Open timsnyder opened 2 years ago
So far, no issues for me, but a newer/regressed form of this flow would be great. As the RISC-V International has adopted Sparta as the framework of choice, I expect more people will want to use it.
https://github.com/sparcians/map/discussions/446#discussioncomment-7936895 points out a case where create_conda_env.sh
is broken. IT's also definitely broken for me when I try to run it.
I think it just recently broke too. I had a colleague at work try it in WSL and it was more work than expected. Back to just listing requirements? That'd suck.
Ok. I fixed it for myself this past weekend. I'll try to push something this week that fixes it.
Much appreciated!
I have the same problem, was able to run it around a month ago and now it's just broken and hangs forever.
Mine was https://github.com/sparcians/map/pull/488#issuecomment-2040220255. After the fix, the script worked correctly.
The availability of packages in conda-forge and changes in conda/mamba dependency solver impacts the ability of
scripts/create_conda_env.sh
to solve the environment (e.g. #145 ). Plus, I know there are assumptions in there that can be broken by edits toconda.recipe/
followed byconda smithy rerender
(e.g. #227 & #228)I don't think we should regress that script all the time with the rest of map but we should:
conda.recipe
mamba
is the reimplementation of the dependency solver in C++For this, I would create a GitHub Actions workflow so that it doesn't conflict with the auto-generated CI stuff.
@knute-sifive, @MichaelSchoenfelder-SiV , @benoy-sifive I don't know how important you guys think this is. Since I haven't been doing things with sparta and projects that use it, I'm not sure how often people have run into issues with my janky script or not. Actually, I'm just realizing after looking at it that the variant yamls have gotten simplified and need to be updated again.. lol.