This repository contains conda recipes, Docker images and tools for building a
set of conda packages for the 32-bit ARM
architecture. Specifically 32-bit devices using version 7 or newer of the ARM
architecture with hardware floating point.
The major Linux distributions refer to this as
armhf or
armhfp.
In conda this is best described by the linux-armv7l platform.
Packages from this work are available in the c4armv7l channel. The final goal is to produce a miniconda like installer which can be used as a starting point for the conda-forge community to build a more complete set of packages for this platform.
All packages were build inside a CentOS 7 Docker container and should support Linux distributions which have a glibc version of 2.17 or newer. These packages should work on all major distribution which support the armhf or armhfp architectures including Debian Jessie, Ubuntu 16.04, and Fedora 28.
| package | version |
|---|---|
| glibc | 2.17 |
| gcc | 8.3.0 |
| binutils | 2.29.1 |
| OpenSSL | 1.1.1c |
| Python | 3.7.3 |
| pip | 10.0.1 |
| conda | 4.7.5 |
| conda-build | 3.17.8 |
| constructor | 2.2.0 |
| anaconda-client | 1.6.14 |
The order in which recipes were built can be found in the build_order.md document.
Earlier this year I, Jonathan Helmus, create a similar set of packages for the 64-bit ARM platform (aarch64 or arm64) in the conda4aarch64 repository
The conda-forge community has begun to build packages for the linux-aarch64
platform as part of the ARM/PowerPC migration. This build out can be tracked
on the conda-forge status page.
Any help porting the recipes in this repository to conda-forge would be most appreciated.
A discussions around conda-forge's support of aarch64 as well as other ARM
platforms can be found in
this GitHub issue.
In have also worked on Berryconda, a
conda based Python distribution for the Raspberry Pi single board computers.
Two flavors of the distribution support the armv7l and armv6l Raspberry Pis.