Add namaster package

[tskisner]

No description provided.

[iparask]

@tskisner I am happy with this change. Do you think we can merge it and trigger an env creation?

[tskisner]

The problem (and why this is not merged yet) is there is something unrelated failing in the CI tests. I agree we need this merged and a new tag. I'll work on packaging tasks later today.

[Bai-Chiang]

I think it's unrelated to namaster package. I've used my own configuration [1] to setup SOOPERCOOL environment. This configuration worked before, but stopped working before I went to CMB summer school. Currently even the default config won't work, pixell package would fail compiling (both python 3.10 and 3.11 would fail).

[1] https://github.com/Bai-Chiang/SO_pipeline_test/blob/main/soconda_config/soopercool/packages_conda.txt

[iparask]

At Della, at least, namaster tries to upgrade numpy to version 2. I'm trying to install namaster with pip after everything has finished for python 3.10. Is there a chance your numpy version is upgraded and causes the failures?

[iparask]

@Bai-Chiang can you also share the error you see from pixel?

[Bai-Chiang]

GitHub can't render the entire log_pixell, here is the pastebin link:

https://pastebin.com/fw11smw4

I'm trying to build pixell manually see if I can reproduce error.

It happens when I building packages [1] before installing any packages [2]. If I check the env when this error happened, numpy hasn't installed yet. The numpy version used when building pixell is 1.26

[1]

soconda/soconda.sh

Line 241 in 1638122

# Build local packages. These are built in an isolated environment with

[2]

soconda/soconda.sh

Line 276 in 1638122

# Install upstream conda packages.

[Bai-Chiang]

Update:
I could build and install pixell manually.
Here is the env.yaml file: https://pastebin.com/3nm8ATEu

Clone pixell repo
git clone [email protected]:simonsobs/pixell.git && cd pixell && git switch --detach tags/v0.23.14

Create conda env with
conda env create -f env.yaml && conda activate pixell-debug-3.11

Build locally with
python -m pip install --no-binary ":all:" --ignore-installed --no-deps -vvv --ignore-installed --prefix "${CONDA_PREFIX}" .

[tskisner]

Since everyone is happy with this, let's merge and then I'll rebase #42, which has version updates and fixes pixell compilation with conda compilers.