libera/#sbcl - IRC Chatlog
Search
7:19:45
resttime
Not sure if it's a bug or my machine but I tried to build the latest SBCL from the repo just now with `sh make.sh --fancy` and got an error during the build on the line https://sourceforge.net/p/sbcl/sbcl/ci/master/tree/tools-for-build/grovel-headers.c#l228 that `O_LARGEFILE` is an undefined constant used forthe first time. I fixed this by adding -D_GNU_SOURCE to the CFLAG's in the
13:06:32
nij-
This line calls +all-spaces+. https://github.com/sbcl/sbcl/blob/0492f5bc2c807448d9d6cbd5d22af37bb9fcfae7/src/code/room.lisp#L1057 But when I evaluate +all-space+ in the package "SB-VM" manually, it says it's unbound. What's going on?
13:11:36
mfiano
Uhh I don't know if you should be `in-package`-ing around SBCL internals. That might give surprising results.
13:12:23
scymtym
+ALL-SPACES+ is constant-ish so the system may have inlined the value at all uses and removed the definition before dumping the final image
13:14:11
scymtym
mfiano: if you two ran the same SBCL version, sure. but this kind of thing changes from time to time
13:14:25
scymtym
either way, if that wasn't clear. internals like this are not meant to be used by user code
15:35:52
phoe
oh wait, interning into it doesn't cause sbcl to complain either; I must be remembering wrong
15:36:42
|3b|
yeah, i guess :: might hit a package lock where in-package wouldn't, but interning still wouldn't do much
15:38:53
|3b|
ACTION suspects anything wrong about (in-package :sb-whatever) is true of in-package in general for packages you don't control
15:39:52
|3b|
but you can if you want to, and either know the problems or want to discover them for yourself :)
15:40:32
|3b|
well, if you don't bind *package* locally, you could run a thread to cycle it through all of them, which would be sort of like all at once?
15:40:45
phoe
(let ((*package* t)) (intern "FOO")) ;; interns FOO into all packages at the same time