Closed murrayE closed 3 years ago
Hi Murray, it looks like you installed SageMath-9.3.rc2 instead of the most recent release which is SageMath-9.3.rc4.
Can you please try with that latest (but soon to be obsolete) release?
One other thing that you could also try is to first run the app but select the terminal option. That should create the symlink in /var/tmp. However, running the jupyter from conda rather than using the built-in jupyter server has not been tested and might fail for other reasons even if the symlink exists. And it really would be best to debug the rc4 release.
So how do I start jupyter with sagemath-9.3.app if I do not use the conda environment?
You double-click the app icon to start it, select the radio button for using the notebook interface, choose a directory for storing notebooks, then click the "Launch" button.
On Thu, May 20, 2021, 10:56 AM murrayE @.***> wrote:
So how do I start jupyter with sagemath-9.3.app if I do not use the conda environment?
— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub https://github.com/3-manifolds/Sage_macOS/issues/5#issuecomment-845288212, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAJ6CPYUFEAN5IDMTKH2AK3TOU5KXANCNFSM45HK25KQ .
Sagemath 9.3 rc4 is working in jupyter. At first I thought not, but the problem was the notebook was trying kernel "Sage 9.3" rather than "Sage 9.3 rc4".
Great! Thanks!
However, I think that you have found a bug in the "extras" package even if it is not affecting your launch. You should have a file /usr/local/share/jupyter/kernels/sagemath_9_3_all/kernel.json on your system, installed by that package. On my system I notice that it seems to contain information about the SageMath-9.2 app, meaning that I forgot to update that file in the package when I released the SageMath-9.3.rc4 app.
I will leave this ticket open for now, and close it when I fix the package for the final SageMath-9.3 release, which should happen in early June.
the notebook was trying kernel "Sage 9.3" rather than "Sage 9.3 rc4".
I don't actually understand how you are running the app in order to make such an option available. I don't think that running it in the way that I expect could possibly lead to that. If you could provide details it would be appreciated.
I think I know what is going on. The 9.3 rc4 pkg installs into
/usr/local/share/jupyter/kernels/sagemath_9_3_all
an out of datekernel.json
file with
{"argv": ["/var/tmp/sage-9.3.rc2-current/local/bin/sage",
"--python", "-m", "sage.repl.ipython_kernel",
"-f", "{connection_file}"],
"display_name": "Sage 9.3",
"language": "sage"}
When I start Jupyter via SageMath-9-3.app and create a new notebook, there are two choices for the Sage kernel: Sage 9.3
and Sage 9.3.rc4
. The former is tied to the above kernel.json
file and doesn't work since argv
points to a nonexistent location. The latter works just fine.
I do have file /usr/local/share/jupyter/kernels/sagemath_9_3_all/kernel.json. In it I have files:
kernel.json logo-32x32.png logo-64x64.png logo.svg
And file kernel.json consists of the one line:
{"argv": ["/var/tmp/sage-9.3.rc2-current/local/bin/sage", "--python", "-m", "sage.repl.ipython_kernel", "-f", "{connection_file}"], "display_name": "Sage 9.3", "language": "sage"}
On my system I’ve now edited to correct the link to become "/var/tmp/sage-9.3.rc4-current/local/bin/sage” and to display the name as “Sage 9.3”.
On 20 May2021, at 2:40 PM, Marc Culler @.***> wrote:
Great! Thanks!
However, I think that you have found a bug in the "extras" package even if it is not affecting your launch. You should have a file /usr/local/share/jupyter/kernels/sagemath_9_3_all/kernel.json on your system, installed by that package. On my system I notice that it seems to contain information about the SageMath-9.2 app, meaning that I forgot to update that file in the package when I released the SageMath-9.3.rc4 app.
I will leave this ticket open for now, and close it when I fix the package for the final SageMath-9.3 release, which should happen in early June.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/3-manifolds/Sage_macOS/issues/5#issuecomment-845370899, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABILR2NH33SNY3LT3KTY2ADTOVJTDANCNFSM45HK25KQ.
——
Murray Eisenberg @. @.>
Professor Emeritus
Mathematics & Statistics Dept.
University of Massachusetts
Amherst, MA USA
Perfect! So that will get fixed in the real SageMath-9.3 release. In the meantime, editing that file as you did seems to be the best workaround. And I will continue to leave this ticket open to remind me.
Now fixed in the real release.
(This is from https://trac.sagemath.org/ticket/31814, where it might not be noticed because that trac system didn't allow me to specify sage 9.3 as the version!)
I sage-9.3-kernelStart_unhandled_error.txt installed Sage 9.3 app under macOS Big Sur (11.3.1) on Intel iMac. From SageMath-9-3.app I can run sage command-line in Terminal.
But I cannot seem to be able to use Sage from jupyter. I try to pen SageMath-9.3.app and choose "Jupyter notebook from folder" option, then Launch. I get error "Safari Can't Connect to the Server ... can't open the page 'localhost:8888/tree?token=......' because Safari can't connect to the server 'localhost'".
Further info: To start jupyter notebook, I have to start a conda environment in Terminal, then give command "jupyter notebook", which opens a jupyter notebook in Safari.
When the jupyter .ipynb notebook is open in Safari, it shows "Sage 9.3" as the kernel but I immediately see a Kernel error, which persists even if I try to restart the kernel or reconnect to it.
Part of the problem might be that the python being accessed is from the anaconda3 distribution, since I need that to even use the conda environment in Terminal in order to launch jupyter notebook.
The "Unhandled error" for the kernel ends with:
FileNotFoundError: [Errno 2] No such file or directory: '/var/tmp/sage-9.3.rc2-current/local/bin/sage' The complete text of the error message is attached.
Note that I can run, for example, a Julia 1.6.1 kernel from a jupyter notebook.