I'd like to open up the discussion of "fixing" the yt coordinate
systems, as we move nearer and nearer on 3.0. Jeff, Nathan and
Britton have brought this up a couple times, that the x/y/z ordering
is not consistent with what they expect, and I'd like to figure out if
we can fix that now -- it's as good a time as any for bandaid ripping.
It may just be a matter of the transposition of buffers and the
x_dict and y_dict, or it might be more complex, although I suspect it
won't be much more than fixing those two items.
I'm inclined to remove plot collection before yt 3.0 is released.
PhasePlot and ProfilePlot both work in 2.6, and we've actually seen
quite a bit of migration to them.
If there's enough support, I'll bring this to a YTEP and we can hash
it out there, but before I went to that point I figured I'd look for a
handful of +1's here.
While we did benefit for a long time from ShiningPanda, we have lately
been relying on Jenkins hosted by Kacper, and hopefully on Polyphemus
in the future. Right now we still use ShiningPanda for building the
blog, but that will be easy to fix. Sad to see them go.
---------- Forwarded message ----------
From: Alexis Tabary <alexis.tabary(a)shiningpanda.com>
Date: Sat, Mar 29, 2014 at 8:15 AM
Subject: ShiningPanda CI, clap de fin
To: Alexis Tabary <alexis.tabary(a)shiningpanda.com>, Olivier Mansion
Today brings some sad news: we are shutting down ShiningPanda CI.
Jenkins instances will remain available until April 30th, at which
point they will all go offline.
This decision became inevitable as we failed to keep ShiningPanda CI
relevant in an increasingly competitive environment: our user base has
been eroding for a rather long while, so we decided to move forward
and simply halt the service.
We are incredibly thankful to all the people who have been relying on
ShiningPanda for integration and deployment, so we want to ensure as
smooth a transition as possible. All users who wish to download
Jenkins configuration files, or build history, should contact us. We
might also be able to help with migrations to other services, in-house
or hosted. Just send us an email.
As for us - the team behind ShiningPanda CI - for now we are focusing
on requires.io, a web service to keep track of dependencies on pypi.
Thank you all for supporting us for so long.
Olivier & Alexis
ShiningPanda Founding Partner
Monitor your Python dependencies
ShiningPanda Consulting Services
Build and Release Management
I'm writing with a quick wrapup of the yt workshop we had at UCSC this
week, sponsored by UC-HiPACC.
All in all, I am extremely pleased with everything that got done.
Over the few days, we had a huge number of improvements all over the
code base, and I was absolutely delighted by the new contributors and
the new improvements. A few highlights:
* SPH frontends (Gabriel Altay and Ben Keller) have been vastly
improved, including ionization fields and a new system for
autodetecting Tipsy frontends (load() now works!) along with
* The GPU volume rendering interface (Alex Bogert, John Holdener and
Nick Smith) has been well-defined, and will be interoperable with the
yt software renderer.
* Halo Finder / Analysis (Hilary Egan, Brian Crosby) has been
streamlined, the mass function code cleaned up, and the FOF+Rockstar
code has been implemented more formally
* The NMSU-ART frontend has been cleaned up, double-checked, and
particle-type-enabled. (Kenza Arraki) Also, new answer tests!
* Software VR has been ripped out and started to be reconstructed
(Sam Skillman) with an eye to clarity and simplicity in the interface.
* Docs and polish (Nathan Goldbaum) have been worked on a
considerable amount -- in fact, the docs now build again for 3.0, and
we're getting awfully close to a full set of them that cover the new
* IRATE halo catalogs can now be loaded in and examined (Miguel
Rocha), thus furthering the ability of yt to read, write, and analyze
halo catalogs as first class datasets.
* A new binding-energy check has been started (Andy Elvin)
* Boxlib 2D polar has been enabled (Elizabeth Lovegrove)
Anyway, it was really fun, productive -- and in many cases,
documented! With the docs now building on a regular basis, we'll
continue to check things in and update the documentation as new
features are added and existing features are polished. Thanks again
to Ji-hoon, Miguel, and Nathan for their organization, and thanks to
Joel Primack, Sue Grasso and all of UC-HiPACC for their support of the
workshop as a whole.
It is my esteemed pleasure to announce that PyTAPS - the Python wrapper to
iTAPS, MOAB, CGM, and Lasso is now available on bitbucket . It is my
hope that this will serve as a launching point for further improvements to
PyTAPS and further collaborations using MOAB.
Vijay and I will be working towards and to enable this goal. Personally, I
am very excited to see this happen. If you are interested in helping PyTAPS
evolve or know someone who is, please let us know!
New issue 821: Numpy not found in install_script.sh
results in an error:
Downloading xray_emissivity.h5 from yt-project.org
Edit setup.cfg to change the build options
matplotlib: yes [1.3.0]
python: yes [2.7.6 (default, Mar 27 2014, 16:56:17) [GCC
4.8.1 20130909 [gcc-4_8-branch revision 202388]]]
platform: yes [linux2]
REQUIRED DEPENDENCIES AND EXTENSIONS
Requires numpy 1.5 or later to build. (Numpy not found)
Ive already installed a different python distribution and hence assume an conflict with the already definied PATHs:
Anyone can help me to fix that? Could I come along with a manual installation of yt?
New issue 820: "all" particle type fields are dimensionless
All particles fields that are accessed with "all" are now being returned as dimensionless. For example, with an enzo-3.0 dataset:
data["all", "particle_mass"] returns dimensionless, but
data["DarkMatter", "particle_mass"] returns the proper units.
Today I found a cool new free service that the readers of this list might
be interested in.
Take a look at this page:
This is a free continuous integration service that supports bitbucket and
mercurial repositories. If you're familiar with Travis CI, this is very
similar, but they don't require you host your code using git on github.
It looks like they're mostly in the business of testing ruby and node
webapps, but it's still possible to set up a scientific python stack using
anaconda and then test python code.
To get a basic yt testing setup working, I used the following configuration:
chmod +x ./Anaconda-1.9.1-Linux-x86_64.sh
PYTHONPATH= PATH=/home/magnum/anaconda/bin:$PATH python setup.py develop
Testing suite commands:
PYTHONPATH= PATH=/home/magnum/anaconda/bin:$PATH nosetests
Unfortuantely it's not perfect - there are some resource limitations that
prevent it from running the full yt unit tests. That said, I've contacted
their support and they're interested in seeing if they can configure
something on their end to prevent the resource limits from kicking in in
In any case, I think this is generically useful - particularly among this
group, who tend to use mercurial and bitbucket a lot more often than a
random sample of developers. If you want to test a small side project,
this might be just what you need.
Hope that's helpful,