freenode/#mezzano - IRC Chatlog
Search
5:15:03
fitzsim
it seems like nibbles:ub32ref/le is getting passed too many arguments (5 instead of 2)
5:19:52
fitzsim
the uses of load.i32 are in longjmp and nanosleep, both of which fail to compile for me
5:20:22
fitzsim
if I replace in load.i32 (the octet-vector (llvm-context-memory llvm-context)) with just e.g., (make-array 100), the both compile
11:11:51
froggey
fitzsim: I think you're running into https://github.com/sharplispers/nibbles/issues/5
13:16:35
fitzsim
froggey: I downgraded to nibbles 0.13 in ~/quicklisp/load-projects and Iota loads perfectly
13:18:35
fitzsim
I am running SBCL 1.3.11-1.el7, so not exactly the same version as in that bug report
13:27:47
fitzsim
I'm trying it out with a configure script, and it's failing immediately because the "C compiler can't create executables"
13:29:47
fitzsim
but would it be possible to have Iota generate an a.out that runs with ./a.out (via an sbcl wrapper script or something)?
13:32:21
fitzsim
I guess it's not really an Iota linker, so much as tying in the iota, clang, and llvm-link steps into one command
13:33:18
froggey
I'll write one eventually. for doom & quake I can get away with hacking the configure scripts & makefiles
13:34:49
froggey
a linker would have to deal with archives, library search paths, that kind of thing. kind of tedious
14:03:06
fitzsim
read through them; the hacks are pretty minimal given the result is running Doom and Quake on Mezzano!
14:08:14
froggey
yeah, most of the work was done by emscripten & google's pnacl. that's where all the complicated lowering passes came from