| IRC logs at
Set by AlexDaniel on 12 June 2018.
00:09 sena_kun left 01:40 nebuchadnezzar left 06:00 domidumont joined 06:01 nebuchadnezzar joined 06:03 vrurg_ joined 06:05 vrurg left 07:34 MasterDuke joined
MasterDuke timotimo: i got distracted by quake champions (at which i'm so very out of practice). but i might have some time today to look at the zen vs whatever thing today 07:37
re the lost optimization, not sure i know enough to quickly figure out what parts can be re-implemented, maybe can look at that after zen vs whatever 07:40
08:09 sena_kun joined 08:13 brrt joined 08:41 domidumont left 09:06 MasterDuke left 09:25 MasterDuke joined 10:09 brrt left 11:01 MasterDuke left 11:28 MasterDuke joined 11:45 Altai-man joined 11:47 sena_kun left
timotimo that's okay, i was suggesting to un-implement it for the version that was still faster to see how strong the impact is compared to the whole "new hash" merge 12:03
Geth MoarVM/hash-start: 4 commits pushed by (Nicholas Clark)++ 13:00
13:01 Kaeipi left 13:05 Kaiepi joined
Geth MoarVM: nwc10++ created pull request #1342:
Hash start
nwc10 I have to make a PR to get the CI tests?
jnthn I think so 13:21
nwc10 OK. I was assuming that I was going to "merge" that (fast-forward) if the CI is happy 13:22
have not tested this, but it should not conflict with new-disp 13:23
hey, I can check that...
[merge clean, MoarVM, nqp building...] 13:24
13:26 domidumont joined 13:32 Geth left 13:33 Geth joined 13:44 brrt joined 13:56 lucasb joined 14:31 brrt left
Geth MoarVM/new-disp: 8c6fee8cb9 | (Jonathan Worthington)++ | 3 files
Stub in dispatcher set/get state syscalls

Which will be part of the resumed dispatch mechanism.
MoarVM/new-disp: e1cdd54642 | (Jonathan Worthington)++ | 4 files
Store captures saved as dispatch state

And note that usage it the recording dump output.
[Coke] wonders why he thought disp was "display" for a few seconds. 15:15
timotimo moarvm gets support for displaying performance metrics on an LCD you plug in 15:17
you know how back in the dos days there was like a common "secondary monochrome monitor" thing that some programs supported for debug info?
i can't find it again 15:21
Another use for the MDA was as a secondary display for debugging. Applications like SoftICE[8] and the Windows debugger[9] permitted the simultaneous use of an MDA and another graphics card, with the MDA displaying a debugger interface while the other card was showing the primary display. 15:26
i actually used a super cheap LCD that was meant to be a digital picture frame that has custom firmware for it to have a status display that works even when my GPU is b0rked or whatever 15:27
but i can't find it any more since the move
[Coke] ... thank you for such a compelling backstory to support my dumb brain :) 15:31
15:46 sena_kun joined 15:47 Altai-man left 15:48 squashable6 left 15:49 squashable6 joined
Geth MoarVM/master: 4 commits pushed by (Nicholas Clark)++ 16:04
16:33 brrt joined
brrt \o 16:56
nwc10 o/ 16:57
17:02 zakharyas joined 17:16 MasterDuke left
Geth MoarVM/expr-jit-invoke: 78afc7c934 | (Bart Wiegmans)++ | 3 files
[JIT] Implement invoke sequence in expr JIT
MoarVM/expr-jit-invoke: 62a3a6ba50 | (Bart Wiegmans)++ | src/jit/graph.c
[JIT] Eliminate special consume_invoke opcode

Except for nativecall_invoke, we no longer need this part of the legacy JIT.
MoarVM/expr-jit-invoke: 7f183cf041 | (Bart Wiegmans)++ | 7 files
[JIT] WIP on nativeinvoke in the expression JIT
brrt that had bitrotted quite a bit
[Coke] brrt++ 17:38
brrt I think the thing that I encountered there was that the rakudo compiler did (imho) a weird and tightly-bound thing that I didn't want it to do, but I also didn't know how to fix it 17:41
timotimo you mean how it compiles stuff for nativecalled subs and methods? 18:02
Geth MoarVM/expr-jit-devirtualize: 5 commits pushed by (Bart Wiegmans)++ 18:09
brrt yeah, that's what I mean 18:10
18:19 brrt left 18:39 brrt joined
Geth MoarVM/jit-expr-optimizer: 4 commits pushed by (Bart Wiegmans)++ 18:46
brrt mind that I don't actually think the expr optimizer in it's current form will work 18:49
nwc10 brrt: Unhandled exception: Unsupported serialization format version 23 (current version is 21) 18:50
can I test this someone?
18:56 domidumont left
brrt nwc10: the optimizer code? 18:56
18:56 domidumont joined
brrt I suppose you can but I recall that it was broken 18:56
on account of it not being able to handle control flow
expr IR is too dumb 18:57
18:57 domidumont left, domidumont joined
brrt not unheard-of-dumb, mind you, but too dumb for what I'm aiming to do 18:57
18:57 domidumont left
nwc10 jit-expr-optimizer 18:57
I'm distracted by 18:59
Kaiepi i picked out udns to handle dns queries for v6.e's dns resolution stuff a while, but it's proving not to be an ideal library to use either 19:07
portability's my main concern, since it won't build on windows unless the master branch is used, but hasn't been contributed to since 2017...
no one single library i could find for this seems to work for how i envisioned this would work 19:08
ldns comes very close, but isn't supported on windows... which has its own dns query api that can be used, and has what appears to be similar enough configuration for it to work comparably with ldns 19:09
so i'm thinking of doing something similar to what nativecall does, where ldns would be the default library to use in most cases, falling back to windns 19:11
19:15 MasterDuke joined
MasterDuke nwc10: virtually? i thought i got emails about, but haven't in a while 19:16
brrt nwc10: i haven't actually tested it myself
nwc10 MasterDuke: yes, jitsi 19:17
MasterDuke hm. i've been meaning to dig up a mic for this desktop for a while...
brrt it segfaults pretty quickly on nqp :-D 19:19
nwc10 oops
I'm on the jitsi app on a tablet
19:45 Altai-man joined 19:47 sena_kun left 21:17 zakharyas left 21:42 brrt left 22:57 leont left 23:20 Altai-man left 23:21 MasterDuke left 23:40 Kaiepi left 23:47 Kaiepi joined