[jira] [Commented] (CALCITE-1814) Improve Derby support in Calcite

["Rick Hillegas (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/CALCITE-1814?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16030327#comment-16030327 ] 

Rick Hillegas commented on CALCITE-1814:
----------------------------------------

Responses to issues posed by Julian on the dev list:

Q: Were you planning on using Derby embedded or via client-server? I
would imagine that it makes no difference to the SQL dialect, a small
difference to connect strings, but possibly a larger difference to the
test infrastructure.

A: For testing, I was hoping to use an in-memory embedded Derby database. Does that sound appropriate?

Q: One option to consider, if Derby supports it: create an embedded Derby
database for the most important data sets (foodmart [1] and scott
[2]). We do this for hsqldb: we wrap each database in a jar (which is
1% code and 99% data) and deploy the databases to Maven central. If
you are able to take that option, people will be able to run against
Derby without any setup.

A: I think I'm following you. It is certainly possible to snapshot a Derby database into a jar file and unjar it later for read/write usage. If the database is only needed for read-only usage, then it can be accessed without having to be unjar'd.

> Improve Derby support in Calcite
> --------------------------------
>
>                 Key: CALCITE-1814
>                 URL: https://issues.apache.org/jira/browse/CALCITE-1814
>             Project: Calcite
>          Issue Type: Improvement
>          Components: jdbc-adapter
>    Affects Versions: 1.13.0
>            Reporter: Rick Hillegas
>            Assignee: Julian Hyde
>
> Make Derby as well supported as h2, hsqldb, MySQL, and PostgreSQL are. See, for instance: https://calcite.apache.org/docs/howto.html#running-tests
> Michael Mior suggests the following improvement:
> 1) If you're looking to add integration tests for
> Derby, the first step would be to get Derby running in the test VM image (
> https://github.com/vlsi/calcite-test-dataset). You should be able to model
> the tests after the existing integration tests for MySQL and PostgreSQL. A
> JIRA to track this would be a good idea.
> Julian Hyde suggests this additional improvement:
> 2) ne option to consider, if Derby supports it: create an embedded Derby
> database for the most important data sets (foodmart [1] and scott
> [2]). We do this for hsqldb: we wrap each database in a jar (which is
> 1% code and 99% data) and deploy the databases to Maven central. If
> you are able to take that option, people will be able to run against
> Derby without any setup.
> [1] https://github.com/julianhyde/foodmart-data-hsqldb
> [2] https://github.com/julianhyde/scott-data-hsqldb



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)