[
https://issues.apache.org/jira/browse/CALCITE-1331?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15394386#comment-15394386
]
Julian Hyde commented on CALCITE-1331:
--------------------------------------
In my opinion, these are some of Calcite's key contributions:
* Open source friendly. Being written in Java, with Apache license, in ASF. So
is the idea that code is divided into re-usable units (e.g. rules).
* More than one engine (Volcano, Hep) that run off the same rules. You can
construct a multi-phase optimization process, using different engines, cost
models and sets of traits (physical properties) in each phase. Rules can be
re-used across phases (e.g. one phase might pull up projects, another might
push them down).
* Streaming, and in particular integration of streaming and non-streaming data.
* The "calling convention" trait, and plans that cross multiple engines.
My motivations creating Calcite:
* Integrating in-memory data (and a programming language model like LINQ) with
a database-like approach (based on relational algebra)
* Hybrid queries (crossing multiple systems and memory)
* Materialized views, especially for OLAP applications, and especially dynamic
(i.e. caching query results or intermediate results, and disabled when segments
are flushed out of memory)
> Submit Calcite paper to CIDR 2017
> ---------------------------------
>
> Key: CALCITE-1331
> URL: https://issues.apache.org/jira/browse/CALCITE-1331
> Project: Calcite
> Issue Type: Bug
> Reporter: Julian Hyde
> Assignee: Julian Hyde
>
> Submit a paper on Calcite to CIDR 2017. Submission is [due
> 8/21|http://cidrdb.org/cidr2017/cfp.html].
> [~jcamachorodriguez] has [started work on the paper LaTeX
> format|https://github.com/jcamachor/calcite-latex].
> We might be able to pull in content from [previous
> talks|http://calcite.apache.org/community/#talks].
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
