Open cosmicBboy opened 2 months ago
You should remove your token! :)
Regarding the environment.yml - it might only work with micromamba
. Or if you want to try something newer and easier, you can give pixi
a go. I put together this pixi.toml
file today: https://gist.github.com/wolfv/87d54ca24bbfb3c4ec34c9688f5c3de0
Once you have pixi, and the pixi.toml
you can run pixi run serve
to start a server that builds everything.
PS: I redacted your token but you shold still rotate it! I believe people can see hte edits on Github
You should remove your token! :)
How do I rotate this? Googling around yields no useful info on this
Additional context: this command was working for me yesterday. I tried a fresh miniforge install, but that didn't get around the issue.
The weird thing is this was working for me yesterday on mamba
🤷♂️ .
@cosmicBboy you should go to anaconda.org (or from where you got the token), and remove it (or recreate).
Yeah, weird, I believe it might be connected to some server issues then. @DerThorsten or @atrawog migth have more info.
At some point the subdir name also changed from emscripten-32
to emscripten-wasm32
.
Not sure if it is related: I got the same error on Read the Docs last week. The mamba command is executed by jupyterlite_xeus
in jupyterlite_xeus/create_conda_env.py
. I reinstall mamba through the environment file, which just works.
channels:
- conda-forge
dependencies:
+ - mamba
I don't know the reason though... But it just works.
I believe our issues are related https://github.com/emscripten-forge/recipes/issues/927. The temporary workaround that I'm using is to downgrade packages that were releasing on emscripten-forge since around the first week of May.
@wolfv it turns out this is a problem with the rattler-build build packages.
In the index.json
of the rattler-build packages is says:
"platform": "emscripten",
"subdir": "emscripten-wasm32"
but the platform is expected to be emscripten-wasm32
@LiamBindle now that we know the issue we are working to fix this as soon as possible!
btw I ran into this bug once, then I reinstalled mamba/micromamba in the host environment...now I cannot reproduce it anymore
I reinstall mamba through the environment file, which just works.
Could that mean that the mamba version specified in your readthedocs config for mambaforge is the culprit? We have this in our docs https://github.com/jupyterlite/xeus/blob/main/.readthedocs.yaml#L6
For the record, installing micromamba seems to also be a solution https://github.com/jupyterlite/xeus/pull/89
But this is definitely just a workaround for now
Hmm, it's correct though that teh platform is the first part of the subdir.
It's:
platform: emscripten
arch: wasm32
subdir: emscripten-wasm32
Didn't we patch conda in some way to allow emscripten installation?
Didn't we patch conda in some way to allow emscripten installation?
that was just for building packages with boa and conda-build.
When using conda to install emscripten-wasm32 package we add the target platform to a .condarc
and point conda/mamba to this rc via an env variable (CONDARC
I believe)
mhhh...I am more and more confused about this issue. So it seems that it appeared once we swtichd to rattler-build. But the first time we could upload rattler build packages was also when we updated the server.
Troubleshooting docs
Anaconda default channels
How did you install Mamba?
Mambaforge or latest Miniforge
Search tried in issue tracker
'emscripten' is not a valid Platform
Latest version of Mamba
Tried in Conda?
I have this problem with Conda as well, without using Mamba
Describe your issue
I'm trying to create a xeus python kernel and have the following environment file:
When I try to create an environment from this file:
I get the following:
Additional context: this command was working for me yesterday. I tried a fresh miniforge install, but that didn't get around the issue.
mamba info / micromamba info
Logs
No response
environment.yml
No response
~/.condarc