Hi John,

If you take a look at Matt's fork, you'll see that I recently updated that unit test to compare to 12 decimal places instead of requiring bitwise identicality.  See this PR: https://bitbucket.org/MatthewTurk/yt/pull-request/6/unit-consistency-enzo-cosmology-units/diff

I think if you merge with Matt (specifically, the unitrefactor bookmark on https://bitbucket.org/MatthewTurk/yt) that issue should go away.  Not sure if new issues will crop up.


On Tue, Dec 31, 2013 at 3:45 PM, John Zuhone <jzuhone@gmail.com> wrote:
Hi all,

Good news on the Windows/yt front. I'm happy to announce that as of
right now only one unit test is failing. The failure is subtle, but I
thought I would run it by you all to see what we should do about it.

The failure in question is a comparison test in
yt/utilities/tests/test_units.py, in "test_power":

    assert u2.cgs_value == (pc_cgs**2 * mK_cgs**4)**2

basically, the unit cgs_value is exactly the combination of these
factors, so even though they are floats, one would expect that the
equality should hold, and on our other platforms it does. However, on
Windows, the comparison is failing on the last digit, with a relative
difference of ~4e-16 between the two values. So, while I'm not
inclined to be very concerned about it, I'm curious to see what you
guys think about how it should be handled.

Happy New Year!

John Z

John ZuHone

Postdoctoral Researcher
NASA/Goddard Space Flight Center

yt-dev mailing list