Replies: 1 comment
-
Same, for DPDK, it was decided to migrate to It was done during a transition period during which both compilation framework were maintained and the the initial one was phased out. |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
autoconf was a key element in hiding the xz backdoor (
build-to-host.m4
, cf. https://www.sentinelone.com/blog/xz-utils-backdoor-threat-actor-planned-to-inject-further-vulnerabilities/ ) This has prompted other FOSS projects to reevaluate their choice of build system, and I think we should do that too.First, to be fair, the hiding was done in the dist tarball, which included the extra
build-to-host.m4
file, this is the one thing of the backdoor that was never in xz git. We stopped providing dist tarballs, so this doesn't quite apply for us. (Unless other people create and distribute a dist tarball, which is in fact the case for some distros.)That said,
m4/*.m4
in our git would be the place to hide a backdoor in just the same. I'm not confident that we would catch in review a PR importing a slightly modified version of some common autoconf m4 macro.As we've seen on the community call, discussing this without pointing at some other system to contrast against. @qlyoung and I have spent time on this several (4?) years ago and ended up with meson as the most appropriate choice.
Meson has a list of projects using it, it notably includes
systemd
,libvirt
,lxc
and a whole bunch of GNOME packages. [cf. https://mesonbuild.com/Users.html]A minor non-security related improvement of meson over autoconf would be that the double-build-directory (
hosttools
) hack to buildclippy
(cf.frr/configure.ac
Lines 109 to 177 in 84d1fb1
(I haven't looked what else meson does better, this specific thing I just happen to remember from evaluating it those few years back.)
Beta Was this translation helpful? Give feedback.
All reactions