00:02 geekosaur joined 01:39 FROGGS_ joined 01:48 ilbot3 joined 06:37 domidumont joined 06:42 domidumont joined
FROGGS_ o/ 07:11
nwc10 \o 07:12
08:16 zakharyas joined 08:48 zakharyas joined
jnthn moarning o/ 08:59
FROGGS morning jnthn
jnthn grmbl...persistent cold is persistent
FROGGS jnthn: ne too now
me*
jnthn Feels like nearly everyone I talk to has it :P 09:00
FROGGS /o\
jnthn Guess it's the time of year
lizmat only has a light sniff 09:03
jnthn Mine has never got overly bad 09:06
It just won't go away
dalek arVM: 78fade8 | jnthn++ | src/6model/serialization.c:
Implement serialization of SCRef.
09:41
jnthn That clears up the immediate error when pre-comping OO::Monitors 09:43
Unfortuantely, it just causes it to fail with a different error
10:08 zakharyas joined
dalek arVM: df0cffc | jnthn++ | src/6model/reprs/MVMCode. (2 files):
Split up resolution of routine file/line.

Into a version that gets a hash for high-level use, and one with out parameters for C-level use.
10:10
arVM: 36e381e | jnthn++ | src/6model/serialization.c:
Include file/line of unserializable closure.

To aid with debugging such problems.
jnthn Odd... Serialization Error: missing static code ref for closure '' (src/Perl6/World.nqp:2180) 10:13
Which is the compiler thunk that's meant to be replaced by runtime 10:14
Uh, by serialization time even
timotimo hum, some reference not being marked and a gc moves it? :\ 10:18
probably a long shot
jnthn Unlikely
Doubt it's VM-level
timotimo mhm 10:20
gist.github.com/timo/a88c8882889de...95996734be 10:26
we may be able to get our minimum memory use for perl6 -e '' down by a lot if we make the first pages in each fsa smaller 10:27
like, we would do a quarter, a quarter, half, normal size 10:29
also, i wonder if we should add one or two additional bits to the size mak above 256 bytes? 10:30
huh, /me apparently totally wrong about the savings from making the first page smaller 10:33
(to be fair, this also includes redzones, so we've got an additional 8 bytes per element in every size class 10:47
jnthn lunch time...still debugging the OO::Monitors precomp fail, though got it golfed a good way 10:53
timotimo good news 10:55
(what you just said, i mean)
dalek arVM/p6opaque_use_fsa: 5a49333 | timotimo++ | src/6model/reprs/P6opaque.c:
use FSA in p6opaque's repr_data

also store a bunch of things more compactly
11:11
timotimo ^- with this i regularly get below 62000 maxresidentk, whereas before i always got above 62000 maxresidentk
dalek arVM/fsa_cleanup_stats: bce5db1 | timotimo++ | src/moar.c:
give overview of FSA usage at begin&end of --full-cleanup
11:13
timotimo hm, i think the latter hangs off of the former branch, but that's not necessary
it's also interesting to put a printf at the beginning of deserialize_repr_data that outputs the debug_name of the thing we have 11:14
i'm really not fond of the whole P6OMAX(something, 1) stuff we have there 11:27
12:09 zakharyas joined
jnthn back 12:12
dalek arVM: 91fe744 | MasterDuke17++ | src/6model/reprs/P6opaque.c:
Add debug info when failing to bind an attribute
14:08
arVM: eb46ed2 | jnthn++ | src/6model/reprs/P6opaque.c:
Merge pull request #435 from MasterDuke17/more_debug_info_for_bind_attribute()

Add debug info when failing to bind an attribute
18:38 vendethiel joined 20:04 leego joined 20:44 vendethiel joined 20:51 leego joined 21:26 travis-ci joined
travis-ci MoarVM build passed. Jonathan Worthington 'Include file/line of unserializable closure. 21:26
travis-ci.org/MoarVM/MoarVM/builds/170728134 github.com/MoarVM/MoarVM/compare/7...e381e2e4cf
21:26 travis-ci left 21:42 ggoebel joined
timotimo something's wrong with travis-ci 22:50
the mac jobs are all not starting 22:51
23:09 vendethiel joined