<gtklocker>
only the second option works but only for a single invocation of darwin-rebuild switch, then permissions on /nix/store get reset to root:nixbld
<gtklocker>
any idea what may be going on?
<LnL>
what would make that check more clear?
<LnL>
you can't just transition between a single <> multi user install without fixing permissions of the store
<LnL>
so unless you disable services.nix-daemon you shouldn't use that second case since that's exactly what the check is for
<gtklocker>
sure I do want to fix permissions
<gtklocker>
it's just that the first suggestion does not work
<gtklocker>
and I do use the daemon
<gtklocker>
so if I take the first suggestion and then rerun darwin-rebuild switch I get
<gtklocker>
ok so permissions for the parent is what differs for me
<gtklocker>
but it probably shouldn't be a problem if nix daemon has root
<LnL>
yeah, but that's not what is broken
<gtklocker>
so wait if I unset NIX_REMOTE you would still expect this to pick up on the daemon automatically?
<LnL>
it's the client that tries to figure out if it should operate on the store directly or talk to the daemon because it doesn't have enough permissions
<gtklocker>
and not try to do a local lock?
<gtklocker>
I see
<LnL>
you can specify explicitly what it should do (with NIX_REMOTE or --store) but by default it should just "figure it out"
<LnL>
I added that check to help but seems like that doesn't entirely match the condition that nix uses itself
<gtklocker>
yeah
<gtklocker>
btw for the whole of /nix would root:wheel be typical?
<LnL>
I thought it was the permissions of /nix/var/nix/db/db.sqlite specifically
<LnL>
yeah, but there are a few exceptions
<LnL>
like /nix/var/nix/profiles/per-user /nix/var/nix/gcroots/per-user
<gtklocker>
I see
philr has quit [Ping timeout: 258 seconds]
Chiliparrot has joined #nix-darwin
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]