Separate code paths for Java 8 and Java 9

[Andrey Kuznetsov <st...@gmail.com>]

Igniters,

As [1] proposes, a couple of new modules with version-specific code should
be created. While working on issues from [2] umbrella I saw no significant
reasons to emerge separate modules; most requirements can be covered by
tiny conditional code pieces in static init blocks of existing utility
classes.

Could someone share the ideas on the content of potential version-specific
modules?

[1] https://issues.apache.org/jira/browse/IGNITE-6730
[2] https://issues.apache.org/jira/browse/IGNITE-6728

-- 
Best regards,
  Andrey Kuznetsov.

Re: Separate code paths for Java 8 and Java 9

[Vladimir Ozerov <vo...@gridgain.com>]

Hi Andrey,

If it is possible to resolve all mentioned issues with static code blocks,
then this is definitely preferrable approach. Let's fix simple issues first
and see what to do with complex ones. Notably:
1) https://issues.apache.org/jira/browse/IGNITE-6740 - this call is on a
hot path, so reflection is not an option;
2) https://issues.apache.org/jira/browse/IGNITE-6736 - the only possible
way to fix it is to resort to ReentrantLock instead of synchronized. But
this could cause bloat and performance degradation for non-contented cases,
when synchronized might be faster.

On Wed, Dec 13, 2017 at 8:18 PM, Andrey Kuznetsov <st...@gmail.com> wrote:

> Igniters,
>
> As [1] proposes, a couple of new modules with version-specific code should
> be created. While working on issues from [2] umbrella I saw no significant
> reasons to emerge separate modules; most requirements can be covered by
> tiny conditional code pieces in static init blocks of existing utility
> classes.
>
> Could someone share the ideas on the content of potential version-specific
> modules?
>
> [1] https://issues.apache.org/jira/browse/IGNITE-6730
> [2] https://issues.apache.org/jira/browse/IGNITE-6728
>
> --
> Best regards,
>   Andrey Kuznetsov.
>