[
https://issues.apache.org/jira/browse/METAMODEL-189?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15028981#comment-15028981
]
Kasper Sørensen commented on METAMODEL-189:
-------------------------------------------
I would certainly say that yes, it is inline with development objects to
"design schema for a schemaless database". That same is the case for
ElasticSearch, CouchDB, HBase and other databases that we support. The reason
for this is pretty central to MM: That we try to build a data abstraction layer
which is uniform and thus makes apples and oranges behave in similar ways.
But that said, there should also be room for sensible deviations of that rule.
I think that especially the case of insert could be a case where we relax the
requirement for existance of fields.
The unfortunate thing would still be that there is a slight difference then.
And some of the things we do get with a schema is better validation and
auto-correction options. For instance we can better auto-correct text-casing if
the user types "foo" but actually the column name is "FOO". That's an example
of something that MetaModel right now does for MongoDB users which (I think)
help them. And if we go for a schema-less insert style then I guess that would
go away.
> Metamodel tries to look into mongodb schema before insert or query
> ------------------------------------------------------------------
>
> Key: METAMODEL-189
> URL: https://issues.apache.org/jira/browse/METAMODEL-189
> Project: Apache MetaModel
> Issue Type: Bug
> Reporter: Hosur Narahari
>
> MongoDB is schemaless database but metamodel checks schema before insert or
> query of collections. This causes failure to insert or query data even if it
> is a valid one.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
