brrt for my own record-keeping; after tiler-linearisation i removed the special-purpose logic for the 'magic' registers, which trips up the register allocator 09:52
might as well move it to offline at this stage
timotimo heyo brrt :) 10:19
brrt hi timotimo
i wonder if the JIT is just especially unforgiving to to quick hacks and cheating, or whether i'm simply doing something wrong 10:20
timotimo i can verily imagine the first be true 10:22
brrt hmm
maybe it is interesting to consider a more specific jit testing system 10:23
moritz it's a bit like working with the GC; you cheat, it cheats you 10:24
brrt right :-0 10:25
that is an excellent comparison