These are chat archives for spyder-ide/public

7th
Sep 2018
Carlos Cordoba
@ccordoba12
Sep 07 2018 03:43 UTC
@paulgureghian_twitter, your error is solved by downgrading ipykernel with the command conda install ipykernel=4.8.2
compressionist
@compressionist
Sep 07 2018 18:44 UTC
Hi, Im getting console error (... raise ValueError('field 6 out of range (need a 48-bit value)'...) while staring Spyder for python 2 on mac high sierra. tried everything (from this post etc.) still getting this error. What Spyder/kernel/conda version should I use?
compressionist
@compressionist
Sep 07 2018 19:11 UTC
I've found the solution: "conda update python python.app"
now I get only warnings: "[warn] kq_init: detected broken kqueue; not using.: Undefined error: 0"
CAM Gerlach
@CAM-Gerlach
Sep 07 2018 19:12 UTC
@compressionist Yup, quite a few people ran into that problem a good many months ago, and we were able to successfully resolve it in exactly that manner.
That warning is from Qt and can be safety ignored
compressionist
@compressionist
Sep 07 2018 19:13 UTC
Ok, Thanks
CAM Gerlach
@CAM-Gerlach
Sep 07 2018 19:16 UTC
Glad you got it working!
compressionist
@compressionist
Sep 07 2018 19:24 UTC
Yeah, I'm pretty much new on python Mac development. On Win/Centos7/Ubuntu almost now issue with Spyder. So, overal it is very good product. Thanks.
Carlos Cordoba
@ccordoba12
Sep 07 2018 19:36 UTC

Hi, Im getting console error (... raise ValueError('field 6 out of range (need a 48-bit value)'...) while staring Spyder for python 2 on mac high sierra

@compressionist, that bug is in Python itself, not in Spyder. It was fixed in Python 3.6.5 but it was not backported to Python 2.7. So I'm afraid you won't be able to use that version with Spyder

Paul Gureghian
@paulgureghian_twitter
Sep 07 2018 19:37 UTC
Capture12.PNG
I removed and created a new py 3 env, and I still get this error. thanks
CAM Gerlach
@CAM-Gerlach
Sep 07 2018 19:38 UTC
@ccordoba12 Well they said it did apparently work for them, so...
@paulgureghian_twitter Did you do conda install ipykernel=4.8.2 as @ccordoba12 asked you to?
Carlos Cordoba
@ccordoba12
Sep 07 2018 19:39 UTC
@CAM-Gerlach, as far as I know, the fix for that ValueError was not backported
Paul Gureghian
@paulgureghian_twitter
Sep 07 2018 19:39 UTC
doing it now, lets see what happens
that seems to work. do i have to keep ipykernel at 4.8.2 ? how to update the env without updating ipykernel ?
CAM Gerlach
@CAM-Gerlach
Sep 07 2018 19:43 UTC
@ccordoba12 Hmm, so not sure how it fixed it unless it somehow updated all the way to Python 3...maybe Anaconda backported it in their version, or fixed it with some sort of shim in their Python.app wrapper?
@paulgureghian_twitter ipykernel shouldn't update by itself unless something needs it to update or you run conda update --all
So just update what you need, and leave the rest. That's why isolated environment are great, since you can create as many as you like and update some but not others, and some packages but not others, each according to your needs.
Carlos Cordoba
@ccordoba12
Sep 07 2018 19:44 UTC
@paulgureghian_twitter, yes, please leave it for now. We're waiting for ipykernel 4.9.1 that should fix that problem
Paul Gureghian
@paulgureghian_twitter
Sep 07 2018 19:45 UTC
i do "conda update --all every morning. is there another way ? do them all individually ?
CAM Gerlach
@CAM-Gerlach
Sep 07 2018 19:47 UTC

i do "conda update --all every morning. is there another way ? do them all individually ?

Unless you really know what you're doing, that's what we call A Bad Idea™ since with so many variables and inter-dependencies involved, it only takes two of them having a problem with one another to cause chaos. If you want to be safe, only update all your packages when a new version of the anaconda metapackage is released (conda update anaconda), aside from updating conda itself, spyder, and specific packages if you really need a specific bug fix or feature sooner (like I sometimes do with pandas or other specific stuff.

Paul Gureghian
@paulgureghian_twitter
Sep 07 2018 19:48 UTC
so , only update if i have errors ?
Carlos Cordoba
@ccordoba12
Sep 07 2018 19:49 UTC
It's better to update only there are new versions of important packages (like Pandas, Matplotlib, IPython, etc)
Or once a month, or every two months. Else you'll get these zero-day errors kind of frequently
Paul Gureghian
@paulgureghian_twitter
Sep 07 2018 19:50 UTC
is 'conda update --all" the recommended method ?
Carlos Cordoba
@ccordoba12
Sep 07 2018 19:51 UTC
Yes, it is. You just don't need to run it every day
Paul Gureghian
@paulgureghian_twitter
Sep 07 2018 19:52 UTC
when you say anaconda metapackage, does that mean ANACONDA, as in the whole distribution ?
Carlos Cordoba
@ccordoba12
Sep 07 2018 19:53 UTC
Yes, that's right. anaconda is like any other package, but composed of a fixed set of packages
CAM Gerlach
@CAM-Gerlach
Sep 07 2018 19:53 UTC
Basically anaconda is a metapackage, essentially a package that doesn't do anything by itself but specifies a specific set of dependencies, in this case all the default packages included with the Anaconda distribution and their precise versions, which are tested to work well together.
If you want to be conservative, you can just update that, plus any important packages, or you can conda update --all at a reasonable frequency like @ccordoba12 suggested.
Paul Gureghian
@paulgureghian_twitter
Sep 07 2018 20:02 UTC
thanks. do you like chrome browsers new UI ? I dont like it much , but will get used to it i guess.
CAM Gerlach
@CAM-Gerlach
Sep 07 2018 20:16 UTC
I think you might have posted that in the wrong chat tab?