[teiid-issues] [JBoss JIRA] Commented: (TEIID-902) Restructure planning

[Steven Hawkins (JIRA)]

    [ https://jira.jboss.org/jira/browse/TEIID-902?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12499258#action_12499258 ] 

Steven Hawkins commented on TEIID-902:
--------------------------------------

Changed the resolving logic to not recursively resolve, rather all typical command expansion happens in the RelationalPlanner, which now encompasses the logic from MergeTreeNodeProcessor, GenerateCanonical, and the view expansion logic form SimpleQueryResolver.  Command caching works as described above.  

Currently this means that our old logic for updatingModelCount will not work since it's based on the resolved command form, which no longer contains the entire command tree.  Also the optimization that remove stored procedure wraps no longer works.  However, I'm inclined to leave this optimization out, since it was added only to support our notion of "storedqueries", which had designer support some time in the past.  Now it's unlikely for someone to use a stored procedure for just a stored query and it's also unlikely that there would be any performance benefit from removing the procedural wrapper (that would only be good if the stored procedure were a subquery in a relational query).

> Restructure planning
> --------------------
>
>                 Key: TEIID-902
>                 URL: https://jira.jboss.org/jira/browse/TEIID-902
>             Project: Teiid
>          Issue Type: Task
>          Components: Query Engine
>    Affects Versions: 7.0
>            Reporter: Steven Hawkins
>            Assignee: Steven Hawkins
>             Fix For: 7.0
>
>
> All transformation queries should be cached - not just views.
> All cached queries should already be validated - and not revalidated with every request.
> Cached queries should be stored individually - not as full and separate trees as they are now, where the same view command exists can exist multiple times in cache.
> The choice about whether to use a materialized view or not should also be moved to planning time - not a resolver activity.
> We should remove the concept of non-embedded command from everything except planning logic.  It's error-prone and confusing in a resolving/rewrite context.

-- 
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: https://jira.jboss.org/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira