Hi!
On Mar 1, 2009, at 7:47 PM, Stewart Smith wrote:
But if we're going to have virtual columns, we'll have to use the
parser
for that func for the virtual column.
Verification of the object created, while today happens in the parser,
could very well happen later.
The fact that we check the object, as we parse the object, is one of
the ways we don't really have a classic design. Delaying verification
until after we parsed... this would be be a lot more common (and it
would make the job of parser writers simpler by allowing them to focus
on what they do...).
When CREATE TABLE is rewritten to just output a Table::Proto we should
move verification for it out and make it a second stage (for the
moment it could live in the *COM bits of sql_parse that need to deal
with CREATE TABLE). After that we need to do the same for the rest of
the DDL.
Unfortunately, SELECT/DML is going to be a much bigger issue.
Cheers,
-Brian
--
_______________________________________________________
Brian "Krow" Aker, brian at tangent.org
Seattle, Washington
http://krow.net/ <-- Me
http://tangent.org/ <-- Software
_______________________________________________________
You can't grep a dead tree.
_______________________________________________
Mailing list: https://launchpad.net/~drizzle-discuss
Post to : [email protected]
Unsubscribe : https://launchpad.net/~drizzle-discuss
More help : https://help.launchpad.net/ListHelp
