| IRC logs at
Set by AlexDaniel on 12 June 2018.
00:11 pamplemousse_ left 00:34 MasterDuke joined, MasterDuke left, MasterDuke joined 00:52 TimToady left, TimToady joined 01:54 dogbert17 joined 03:54 MasterDuke left 06:34 domidumont joined 07:55 zakharyas joined 08:04 domidumont left 08:26 domidumont joined 08:34 domidumont left 08:52 AlexDaniel joined 09:26 brrt joined 09:35 brrt` joined 09:37 brrt left 10:45 brrt` left 11:56 zakharyas left 12:32 MasterDuke joined 12:36 MasterDuke left 13:17 brrt` joined 13:48 zakharyas joined 14:32 brrt` left 15:06 brrt` joined 15:26 sena_kun joined 15:34 robertle joined
Geth MoarVM/more-pea: e1b9faee7d | (Jonathan Worthington)++ | src/spesh/pea.c
Fix a silly thinko
15:56 brrt` left
Geth MoarVM/more-pea: 5613532104 | (Jonathan Worthington)++ | src/spesh/pea.c
Add missing REPR check

Otherwise we SEGV if it's not a P6opaque.
MoarVM/more-pea: 273cdfe784 | (Jonathan Worthington)++ | src/jit/x64/emit.dasc
Fix JIT of sp_neg_bi and sp_abs_bi
16:31 brrt` joined
japhb Oooh, looks like a MoarVM coding day for jnthn++ 16:35
japhb is really looking forward to seeing that branch land, because I'm very curious how it will affect the performance of Terminal::Print animations 16:36
jnthn Well, not so much coding day as coding late-afternoon. Picking off some of the smaller issues. Do need a day to pick off the larger ones. 16:46
Geth MoarVM/more-pea: b57515e7c1 | (Jonathan Worthington)++ | src/spesh/pea.c
Correct wrong union access
jnthn These are all very silly 17:06
Seems we now pass all S02 with MVM_SPESH_BLOCKING on in that branch, though 17:07
Current spectest run is looking a lot better, though 17:12
Curious. A good number of the failures involve threading tests
17:13 pamplemousse joined 17:15 lucasb joined
jnthn ah, I'm missing a probably important fix from MoarVM master... :S 17:17
Geth MoarVM/more-pea: 47 commits pushed by (Jonathan Worthington)++
17:19 brrt` is now known as brrt
jnthn Yeah, less fails with that rebase and HEAD of nqp/rakudo 17:24
Though still some. More work ahead. :) 17:25
And this is all without the DU chain checker on, which will carp aplenty and also needs to be satisfied.
nwc10 indeed it does :-/ 17:26
japhb Still, this sounds like quite a bit of progress. :-) 17:37
nwc10 enjoy food/beer/evening off? 17:38
jnthn Well, certainly going to go and enjoy food now. :) 17:40
I'm guessing the bunch of concurrency ones that fail are going to have a common root cause 17:41
nwc10 with the DU chain checker turned off NQP builds with ASAN 17:42
jnthn Thankfully thre's some non-concurrent ones to look at first, that might be easier targets.
nwc10: Yes, it's one of those cases where the problem is close to harmless, but we really should do the bookkeeping right
But it's not quite trivial to do, alas 17:43
(Needs renames and PHI insertions)
nwc10 and passes make test
jnthn with NODELAY, ooc?
nwc10 er, yes.
+#define FSA_SIZE_DEBUG 1
jnthn Only dared try BLOCKING so far :)
nwc10 MVM_SPESH_BLOCKING=1 17:44
jnthn Alright, off to nom some saussage and potato salad
nwc10 enjoy
japhb Mmmm, now I'm hungry again
jnthn Made with vinegar rather than majonaise, which iiuc meets the approval of Austria and some, but not all, parts of Germany? :) 17:45
I did listen in on an argument about this important matter once, and believe there's some regional disagreement. :)
o/ 17:47
17:48 zakharyas left 18:04 brrt left 18:47 brrt joined 18:53 domidumont joined 18:54 domidumont left
brrt jnthn: belgian mayonaise is, I think, made with either vinegar (or lemon juice) 19:09
19:09 brrt left 19:10 brrt joined
brrt jnthn: I believe belgian-style mayonaise can be made with vinegar, solving the issue? 19:10
19:11 pamplemousse left 19:21 pamplemousse joined
brrt \o pamplemousse 19:24
pamplemousse \o
brrt how are you doing? 19:26
pamplemousse Doing alright. Making progress on the linker. I've been working on a blog post which I'm planning on posting tonight or tomorrow about the progress I've made so far. 19:28
japhb ++pamplemousse # Impending progress post :-) 19:30
brrt yes, I'll be really interested 19:32
as well
pamplemousse I'll post a link as soon as it's finished! 19:33
brrt cool
Any problems you've been having? 19:34
pamplemousse No big problems so far, just keep running into small issues with alignment and making sure I'm putting the right things in the right places in the ELF file. 19:37
brrt are you using the elf.h header in linux? 19:38
pamplemousse Yes 19:39
The alignment issues were when I was reading things in from the MBC file but I found my mistake 19:41
nine jnthn: yes, vinegar is obviously the correct choice. I don't know what's there to argue about ;) 19:55
brrt I see... We can probably make that conditionally dependent to compile 20:27
jnthn brrt: Wow, I never heard of mayo made from vinegar. :) 20:29
sena_kun .oO ( we can make mayo ingredients conditionally dependent on chief's opinion at food's compiling time ) 20:30
japhb sena_kun: If making mayo is compiling, is making the potato salad linking? 20:43
sena_kun japhb, I am afraid to think what eating means in this metaphor! (inb4: probably just using the software) 20:45
20:56 AlexDaniel left 20:57 AlexDaniel joined
brrt belgium is like that 21:41
pamplemousse: be sure to broadcast your blogpost here when it's done :-) 21:43
pamplemousse Will do! 21:44
brrt thank you :-) 21:56
22:15 Voldenet left 22:19 brrt left 22:26 Voldenet joined, Voldenet left, Voldenet joined 22:31 Voldenet left 22:35 Voldenet joined, Voldenet left, Voldenet joined 22:56 Voldenet left 22:58 ggoebel left
Geth MoarVM: 9ed86e4c95 | (Jonathan Worthington)++ | src/6model/6model.h
Eliminate an unused collectable flag
MoarVM: 7601c9b3d4 | (Jonathan Worthington)++ | src/6model/6model.h
Update comment to match current reality
23:02 Voldenet joined, Voldenet left, Voldenet joined 23:04 pamplemousse left
Geth MoarVM: 09fa0d2839 | (Jonathan Worthington)++ | src/core/frame.c
Remove out-dated frame refcount mention
23:17 sena_kun left
Geth MoarVM: 1c279a231b | (Jonathan Worthington)++ | src/core/frame.c
Don't needlessly install VMNulls into ->work

A while back, we put analysis in spesh to make sure that we never read any object registers before writing them, and in the case that happens to explicitly `null` them at frame entry, so we don't need to poke null values in at frame creation time. However, we were still running the loop to install the VMNulls! Don't do that. Should cheap entry to all spesh'd/JIT-compiled frames.
jnthn sleep o/
23:56 pamplemousse joined