[
https://issues.apache.org/jira/browse/CALCITE-5071?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17516106#comment-17516106
]
Julian Hyde commented on CALCITE-5071:
--------------------------------------
I agree with [~zabetak]. {{Schema.getTableMap}} is only supposed to contain
tables that are DIRECT descendants of that schema, and their names are unique
(modulo case-sensitivity), so there's no problem putting them in a map.
In order to find both direct and indirect descendants (e.g. all tables named
'FOO' in all sub-schemas and sub-sub-schemas) then you would need to do a
tree-walk. The result may contain several tables named 'FOO', but each would
have a unique path (e.g. '/schema1/subSchema2/subSubSchema3/FOO'). If it's not
clear, there's a strong correspondence with how file systems work.
> Schema.getTableMap cannot handle tables with the same name in
> "Schema->SubSchema->Table" mode
> ---------------------------------------------------------------------------------------------
>
> Key: CALCITE-5071
> URL: https://issues.apache.org/jira/browse/CALCITE-5071
> Project: Calcite
> Issue Type: Improvement
> Reporter: itxiangkui
> Priority: Major
>
> When I define catalog(schema) -> database(subschema) -> table three-level
> warehouse management mode, RootSchema is defined as Catalog, the example is
> as follows:
> - es
> -- db1
> --- table1
> - mysql
> -- db2
> --- table1
> When I use the !tables command under Sqlline, that will trigger the call
> getTableMap()
> RootSchema.getTableNames()
> RootSchema.getTable(String name)
> three methods,
> I extended RootSchema.getTableMap() and found that table1 with the same name
> will appear, then the !tables result set can only display one of the table
> items
> Can we consider getTableMap() to return Guava Multimap and similar data
> structures....
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
