Injections in entry processor

classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

Injections in entry processor

Valentin Kulichenko
Igniters,

I noticed that we don't inject resources to entry processors. This doesn't
look consistent, because do this everywhere else (closures, jobs,
listeners, etc.). But at the same time I believe it will cause performance
degradation, because we will have to inject on each operation.

Any thoughts on that? Maybe we should support it, but make it optional with
a configuration flag?

-Val
Reply | Threaded
Open this post in threaded view
|

Re: Injections in entry processor

dsetrakyan
Agree. I think we can avoid the performance degradation… If we cache the
fact that EP does not have any resource annotations, then we can skip the
injection without any performance impact, no?

On Thu, Feb 4, 2016 at 5:48 PM, Valentin Kulichenko <
[hidden email]> wrote:

> Igniters,
>
> I noticed that we don't inject resources to entry processors. This doesn't
> look consistent, because do this everywhere else (closures, jobs,
> listeners, etc.). But at the same time I believe it will cause performance
> degradation, because we will have to inject on each operation.
>
> Any thoughts on that? Maybe we should support it, but make it optional with
> a configuration flag?
>
> -Val
>
Reply | Threaded
Open this post in threaded view
|

Re: Injections in entry processor

Valentin Kulichenko
Makes sense! I create the ticket:
https://issues.apache.org/jira/browse/IGNITE-2560

-Val

On Thu, Feb 4, 2016 at 6:22 PM, Dmitriy Setrakyan <[hidden email]>
wrote:

> Agree. I think we can avoid the performance degradation… If we cache the
> fact that EP does not have any resource annotations, then we can skip the
> injection without any performance impact, no?
>
> On Thu, Feb 4, 2016 at 5:48 PM, Valentin Kulichenko <
> [hidden email]> wrote:
>
> > Igniters,
> >
> > I noticed that we don't inject resources to entry processors. This
> doesn't
> > look consistent, because do this everywhere else (closures, jobs,
> > listeners, etc.). But at the same time I believe it will cause
> performance
> > degradation, because we will have to inject on each operation.
> >
> > Any thoughts on that? Maybe we should support it, but make it optional
> with
> > a configuration flag?
> >
> > -Val
> >
>