Bug #23106 | Invalid View definition error is too vague | ||
---|---|---|---|
Submitted: | 9 Oct 2006 13:32 | Modified: | 18 Jan 2007 0:17 |
Reporter: | Mark Leith | Email Updates: | |
Status: | Won't fix | Impact on me: | |
Category: | MySQL Server: Errors | Severity: | S3 (Non-critical) |
Version: | OS: | ||
Assigned to: | Evgeny Potemkin | CPU Architecture: | Any |
Tags: | error messages, errors, Views |
[9 Oct 2006 13:32]
Mark Leith
[19 Oct 2006 19:20]
Chad MILLER
We discussed this in the support/development conference-call today. Here are my opinions: If we intend to make a statement about the existence of a table, view, column, et c., then we should go through the same permission mechanism as the mundane test of its existence. That is, abstracting a query behind a view MUST NOT confer any new privileges that the DEFINER/INVOKER would not normally have, and we MUST NOT report error messages that the INVOKER could not normally see (and to chase down the difference between what this invoker should be able to see versus what the definer should be able to see could be difficult and expensive (imagine daisy-chained views)). For SQL SECURITY DEFINER views, the vague error message is not a bug. For SQL SECURITY INVOKER views, the error message technically could perhaps be better, but would have the side-effect of making the error messages a user should expect depend on SQL SECURITY setting of the view(s).