[Tarantool-patches] [PATCH 2/2] fiber: destroy fiber.storage created by iproto

Konstantin Osipov kostja.osipov at gmail.com
Tue Dec 10 11:21:02 MSK 2019


* Vladislav Shpilevoy <v.shpilevoy at tarantool.org> [19/12/10 10:22]:
> > The danger of reporting, self-assigning and fixing a bug is that
> > ... it might be not a bug.
> 
> I agree. Except when something is obviously a bug which is
> clearly the case.
> 
> > 
> > Users should know that iproto fibers are pooled, not
> > created/destroyed on demand. So fiber local storage is simply not
> > making any sense for them.
> 
> Here I disagree. Users should not know anything about whether
> the fibers are pooled, when, and how.
> 
> If so, then please, give an example, when such a strange
> fiber.storage is needed? I just really can't come up with a
> usage case.
> 
> I did a public poll to understand what a behaviour is expected,
> and not a single person expects, that pooled vs not pooled fiber
> would affect its behaviour. Pool is like a cache. It should not
> be visible. You literally said it in another thread about SQL
> PREPARE.

You had a discussion with the community on the channel and many
people favour your fix. This is how you handle this! 

I will send a review.

-- 
Konstantin Osipov, Moscow, Russia


More information about the Tarantool-patches mailing list