| IRC logs at
Set by AlexDaniel on 12 June 2018.
buggable 🎺🎺🎺 It's time for the monthly Accidental /win Lottery 😍😍😍 We have 2 ballots submitted by 2 users! DRUM ROLL PLEASE!... 00:00
And the winning number is 10! Congratulations to samcv, nine! You win a roll of duck tape!
00:06 stmuk joined 00:08 stmuk_ left 03:33 p6lert left, dalek joined, Geth joined, p6lert joined, synopsebot joined, SourceBaby left 03:34 SourceBaby joined 05:22 Kaiepi left 05:23 Kaiepi joined 09:36 robertle joined 09:59 lizmat left 10:29 Ven`` joined 10:33 Ven`` left 10:44 lizmat joined 13:48 brrt joined
brrt .tell lizmat gc documentation here: 13:48
yoleaux brrt: I'll pass your message to lizmat.
lizmat brrt: thanks! 13:49
yoleaux 13:48Z <brrt> lizmat: gc documentation here:
brrt :-) 14:16
14:27 lizmat left 14:50 zakharyas joined 15:15 lizmat joined
lizmat brrt: opinions about ? 15:44
brrt i was actually looking through that 15:50
i wanted to give that a proper review, but it looks alright
lizmat .ask jnthn is still up to date, or could it use an update / refinement ? 15:51
yoleaux lizmat: I'll pass your message to jnthn.
lizmat brrt: so you're ok with merging ?
brrt ehm. well, if it passes the test suite, then my hand-review is not likely to contribute much 15:54
but if it's all the same to you please give me 15 more minutes :-)
lizmat yeah, I can wait 15 minutes
16:00 zakharyas left 16:01 zakharyas joined
brrt so I learned is that we actually know a lot about the templates that we're not currently exploiting 16:06
lizmat is intrigued
brrt ok, for instance, we're computing the operand types and sizes at runtime 16:07
we don't actually need to do that
and we're inserting casts
also at runtime
and that as well can be eliminated, since we know (per oplist) the proper size of all things going in 16:08
and types
lizmat that sounds like an interesting optimization opportunity ? 16:14
brrt well, the result of it will probably be very minor since it is compiler internal 16:18
what it does do however is make it much easier to assert that a given template actually is correct 16:19
that is kind of a valuable thing, I think
lizmat indeed :-) 16:20
brrt I reviewed and left comments on a bunch of *very* minor things 16:50
16:56 brrt left 17:06 zakharyas left 17:16 zakharyas joined 17:36 domidumont joined 17:41 robertle left 17:43 domidumont left 17:44 domidumont joined 17:45 domidumont left 17:56 robertle joined 18:25 zakharyas left 18:38 brrt joined 18:41 Kaiepi left 19:21 domidumont joined 19:28 domidumont left 20:38 ggoebel left 20:39 brrt left 21:07 robertle left 21:59 MasterDuke joined
Geth MoarVM: 9f005086e3 | (Samantha McVey)++ | 11 files
Used fixed size allocator for hashes. Nice speedup for table expansion
MoarVM: 6646038f59 | (Samantha McVey)++ | src/strings/ops.c
Factor out our hashing function and speed up hashing flat strings

Factor out our string hashing code into one function which adds three graphemes and another one which finishes the hash adding 0, 1 or 2 further graphemes.
This makes it easier for us to provide faster hashing for flat 8 or 32 bit strings by not having to use a grapheme iterator.
23:41 MasterDuke left