[ https://issues.jboss.org/browse/TEIID-4928?page=com.atlassian.jira.plugin... ] Juraj Duráni edited comment on TEIID-4928 at 5/24/17 8:24 AM: -------------------------------------------------------------- [~kylin] How is it compatible? If _NAMEINSOURCE_ is not defined for column, generated N1QL is missing back quotes \(`\) around _ShortValue_ which, from my testing, does not work. Not adding table name to N1QL is a default design pattern? I think I am not confused about converting SQL to N1QL. What I am worried about is *...FROM null...* in generated N1QL if _NAMEINSOURCE_ is not defined. Obviously it is. Not recommended does not mean not supported. Defining DDL metadata in source mode has some advantages over NATIVE metadata. Moreover, what if one bucket contains documents with more than one "schema"? How can you be so sure that you will fetch enough documents to infer all related schemas? Should user set *importer.sampleSize* property to 1,000,000? Or even more? What is Teiid's "oracle" according to which you fetch documents from Couchbase? If you let user define her/his own schema, you will avoid user's frustration of Teiid not being able to infer proper schemas/tables from sample documents. And of course, this should be as simple as possible. Without requiring user to define any of OPTIONS. In fact, after talking with other QE we would not recommend using NATIVE metadata for Couchbase. was (Author: jdurani): [~kylin] How is it compatible? If _NAMEINSOURCE _is not defined for column, generated N1QL is missing back quotes \(`\) around _ShortValue_ which, from my testing, does not work. Not adding table name to N1QL is a default design pattern? I think I am not confused about converting SQL to N1QL. What I am worried about is *...FROM null...* in generated N1QL if _NAMEINSOURCE_ is not defined. Obviously it is. Not recommended does not mean not supported. Defining DDL metadata in source mode has some advantages over NATIVE metadata. Moreover, what if one bucket contains documents with more than one "schema"? How can you be so sure that you will fetch enough documents to infer all related schemas? Should user set *importer.sampleSize* property to 1,000,000? Or even more? What is Teiid's "oracle" according to which you fetch documents from Couchbase? If you let user define her/his own schema, you will avoid user's frustration of Teiid not being able to infer proper schemas/tables from sample documents. And of course, this should be as simple as possible. Without requiring user to define any of OPTIONS. In fact, after talking with other QE we would not recommend using NATIVE metadata for Couchbase. -- This message was sent by Atlassian JIRA (v7.2.3#72005)