[
https://issues.apache.org/jira/browse/CALCITE-2345?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16507119#comment-16507119
]
Julian Hyde commented on CALCITE-2345:
--------------------------------------
bq. I took the liberty to create zips-mini.json for ES adapter test (see 716).
It is a subset of zips but reduced to 149 records: top 3 cities by population
for each state.
Perfect.
bq. 1. Should all adapters use identical dataset
Only if it makes sense for that adapter. Mongo and ES and Drill could use a
flat-ish JSON data set; Druid could not.
bq. 2. Should unit and integration tests share same dataset?
Yes, if possible. You can have extra integration tests with extra-large data
sets, but we want to cover as much functionality as we can in the unit tests
(i.e.the tests that are run every commit).
bq. 3. How to share this static file among calcite test modules ? (currently it
is cloned)
Cloned will do for now.
Or maybe you could put it in core/src/test/resources, then other Calcite
modules can use if via test-jar.
bq. 4. For document databases, should we continue using zips schema or create
something more complicated
Both. Mini-zips is good enough for 80% of functionality, but a rich nested data
set would be good. Preferably with nesting (instances of an element of the same
type within itself). A few KB should be sufficient.
> add tests using Fongo to Mongo Adapter
> --------------------------------------
>
> Key: CALCITE-2345
> URL: https://issues.apache.org/jira/browse/CALCITE-2345
> Project: Calcite
> Issue Type: Improvement
> Reporter: Andrei Sereda
> Assignee: Michael Mior
> Priority: Major
>
> Better test coverage for unit tests using
> [Fongo|https://github.com/fakemongo/fongo] which is in-memory implementation
> of Mongo API.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
