help needed: hsqldb shutdown race condition, how to fix without deadlock

[Lionel Elie Mamane]

On Wed, Jun 03, 2015 at 11:08:36PM +0200, Michael Stahl wrote:

> hmm... how about we solve the JDBC performance problem by, the first
> time a Java component is created, the component-factory stuff posts
> a message to the main thread via PostUserEvent with a handler that
> does a "dummy" JNI call into some Java function that calls back via
> JNI into another function that calls Application::Yield again; now
> JNI calls will never *enter* the JVM on the main thread because it's
> always *in* the JVM already - just needs cleaning up on quit, what
> could possibly go wrong...

That kind of idea is intellectually interesting to me (something like
that was in the laundry list in
https://bugs.documentfoundation.org/show_bug.cgi?id=57872#c0 ), but I
don't think I can implement it :)

-- 
Lionel