[
https://issues.apache.org/jira/browse/CALCITE-498?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14236236#comment-14236236
]
Julian Hyde commented on CALCITE-498:
-------------------------------------
There are two major tasks here:
1. Make the parser extensible (at run time, without changing the grammar)
2. Convert SqlTypeName from an enum into a class (or perhaps interface) and
rename the existing enum to BuiltInSqlTypeName.
This might be a lot of work. Since it is so rare that a new DB adds types we've
not seen before, it's simpler to just add them when they occur. We'd assign
them a numeric ID, something not used in java.sql.Types.
> Allow non standard SQL type names to be defined
> -----------------------------------------------
>
> Key: CALCITE-498
> URL: https://issues.apache.org/jira/browse/CALCITE-498
> Project: Calcite
> Issue Type: Improvement
> Reporter: James Taylor
> Assignee: Julian Hyde
> Labels: phoenix
>
> In order to provide a means of mapping to already existing data, systems may
> need to define non standard types. For example, Phoenix defines its
> UNSIGNED_* types
> (http://phoenix.apache.org/language/datatypes.html#unsigned_time_type) which
> serialize/deserialze using the standard HBase serialization utility
> functions. This enables Phoenix to be used to query HBase data which was not
> created using Phoenix.
> It'd be useful as an escape mechanism, if Calcite allowed such types to be
> defined and referenced in DDL statements and CAST operations.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
