03:45 FROGGS_ joined 03:51 vendethiel joined 06:19 vendethiel joined 06:24 lizmat joined 06:48 FROGGS[mobile] joined 06:55 lizmat_ joined 07:56 lizmat joined 09:06 lizmat joined, vendethiel joined 09:15 rurban joined 09:26 FROGGS[mobile] joined 10:59 lizmat joined 11:00 vendethiel joined 12:10 lizmat joined 14:04 agentzh_ joined 15:07 FROGGS[mobile] joined 15:24 tadzik joined 15:25 lizmat joined 15:31 agentzh_ joined 15:55 agentzh_ joined 16:37 zakharyas joined 17:41 zakharyas joined
FROGGS I think I found a codegen bug :o( 19:00
TimToady \o/ 19:01
FROGGS a 'bitand_i' op that I emit in the charrange method in the regex compiler ends up being a no_op
that is, I see that no_op in the --dump 19:02
AHHHHHHHHHHHHHH 19:13
TimToady \o/?
FROGGS m: say "a" ~~ /:i <[b..c]>/
camelia rakudo-moar a040b1: OUTPUT«Nil␤»
FROGGS m: say "d" ~~ /:i <[b..c]>/
camelia rakudo-moar a040b1: OUTPUT«「d」␤»
FROGGS -.-
19:14 zakharyas joined
FROGGS it was not a code gen bug, it was a bug in the person wo wrote the code that I just copied 19:14
vendethiel lol 19:15
well :-)
FROGGS is not smart when he copies old bugs from his self
github.com/perl6/nqp/blob/master/s...T.nqp#L541
this op should read 'band_i'
19:16 agentzh_ joined
FROGGS so beware, typos there are just translated to no_op 19:16
nwc10 is there a reason why it's that way? 19:26
FROGGS no idea
nwc10 It seems LTA to silenty carry on, but I'm wondering if there's a particular bootstrapping (or something) use case that the behaviour makes easier
FROGGS hmmm, I am unable to find the place where we translate op names to their index... 19:32
dalek arVM/ignoremark: d0f139b | FROGGS++ | src/6model/reprs/NFA.c:
remove debug output
19:33
arVM/ignoremark: 99435bf | FROGGS++ | src/6model/reprs/NFA.c:
fix leftover from refactoring ignoremark in NFA
19:39
arVM: 4c35019 | FROGGS++ | / (10 files):
implement ordbaseat and eqatim_s for ignoremark
20:38
arVM: c985680 | FROGGS++ | src/core/interp.c:
chars are not equal beyond end of string
arVM: f63f1c5 | FROGGS++ | src/6model/reprs/NFA.c:
remove debug output
arVM: 65c89e7 | FROGGS++ | src/6model/reprs/NFA.c:
fix leftover from refactoring ignoremark in NFA
FROGGS is that new? 20:46
src/jit/graph.c: In function ‘jgb_consume_reprop’:
src/jit/graph.c:767:21: warning: pointer type mismatch in conditional expression [enabled by default]
: ((MVMObject*)type_facts->type)->st->REPR->pos_funcs.at_pos;
timotimo FROGGS: yes, it's due to the jit_devirtualize_reprop branch merge; jnthn fixed it to not bust the msvc build any more, but that causes this warning 20:48
FROGGS ahh 20:49
20:58 colomon joined 21:14 cygx joined
cygx o/ 21:14
FROGGS: you get rid of the warning, you need to add a cast (void *) after both ? and : 21:15
FROGGS cygx: that's what I think too
cygx FROGGS: it's actually mandatory as function pointers don't auto-convert to void* 21:16
I suspect that goes back to segmented memory modes where function pointers could be FAR 21:17
FROGGS turns on his Windows machine Justin Case 21:18
timotimo 224637 FROGGS │ src/jit/graph.c:767:21: warning: pointer type mismatch in conditional expression [enabled by 21:21
sorry 21:22
cat on mouse pad
FROGGS that how it is meant to be :o) 21:23
dalek arVM: fe125e5 | FROGGS++ | src/jit/graph.c:
silence warning about pointer type mismatch
21:29
22:01 cygx left 22:18 jepeway joined 22:46 lizmat joined 23:15 danaj joined 23:45 agentzh_ joined