infinisil changed the topic of #nix-lang to: Channel for discussing Nix as a language - https://nixos.org/nix/manual/#chap-writing-nix-expressions - Logs: https://logs.nix.samueldr.com/nix-lang/
ddellacosta has quit [Ping timeout: 265 seconds]
ddellacosta has joined #nix-lang
ddellacosta has quit [Ping timeout: 260 seconds]
__monty__ has joined #nix-lang
Synthetica has joined #nix-lang
qyliss has quit [Quit: bye]
qyliss has joined #nix-lang
qyliss has quit [Quit: bye]
qyliss has joined #nix-lang
qyliss has quit [Remote host closed the connection]
qyliss has joined #nix-lang
qyliss has quit [Quit: bye]
qyliss has joined #nix-lang
kqb has joined #nix-lang
kqb has quit [Client Quit]
<puck> woops found a way to get double INF in Nix, and it doesn't roundtrip, obviously
<puck> nix-repl> let b = builtins.fromJSON "1e-323"; in builtins.fromJSON (builtins.toJSON (1.0 / b))
<puck> yes that is dividing by approximately zero
<puck> also that means i can make a value where you can't use builtins.lessThan to determine its exact location
<puck> since nan < N is false, but nan > N also is false
<puck> > let inf = 1.0 / (builtins.fromJSON ("0.1E-306")) * 100; nan = inf / inf; in [ (nan > nan) (nan < nan) (nan == nan) ]
<{^_^}> [ <CODE> <CODE> <CODE> ]
<puck> > let inf = 1.0 / (builtins.fromJSON ("0.1E-306")) * 100; nan = inf / inf; v = [ (nan > nan) (nan < nan) (nan == nan) ]; in builtins.deepSeq v v
<{^_^}> [ false false false ]
<puck> betterl
__monty__ has quit [Quit: leaving]