bigtop-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Roman Shaposhnik <ro...@shaposhnik.org>
Subject Re: [DISCUSS] Working on the BOM.next (0.9.0 or 1.0?)
Date Mon, 13 Oct 2014 03:22:19 GMT
Hi!

Sorry for coming to the party rather late (but at least
I had some time to think about it ;-)).

On Sat, Oct 4, 2014 at 4:10 PM, Konstantin Boudnik <cos@apache.org> wrote:
> Now as 0.8.0 is practically done I want to kick of the discussion about next
> release's scope and aim. We already have discussed a number of things that we
> want to address in this round, including
>     - CI improvements
>     - build system enhancements
>     - overhaul of the testing experience
>     - TBD

Yup. All of the above. Personally, whatever limited cycles I've
got I'll sink them into CI and Docker/OSv work. Not to mention
the usual upgrades of the BOM.

Not sure what other have as the main goal for the next few months.

As a general suggestion: I'd really like to see us get together
sometime soon (perhaps Strata or right after it in the valley)
brainstorm some of these things that we'd like to do and put
them on a wiki (or better yet into JIRAs).

Now, if I were a king for a day (read: I had extra resources,
not just my copious free time) I would really like to tackle
a problem of tracking upstream releases of various components
and essentially offering Bigtop feedback whenever they
do their RCs. I've tried doing that on and off with Hadoop,
but since it was pretty time consuming had to drop it :-(

> A very important topic to cover here is, of course, the scope of the next
> release. Bigtop has a very powerful ability to help new technologies to thrive
> by wider promoting them across the Apache communities and beyond. However,
> without deliberate help from the perspective projects we can not keep on
> supporting them in the Apache Hadoop Stack: after all we are a bunch of
> volunteers spending our own time on things we like to do. Hence, we need to
> carefully prioritize what we realistically can or can not support.

I think historically what I've seen is that upstream communities are more
than happy to receive feedback but most of the time they don't have
cycles to contribute to Bigtop directly.

Currently I feel like the size of our stack (in terms of # of components)
is not (yet) over the top. Feels like we can keep managing it just fine.

> The other side of this coin is to make this project fun where people want to
> contribute something they feel passionate about. Let's say, I am the only one
> contributing fixes and supporting component X. However, for whatever reason I
> couldn't possibly care less about having component X on board. The question is
> - shall I use my limited time on the face of this planet for something that
> makes me tick, instead of wasting it on the dull chores around component X?
> Let's weigh carefully on what we want to carry on and what we'll be dropping
> this time.

That's a great point!

> Supported OS spectrum: let's drop the dead weight and carefully consider what
> to add.

I'd say we only have to support one stable (LTS type) of OS and one cutting
edge in every "family". But if push comes to shove, I'd rather optimize in favor
of cutting edge since that's where the fun is (at least for me).

> Release cycle: with new docker based CI we should be better equipped for
> regular builds testing and build environment which will be _exactly_ the same
> for developers and official Jenkins. Hence, can we have next release in the
> by the end of 2014? We already know that 0.8.0 has a few issues that we want
> to improve upon quickly.

I'd very strongly suggest we try 6 months this time, but be very, very
diligent about
it. IOW, the next release is *guaranteed* to happen in exactly 6
months from now.
We may, of course, still have intermediate releases if anybody's interested.

> User facing documentation, wiki, website: let's see where we can get some help
> with this. Shall we try recruiting fresh-grads and students who wants to start
> in on of the projects we support, but they don't know how? How we can increase
> the reach?

If only I knew how to recruit more users :-( I've tried everything by now and it
still feels like we've got a core community of developers, but not so
much users.

Thanks,
Roman.

Mime
View raw message