hadoop-pig-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "pi song" <pi.so...@gmail.com>
Subject Re: we need a release or a branch
Date Tue, 18 Mar 2008 13:18:04 GMT
Dear Stefan,

As someone who is under pressure of maintaining a big production system,
intuitively what you would do? try to stabilize or introduce new changes? So
in this point I understand people at Yahoo. They have "fear" due to their
work commitments.

I think this is just a matter of time. When we have more diverse people as
key people, they shouldn't be in the same mode all the time.

But one thing that I want to mention that I don't feel right is that still
some design decisions are made in meeting rooms instead of on JIRA issues.
This is not what open source developers are after. Please make sure that we
know as much as you know. The concept of open source is to drive the project
through the community's decisions to serve the community's needs not just
individual's (a person or a company).

Regarding the release, to some degree the current application is stable
enough for most operations. I agree that we should make a release out of it
(0.0.1-alpha whatever) so that people in production won't have problems and
people in development can move forward more quickly. I like the even/odd
version numbering system like in Linux kernel so that people can drive fast
when appropriate.

Cheers,
Pi

On Tue, Mar 18, 2008 at 1:17 PM, Stefan Groschupf <sg@101tec.com> wrote:

> Dear Developers,
>
> I think we need a 0.0.1 release or a branch! Let me explain why I
> think so, I'm sorry if I'm too honest but I prefer an open
> constructive discussion.
> I wonder what the motivation was to publish pig as open source. My
> understanding of an open source projet is that the ultimate goal is to
> establish a user and developer community and split the workload but
> also drive the development by the community needs and not just from
> one perspective.
> In the moment I do not see the open source community that much
> involved.  I understand and agree the need of an shared goal and a
> decision maker. However I thought that is defined in
> http://wiki.apache.org/pig/ProposedRoadMap
> . What disappoint me as someone that invested significant time to
> contribute improvements is that things move too slow from my point of
> view.
>
> As mentioned me and my colleagues are happy to help full-time to fix
> issues since we want to use pig in production and I see a couple more
> people in the list that working hard to contribute patches.
>
> But currently to get a patch to trunk takes a couple weeks so instead
> we considering now creating a own fork of pig or moving to cascading
> or jaql. Don't get me wrong pig functionality is great, the work to
> guys did so far is great and we really appreciate your hard work.
> However from a java point of  view there is significant space for
> improvement in the code, have a look to hbase test coverage, javadoc
> or code style.
>
> So I wondered what makes pig moving that slow and what is the reason
> why instead of moving forward there are concerns about removing a
> unused class (HConfiguration) or improving the build script
> incrementally.
> I guess the problem is that pig is already in use in a kind of
> production scenarios.
>
> Therefore I suggest we create a prerelease or an brunch to keep this
> users happy and start real development in trunk.
>
> Thoughts?
>
> Thanks for being patient with me. :-)
> Stefan
>
>
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> 101tec Inc.
> Menlo Park, California, USA
> http://www.101tec.com
>
>
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message