diakopter hey, why can't an integer literal with pre- or post-decrement be invisbily transformed to (BEGIN my$=6)-- , as in: say 44 while 6-- 01:03
m: say 44 while (BEGIN my$=6)-- 01:04
camelia rakudo-moar ac0dcd: OUTPUT«44␤44␤44␤44␤44␤44␤»
lizmat diakopter: because phasers that run only once (like BEGIN and INIT) freeze their result 04:26
m: say now - INIT now
camelia rakudo-moar ac0dcd: OUTPUT«0.0015150␤»
lizmat m: say now - BEGIN now
camelia rakudo-moar ac0dcd: OUTPUT«0.0097811␤»
brrt good * #moarvm 08:28
ok, i've figured out a number of ways why the current iteration of inserting-spills doesn't work
a): the spilled values (which refer to locals) displace the register value descriptors 08:31
so we point the register to the local address, which is obviously out of range for the register indexes 08:34
b): we don't actually insert the 'load' code yet
then there is more stuff:
- the creation of a value descriptor, and assigning its (virtual) live range, are currently a single step, and they shouldn't be 08:35
jnthn morning o/ 08:36
brrt good moarning jnthn
lizmat looks interesting, have only scanned it: kcsrk.info/ocaml/multicore/2016/06/11/lock-free/ 10:38
nwc10 jnthn: ASAN tolerates your branch 11:23
brrt which branch is that 11:49
nwc10 origin/return-without-lexotic 11:51
in nqp and rakudo
brrt oh, that one, that is awesome work
FROGGS[mobile] o/ 16:40

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!