Sam James posted on Wed, 04 Mar 2026 02:42:09 +0000 as excerpted: > One of the arts of working on Gentoo is balancing many different > concerns and becoming used to people doing unexpected things, and trying > to have both reasonable defaults while allowing people to do their own > thing.
Just to say, as one such user doing unexpected things (reverse usr-merge with /usr -> . anyone? of course the not recommended USE="-* ..." is old news here, as is -* in /etc/portage/profile/packages to kill the entire @system set (with the full -package list to negate @system before portage began supporting -*, my comment in the file from the -* simplification update is dated 2017)). Even where not explicitly supported and in fact directly tested against in global profile scope, like an otherwise systemd profile on reverse usr- merge, Gentoo on "user did the unexpected" systems is /shockingly/ easy to maintain locally by a reasonably experienced gentooer, with very few patches (one to fix that profile test, a couple others to various packages or their ebuilds where upstream doesn't do the right thing if /usr is a symlink) actually needed and once they're setup it "just works". And what's nice, as long as I've been upfront with my "unexpecteds", my bug reports haven't been arbitrarily closed or otherwise discriminated against just because I'm "doing something that's not technically supported", where in fact that has nothing at all to do with the bug I actually reported. That wouldn't be the case with all distros, and gentoo deserves credit for it. Then there's the gentoo dev, floppym as it happens, that recently helped me spot a 20+ year latent local config bug that ultimately traced to the way I migrated users from Mandrake when I became a Gentooer back in 2004! I had /no/ clue, why should I after 20 years of it working without issue, and the available documentation (beyond the systemd code itself and the git log of the commit introducing it) simply didn't mention that as a trigger case and I was /all/ up the wrong tree following some case the docs did mention but that it turned out was only one case of the two, the other undocumented, but it all clicked into place when he mentioned the human user vs. system user config in /etc/login.defs and wondered if my human user UID might somehow be configured as a system user, and I recalled worrying about effects some 20 years earlier in that migration, effects that didn't actually trigger for 20 years! You sometimes read stories about old-timer Linux admins "doing the impossible" to rescue systems because it turns out it's not impossible after all if you don't reboot and lose existing running executables still functional after glibc goes poof in an upgrade gone bad, things like that. >From where I sit floppym just ascended to the same near-$deity status in the parallel distro-dev space when he made that connection, while surely having literally no idea I had that sort of migration history... 20 years ago! Legend indeed! A very personal and public THANKS MAN! as he sure deserves it, but despite the legend, we know distros the size of Gentoo aren't all one person, and there's thanks to go around for all the other devs (and other testers and bug reporters and wiki and forums contributors too). -- Duncan - List replies preferred. No HTML msgs. "Every nonfree program has a lord, a master -- and if you use the program, he is your master." Richard Stallman
