Quantcast

GPars 1, 2, and 3

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

GPars 1, 2, and 3

Russel Winder-3
Once JAXFinance and DevoxxUK are over I will (finally) be able to get
back to progressing GPars 2.0 to releasability. I also have to release
Gant to try and find out how to release via
JCenter/Bintray/Artifactory, good knowledge for GPars. Something nearly
got done in January but that effort fell at the last hurdle.

I also note that JDK9 may have some features that require a rethink of
some GPars implementation detail for a JDK and later specific GPars.
cf. http://gee.cs.oswego.edu/dl/html/j9mm.html

So the overall plot seems to be:

GPars 1.X effectively abandoned
GPars 2.x JDK8 and above only due 2017-07
GPars 3.x JDK9 and above in planning.

It might be worth thinking about whether GPars has a real future since
so few people seem to be interested in actively working on it. With
Quasar (Parallel Universe thing not Quasar Framework) there is a
fibers, actors, CSP framework with resources behind it. OK so no proper
dataflow out of the box, but it could be added.

--
Russel.
=============================================================================
Dr Russel Winder      t: +44 20 7585 2200   voip: sip:[hidden email]
41 Buckmaster Road    m: +44 7770 465 077   xmpp: [hidden email]
London SW11 1EN, UK   w: www.russel.org.uk  skype: russel_winder

signature.asc (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: GPars 1, 2, and 3

Jochen Theodorou
On 14.03.2017 12:06, Russel Winder wrote:
[...]
> I also note that JDK9 may have some features that require a rethink of
> some GPars implementation detail for a JDK and later specific GPars.
> cf. http://gee.cs.oswego.edu/dl/html/j9mm.html

yes, VarHandles are something to take a look at for sure.

> So the overall plot seems to be:
>
> GPars 1.X effectively abandoned
> GPars 2.x JDK8 and above only due 2017-07
> GPars 3.x JDK9 and above in planning.

sounds like a plan to me

> It might be worth thinking about whether GPars has a real future since
> so few people seem to be interested in actively working on it. With
> Quasar (Parallel Universe thing not Quasar Framework) there is a
> fibers, actors, CSP framework with resources behind it. OK so no proper
> dataflow out of the box, but it could be added.

you mean http://www.paralleluniverse.co/quasar/ ?
Fibers there work based on agents and bytecode rewriting? CSP I have not
seen, but I did not look to closely. License wise it might be ok, since
it offers EPL 1.0 dual with GPL3 (the later would not be ok for us).

But frankly, if I am right about the fibers, then this part is no option
for me. That leaves the CSP part, which I did not see and actors. Is it
really worth it then?

bye Jochen

Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: GPars 1, 2, and 3

Jochen Theodorou
On 16.03.2017 22:54, Jochen Theodorou wrote:

> On 14.03.2017 12:06, Russel Winder wrote:
> [...]
>> It might be worth thinking about whether GPars has a real future since
>> so few people seem to be interested in actively working on it. With
>> Quasar (Parallel Universe thing not Quasar Framework) there is a
>> fibers, actors, CSP framework with resources behind it. OK so no proper
>> dataflow out of the box, but it could be added.
>
> you mean http://www.paralleluniverse.co/quasar/ ?
> Fibers there work based on agents and bytecode rewriting? CSP I have not
> seen, but I did not look to closely. License wise it might be ok, since
> it offers EPL 1.0 dual with GPL3 (the later would not be ok for us).
>
> But frankly, if I am right about the fibers, then this part is no option
> for me. That leaves the CSP part, which I did not see and actors. Is it
> really worth it then?

found Dataflow, reactive streams and channels now... also that it seems
to have a dependency on Guava... nothing against Guava itself, but there
are too many incompatible version going around and it is a big
dependency of over 1MB

bye Jochen

Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: GPars 1, 2, and 3

jnorthr
if you need more recent gpars documentation, i'm working on v2.0 of it that you can see here: http://gpars.aws.ie.a9sapp.eu
 the download page does show gpars 2.0 but it's just copies of 1.2.1 that i've renamed as placeholders.
since jsr-166Y may be dropped there are a number of gpars features that may disappear, so we're hoping to clarify this issue before summer release.
😊

On 16 March 2017 at 23:13, Jochen Theodorou <[hidden email]> wrote:
On 16.03.2017 22:54, Jochen Theodorou wrote:
On 14.03.2017 12:06, Russel Winder wrote:
[...]
It might be worth thinking about whether GPars has a real future since
so few people seem to be interested in actively working on it. With
Quasar (Parallel Universe thing not Quasar Framework) there is a
fibers, actors, CSP framework with resources behind it. OK so no proper
dataflow out of the box, but it could be added.

you mean http://www.paralleluniverse.co/quasar/ ?
Fibers there work based on agents and bytecode rewriting? CSP I have not
seen, but I did not look to closely. License wise it might be ok, since
it offers EPL 1.0 dual with GPL3 (the later would not be ok for us).

But frankly, if I am right about the fibers, then this part is no option
for me. That leaves the CSP part, which I did not see and actors. Is it
really worth it then?

found Dataflow, reactive streams and channels now... also that it seems to have a dependency on Guava... nothing against Guava itself, but there are too many incompatible version going around and it is a big dependency of over 1MB

bye Jochen


Loading...