conda-forge / conda-forge/conda-forge.github.io
Document ecryptfs considerations in the "local testing" docs
- Dominant language
- JavaScript
- Stars
- 170
- Forks
- 320
- Avg merge
- 2d 10h
- Merged PRs (30d)
- 5
Description
### Where should the content be added?
in the ["local testing" maintainer docs](https://conda-forge.org/docs/maintainer/updating_pkgs/#testing-changes-locally)
### What should be added?
Add a troubleshooting note -- if you receive "File name too long" when running `build-locally.py`, it's likely you're working in an encrypted file system. Clone the feedstock to an unencrypted location (e.g. `/tmp/my-feedstock`? `/home/shared/my-feedstock`?) and try again!
### Additional information
Currently, if the user has an encrypted home directory with `ecryptfs`, they will receive "File name too long" when building locally. This is a well-known problem, but I did a fresh install of Linux on my laptop a few months ago, and I didn't even remember I had an encrypted home directory until I tried to do a local build and it failed. When users search for documentation about this, they may find [a GitHub issue on the conda-build repo](https://github.com/conda/conda-build/issues/1482) as the first result. The suggestion made there is to "install miniconda to a location that is not encrypted". That doesn't help with `build-locally.py`, which uses Docker volumes to `$PWD`; so the user really needs to be working on a **feedstock** in an unencrypted location; they don't even need miniconda installed at all. The user may waste time following this unintentional red herring!
Contributor guide
Assessment
This issue has not been assessed yet.