https://www.puppet.com/community logo
Join Slack
Powered by
# voxpupuli
  • v
    ewoud: basically, it requires the equivalent of libc-dev installed, which very small images might not have
  • v
    meena: ,/
  • v
    o/*
  • v
    damned, the morning start bad… I don't have your issue on 13.1-RELEASE at Hetzner…
  • v
    facter virtual #=> kvm
  • v
    Do you mean you have the issue in the sto^Wborrowed code? so in your code?
  • v
    smortex: I had a very slimmed down 13.1…built from PkgBase installed, and without -dev packages. libc.so is in FreeBSD-clibs-dev
  • v
    meena: Ok, better understand. Do you think we can / should do something that could improve the situation?
  • v
    smortex: i would have to look into Ruby's ffi to see if something can be done
  • v
    smortex: https://github.com/ffi/ffi/wiki/Basic-Usage#hello-world-using-ffi maybe this already fixes the issue
  • v
  • v
    Hetzner <3
  • v
  • v
    lol… /usr/lib/libc.so: ASCII text
  • v
    my mind hurt… wtf?
  • v
    and indeed it seems to point to the right .so.7 library
  • v
    yupp
  • v
    so the fix is: ffi_lib ['c', 'libc.so.7']
  • v
    and when FreeBSD bumps that again in fifteen years, then you do, ffi_lib ['c', 'libc.so.8', 'libc.so.7'] for a year or so
  • v
    you'd think there's a better way, like doing so in packaging
  • v
    https://github.com/ffi/ffi/wiki/Loading-Libraries#in-memory-libraries wouldn't Ruby load libc already so it's already in-memory?
  • v
    at least on Fedora 37 /usr/bin/ruby-mri is linked against libc.so.6
  • v
    possibly also more efficient
  • v
    duh, good point. Will try
  • v
    interesting that it also appears to use ffi_lib :FFILibrary:LIBC in some places, but not for bsd and freebsd
  • v
    s/it/facter/
  • v
    on the same Fedora 37 that resolves to libc.so.6
  • v
    :FFILibrary:LIBC is a lot of code
  • v
    the way most of facter is written, it seems to want to be as efficient as possible, so using that seems to defeat that idea
  • v
    it's used for hostname and load_averages so it's already loaded in many cases?
1...241242243...648Latest