[Solved] Sagemath, jupyter, and the bloody JSON encoder...

gauthma · 2017-07-07 12:58:11

When using Sage's notebook, via jupyter, whenever evaluating a cell, regardless of its content (as long as it is valid code), I get the following warning:

/usr/lib/python2.7/json/encoder.py:207: DeprecationWarning: Interpreting naive datetime as local 2017-07-07 12:19:20.338945. Please add timezone info to timestamps.
chunks = self.iterencode(o, _one_shot=True)

A bit of googling indicated that this was indeed a bug, that has in the meantime, been fixed. However, no matter how much I updated, I keep stumbling upon across it. So to see if I could compile a list of steps to reproduce the problem, I uninstalled everything Sage plus dependencies (and also removed ~/.sage and ~/.jupyter)

# pacman -Rs sagemath sage-notebook sagemath-jupyter

Then I installed sagemath-jupyter, which pulls the necessary dependencies for sage et al.

# pacman -S sagemath-jupyter

Now start the notebook:

$ jupyter notebook

Then with kernel sagemath selected, type (eg)

print("foo")

and hit Shift+Enter (for evaluation). You get the above warning. The reason I post this in the forum, as opposed to a bug report, is because I found so little information about this bug in Archlinux that I can't help but wonder if I misconfigured something (and then forgot about it until I needed the notebook again...)

Thank you in advance.

EDIT TO ADD: I also have installed kernels for Python 2 and 3, but this warning is NOT shown when using either of those.

Last edited by gauthma (2017-07-10 13:05:29)

arojas · 2017-07-07 14:29:23

Add

warnings.filterwarnings(action='ignore')

to ~/.sage/init.sage

gauthma · 2017-07-07 16:59:53

arojas wrote:

Add
warnings.filterwarnings(action='ignore')
to ~/.sage/init.sage

Thanks, this definitely makes working in the notebook more bearable... but of course does not fix the underlying problem... is a bug report appropriate?

arojas · 2017-07-07 17:35:51

Well, you said it has been fixed already, so it will make its way into the repos eventually. If you can pinpoint where the fix is you can open a backport request, otherwise it can be a mess to dig into the gazillion python modules involved here to figure out where the issue is. A full backtrace of the warning would definitely help.

gauthma · 2017-07-07 17:55:41

Well, my understanding of this is that the problem (which is with jupyter only) is fixed. And in fact, when using jupyter with a Python 2 or 3 kernel, I don't get that warning.

It only happens when using sage, which is I think the problem must be in sage's kernel. I originally thought that the link said the problem with sage was fixed; it was only when I searched again for that link above that I noticed that the underlying problem was about jupyter alone -- that is the one that is fixed. Sorry about that

Which means submitting a bug report (about the sage kernel) might not be the worst of ideas...

arojas · 2017-07-08 00:01:17

According to that link the fix is in ipykernel 4.6.1, which is not in our repos yet

gauthma · 2017-07-10 13:05:15

Well, I guess that settles it then

Arch Linux

#1 2017-07-07 12:58:11

[Solved] Sagemath, jupyter, and the bloody JSON encoder...

#2 2017-07-07 14:29:23

Re: [Solved] Sagemath, jupyter, and the bloody JSON encoder...

#3 2017-07-07 16:59:53

Re: [Solved] Sagemath, jupyter, and the bloody JSON encoder...

#4 2017-07-07 17:35:51

Re: [Solved] Sagemath, jupyter, and the bloody JSON encoder...

#5 2017-07-07 17:55:41

Re: [Solved] Sagemath, jupyter, and the bloody JSON encoder...

#6 2017-07-08 00:01:17

Re: [Solved] Sagemath, jupyter, and the bloody JSON encoder...

#7 2017-07-10 13:05:15

Re: [Solved] Sagemath, jupyter, and the bloody JSON encoder...

Board footer