You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@mahout.apache.org by Pat Ferrel <pa...@occamsmachete.com> on 2015/07/17 19:14:32 UTC
11 vs 10.x
We’re planning a 10.3 release I guess?
I’m about to push some small changes to master for 11.x should I also push to 10.3?
The most significant change is an optimization using drm.par(auto=true) which speeds up itemsimilarity and rowsimilarity by a fair bit.
When do we plan a release of 11? This IMO is far more important than 10 since not only are all the major distros supporting Spark 1.3.1 but 1.4 is just about ripe for release.
Re: 11 vs 10.x
Posted by Pat Ferrel <pa...@occamsmachete.com>.
From my POV Spark 1.3 has been a painful wait and therefor may be an exception to any general rule. We get the features there anyway.
Can we please release 11 soon?
On Jul 17, 2015, at 11:49 AM, Dmitriy Lyubimov <dl...@gmail.com> wrote:
0.11 is also good. They are practically equivalent. if it is just about 1.3
support then .11 can be released any day.
but i do think that we can't just flip our major number just to accomodate
spark version. i'd prefer to see some new mahout stuff between major
releases too.
On Fri, Jul 17, 2015 at 11:46 AM, Dmitriy Lyubimov <dl...@gmail.com>
wrote:
> we are still at 10.2
>
> On Fri, Jul 17, 2015 at 10:14 AM, Pat Ferrel <pa...@occamsmachete.com>
> wrote:
>
>> We’re planning a 10.3 release I guess?
>>
>> I’m about to push some small changes to master for 11.x should I also
>> push to 10.3?
>>
>> The most significant change is an optimization using drm.par(auto=true)
>> which speeds up itemsimilarity and rowsimilarity by a fair bit.
>>
>> When do we plan a release of 11? This IMO is far more important than 10
>> since not only are all the major distros supporting Spark 1.3.1 but 1.4 is
>> just about ripe for release.
>
>
>
Re: 11 vs 10.x
Posted by Dmitriy Lyubimov <dl...@gmail.com>.
0.11 is also good. They are practically equivalent. if it is just about 1.3
support then .11 can be released any day.
but i do think that we can't just flip our major number just to accomodate
spark version. i'd prefer to see some new mahout stuff between major
releases too.
On Fri, Jul 17, 2015 at 11:46 AM, Dmitriy Lyubimov <dl...@gmail.com>
wrote:
> we are still at 10.2
>
> On Fri, Jul 17, 2015 at 10:14 AM, Pat Ferrel <pa...@occamsmachete.com>
> wrote:
>
>> We’re planning a 10.3 release I guess?
>>
>> I’m about to push some small changes to master for 11.x should I also
>> push to 10.3?
>>
>> The most significant change is an optimization using drm.par(auto=true)
>> which speeds up itemsimilarity and rowsimilarity by a fair bit.
>>
>> When do we plan a release of 11? This IMO is far more important than 10
>> since not only are all the major distros supporting Spark 1.3.1 but 1.4 is
>> just about ripe for release.
>
>
>
Re: 11 vs 10.x
Posted by Dmitriy Lyubimov <dl...@gmail.com>.
we are still at 10.2
On Fri, Jul 17, 2015 at 10:14 AM, Pat Ferrel <pa...@occamsmachete.com> wrote:
> We’re planning a 10.3 release I guess?
>
> I’m about to push some small changes to master for 11.x should I also push
> to 10.3?
>
> The most significant change is an optimization using drm.par(auto=true)
> which speeds up itemsimilarity and rowsimilarity by a fair bit.
>
> When do we plan a release of 11? This IMO is far more important than 10
> since not only are all the major distros supporting Spark 1.3.1 but 1.4 is
> just about ripe for release.