Possible bug: Nothing auto-releases statement

[schveiguy]

Command has a function releaseStatement which is supposed to clear the server resource handle for the prepared statement.

There's nothing that automatically does this. Shouldn't the command have a destructor that does this? ResultSequence purges the command, but does not release the statement (and it shouldn't, because you may go and re-execute a new result sequence!). In addition, there's no way to tell if it needs to release a statement (no way to see if it's a prepared statement or not).

I don't know what the protocol specifies. Maybe statements expire after some time and it's not required to clean it up? This is why I say "possible bug".

[Abscissa]

The protocol docs don't say much about it, but according to the SQL docs, it does NOT get cleared until the session is terminated (I assume that means connection):

http://dev.mysql.com/doc/refman/5.7/en/deallocate-prepare.html
http://dev.mysql.com/doc/refman/5.7/en/sql-syntax-prepared-statements.html

On the bright side, prepared statements are specific to a session, and the (adjustable) prepared statement limit is per-session, not per-server. So I don't think it's a big problem since the user is still in control of it. But still, sounds like these really should get auto-released when possible.

[Abscissa]

This should be fixed in the new design, too. The new interface for prepared statements, Prepared, is reference-counted and releases itself in its dtor.

[Abscissa]

Fixed in v1.0.0

[schveiguy]

Awesome. Sorry haven't had a chance to look at this, I've had to stick with the old API for now. But I will invest some time soon to try and migrate all my stuff over.