You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@hadoop.apache.org by Glen Mazza <gm...@talend.com> on 2013/01/18 15:48:54 UTC
Cohesion of Hadoop team?
Hi, looking at the derivation of the 0.23.x & 2.0.x branches on one
hand, and the 1.x branches on the other, as described here:
http://mail-archives.apache.org/mod_mbox/hadoop-user/201301.mbox/%3CCD0CAB8B.1098F%25evans%40yahoo-inc.com%3E
One gets the impression the Hadoop committers are split into two teams,
with one team working on 0.23.x/2.0.2 and another team working on 1.x,
running the risk of increasingly diverging products eventually competing
with each other. Is that the case? Is there expected to be a Hadoop
3.0 where the results of the two lines of development will merge or is
it increasingly likely the subteams will continue their separate routes?
Thanks,
Glen
--
Glen Mazza
Talend Community Coders - coders.talend.com
blog: www.jroller.com/gmazza
Re: Cohesion of Hadoop team?
Posted by Andrew Purtell <ap...@apache.org>.
It might also be helpful to differentiate between what commercial entities
might do with release lines in ASF Hadoop and the relationship between the
branches within ASF Hadoop itself. ASF Hadoop 1 and ASF Hadoop 2 are two
versions of the same product, they do not compete with each other.
On Fri, Jan 18, 2013 at 6:48 AM, Glen Mazza <gm...@talend.com> wrote:
> One gets the impression the Hadoop committers are split into two teams,
> with one team working on 0.23.x/2.0.2 and another team working on 1.x,
> running the risk of increasingly diverging products eventually competing
> with each other.
--
Best regards,
- Andy
Problems worthy of attack prove their worth by hitting back. - Piet Hein
(via Tom White)
Re: Cohesion of Hadoop team?
Posted by Matt Foley <ma...@apache.org>.
The policy should be stated :-) It is: "All new features must be
committed to trunk before or simultaneously with being committed to a
lower-numbered branch."
Thus, we don't forbid new features from being added to the Hadoop-1 branch,
but they must be back-ports of features already added to trunk/Hadoop-2.
Hadoop-1 continues to live because it is very stable and the community
needs time to stabilize the great new major features in Hadoop-2. But most
of us are looking forward to moving to Hadoop-2 when it is ready. By
keeping Hadoop-1 in good shape in the meantime, we assure that the
transition can be smooth and reasonably comfortable for the end-user
companies that depend on Hadoop for business-critical use.
--Matt
Hadoop-1 release manager
On Fri, Jan 18, 2013 at 9:54 AM, Arun C Murthy <ac...@hortonworks.com> wrote:
> Absolutely not true.
>
> We have a clear policy in place which ensures there is no feature
> disparity between hadoop-1 and hadoop-2. So, hadoop-2 is a super-set of
> functionality in hadoop-1 and there is no disparity.
>
> hth,
> Arun
>
> On Jan 18, 2013, at 6:48 AM, Glen Mazza wrote:
>
> Hi, looking at the derivation of the 0.23.x & 2.0.x branches on one hand,
> and the 1.x branches on the other, as described here:
>
> http://mail-archives.apache.org/mod_mbox/hadoop-user/201301.mbox/%3CCD0CAB8B.1098F%25evans%40yahoo-inc.com%3E
>
> One gets the impression the Hadoop committers are split into two teams,
> with one team working on 0.23.x/2.0.2 and another team working on 1.x,
> running the risk of increasingly diverging products eventually competing
> with each other. Is that the case? Is there expected to be a Hadoop 3.0
> where the results of the two lines of development will merge or is it
> increasingly likely the subteams will continue their separate routes?
>
> Thanks,
> Glen
>
> --
> Glen Mazza
> Talend Community Coders - coders.talend.com
> blog: www.jroller.com/gmazza
>
>
> --
> Arun C. Murthy
> Hortonworks Inc.
> http://hortonworks.com/
>
>
>
Re: Cohesion of Hadoop team?
Posted by Matt Foley <ma...@apache.org>.
The policy should be stated :-) It is: "All new features must be
committed to trunk before or simultaneously with being committed to a
lower-numbered branch."
Thus, we don't forbid new features from being added to the Hadoop-1 branch,
but they must be back-ports of features already added to trunk/Hadoop-2.
Hadoop-1 continues to live because it is very stable and the community
needs time to stabilize the great new major features in Hadoop-2. But most
of us are looking forward to moving to Hadoop-2 when it is ready. By
keeping Hadoop-1 in good shape in the meantime, we assure that the
transition can be smooth and reasonably comfortable for the end-user
companies that depend on Hadoop for business-critical use.
--Matt
Hadoop-1 release manager
On Fri, Jan 18, 2013 at 9:54 AM, Arun C Murthy <ac...@hortonworks.com> wrote:
> Absolutely not true.
>
> We have a clear policy in place which ensures there is no feature
> disparity between hadoop-1 and hadoop-2. So, hadoop-2 is a super-set of
> functionality in hadoop-1 and there is no disparity.
>
> hth,
> Arun
>
> On Jan 18, 2013, at 6:48 AM, Glen Mazza wrote:
>
> Hi, looking at the derivation of the 0.23.x & 2.0.x branches on one hand,
> and the 1.x branches on the other, as described here:
>
> http://mail-archives.apache.org/mod_mbox/hadoop-user/201301.mbox/%3CCD0CAB8B.1098F%25evans%40yahoo-inc.com%3E
>
> One gets the impression the Hadoop committers are split into two teams,
> with one team working on 0.23.x/2.0.2 and another team working on 1.x,
> running the risk of increasingly diverging products eventually competing
> with each other. Is that the case? Is there expected to be a Hadoop 3.0
> where the results of the two lines of development will merge or is it
> increasingly likely the subteams will continue their separate routes?
>
> Thanks,
> Glen
>
> --
> Glen Mazza
> Talend Community Coders - coders.talend.com
> blog: www.jroller.com/gmazza
>
>
> --
> Arun C. Murthy
> Hortonworks Inc.
> http://hortonworks.com/
>
>
>
Re: Cohesion of Hadoop team?
Posted by Matt Foley <ma...@apache.org>.
The policy should be stated :-) It is: "All new features must be
committed to trunk before or simultaneously with being committed to a
lower-numbered branch."
Thus, we don't forbid new features from being added to the Hadoop-1 branch,
but they must be back-ports of features already added to trunk/Hadoop-2.
Hadoop-1 continues to live because it is very stable and the community
needs time to stabilize the great new major features in Hadoop-2. But most
of us are looking forward to moving to Hadoop-2 when it is ready. By
keeping Hadoop-1 in good shape in the meantime, we assure that the
transition can be smooth and reasonably comfortable for the end-user
companies that depend on Hadoop for business-critical use.
--Matt
Hadoop-1 release manager
On Fri, Jan 18, 2013 at 9:54 AM, Arun C Murthy <ac...@hortonworks.com> wrote:
> Absolutely not true.
>
> We have a clear policy in place which ensures there is no feature
> disparity between hadoop-1 and hadoop-2. So, hadoop-2 is a super-set of
> functionality in hadoop-1 and there is no disparity.
>
> hth,
> Arun
>
> On Jan 18, 2013, at 6:48 AM, Glen Mazza wrote:
>
> Hi, looking at the derivation of the 0.23.x & 2.0.x branches on one hand,
> and the 1.x branches on the other, as described here:
>
> http://mail-archives.apache.org/mod_mbox/hadoop-user/201301.mbox/%3CCD0CAB8B.1098F%25evans%40yahoo-inc.com%3E
>
> One gets the impression the Hadoop committers are split into two teams,
> with one team working on 0.23.x/2.0.2 and another team working on 1.x,
> running the risk of increasingly diverging products eventually competing
> with each other. Is that the case? Is there expected to be a Hadoop 3.0
> where the results of the two lines of development will merge or is it
> increasingly likely the subteams will continue their separate routes?
>
> Thanks,
> Glen
>
> --
> Glen Mazza
> Talend Community Coders - coders.talend.com
> blog: www.jroller.com/gmazza
>
>
> --
> Arun C. Murthy
> Hortonworks Inc.
> http://hortonworks.com/
>
>
>
Re: Cohesion of Hadoop team?
Posted by Matt Foley <ma...@apache.org>.
The policy should be stated :-) It is: "All new features must be
committed to trunk before or simultaneously with being committed to a
lower-numbered branch."
Thus, we don't forbid new features from being added to the Hadoop-1 branch,
but they must be back-ports of features already added to trunk/Hadoop-2.
Hadoop-1 continues to live because it is very stable and the community
needs time to stabilize the great new major features in Hadoop-2. But most
of us are looking forward to moving to Hadoop-2 when it is ready. By
keeping Hadoop-1 in good shape in the meantime, we assure that the
transition can be smooth and reasonably comfortable for the end-user
companies that depend on Hadoop for business-critical use.
--Matt
Hadoop-1 release manager
On Fri, Jan 18, 2013 at 9:54 AM, Arun C Murthy <ac...@hortonworks.com> wrote:
> Absolutely not true.
>
> We have a clear policy in place which ensures there is no feature
> disparity between hadoop-1 and hadoop-2. So, hadoop-2 is a super-set of
> functionality in hadoop-1 and there is no disparity.
>
> hth,
> Arun
>
> On Jan 18, 2013, at 6:48 AM, Glen Mazza wrote:
>
> Hi, looking at the derivation of the 0.23.x & 2.0.x branches on one hand,
> and the 1.x branches on the other, as described here:
>
> http://mail-archives.apache.org/mod_mbox/hadoop-user/201301.mbox/%3CCD0CAB8B.1098F%25evans%40yahoo-inc.com%3E
>
> One gets the impression the Hadoop committers are split into two teams,
> with one team working on 0.23.x/2.0.2 and another team working on 1.x,
> running the risk of increasingly diverging products eventually competing
> with each other. Is that the case? Is there expected to be a Hadoop 3.0
> where the results of the two lines of development will merge or is it
> increasingly likely the subteams will continue their separate routes?
>
> Thanks,
> Glen
>
> --
> Glen Mazza
> Talend Community Coders - coders.talend.com
> blog: www.jroller.com/gmazza
>
>
> --
> Arun C. Murthy
> Hortonworks Inc.
> http://hortonworks.com/
>
>
>
Re: Cohesion of Hadoop team?
Posted by Arun C Murthy <ac...@hortonworks.com>.
Absolutely not true.
We have a clear policy in place which ensures there is no feature disparity between hadoop-1 and hadoop-2. So, hadoop-2 is a super-set of functionality in hadoop-1 and there is no disparity.
hth,
Arun
On Jan 18, 2013, at 6:48 AM, Glen Mazza wrote:
> Hi, looking at the derivation of the 0.23.x & 2.0.x branches on one hand, and the 1.x branches on the other, as described here:
> http://mail-archives.apache.org/mod_mbox/hadoop-user/201301.mbox/%3CCD0CAB8B.1098F%25evans%40yahoo-inc.com%3E
>
> One gets the impression the Hadoop committers are split into two teams, with one team working on 0.23.x/2.0.2 and another team working on 1.x, running the risk of increasingly diverging products eventually competing with each other. Is that the case? Is there expected to be a Hadoop 3.0 where the results of the two lines of development will merge or is it increasingly likely the subteams will continue their separate routes?
>
> Thanks,
> Glen
> --
> Glen Mazza
> Talend Community Coders - coders.talend.com
> blog: www.jroller.com/gmazza
--
Arun C. Murthy
Hortonworks Inc.
http://hortonworks.com/
Re: Cohesion of Hadoop team?
Posted by Andrew Purtell <ap...@apache.org>.
It might also be helpful to differentiate between what commercial entities
might do with release lines in ASF Hadoop and the relationship between the
branches within ASF Hadoop itself. ASF Hadoop 1 and ASF Hadoop 2 are two
versions of the same product, they do not compete with each other.
On Fri, Jan 18, 2013 at 6:48 AM, Glen Mazza <gm...@talend.com> wrote:
> One gets the impression the Hadoop committers are split into two teams,
> with one team working on 0.23.x/2.0.2 and another team working on 1.x,
> running the risk of increasingly diverging products eventually competing
> with each other.
--
Best regards,
- Andy
Problems worthy of attack prove their worth by hitting back. - Piet Hein
(via Tom White)
Re: Cohesion of Hadoop team?
Posted by Andrew Purtell <ap...@apache.org>.
It might also be helpful to differentiate between what commercial entities
might do with release lines in ASF Hadoop and the relationship between the
branches within ASF Hadoop itself. ASF Hadoop 1 and ASF Hadoop 2 are two
versions of the same product, they do not compete with each other.
On Fri, Jan 18, 2013 at 6:48 AM, Glen Mazza <gm...@talend.com> wrote:
> One gets the impression the Hadoop committers are split into two teams,
> with one team working on 0.23.x/2.0.2 and another team working on 1.x,
> running the risk of increasingly diverging products eventually competing
> with each other.
--
Best regards,
- Andy
Problems worthy of attack prove their worth by hitting back. - Piet Hein
(via Tom White)
Re: Cohesion of Hadoop team?
Posted by Glen Mazza <gm...@talend.com>.
On 01/18/2013 11:58 AM, Suresh Srinivas wrote:
>
>
>
> On Fri, Jan 18, 2013 at 6:48 AM, Glen Mazza <gmazza@talend.com
> <ma...@talend.com>> wrote:
>
> Hi, looking at the derivation of the 0.23.x & 2.0.x branches on
> one hand, and the 1.x branches on the other, as described here:
> http://mail-archives.apache.org/mod_mbox/hadoop-user/201301.mbox/%3CCD0CAB8B.1098F%25evans%40yahoo-inc.com%3E
>
> One gets the impression the Hadoop committers are split into two
> teams, with one team working on 0.23.x/2.0.2 and another team
> working on 1.x, running the risk of increasingly diverging
> products eventually competing with each other. Is that the case?
>
>
> I am not sure how you came to this conclusion. The way I see it is,
> all the folks are working on trunk. Subset of this work from trunk is
> pushed to older releases such as 1.x or 0.23.x. In Apache Hadoop,
> features always go to trunk first before going to any older releases
> 1.x or 0.23.x. That means trunk is a superset of all the features.
>
> Is there expected to be a Hadoop 3.0 where the results of the two
> lines of development will merge or is it increasingly likely the
> subteams will continue their separate routes?
>
>
> 2.0.3-alpha, which is the latest release based off of trunk, that is
> in final stage of completion should have all the features that all the
> other releases have. Let me know if there are any exceptions to this
> that you know of.
I had entered a JIRA here:
https://issues.apache.org/jira/browse/HADOOP-9206 . The instructions
for single-node setup on 1.1.x are radically different from the
instructions for 0.23 and 2.0.2; furthermore, the JARs and folder
structure of what you get from the 1.1.x download and what you get with
either 0.23.x or 2.0.x-alpha is also considerably different. The deltas
here, along with Bobby Evans' explanation of the version histories I
linked to above, gave me the impression that 1.x has one team working on
it while the other branches have another. If that was the case (as
you're not clarifying, it's not) I was then wondering when all
committers would be more or less on the same page again.
Thanks for the clarification.
Glen
>
> Thanks,
> Glen
>
> --
> Glen Mazza
> Talend Community Coders -coders.talend.com <http://coders.talend.com>
> blog:www.jroller.com/gmazza <http://www.jroller.com/gmazza>
>
>
>
>
> --
> http://hortonworks.com/download/
--
Glen Mazza
Talend Community Coders - coders.talend.com
blog: www.jroller.com/gmazza
Re: Cohesion of Hadoop team?
Posted by Glen Mazza <gm...@talend.com>.
On 01/18/2013 11:58 AM, Suresh Srinivas wrote:
>
>
>
> On Fri, Jan 18, 2013 at 6:48 AM, Glen Mazza <gmazza@talend.com
> <ma...@talend.com>> wrote:
>
> Hi, looking at the derivation of the 0.23.x & 2.0.x branches on
> one hand, and the 1.x branches on the other, as described here:
> http://mail-archives.apache.org/mod_mbox/hadoop-user/201301.mbox/%3CCD0CAB8B.1098F%25evans%40yahoo-inc.com%3E
>
> One gets the impression the Hadoop committers are split into two
> teams, with one team working on 0.23.x/2.0.2 and another team
> working on 1.x, running the risk of increasingly diverging
> products eventually competing with each other. Is that the case?
>
>
> I am not sure how you came to this conclusion. The way I see it is,
> all the folks are working on trunk. Subset of this work from trunk is
> pushed to older releases such as 1.x or 0.23.x. In Apache Hadoop,
> features always go to trunk first before going to any older releases
> 1.x or 0.23.x. That means trunk is a superset of all the features.
>
> Is there expected to be a Hadoop 3.0 where the results of the two
> lines of development will merge or is it increasingly likely the
> subteams will continue their separate routes?
>
>
> 2.0.3-alpha, which is the latest release based off of trunk, that is
> in final stage of completion should have all the features that all the
> other releases have. Let me know if there are any exceptions to this
> that you know of.
I had entered a JIRA here:
https://issues.apache.org/jira/browse/HADOOP-9206 . The instructions
for single-node setup on 1.1.x are radically different from the
instructions for 0.23 and 2.0.2; furthermore, the JARs and folder
structure of what you get from the 1.1.x download and what you get with
either 0.23.x or 2.0.x-alpha is also considerably different. The deltas
here, along with Bobby Evans' explanation of the version histories I
linked to above, gave me the impression that 1.x has one team working on
it while the other branches have another. If that was the case (as
you're not clarifying, it's not) I was then wondering when all
committers would be more or less on the same page again.
Thanks for the clarification.
Glen
>
> Thanks,
> Glen
>
> --
> Glen Mazza
> Talend Community Coders -coders.talend.com <http://coders.talend.com>
> blog:www.jroller.com/gmazza <http://www.jroller.com/gmazza>
>
>
>
>
> --
> http://hortonworks.com/download/
--
Glen Mazza
Talend Community Coders - coders.talend.com
blog: www.jroller.com/gmazza
Re: Cohesion of Hadoop team?
Posted by Glen Mazza <gm...@talend.com>.
On 01/18/2013 11:58 AM, Suresh Srinivas wrote:
>
>
>
> On Fri, Jan 18, 2013 at 6:48 AM, Glen Mazza <gmazza@talend.com
> <ma...@talend.com>> wrote:
>
> Hi, looking at the derivation of the 0.23.x & 2.0.x branches on
> one hand, and the 1.x branches on the other, as described here:
> http://mail-archives.apache.org/mod_mbox/hadoop-user/201301.mbox/%3CCD0CAB8B.1098F%25evans%40yahoo-inc.com%3E
>
> One gets the impression the Hadoop committers are split into two
> teams, with one team working on 0.23.x/2.0.2 and another team
> working on 1.x, running the risk of increasingly diverging
> products eventually competing with each other. Is that the case?
>
>
> I am not sure how you came to this conclusion. The way I see it is,
> all the folks are working on trunk. Subset of this work from trunk is
> pushed to older releases such as 1.x or 0.23.x. In Apache Hadoop,
> features always go to trunk first before going to any older releases
> 1.x or 0.23.x. That means trunk is a superset of all the features.
>
> Is there expected to be a Hadoop 3.0 where the results of the two
> lines of development will merge or is it increasingly likely the
> subteams will continue their separate routes?
>
>
> 2.0.3-alpha, which is the latest release based off of trunk, that is
> in final stage of completion should have all the features that all the
> other releases have. Let me know if there are any exceptions to this
> that you know of.
I had entered a JIRA here:
https://issues.apache.org/jira/browse/HADOOP-9206 . The instructions
for single-node setup on 1.1.x are radically different from the
instructions for 0.23 and 2.0.2; furthermore, the JARs and folder
structure of what you get from the 1.1.x download and what you get with
either 0.23.x or 2.0.x-alpha is also considerably different. The deltas
here, along with Bobby Evans' explanation of the version histories I
linked to above, gave me the impression that 1.x has one team working on
it while the other branches have another. If that was the case (as
you're not clarifying, it's not) I was then wondering when all
committers would be more or less on the same page again.
Thanks for the clarification.
Glen
>
> Thanks,
> Glen
>
> --
> Glen Mazza
> Talend Community Coders -coders.talend.com <http://coders.talend.com>
> blog:www.jroller.com/gmazza <http://www.jroller.com/gmazza>
>
>
>
>
> --
> http://hortonworks.com/download/
--
Glen Mazza
Talend Community Coders - coders.talend.com
blog: www.jroller.com/gmazza
Re: Cohesion of Hadoop team?
Posted by Glen Mazza <gm...@talend.com>.
On 01/18/2013 11:58 AM, Suresh Srinivas wrote:
>
>
>
> On Fri, Jan 18, 2013 at 6:48 AM, Glen Mazza <gmazza@talend.com
> <ma...@talend.com>> wrote:
>
> Hi, looking at the derivation of the 0.23.x & 2.0.x branches on
> one hand, and the 1.x branches on the other, as described here:
> http://mail-archives.apache.org/mod_mbox/hadoop-user/201301.mbox/%3CCD0CAB8B.1098F%25evans%40yahoo-inc.com%3E
>
> One gets the impression the Hadoop committers are split into two
> teams, with one team working on 0.23.x/2.0.2 and another team
> working on 1.x, running the risk of increasingly diverging
> products eventually competing with each other. Is that the case?
>
>
> I am not sure how you came to this conclusion. The way I see it is,
> all the folks are working on trunk. Subset of this work from trunk is
> pushed to older releases such as 1.x or 0.23.x. In Apache Hadoop,
> features always go to trunk first before going to any older releases
> 1.x or 0.23.x. That means trunk is a superset of all the features.
>
> Is there expected to be a Hadoop 3.0 where the results of the two
> lines of development will merge or is it increasingly likely the
> subteams will continue their separate routes?
>
>
> 2.0.3-alpha, which is the latest release based off of trunk, that is
> in final stage of completion should have all the features that all the
> other releases have. Let me know if there are any exceptions to this
> that you know of.
I had entered a JIRA here:
https://issues.apache.org/jira/browse/HADOOP-9206 . The instructions
for single-node setup on 1.1.x are radically different from the
instructions for 0.23 and 2.0.2; furthermore, the JARs and folder
structure of what you get from the 1.1.x download and what you get with
either 0.23.x or 2.0.x-alpha is also considerably different. The deltas
here, along with Bobby Evans' explanation of the version histories I
linked to above, gave me the impression that 1.x has one team working on
it while the other branches have another. If that was the case (as
you're not clarifying, it's not) I was then wondering when all
committers would be more or less on the same page again.
Thanks for the clarification.
Glen
>
> Thanks,
> Glen
>
> --
> Glen Mazza
> Talend Community Coders -coders.talend.com <http://coders.talend.com>
> blog:www.jroller.com/gmazza <http://www.jroller.com/gmazza>
>
>
>
>
> --
> http://hortonworks.com/download/
--
Glen Mazza
Talend Community Coders - coders.talend.com
blog: www.jroller.com/gmazza
Re: Cohesion of Hadoop team?
Posted by Suresh Srinivas <su...@hortonworks.com>.
On Fri, Jan 18, 2013 at 6:48 AM, Glen Mazza <gm...@talend.com> wrote:
> Hi, looking at the derivation of the 0.23.x & 2.0.x branches on one hand,
> and the 1.x branches on the other, as described here:
>
> http://mail-archives.apache.org/mod_mbox/hadoop-user/201301.mbox/%3CCD0CAB8B.1098F%25evans%40yahoo-inc.com%3E
>
> One gets the impression the Hadoop committers are split into two teams,
> with one team working on 0.23.x/2.0.2 and another team working on 1.x,
> running the risk of increasingly diverging products eventually competing
> with each other. Is that the case?
>
I am not sure how you came to this conclusion. The way I see it is, all the
folks are working on trunk. Subset of this work from trunk is pushed to
older releases such as 1.x or 0.23.x. In Apache Hadoop, features always go
to trunk first before going to any older releases 1.x or 0.23.x. That means
trunk is a superset of all the features.
Is there expected to be a Hadoop 3.0 where the results of the two lines of
> development will merge or is it increasingly likely the subteams will
> continue their separate routes?
>
2.0.3-alpha, which is the latest release based off of trunk, that is in
final stage of completion should have all the features that all the other
releases have. Let me know if there are any exceptions to this that you
know of.
>
> Thanks,
> Glen
>
> --
> Glen Mazza
> Talend Community Coders - coders.talend.com
> blog: www.jroller.com/gmazza
>
>
--
http://hortonworks.com/download/
Re: Cohesion of Hadoop team?
Posted by Arun C Murthy <ac...@hortonworks.com>.
Absolutely not true.
We have a clear policy in place which ensures there is no feature disparity between hadoop-1 and hadoop-2. So, hadoop-2 is a super-set of functionality in hadoop-1 and there is no disparity.
hth,
Arun
On Jan 18, 2013, at 6:48 AM, Glen Mazza wrote:
> Hi, looking at the derivation of the 0.23.x & 2.0.x branches on one hand, and the 1.x branches on the other, as described here:
> http://mail-archives.apache.org/mod_mbox/hadoop-user/201301.mbox/%3CCD0CAB8B.1098F%25evans%40yahoo-inc.com%3E
>
> One gets the impression the Hadoop committers are split into two teams, with one team working on 0.23.x/2.0.2 and another team working on 1.x, running the risk of increasingly diverging products eventually competing with each other. Is that the case? Is there expected to be a Hadoop 3.0 where the results of the two lines of development will merge or is it increasingly likely the subteams will continue their separate routes?
>
> Thanks,
> Glen
> --
> Glen Mazza
> Talend Community Coders - coders.talend.com
> blog: www.jroller.com/gmazza
--
Arun C. Murthy
Hortonworks Inc.
http://hortonworks.com/
Re: Cohesion of Hadoop team?
Posted by Suresh Srinivas <su...@hortonworks.com>.
On Fri, Jan 18, 2013 at 6:48 AM, Glen Mazza <gm...@talend.com> wrote:
> Hi, looking at the derivation of the 0.23.x & 2.0.x branches on one hand,
> and the 1.x branches on the other, as described here:
>
> http://mail-archives.apache.org/mod_mbox/hadoop-user/201301.mbox/%3CCD0CAB8B.1098F%25evans%40yahoo-inc.com%3E
>
> One gets the impression the Hadoop committers are split into two teams,
> with one team working on 0.23.x/2.0.2 and another team working on 1.x,
> running the risk of increasingly diverging products eventually competing
> with each other. Is that the case?
>
I am not sure how you came to this conclusion. The way I see it is, all the
folks are working on trunk. Subset of this work from trunk is pushed to
older releases such as 1.x or 0.23.x. In Apache Hadoop, features always go
to trunk first before going to any older releases 1.x or 0.23.x. That means
trunk is a superset of all the features.
Is there expected to be a Hadoop 3.0 where the results of the two lines of
> development will merge or is it increasingly likely the subteams will
> continue their separate routes?
>
2.0.3-alpha, which is the latest release based off of trunk, that is in
final stage of completion should have all the features that all the other
releases have. Let me know if there are any exceptions to this that you
know of.
>
> Thanks,
> Glen
>
> --
> Glen Mazza
> Talend Community Coders - coders.talend.com
> blog: www.jroller.com/gmazza
>
>
--
http://hortonworks.com/download/
Re: Cohesion of Hadoop team?
Posted by Suresh Srinivas <su...@hortonworks.com>.
On Fri, Jan 18, 2013 at 6:48 AM, Glen Mazza <gm...@talend.com> wrote:
> Hi, looking at the derivation of the 0.23.x & 2.0.x branches on one hand,
> and the 1.x branches on the other, as described here:
>
> http://mail-archives.apache.org/mod_mbox/hadoop-user/201301.mbox/%3CCD0CAB8B.1098F%25evans%40yahoo-inc.com%3E
>
> One gets the impression the Hadoop committers are split into two teams,
> with one team working on 0.23.x/2.0.2 and another team working on 1.x,
> running the risk of increasingly diverging products eventually competing
> with each other. Is that the case?
>
I am not sure how you came to this conclusion. The way I see it is, all the
folks are working on trunk. Subset of this work from trunk is pushed to
older releases such as 1.x or 0.23.x. In Apache Hadoop, features always go
to trunk first before going to any older releases 1.x or 0.23.x. That means
trunk is a superset of all the features.
Is there expected to be a Hadoop 3.0 where the results of the two lines of
> development will merge or is it increasingly likely the subteams will
> continue their separate routes?
>
2.0.3-alpha, which is the latest release based off of trunk, that is in
final stage of completion should have all the features that all the other
releases have. Let me know if there are any exceptions to this that you
know of.
>
> Thanks,
> Glen
>
> --
> Glen Mazza
> Talend Community Coders - coders.talend.com
> blog: www.jroller.com/gmazza
>
>
--
http://hortonworks.com/download/
Re: Cohesion of Hadoop team?
Posted by Arun C Murthy <ac...@hortonworks.com>.
Absolutely not true.
We have a clear policy in place which ensures there is no feature disparity between hadoop-1 and hadoop-2. So, hadoop-2 is a super-set of functionality in hadoop-1 and there is no disparity.
hth,
Arun
On Jan 18, 2013, at 6:48 AM, Glen Mazza wrote:
> Hi, looking at the derivation of the 0.23.x & 2.0.x branches on one hand, and the 1.x branches on the other, as described here:
> http://mail-archives.apache.org/mod_mbox/hadoop-user/201301.mbox/%3CCD0CAB8B.1098F%25evans%40yahoo-inc.com%3E
>
> One gets the impression the Hadoop committers are split into two teams, with one team working on 0.23.x/2.0.2 and another team working on 1.x, running the risk of increasingly diverging products eventually competing with each other. Is that the case? Is there expected to be a Hadoop 3.0 where the results of the two lines of development will merge or is it increasingly likely the subteams will continue their separate routes?
>
> Thanks,
> Glen
> --
> Glen Mazza
> Talend Community Coders - coders.talend.com
> blog: www.jroller.com/gmazza
--
Arun C. Murthy
Hortonworks Inc.
http://hortonworks.com/
Re: Cohesion of Hadoop team?
Posted by Arun C Murthy <ac...@hortonworks.com>.
Absolutely not true.
We have a clear policy in place which ensures there is no feature disparity between hadoop-1 and hadoop-2. So, hadoop-2 is a super-set of functionality in hadoop-1 and there is no disparity.
hth,
Arun
On Jan 18, 2013, at 6:48 AM, Glen Mazza wrote:
> Hi, looking at the derivation of the 0.23.x & 2.0.x branches on one hand, and the 1.x branches on the other, as described here:
> http://mail-archives.apache.org/mod_mbox/hadoop-user/201301.mbox/%3CCD0CAB8B.1098F%25evans%40yahoo-inc.com%3E
>
> One gets the impression the Hadoop committers are split into two teams, with one team working on 0.23.x/2.0.2 and another team working on 1.x, running the risk of increasingly diverging products eventually competing with each other. Is that the case? Is there expected to be a Hadoop 3.0 where the results of the two lines of development will merge or is it increasingly likely the subteams will continue their separate routes?
>
> Thanks,
> Glen
> --
> Glen Mazza
> Talend Community Coders - coders.talend.com
> blog: www.jroller.com/gmazza
--
Arun C. Murthy
Hortonworks Inc.
http://hortonworks.com/
Re: Cohesion of Hadoop team?
Posted by Suresh Srinivas <su...@hortonworks.com>.
On Fri, Jan 18, 2013 at 6:48 AM, Glen Mazza <gm...@talend.com> wrote:
> Hi, looking at the derivation of the 0.23.x & 2.0.x branches on one hand,
> and the 1.x branches on the other, as described here:
>
> http://mail-archives.apache.org/mod_mbox/hadoop-user/201301.mbox/%3CCD0CAB8B.1098F%25evans%40yahoo-inc.com%3E
>
> One gets the impression the Hadoop committers are split into two teams,
> with one team working on 0.23.x/2.0.2 and another team working on 1.x,
> running the risk of increasingly diverging products eventually competing
> with each other. Is that the case?
>
I am not sure how you came to this conclusion. The way I see it is, all the
folks are working on trunk. Subset of this work from trunk is pushed to
older releases such as 1.x or 0.23.x. In Apache Hadoop, features always go
to trunk first before going to any older releases 1.x or 0.23.x. That means
trunk is a superset of all the features.
Is there expected to be a Hadoop 3.0 where the results of the two lines of
> development will merge or is it increasingly likely the subteams will
> continue their separate routes?
>
2.0.3-alpha, which is the latest release based off of trunk, that is in
final stage of completion should have all the features that all the other
releases have. Let me know if there are any exceptions to this that you
know of.
>
> Thanks,
> Glen
>
> --
> Glen Mazza
> Talend Community Coders - coders.talend.com
> blog: www.jroller.com/gmazza
>
>
--
http://hortonworks.com/download/
Re: Cohesion of Hadoop team?
Posted by Andrew Purtell <ap...@apache.org>.
It might also be helpful to differentiate between what commercial entities
might do with release lines in ASF Hadoop and the relationship between the
branches within ASF Hadoop itself. ASF Hadoop 1 and ASF Hadoop 2 are two
versions of the same product, they do not compete with each other.
On Fri, Jan 18, 2013 at 6:48 AM, Glen Mazza <gm...@talend.com> wrote:
> One gets the impression the Hadoop committers are split into two teams,
> with one team working on 0.23.x/2.0.2 and another team working on 1.x,
> running the risk of increasingly diverging products eventually competing
> with each other.
--
Best regards,
- Andy
Problems worthy of attack prove their worth by hitting back. - Piet Hein
(via Tom White)