01:22 Zoffix joined 01:57 ilbot3 joined
Geth MoarVM: 7942fdd1f1 | (Samantha McVey)++ | docs/ChangeLog
Update ChangeLog for release
 02:31
MoarVM: f00e975736 | (Samantha McVey)++ | docs/ChangeLog
Remove accidently included line in ChangeLog
 02:33
samcv jnthn: if you are around want to check out the tentative changelog?
02:47 travis-ci joined
travis-ci MoarVM build failed. Samantha McVey 'Update ChangeLog for release' 02:47
travis-ci.org/MoarVM/MoarVM/builds/370868976 github.com/MoarVM/MoarVM/compare/0...42fdd1f10f
02:47 travis-ci left 02:53 travis-ci joined
travis-ci MoarVM build failed. Samantha McVey 'Update ChangeLog for release' 02:53
travis-ci.org/MoarVM/MoarVM/builds/370868976 github.com/MoarVM/MoarVM/compare/0...42fdd1f10f
02:53 travis-ci left
samcv not sure why that test is failing on travis... 02:55
it's this one: lstat_time doesn't follow symlink
03:00 travis-ci joined
travis-ci MoarVM build failed. Samantha McVey 'Remove accidently included line in ChangeLog' 03:00
travis-ci.org/MoarVM/MoarVM/builds/370869338 github.com/MoarVM/MoarVM/compare/7...0e9757362a
03:00 travis-ci left
Zoffix samcv: FWIW, it used to be fudged until very recently: github.com/perl6/nqp/commit/6c982a...e176854c06 03:00
Perhaps, it need to be re-fudged
samcv hm
Geth MoarVM: 1e967475f4 | (Samantha McVey)++ | tools/release.sh
Make sure libatomic_ops submodule gets included in release
 05:40
MoarVM: 8a90168c72 | (Samantha McVey)++ | VERSION
Release 2018.04
MoarVM: 577857ae02 | (Samantha McVey)++ | docs/release_guide.md
Add instructions to the release guide about gpg signing

Give instructions about how to gpg sign the releases.		 05:45
05:54 domidumont joined
AlexDaniel` samcv++ 05:54
06:00 domidumont joined 06:31 robertle joined 07:36 domidumont1 joined
nwc10 good *, #moarvm 07:47
samcv good * nwc10 07:48
08:22 zakharyas joined 08:31 zakharyas joined 08:38 brrt joined 08:42 zakharyas joined
jnthn samcv++ 09:35
samcv: Was already sleeping when you asked, sorry :)
09:46 zakharyas joined 10:59 zakharyas joined
Zoffix samcv++ 11:03
11:17 brrt joined 11:54 Zoffix left 12:03 zakharyas joined 12:06 zakharyas joined 12:14 domidumont joined 12:26 ingy joined, avar joined 12:38 ZofBot joined 12:44 releasable6 joined 12:58 committable6 joined 13:04 zakharyas joined
brrt good * 13:18
13:23 quotable6 joined, benchable6 joined, evalable6 joined, coverable6 joined, greppable6 joined, bloatable6 joined
brrt .ask jnthn if you maybe have any experience with ml64 or soemsuch assemlber for microsoft 13:26
yoleaux brrt: I'll pass your message to jnthn.
14:34 Kaiepi joined 14:42 ggoebel joined 14:46 zakharyas joined 15:04 avar joined, ingy joined, moritz joined, leedo joined, [Coke] joined, japhb joined, jsimonet joined, mst joined, krunen joined, SmokeMachine joined, jnthn joined, hoelzro joined, nebuchadnezzar joined, mtj_ joined, sivoais joined, samcv joined, BinGOs joined, TimToady joined, jpf joined, camelia joined, geekosaur joined, dalek joined, Geth joined, synopsebot joined, SourceBaby joined, Util joined, p6lert joined, yoleaux joined, ilmari joined, Voldenet joined, dogbert2_ joined, gtodd joined, tbrowder_ joined, ilmari[m] joined, AlexDaniel` joined, wictory[m] joined, nativecallable6 joined, notable6 joined, unicodable6 joined, reportable6 joined, shareable6 joined, bisectable6 joined, huggable joined, buggable joined, nwc10 joined, timotimo joined, AlexDaniel joined, lizmat joined, dogbert17 joined, MasterDuke joined, brrt joined, ZofBot joined, committable6 joined, quotable6 joined, benchable6 joined, evalable6 joined, coverable6 joined, greppable6 joined, bloatable6 joined, Kaiepi joined, zakharyas joined, ggoebel joined, harrow joined, scovit joined, bartolin joined, btyler joined, nine joined, ChanServ joined 15:05 squashable6 joined 15:19 FROGGS joined
Geth MoarVM/master: 5 commits pushed by (Jonathan Worthington)++ 15:27
MoarVM: 641c7a1d13 | (Timo Paulssen)++ | 2 files
debugserver: add a "handle equivalence class" command

lets you find out if handles actually refer to the same object.		 15:42
MoarVM: cfe1378fcd | (Timo Paulssen)++ | src/debug/debugserver.c
bump minor protocol version for new messagetype
jnthn Some design notes on making spesh "teachable" to help us better deal with various optimization challenges: gist.github.com/jnthn/e51a06c6882f...0a3dd373e6 15:48
Thoughts welcome. I won't publish this anywhere, so don't bother reporting typos. 15:49
It's also possible that all of our method caching might go away in favor of this 15:51
(Eventually) 15:52
timotimo should we remove the "Better optimization around closures 16:00
" part from the roadmap on moarvm.org?
"Today's optimizer does a poor job of, and has an inability to inline, first class functions and closures. Recent work has started to collect the data needed to do better; optimizations to exploit it are still to come."
jnthn ooh, yes 16:01
we do that now :)
timotimo or maybe there's still more to be added in the future and we should just reword that?
16:01 releasable6 joined, evalable6 joined
timotimo whoops, the release page has a part of the document displayed wrong 16:01
the debug server section accidentally has all bullet points in one line
m: my $foo = try await Promise.broken("Oh no!"); say $foo 16:04
camelia (Any)
timotimo is this right?
m: my $foo; try $foo = await Promise.broken("Oh no!"); say $foo
camelia (Any)
timotimo m: my $foo; try $foo = await Promise.broken("Oh no!"); CATCH { .say }; say $foo
camelia (Any)
jnthn Yes
timotimo m: my $foo; $foo = await Promise.broken("Oh no!"); CATCH { .say }; say $foo
camelia An operation first awaited:
in block <unit> at <tmp> line 1

Died with the exception:
Oh no!
in block <unit> at <tmp> line 1

An operation first awaited:
in block <unit> at <tmp> line 1

Died with the exception:
…
jnthn m: my $foo; try $foo = await Promise.broken("Oh no!"); say $!
camelia An operation first awaited:
in block <unit> at <tmp> line 1

Died with the exception:
Oh no!
in block <unit> at <tmp> line 1
timotimo mhm 16:05
16:05 ilmari_ joined 16:08 statisfiable6 joined 16:10 ilmari joined, brrt joined 16:18 ilmari[m] joined
lizmat vorpus.org/blog/notes-on-structure...d-harmful/ # wonder how this applies to Perl 6 16:28
jnthn Looks at least somewhat related to `react` 16:35
Though I'd argue that the list form of await is more convenient for "set two tasks off at once and wait for both" :) 16:36
But in terms of the automated tracking of completion and automatic propagation of errors, that's very much what a react block is about 16:37
16:50 domidumont joined 16:56 wictory[m] joined, AlexDaniel` joined
Geth MoarVM/pluggable-spesh: 615d5f913a | (Jonathan Worthington)++ | 10 files
Stub the spesh plugin ops
 16:57
17:01 robertle joined
brrt i was also looking at that just now and owndering what was actually ned 17:34
new
17:36 chansen_ joined 17:48 coverable6 joined
Geth MoarVM: b87b03e234 | (Timo Paulssen)++ | src/debug/debugserver.c
throw out superfluous debugspam
 18:11
18:47 brrt joined
jnthn brrt: Talking of new, did you see the design notes I linked earlier on the spesh plugin mechanism? :) 18:53
yoleaux 18:33Z <Zoffix> jnthn: we need an "executive" decision: **nqp**'s numerics are parsed with `MVM_coerce_s_n`. It has precision issues, such as parsing 3.145e-20 into 3.14149999999999969e-20 instead of the closest representable num 3.14150000000000029e-20 and it also can't handle denormals (5e-324 parses as 0). This makes 2 sprintf nqp tests fail on 32bit boxes and it's currently proposed to fix that and make a point
18:33Z <Zoffix> jnthn: rest of that message: irclog.perlgeek.de/perl6-dev/2018-...i_16094868 (it's fairly long)
brrt i was looking at them just now 19:00
timotimo i haven't looked deeply into invokedynamic, but i feel reminded of it
brrt i am incidentally looking at the java vm specification recently 19:01
jnthn timotimo: Yes, they're in the same problem space 19:02
timotimo these handlers are almost traced, aren't they?
jnthn Not quite but kinda-ish :) 19:03
timotimo right, any ops that aren't on that list won't be "recorded" 19:04
jnthn Well, mostly all that one is doing there is setting up the guards that should apply in order for the eventual result to be valid
timotimo right, and what those guards apply to is very restricted 19:05
which keeps us from having to trace ops
jnthn *nod*
timotimo i like it in general :) 19:06
jnthn It'll give us some traction on various thorny problems, not least assignment
timotimo in the first example, i don't see how the resolve is referencing "privmeth" from the speshreg op 19:07
ah, the example's use of speshresolve differs from the declaration of speshresolve
nqp::speshresolve(str $name --> obj) 19:08
vs
nqp::speshresolve(nqp::decont(self), 'foo')(self, $bar)
jnthn oops, yes :)
timotimo .o( and the missing ` at the end of a few of those lines making code non-highlighted
jnthn I didn't quite figure out the exact interface yet
(At an NQP level, I mean)
But could just put the name as the first arg 19:09
And compile the rest as call args
timotimo that sounds like it'd work, yeah
19:13 squashable6 joined
brrt hmmm. 19:26
19:29 releasable6 joined
brrt any idea if we still look up the current inline position in the frame, or if we've removed that? 19:35
timotimo "inline position in the frame"? in what situation? 19:39
19:45 huggable joined, ZofBot joined 19:46 buggable joined
brrt which inlined frame(s) the current position of the JIT is in 19:49
anyway, i found it, and we do
is there a maximum number of inlines?
jnthn As in maximum depth? 19:50
No
brrt hmm. i kind of want to avoid malloc()-ing a buffer for getting that 19:52
i can of course avoid doing so...
the thing is, MVM_frame_find_contextual_by_name has to inspect the inline structure, and I want to start hiding that from core 19:53
my initial idea was, let's allocate a buffer and get the full list of inlines that we are 'in' 19:54
but i don't want to do that in a potentially hot path like find_contextual_by_name
timotimo time for alloca?
brrt there are other options :-)
i'm not sure we do alloca cross-platformy these days?
actually, we do use alloca 19:57
samcv has been using it, it seems
ok,
that might make it more reasonable
see, the other alternative is, just pass a start offset to search from, and iterate starting from there 20:01
int i; for (i = MVM_jit_code_get_active_inline(tc, code, frame, 0); i < code->num_inlines; i = MVM_Jit_code_get_active_inlines(tc, code, frame, i)) 20:02
yeah, i think i'll do that 20:06
jnthn: i'll look further into it, but it looks basically good to me 20:08
Geth MoarVM/master: 5 commits pushed by (Samantha McVey)++ 21:11
timotimo cool 21:19
lizmat timotimo: still no time to bump ?
jnthn samcv++ 21:20
Geth MoarVM: ffdd62d487 | (Timo Paulssen)++ | src/debug/debugserver.c
debugserver: allow unrecognized fields in messages

the versioning part of the protocol design says that moar shall accept any and all fields in the message map, even ones that it doesn't know. So we skip anything we see for keys that we don't know about.
MoarVM: 24a91a460b | (Timo Paulssen)++ | src/debug/debugserver.c
use official API for thread lock count
jnthn lizmat: Think samcv just said she'd do so
21:22 benchable6 joined
lizmat jnthn: ok 21:22
Geth MoarVM: 5728e1f399 | (Samantha McVey)++ | 4 files
Fix CJK Unified Ideograph and other derived Unicode names

Conform to Unicode spec Name Derivation Rule Prefix Strings, and ditch the angle brackets. <CJK Ideograph-XXXX> becomes
  "CJK Unified Ideograph-XXXX". Also effects the other non-named
ideographs like Tangut.		 22:10
22:21 Geth joined 22:22 Geth joined
samcv jnthn: i'm getting ASAN errors in t/nqp/106-unicodenames.t 23:17
caused by NFA
jnthn New ones? 23:22
If they're recent, they're perhaps caused by the recent change I did there
Please file an issue if so, and I'll take a look tomorrow
samcv ok
jnthn Well, file an issue anyway, though it's nice if we know if it's the recent change :)
Time for me to rest; 'night o/ 23:24
Geth MoarVM: samcv assigned to jnthn Issue ASAN errors due to NFA changes github.com/MoarVM/MoarVM/issues/846
e7d79d8691 | (Samantha McVey)++ | 3 files

The previous Unicode commit would remove brackets from almost all codepoint names. Instead make sure we only remove the ones we want. Also in the Unicode database use uppercase for the CJK and Tangut names.		 23:25

Powered by Cro and the Raku® Programming Language.

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

Thank you!