Hey Daniel, ke 22.10.2025 klo 16.55 Daniel Gröber ([email protected]) kirjoitti: > On Wed, Apr 30, 2025 at 05:11:11PM +0300, Martin-Éric Racine wrote: > > To me, the problem is Lintian complaining that what's in our > > subdirectory lacks the shebang (but the script that sources them has > > it) i.e. /usr/lib/dhcpcd/dhcpcd-run-hooks has the shebang, but > > everything inside /usr/lib/dhcpcd/dhcpcd-hooks lacks it. > > Lintian is correct here. /usr/lib allows non-executable files for > historical reasons libexec does not. See quoted d-policy "relaxed to a > suggestion" below: > > >> 9.1.1. File System Structur > >> > >> [...] The following exceptions to the FHS apply: > >> > >> The FHS requirement that architecture-independent application-specific > >> static files be located in /usr/share is relaxed to a suggestion. > >> > >> In particular, a subdirectory of /usr/lib may be used by a package (or a > >> collection of packages) to hold a mixture of architecture-independent and > >> architecture-dependent files. However, when a directory is entirely > >> composed of architecture-independent files, it should be located in > >> /usr/share. > > Also FHS 3.0 Sec 4.7: > > >> /usr/libexec includes internal binaries that are not intended to be > >> executed directly by users or shell scripts. > > Sourced scripts are neither "binaries" nor "executed". They are > "interpreted" if anything.
That's the way I interpreted it as well, hence why everything is still in /usr/lib even though the top item (the script that sources the numbered fragments) is an executable shell script. > > > Looking more closesly at dhcpcd-base I'm actually confused. In addtion to > > > /usr/lib/dhcpcd/dhcpcd-hooks/ there's also eg. /usr/share/dhcpcd/hooks/ > > > wich has more numbered scripts: > > > > > > /usr/share/dhcpcd/hooks/10-wpa_supplicant > > > /usr/share/dhcpcd/hooks/15-timezone > > > /usr/share/dhcpcd/hooks/29-lookup-hostname > > > > > > What's the difference? > > > > See https://github.com/NetworkConfiguration/dhcpcd/issues/494 > > > > Those are sample scripts that are not installed by default (but might > > be symbolically linked by the administrator if needed). > > I see. > > Well since you're big on "pedantic respect" if these truly are examples > they outght to be in /usr/share/doc/dhcpcd/examples by policy: > https://www.debian.org/doc/debian-policy/ch-docs.html#examples Agreed. Then again, having recently had a discussion with upstream about what those sample scripts do (provide workarounds that haven't been needed in ages), I'd be tempted to just stop shipping them. > > ke 30.4.2025 klo 15.21 Daniel Gröber ([email protected]) kirjoitti: > > > On Thu, Apr 17, 2025 at 03:24:33PM +0200, Laurent Bigonville wrote: > > > > Would be a good idea to move the dhcpcd-run-hooks and the dhcpcd-hooks > > > > directory > > > > > > So we're talking about these: > > > > > > /usr/lib/dhcpcd/dhcpcd-hooks/01-test > > > /usr/lib/dhcpcd/dhcpcd-hooks/20-resolv.conf > > > /usr/lib/dhcpcd/dhcpcd-hooks/30-hostname > > > > > > > [move] to /usr/libexec > > > > > > To what end? > > > > Pedantic respect of FHS, which Laurent rightfully hopes to achieve. > > How does blind pedantic respect serve our users? That's the endless debate: Do we put respecting standards first or keeping things predictable? Personally, I understand Laurent's point, but I'm not sure that changing this path just for FHS' sake is desirable. Martin-Éric PS: I'll get back to your thread about ifupdown integration sometime next week. You've raised some valid points which I'd like to address when I have enough time to sit down, and that won't be until the end of next week.
