[
https://issues.apache.org/jira/browse/SOLR-7377?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14661700#comment-14661700
]
Joel Bernstein commented on SOLR-7377:
--------------------------------------
One of cool things about Streaming Expressions is that provides a high
performance way to push operations to worker nodes ...
AND
It's a query language that can be used on it's own.
Streaming Expressions can model any operation that be done with the Streaming
API.
The new parallel SQL interface, compiles SQL statements to Streaming API
objects. Those object are then serialized to Streaming Expressions and sent to
worker nodes to handle the map/reduce aggregations in parallel.
> SOLR Streaming Expressions
> --------------------------
>
> Key: SOLR-7377
> URL: https://issues.apache.org/jira/browse/SOLR-7377
> Project: Solr
> Issue Type: Improvement
> Components: clients - java
> Reporter: Dennis Gove
> Priority: Minor
> Fix For: 5.2, Trunk
>
> Attachments: SOLR-7377.patch, SOLR-7377.patch, SOLR-7377.patch,
> SOLR-7377.patch, SOLR-7377.patch, SOLR-7377.patch, SOLR-7377.patch,
> SOLR-7377.patch, SOLR-7377.patch, SOLR-7377.patch
>
>
> It would be beneficial to add an expression-based interface to Streaming API
> described in SOLR-7082. Right now that API requires streaming requests to
> come in from clients as serialized bytecode of the streaming classes. The
> suggestion here is to support string expressions which describe the streaming
> operations the client wishes to perform.
> {code:java}
> search(collection1, q=*:*, fl="id,fieldA,fieldB", sort="fieldA asc")
> {code}
> With this syntax in mind, one can now express arbitrarily complex stream
> queries with a single string.
> {code:java}
> // merge two distinct searches together on common fields
> merge(
> search(collection1, q="id:(0 3 4)", fl="id,a_s,a_i,a_f", sort="a_f asc, a_s
> asc"),
> search(collection2, q="id:(1 2)", fl="id,a_s,a_i,a_f", sort="a_f asc, a_s
> asc"),
> on="a_f asc, a_s asc")
> // find top 20 unique records of a search
> top(
> n=20,
> unique(
> search(collection1, q=*:*, fl="id,a_s,a_i,a_f", sort="a_f desc"),
> over="a_f desc"),
> sort="a_f desc")
> {code}
> The syntax would support
> 1. Configurable expression names (eg. via solrconfig.xml one can map "unique"
> to a class implementing a Unique stream class) This allows users to build
> their own streams and use as they wish.
> 2. Named parameters (of both simple and expression types)
> 3. Unnamed, type-matched parameters (to support requiring N streams as
> arguments to another stream)
> 4. Positional parameters
> The main goal here is to make streaming as accessible as possible and define
> a syntax for running complex queries across large distributed systems.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
