[Yt-dev] Time-based releases

Matthew Turk matthewturk at gmail.com
Thu May 26 19:33:44 PDT 2011


Hi Cameron,

I agree, there is no sense in pushing out a poor release.  But perhaps
it would be worthwhile to try to aim for six-month intervals.

That would place this next release as coming out on October 6.  I
think that it would nominally be ready much sooner, possibly as soon
as the end of June.

The main features I see in this release:

  * Reason (the web GUI)
  * Initial Nyx support, which will be improved substantially over the
month of June
  * Python 2.7 requirement
  * Re-centering functions in halo profiler
  * Isolated plot objects (yt.visualization.profile_plotter contains these)
  * Initial plot window

We could potentially also have the new field containers, as well.
Additionally, yt-in-Paraview is coming along extremely well
(discussion on that has been ongoing on the "Amr" mailing list on the
kitware server), so we could very well have that done.

The reason I'm listing these features is that as we grow as a
community, maintaining modules on supercomputing centers becomes more
and more attractive.  And long waits between releases just means that
people who rely on those modules will become further detached from the
main stream of development.  Right now we have modules on Nautilus,
Kraken, and I was just notified tonight that modules on Odyssey (CfA)
and Trestles are forthcoming.

But, yes.  It needs to be well-tested and not shoved out like some
natty narwhal on its own with no sense of unity.  I like Britton's
point about doing testing.  I will try to finish up the outward facing
aspects of the test runner shortly and then write to the list (and
document) how it's currently set up to work.  One unfortunate aspect
of it is that it's currently challenging to fork or execute
subprocesses, and much of what I am skittish about with the new field
container is related to overriding and whatnot, which are best tested
in clean processor spaces.  So that's another item to add to that
list.

Anyhow, it sounds like we're all sort of on the same page: timed
releases, roughly 6 months apart, that are well-tested, is a good way
to go.  Let's strive for this, and I'd like to start it up by aiming
to have this relatively near-to-completion release out by the end of
June.

Thanks everyone for your feedback!

-Matt

On Thu, May 26, 2011 at 3:14 PM, Cameron Hummels
<chummels at astro.columbia.edu> wrote:
> +1 to semi-annual or annual, but contingent on having a testing mechanism to
> make sure things are *stable* when we release them.  there is no sense in
> pushing out lots of versions unless we're happy to make sure they are stable
> (after all, people can download the unstable from the web at any time), and
> making things stable takes time.
>
> cameron
>
> On 5/26/11 1:54 PM, Britton Smith wrote:
>
> I'm +1 for either Jeff's or Matt's version naming.  Also, quarterly I think
> should be the highest frequency.
>
> Britton
>
> On Thu, May 26, 2011 at 1:51 PM, david collins <antpuncher at gmail.com> wrote:
>>
>> +1 >= quarterly.
>>
>> On Thu, May 26, 2011 at 11:44 AM, j s oishi <jsoishi at gmail.com> wrote:
>> > +1 semiannual. but if they have alliterative animal names, i'm quitting.
>> >
>> > On Thu, May 26, 2011 at 10:38 AM, Matthew Turk <matthewturk at gmail.com>
>> > wrote:
>> >> (moved to yt-dev)
>> >>
>> >> Hi all,
>> >>
>> >> This is a great reason to push on a more regular release schedule.
>> >>
>> >> Do you think it might be worthwhile to aim for time-based releases?
>> >> We've had a spring release, maybe aim for a Summer release, and then
>> >> we could try to just do rolling releases like that?  [+-][01] on
>> >> [quarterly|semiannual|annual] releases?
>> >>
>> >> -Matt
>> >>
>> >> On Thu, May 26, 2011 at 1:35 PM, Anthony Harness
>> >> <anthony.harness at colorado.edu> wrote:
>> >>> Britton,
>> >>>
>> >>> Adding those lines works for the stable install, which I guess is
>> >>> maintained
>> >>> by Hari. My problem was running the unstable install maintained by
>> >>> Stephen,
>> >>> which he informed me does not work in parallel. As Stephen mentioned
>> >>> before,
>> >>> if Hari could update his development install, I could use that
>> >>> instead.
>> >>>
>> >>> Thanks,
>> >>> Anthony
>> >>>
>> >>> _______________________________________________
>> >>> yt-users mailing list
>> >>> yt-users at lists.spacepope.org
>> >>> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>> >>>
>> >>>
>> >> _______________________________________________
>> >> Yt-dev mailing list
>> >> Yt-dev at lists.spacepope.org
>> >> http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org
>> >>
>> > _______________________________________________
>> > Yt-dev mailing list
>> > Yt-dev at lists.spacepope.org
>> > http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org
>> >
>>
>>
>>
>> --
>> Sent from my computer.
>> _______________________________________________
>> Yt-dev mailing list
>> Yt-dev at lists.spacepope.org
>> http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org
>
>
> _______________________________________________
> Yt-dev mailing list
> Yt-dev at lists.spacepope.org
> http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org
>
> _______________________________________________
> Yt-dev mailing list
> Yt-dev at lists.spacepope.org
> http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org
>
>



More information about the yt-dev mailing list