Rename release files so that `uname -m` could be used
ifireball opened this issue · 4 comments
The files released on GitHub use rather strange names for architectures such as amd64
(usually referred to as x86_64
) or ppc64el
(Usually referred to as ppc64le
).
If standard architecture names would be used, it would be possible to download dumb-init
with an architecture-neutral command such as:
wget -O /usr/local/bin/dumb-init \
https://github.com/Yelp/dumb-init/releases/download/v1.2.2/dumb-init_1.2.2_$(uname -m)
This is especially important for writing Dockerfiles that can be built on multiple architectures.
The naming that is used currently is consistent with debian's architecture naming -- you can use dpkg --print-architecture
if you're on a debian-derivative
Any solution for when you are not?
Seems that kernel-based naming would be a more neutral ground.
Using the kernel naming probably makes the most sense for the binaries (I think the .deb naming is best left as-is). I'll make a PR to do that for future versions and for the current version I'll upload renamed copies (leaving the old ones in place as duplicates to avoid breaking anything, of course).
So the renames are:
dumb-init_1.2.2_amd64
=>dumb-init_1.2.2_x86_64
dumb-init_1.2.2_arm64
=>dumb-init_1.2.2_aarch64
dumb-init_1.2.2_ppc64el
=>dumb-init_1.2.2_ppc64le
dumb-init_1.2.2_s390x
=> no change
I went ahead and re-uploaded the existing binaries under their new names for 1.2.2: https://github.com/Yelp/dumb-init/releases/tag/v1.2.2