These are chat archives for spyder-ide/public

8th
Apr 2019
bronzecraft
@bronzecraft
Apr 08 16:27
@CAM-Gerlach Thanks for reply. My spyder is installed by using Anaconda with a version of 3.3 on Windows. I dont know what version of spyder-kernels I am using now. The Python version attached to the anaconda is 3.7, and I wanna to using a Python version of 3.5.1. I changed the interpreter in the preference of spyder, then I restart the kernel, it worked well in old console window, but it was show “an error occured while starting the kernel” when I open a new console window. It confused me much time and I dont know how to solve this problem.
CAM Gerlach
@CAM-Gerlach
Apr 08 16:54

Anaconda with a version of 3.3 on Windows

@bronzecraft That is not a valid Spyder or Anaconda version. Do you mean Spyder 3.3.3? Or 3.3.0?

I wanna to using a Python version of 3.5.1.

I strongly recommend using at least 3.6.x, or if not at least the latest 3.5.x release since 3.5.1 is over 3 years old with many bugs and security holes. Essentially every major package supports at least 3.6 now, and upgrading is a fairly minor matter.

it was show “an error occured while starting the kernel”

Again, which error was reported below that message? Otherwise, as I stated above, we are merely blindly guessing in the dark. As a first step, you need to install the spyder-kernels package in the Python environment you want to use with Spyder (in this case, your 3.5 environment) so Spyder can communicate it and you can use Spyder's advanced features with it. Thanks.

paulgureghian
@paulgureghian_gitlab
Apr 08 18:27
What is the 'rule of thumb' for envs in Anaconda for Spyder other than the 'base'. Do I need an env for every Python/ ML/DL framework I want to use with Spyder ?
Also, does the new version of Anaconda (2019.03) help or hinder Spyder ?
CAM Gerlach
@CAM-Gerlach
Apr 08 18:40

What is the 'rule of thumb' for envs in Anaconda for Spyder

@paulgureghian_gitlab Depends on your workflow, and really doesn't have much to do with Spyder directly. Spyder only needs to be installed in one env; the others are really up to what you're diong. You for every major project/domain you're working on with a coherent set of packages that need to talk to one another, just one working env for everything, or anywhere in between. Depends on how many big-iron framework and non-standard conda-forge/pip packages you use, etc.

Also, does the new version of Anaconda (2019.03) help or hinder Spyder ?

Its not quite that simple. On one hand, it should be tested to work with it and should deliver new features, enhancements and bug fixes for your other packages. In the other hand, particularly if you're in the middle of a project, there's always the risk of something breaking for uncertain gain. We do recommend you always keep Spyder itself up to date (3.3.4 was just released), with conda update spyder spyder-kernels.

paulgureghian
@paulgureghian_gitlab
Apr 08 18:43
Should we update Spyder in Anaconda with the conda terminal command or wait until Navigator has the update ?
CAM Gerlach
@CAM-Gerlach
Apr 08 18:45
Anaconda Navigator is just a thin GUI wrapper around conda, so there should be no difference. I always recommend using Anaconda prompt over Navigator, since the latter seems to have a ton of reported bugs.
paulgureghian
@paulgureghian_gitlab
Apr 08 18:45
OK. Thanks
paulgureghian
@paulgureghian_gitlab
Apr 08 19:01
Just did 'conda update spyder' and the output is 'already installed' . I have 3.3.3
CAM Gerlach
@CAM-Gerlach
Apr 08 19:04
It hasn't been pushed to their channel yet, so just be patient and the Spyder update prompt will tell you when to update. Also, be sure to use the command I specified, conda update spyder spyder-kernels to get the latest updates in that too, as well as conda install spyder-kernels=0 in any other environment you have spyder-kernels installed in.
paulgureghian
@paulgureghian_gitlab
Apr 08 19:04
Ok
paulgureghian
@paulgureghian_gitlab
Apr 08 19:10
Can we compile the update from source into binaries ? I think the source code is hosted on Github ?
CAM Gerlach
@CAM-Gerlach
Apr 08 19:13
You don't need to compile anything, you can run the latest development version (master or 3.x) by following the steps listed in our Contributing guide on Github. That said, if you're not up for bugs, half-baked features and figuring things out yourself without asking us all the time, I'd suggest you simply be patient, wait a few days and update to the new version when its available. After all, the 3.x series has reached the end of the line, with little more than just bugixes, all the feature development for the past year or two has been doing into Spyder 4. So unless you're experiencing a bug (unlikely, since they're mostly Qt 5.12 related which isn't even available on Anaconda yet) then you won't notice any real difference with the 3.3.4 release over 3.3.3.
paulgureghian
@paulgureghian_gitlab
Apr 08 20:06
Wouldn't the latest dev version be the same one which gets packaged and published as the official update ?
CAM Gerlach
@CAM-Gerlach
Apr 08 20:11
No, otherwise it wouldn't be the latest dev version, would it? There have already been 3 significant changes to the 3.x branch since 3.3.4 was released a few days ago, and over 3200 commits to master (Spyder 4), that's over 1/4th the commits in the entire history of Spyder. There are also difference in packaging, performance, behavior, etc. when Spyder knows that its running from a development build rather than a released version (since its intended for developers only, not regular users like yourself...hence the name).
paulgureghian
@paulgureghian_gitlab
Apr 08 20:14
Like BETA testers
CAM Gerlach
@CAM-Gerlach
Apr 08 20:15
Well (not sure why you capitalized it) technically there are Spyder 4 betas for that...
Hence, the name.
paulgureghian
@paulgureghian_gitlab
Apr 08 20:17
It will go from 3.3.4 to 4.0 ?
CAM Gerlach
@CAM-Gerlach
Apr 08 20:18
There will probably be at least one more 3.3.x release (3.3.5), and we have already released 4.0 beta 1 with beta 2 coming soon and beta 3 sometime after.
paulgureghian
@paulgureghian_gitlab
Apr 08 22:49
I installed the new Anaconda and tried to create the envs I had before. Now Python is crashing upon Spyder launch. Whats the command to capture the env details for troubleshooting this ?
CAM Gerlach
@CAM-Gerlach
Apr 08 22:59

I installed the new Anaconda and tried to create the envs I had before.

You mean you uninstalled your Anaconda distribution and did a full reinstall just to update it? That's totally unnecessary, ou only needed to run conda update anaconda or conda install anaconda=2019.03, and even that isn't necessary unless you have a specific reason for upgrading (you only really need to upgrade conda, spyder/spyder-kernels, and any core packages in your workflow that have specific enhancements you need with any regularity).

Whats the command to capture the env details for troubleshooting this ?

If there's error output, set SPYDER_DEBUG=3 before running Spyder will display it. If you don't see anything there, most likely the error is in Qt. Try conda install qt=5.9.6 and see if Spyder launches, and if that doesn't work, conda install pyqt=5.6. This is why we don't recommend people upgrade everything in a working environment unless they need to.

paulgureghian
@paulgureghian_gitlab
Apr 08 23:01
Wish I knew that before I nuked everything
CAM Gerlach
@CAM-Gerlach
Apr 08 23:16
...Or performed a simple Google search/read the docs?
paulgureghian
@paulgureghian_gitlab
Apr 08 23:17
I see a new version and I install it hoping for the best
CAM Gerlach
@CAM-Gerlach
Apr 08 23:18
...
paulgureghian
@paulgureghian_gitlab
Apr 08 23:33
I forgot the command which captures the env details so they can be shared with others
paulgureghian
@paulgureghian_gitlab
Apr 08 23:40
Do I need both qt and pyqt ?
CAM Gerlach
@CAM-Gerlach
Apr 08 23:41
No.

I forgot the command which captures the env details so they can be shared with others

I just replied to this above.

paulgureghian
@paulgureghian_gitlab
Apr 08 23:44
I just have qt 5.9.6 and no pyqt and Spyder 3.3.3 launched with ipython 7.4.0 and it looks good. Can I assume its ok now for me to start adding numpy, sklearn, tensorflow ... ?
CAM Gerlach
@CAM-Gerlach
Apr 08 23:54
Must be a bug with 5.9.7 on your machine. Yeah, go ahead. Best of luck!