[infinispan-dev] jdk8backported package

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

[infinispan-dev] jdk8backported package

Radim Vansa
Hi,

although we're on Java 8, there's still the package
org.infinispan.*.jdk8backported in our codebase. Is there any plan (and
possibility) to remove these and use implementation provided by runtime?
Or have we tweaked them too much, so shall we rather rename them?

Radim

--
Radim Vansa <[hidden email]>
JBoss Performance Team

_______________________________________________
infinispan-dev mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/infinispan-dev
Reply | Threaded
Open this post in threaded view
|

Re: [infinispan-dev] jdk8backported package

William Burns-3
Unfortunately this isn't really possible to remove these classes.  They have additional changes over and above the jdk8 classes.  We could move it into a different package though if we wanted to.

To be honest the only reason this is still really needed is for the BoundedConcurrentHashMapV8.  I have been talking with Dan on and off for a while about even possibly looking into using a different map implementation (such as caffeine or something else equivalent).  We haven't looked into it very seriously though.

We could look into this more if people think we want to tackle that.

 - Will

On Thu, Jan 14, 2016 at 4:34 AM Radim Vansa <[hidden email]> wrote:
Hi,

although we're on Java 8, there's still the package
org.infinispan.*.jdk8backported in our codebase. Is there any plan (and
possibility) to remove these and use implementation provided by runtime?
Or have we tweaked them too much, so shall we rather rename them?

Radim

--
Radim Vansa <[hidden email]>
JBoss Performance Team

_______________________________________________
infinispan-dev mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/infinispan-dev

_______________________________________________
infinispan-dev mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/infinispan-dev
Reply | Threaded
Open this post in threaded view
|

Re: [infinispan-dev] jdk8backported package

Sanne Grinovero-3
On 25 January 2016 at 14:15, William Burns <[hidden email]> wrote:
> Unfortunately this isn't really possible to remove these classes.  They have
> additional changes over and above the jdk8 classes.  We could move it into a
> different package though if we wanted to.

+1 to change package: I have been confused about this too in the recent past.

>
> To be honest the only reason this is still really needed is for the
> BoundedConcurrentHashMapV8.  I have been talking with Dan on and off for a
> while about even possibly looking into using a different map implementation
> (such as caffeine or something else equivalent).  We haven't looked into it
> very seriously though.

Could you remove the other ones which are no longer needed?

>
> We could look into this more if people think we want to tackle that.
>
>  - Will
>
> On Thu, Jan 14, 2016 at 4:34 AM Radim Vansa <[hidden email]> wrote:
>>
>> Hi,
>>
>> although we're on Java 8, there's still the package
>> org.infinispan.*.jdk8backported in our codebase. Is there any plan (and
>> possibility) to remove these and use implementation provided by runtime?
>> Or have we tweaked them too much, so shall we rather rename them?
>>
>> Radim
>>
>> --
>> Radim Vansa <[hidden email]>
>> JBoss Performance Team
>>
>> _______________________________________________
>> infinispan-dev mailing list
>> [hidden email]
>> https://lists.jboss.org/mailman/listinfo/infinispan-dev
>
>
> _______________________________________________
> infinispan-dev mailing list
> [hidden email]
> https://lists.jboss.org/mailman/listinfo/infinispan-dev
_______________________________________________
infinispan-dev mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/infinispan-dev
Reply | Threaded
Open this post in threaded view
|

Re: [infinispan-dev] jdk8backported package

Sebastian Laskawiec
In reply to this post by William Burns-3
Would it be possible to decouple Map implementation from our custom code (like eviction policy or setting max size)? Or the dependency between those two is too strong?

On Mon, Jan 25, 2016 at 3:15 PM, William Burns <[hidden email]> wrote:
Unfortunately this isn't really possible to remove these classes.  They have additional changes over and above the jdk8 classes.  We could move it into a different package though if we wanted to.

To be honest the only reason this is still really needed is for the BoundedConcurrentHashMapV8.  I have been talking with Dan on and off for a while about even possibly looking into using a different map implementation (such as caffeine or something else equivalent).  We haven't looked into it very seriously though.

We could look into this more if people think we want to tackle that.

 - Will

On Thu, Jan 14, 2016 at 4:34 AM Radim Vansa <[hidden email]> wrote:
Hi,

although we're on Java 8, there's still the package
org.infinispan.*.jdk8backported in our codebase. Is there any plan (and
possibility) to remove these and use implementation provided by runtime?
Or have we tweaked them too much, so shall we rather rename them?

Radim

--
Radim Vansa <[hidden email]>
JBoss Performance Team

_______________________________________________
infinispan-dev mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/infinispan-dev

_______________________________________________
infinispan-dev mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/infinispan-dev


_______________________________________________
infinispan-dev mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/infinispan-dev
Reply | Threaded
Open this post in threaded view
|

Re: [infinispan-dev] jdk8backported package

William Burns-3
In reply to this post by Sanne Grinovero-3


On Mon, Jan 25, 2016 at 9:24 AM Sanne Grinovero <[hidden email]> wrote:
On 25 January 2016 at 14:15, William Burns <[hidden email]> wrote:
> Unfortunately this isn't really possible to remove these classes.  They have
> additional changes over and above the jdk8 classes.  We could move it into a
> different package though if we wanted to.

+1 to change package: I have been confused about this too in the recent past.

I am guessing we can just put these in the concurrent package 1 below?  WDYT?
 

>
> To be honest the only reason this is still really needed is for the
> BoundedConcurrentHashMapV8.  I have been talking with Dan on and off for a
> while about even possibly looking into using a different map implementation
> (such as caffeine or something else equivalent).  We haven't looked into it
> very seriously though.

Could you remove the other ones which are no longer needed?

There is only one class that can be really removed and that is once M/R is removed I believe (ConcurrentParallelHashMapV8).
 

>
> We could look into this more if people think we want to tackle that.
>
>  - Will
>
> On Thu, Jan 14, 2016 at 4:34 AM Radim Vansa <[hidden email]> wrote:
>>
>> Hi,
>>
>> although we're on Java 8, there's still the package
>> org.infinispan.*.jdk8backported in our codebase. Is there any plan (and
>> possibility) to remove these and use implementation provided by runtime?
>> Or have we tweaked them too much, so shall we rather rename them?
>>
>> Radim
>>
>> --
>> Radim Vansa <[hidden email]>
>> JBoss Performance Team
>>
>> _______________________________________________
>> infinispan-dev mailing list
>> [hidden email]
>> https://lists.jboss.org/mailman/listinfo/infinispan-dev
>
>
> _______________________________________________
> infinispan-dev mailing list
> [hidden email]
> https://lists.jboss.org/mailman/listinfo/infinispan-dev
_______________________________________________
infinispan-dev mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/infinispan-dev

_______________________________________________
infinispan-dev mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/infinispan-dev
Reply | Threaded
Open this post in threaded view
|

Re: [infinispan-dev] jdk8backported package

William Burns-3
In reply to this post by Sebastian Laskawiec


On Mon, Jan 25, 2016 at 10:09 AM Sebastian Laskawiec <[hidden email]> wrote:
Would it be possible to decouple Map implementation from our custom code (like eviction policy or setting max size)? Or the dependency between those two is too strong?

This is possible.  I had already thought of doing that at some point, but haven't been able to look into it or have a great reason yet (we lose some stuff like better size estimations - since we have access to the underlying array).

Unfortunately the equivalent versions have to stay even if we remove the eviction stuff, since there is no pluggable way to do this with the JVM classes.
 

On Mon, Jan 25, 2016 at 3:15 PM, William Burns <[hidden email]> wrote:
Unfortunately this isn't really possible to remove these classes.  They have additional changes over and above the jdk8 classes.  We could move it into a different package though if we wanted to.

To be honest the only reason this is still really needed is for the BoundedConcurrentHashMapV8.  I have been talking with Dan on and off for a while about even possibly looking into using a different map implementation (such as caffeine or something else equivalent).  We haven't looked into it very seriously though.

We could look into this more if people think we want to tackle that.

 - Will

On Thu, Jan 14, 2016 at 4:34 AM Radim Vansa <[hidden email]> wrote:
Hi,

although we're on Java 8, there's still the package
org.infinispan.*.jdk8backported in our codebase. Is there any plan (and
possibility) to remove these and use implementation provided by runtime?
Or have we tweaked them too much, so shall we rather rename them?

Radim

--
Radim Vansa <[hidden email]>
JBoss Performance Team

_______________________________________________
infinispan-dev mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/infinispan-dev

_______________________________________________
infinispan-dev mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/infinispan-dev

_______________________________________________
infinispan-dev mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/infinispan-dev

_______________________________________________
infinispan-dev mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/infinispan-dev
Reply | Threaded
Open this post in threaded view
|

Re: [infinispan-dev] jdk8backported package

Sanne Grinovero-3
Hi,
sorry for reviving this old thread, but this came up again in the form
of doubts and design questions this morning from people in the
performance team profiling it.

Let's please change the name, I've opened a JIRA to not forget again:
 - https://issues.jboss.org/browse/ISPN-6933

Will, assigned to you as I assume it would take you 2 minutes :)

Thanks,
Sanne


On 25 January 2016 at 20:18, William Burns <[hidden email]> wrote:

>
>
> On Mon, Jan 25, 2016 at 10:09 AM Sebastian Laskawiec <[hidden email]>
> wrote:
>>
>> Would it be possible to decouple Map implementation from our custom code
>> (like eviction policy or setting max size)? Or the dependency between those
>> two is too strong?
>
>
> This is possible.  I had already thought of doing that at some point, but
> haven't been able to look into it or have a great reason yet (we lose some
> stuff like better size estimations - since we have access to the underlying
> array).
>
> Unfortunately the equivalent versions have to stay even if we remove the
> eviction stuff, since there is no pluggable way to do this with the JVM
> classes.
>
>>
>>
>> On Mon, Jan 25, 2016 at 3:15 PM, William Burns <[hidden email]>
>> wrote:
>>>
>>> Unfortunately this isn't really possible to remove these classes.  They
>>> have additional changes over and above the jdk8 classes.  We could move it
>>> into a different package though if we wanted to.
>>>
>>> To be honest the only reason this is still really needed is for the
>>> BoundedConcurrentHashMapV8.  I have been talking with Dan on and off for a
>>> while about even possibly looking into using a different map implementation
>>> (such as caffeine or something else equivalent).  We haven't looked into it
>>> very seriously though.
>>>
>>> We could look into this more if people think we want to tackle that.
>>>
>>>  - Will
>>>
>>> On Thu, Jan 14, 2016 at 4:34 AM Radim Vansa <[hidden email]> wrote:
>>>>
>>>> Hi,
>>>>
>>>> although we're on Java 8, there's still the package
>>>> org.infinispan.*.jdk8backported in our codebase. Is there any plan (and
>>>> possibility) to remove these and use implementation provided by runtime?
>>>> Or have we tweaked them too much, so shall we rather rename them?
>>>>
>>>> Radim
>>>>
>>>> --
>>>> Radim Vansa <[hidden email]>
>>>> JBoss Performance Team
>>>>
>>>> _______________________________________________
>>>> infinispan-dev mailing list
>>>> [hidden email]
>>>> https://lists.jboss.org/mailman/listinfo/infinispan-dev
>>>
>>>
>>> _______________________________________________
>>> infinispan-dev mailing list
>>> [hidden email]
>>> https://lists.jboss.org/mailman/listinfo/infinispan-dev
>>
>>
>> _______________________________________________
>> infinispan-dev mailing list
>> [hidden email]
>> https://lists.jboss.org/mailman/listinfo/infinispan-dev
>
>
> _______________________________________________
> infinispan-dev mailing list
> [hidden email]
> https://lists.jboss.org/mailman/listinfo/infinispan-dev
_______________________________________________
infinispan-dev mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/infinispan-dev