[
https://issues.apache.org/jira/browse/SOLR-10933?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16057631#comment-16057631
]
Susheel Kumar commented on SOLR-10933:
--------------------------------------
You saved my day, Joel. I was struggling on this bit and after noticing this
issue, i changed my variables to single alphabet the problem disappeared.
let(a=fetch(collection1,having(rollup(over=email,
count(email),
select(search(collection1,
q=*:*,
fl="id,business_email",
sort="business_email asc"),
id,
business_email as email)),
eq(count(email),1)),
fl="id,business_email as email",
on="email=business_email"),
b=fetch(collection1,having(rollup(over=email,
count(email),
select(search(collection1,
q=*:*,
fl="id,personal_email",
sort="personal_email asc"),
id,
personal_email as email)),
eq(count(email),1)),
fl="id,personal_email as email",
on="email=personal_email"),
c=hashJoin(get(a),hashed=get(b),on="email"),
d=hashJoin(get(b),hashed=get(a),on="email"),
e=select(get(a),id,email),
get(e)
)
> LetStream variables are not evaluated in proper order
> -----------------------------------------------------
>
> Key: SOLR-10933
> URL: https://issues.apache.org/jira/browse/SOLR-10933
> Project: Solr
> Issue Type: Bug
> Security Level: Public(Default Security Level. Issues are Public)
> Reporter: Joel Bernstein
> Assignee: Joel Bernstein
> Fix For: master (7.0), 6.7
>
> Attachments: SOLR-10933.patch
>
>
> The LetStream is currently using a HashMap to hold its variable mappings.
> This is problematic because the ordering of the variables will be lost as
> they are evaluated. The test cases pass currently because single letter
> variables in ascending order are used which by luck caused the variables to
> be evaluated in order.
> There is a very simple fix for this which is to use a LinkedHashMap to hold
> the variables to ensure they are evaluated in the order that they are
> received.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
