[ERROR (-1005)]: Storage Error: Vertex or edge not found.

Wed, 09 Feb 2022 14:52:30 UTC

Options I could think of are as follows:

1. Use variable or pipeline to fetch ranks and specify to update edge

This is not yet supported:

go from "player126" over follow WHERE dst(edge) == "player116" yield rank(edge) as r | UPDATE EDGE ON follow "player126"->"player116"@$-.r SET degree = 1000 WHEN degree < 1000

[ERROR (-1004)]: SyntaxError: syntax error near `UPDATE'

update_edge_sentence
    // ======== Begin: Compatible with 1.0 =========
    : KW_UPDATE KW_EDGE vid R_ARROW vid KW_OF name_label
      KW_SET update_list when_clause yield_clause {
        auto sentence = new UpdateEdgeSentence($3, $5, 0, $7, $9, $10, $11);
        $$ = sentence;
    }
    | KW_UPSERT KW_EDGE vid R_ARROW vid KW_OF name_label
      KW_SET update_list when_clause yield_clause {
        auto sentence = new UpdateEdgeSentence($3, $5, 0, $7, $9, $10, $11, true);
        $$ = sentence;
    }
    | KW_UPDATE KW_EDGE vid R_ARROW vid AT rank KW_OF name_label
      KW_SET update_list when_clause yield_clause {
        auto sentence = new UpdateEdgeSentence($3, $5, $7, $9, $11, $12, $13);
        $$ = sentence;
    }
    | KW_UPSERT KW_EDGE vid R_ARROW vid AT rank KW_OF name_label
      KW_SET update_list when_clause yield_clause {
        auto sentence = new UpdateEdgeSentence($3, $5, $7, $9, $11, $12, $13, true);
        $$ = sentence;

...

rank: unary_integer { $$ = $1; };

2. Introduce wildcard rank i.e. @*

UPDATE EDGE ON follow "player126"->"player116"@* SET degree = 1000 WHEN degree < 1000;

Question

Could we support 1 or 2 or other options?

[wey-gu]

If either 1 or 2 is doable, I would like to take this task to learn C++ and contributing core ;)

[Sophie-Xie]

If either 1 or 2 is doable, I would like to take this task to learn C++ and contributing core ;)

@CPWstatic Hello, what do you think?:)

[sworduo]

Maybe support rank comparison in where clause is better. Suppose an edge schema named test_edge, then rank(test_edge) indicates the rank of edge test_edge.
In this case, update specified edges can be written as
UPDATE EDGE ON "player126"->"player116" SET degree = 1000 WHEN degree < 1000 and rank(test_edge) >""
Futhermore, the following issue will be resolved toghther.
#3155