ASF GitHub Bot commented on TINKERPOP-1522:

Github user spmallette commented on a diff in the pull request:

    --- Diff: docs/src/upgrade/release-3.4.x.asciidoc ---
    @@ -28,3 +28,8 @@ NEED AN IMAGE
     Please see the 
 for a complete list of all the modifications that are part of this release.
     === Upgrading for Users
    +==== Change in order of select() scopes
    +The order of select scopes has been changed to: maps, side-effects, paths
    +Previously the order was: side-effects, maps, paths - which made it almost 
impossible to select a specific map entry if a side-effect with the same name 
    --- End diff --
    two things: 
    1. Could you go the extra step and add an example of the behavior 
before/after the change. I think people would understand it better if they 
could see the example (remember the source in upgrade is static so you should 
just copy/paste the example from your console)
    2. note that we typically add a link to the relevant JIRA ticket: 

> Order of select() scopes
> ------------------------
>                 Key: TINKERPOP-1522
>                 URL: https://issues.apache.org/jira/browse/TINKERPOP-1522
>             Project: TinkerPop
>          Issue Type: Improvement
>          Components: process
>    Affects Versions: 3.2.3
>            Reporter: Daniel Kuppitz
>            Assignee: Daniel Kuppitz
>            Priority: Major
>              Labels: breaking
> As it currently stands, side-effects have the highest priority when a key is 
> {{select()}}'ed. I just ran into a problem where this behavior was more than 
> disadvantageous:
> {code}
> gremlin> g = TinkerGraph.open().traversal()
> ==>graphtraversalsource[tinkergraph[vertices:0 edges:0], standard]
> gremlin> g.withSideEffect("a", ["a": 
> "marko"]).inject(1).select("a").select("a") // expected result is "marko", 
> not "[a:marko]"
> ==>[a:marko]
> {code}
> In my use-case the map keys were not predictable, hence it's almost 
> impossible to prevent a key name collision. IMO maps (and paths) should take 
> precedence over side-effects.
> It is still possible to get the nested {{a}} key, but I'm pretty sure that 
> the common Gremlin user won't be able to come up with this query:
> {code}
> gremlin> g.withSideEffect("a", ["a": "marko"]).inject(1).select("a").
>            map(unfold().filter(select(keys).is("a")).select(values))
> ==>marko
> {code}

This message was sent by Atlassian JIRA

Reply via email to