08:25 <ardo> farkuhar: I could be wrong, I've just taken a quick look, but it looks like posix does not define the rehashing events, simply what and how to execute a command when it is invoked: https://tinyurl.com/yys2enna

08:26 <ardo> farkuhar: can't you preemptively 'hash -r'?

08:52 <cruxbot> [opt.git/3.7]: xterm: update to 387

11:10 <cruxbot> [opt.git/3.7]: clang: 17.0.2 -> 17.0.3

11:10 <cruxbot> [opt.git/3.7]: compiler-rt: 17.0.2 -> 17.0.3

11:10 <cruxbot> [opt.git/3.7]: libclc: 17.0.2 -> 17.0.3

11:10 <cruxbot> [opt.git/3.7]: lld: 17.0.2 -> 17.0.3

11:10 <cruxbot> [opt.git/3.7]: lldb: 17.0.2 -> 17.0.3

11:10 <cruxbot> [opt.git/3.7]: llvm: 17.0.2 -> 17.0.3

11:10 <cruxbot> [compat-32.git/3.7]: llvm-32: 17.0.2 -> 17.0.3

14:38 <SiFuh> r0ni: ivandi: I upgraded my main system and mc.ext is no longer recognised. I was able to use this to convert my old mc.ext to the new mc.ext.ini https://github.com/yurytch/mc.ext_to_mc.ext.ini however, mc still wasn't working. On arch forum they suggested changing all \ to \\ sed '/Regex=/s/\\/\\\\/g' mc.ext.ini Now it works again

15:28 <farkuhar> ardo: thanks for investigating. It would have been nice if the standard mandated a fresh search of $PATH whenever the previously hashed value points to a nonexistent file. But preemptively running 'hash -r' is a fine solution too.

15:29 <cruxbot> [contrib.git/3.7]: foot-terminfo: 1.16.1 -> 1.16.2

15:29 <cruxbot> [contrib.git/3.7]: foot: 1.16.1 -> 1.16.2

15:45 <cruxbot> [opt.git/3.7]: mariadb: include fix for rc script in package

15:45 <cruxbot> [opt.git/3.7]: libjpeg-turbo: update to 3.0.1

16:29 <cruxbot> [opt.git/3.7]: qemu-all: update to 8.1.2

16:29 <cruxbot> [opt.git/3.7]: qemu: update to 8.1.2

18:13 <farkuhar> The mariadb rc script business got me thinking ... what if we needed to push a fix for one of the service files that are exempted from upgrades? pkgadd.conf carves out exceptions for /etc/rc.d/{net,wlan} because users are supposed to customize them in-place. Was it ever considered to store custom network definitions elsewhere, and let the service files be subject to upgrades?

18:19 <farkuhar> Of course users are expected to run rejmerge, for anything intercepted by a pkgadd.conf UPGRADE NO directive. A distracted sysadmin might postpone that step for another day, though, leaving the old (broken?) file in place.

18:21 <ukky> farkuhar: imho, all service start-stop files should be subject for an update. Gentoo uses /etc/conf.d/{service-name} files, included in corresponding /etc/rc.d/{service-name} service file. We could use something similar to let users customize services.

18:23 <ukky> Though, I am not using SysV init system, I am facing similar problems, i.e. service customization.

18:23 <farkuhar> Not having any UPGRADE NO exceptions under /etc/rc.d would keep the logic of {start,stop,restart,status} separate from the options passed to the daemon. I like ukky's suggestion of a parallel directory /etc/conf.d that users can populate themselves, independent of ports and pkgutils.

18:29 <ukky> Second option, instead of /etc/conf.d/{service-name}, use /etc/rc.d/{service-name}.conf. And UPGRADE NO still could be applied only to ^/etc/rc\.d/[_-0-9a-zA-Z]\+\.conf$

23:13 <cruxbot> [compat-32.git/3.7]: harfbuzz-32: 8.2.1 -> 8.2.2

23:14 <cruxbot> [opt.git/3.7]: harfbuzz: 8.2.1 -> 8.2.2