Welcome to the main channel on the development of MoarVM, a virtual machine for NQP and Rakudo (moarvm.org). This channel is being logged for historical purposes.
Set by lizmat on 24 May 2021.
00:00 reportable6 left 00:03 reportable6 joined 01:53 jgaz left 06:00 reportable6 left 06:01 reportable6 joined 07:21 sena_kun joined 07:40 sena_kun left 08:13 Techcable left 08:16 Techcable joined 09:21 timo left 09:48 lizmat left 09:52 lizmat joined 10:07 timo joined
timo !cursor_start uses the callercode op, which is currently :noinline, can't we relatively easily transform this when inlining in spesh though? 10:27
lizmat oooh... that would be a relative hot spot for regex heavy code 10:28
10:46 squashable6 left 10:48 squashable6 joined
timo i did the trivial implementation and it doesn't explode violently 11:19
Files=1355, Tests=117980, 83 wallclock secs (12.10 usr 2.20 sys + 1393.01 cusr 173.75 csys = 1581.06 CPU)
damn this cpu is fine
lizmat indeed :) 11:21
timo now i don't actually know in what way things would break if callercode would result in the wrong thing 11:24
gist.github.com/timo/e0710330b99df...4b80b7103c can you apply this patch to your moarvm and do some general testing, and maybe measurements with core setting compilation and other grammar-related tasks? check the inlining stats and whatnot? still have lots of dayjob stuff to do 11:40
12:00 reportable6 left 12:02 reportable6 joined 14:05 notable6 left, coverable6 left, statisfiable6 left, bloatable6 left, sourceable6 left, committable6 left, quotable6 left, nativecallable6 left, benchable6 left, unicodable6 left, greppable6 left, evalable6 left, linkable6 left, releasable6 left, reportable6 left, tellable6 left, squashable6 left, shareable6 left, bisectable6 left, tellable6 joined, squashable6 joined 14:06 releasable6 joined, linkable6 joined, notable6 joined, quotable6 joined 14:07 committable6 joined, bloatable6 joined, greppable6 joined, statisfiable6 joined, reportable6 joined, unicodable6 joined, nativecallable6 joined, evalable6 joined, benchable6 joined 14:08 shareable6 joined, sourceable6 joined, bisectable6 joined, coverable6 joined 17:13 sena_kun joined 18:00 reportable6 left 18:02 reportable6 joined 19:36 squashable6 left 19:37 squashable6 joined
Geth MoarVM/callercode_in_inline_is_just_curcode: e15c55036a | (Timo Paulssen)++ | 3 files
callercode no longer prevents inlining

assuming that an inliner's curcode is what an inlinee's callercode is supposed to return, we can just turn callercode into curcode when inlining. There is already a guard that prevents an inlining block from immediately inlining another so we can't for example by accident skip over one level of ... (48 more lines)
MoarVM: timo++ created pull request #1753:
callercode no longer prevents inlining
timo lizmat: have you had time to do some testing?
21:31 sena_kun left
lizmat timo: no 22:11
timo OK! that's no problem :) 22:12
if nothing asplodes in the spec tests, i'm sure it's fine :P :P :P :P
but no really i would like to have someone who's had to deal with inlining issues and wrote spesh code before review this, maybe nine has a moment to poke at it 22:13
moarvm.org's "features" page and the readme have outdated unicode versions, they say 13.1 but in january we got support for 15 afaict 22:15
Geth MoarVM/main: 80559738e1 | (Timo Paulssen)++ | README.markdown
mention we have unicode 15
lizmat sleep& 22:31
23:09 linkable6 left, evalable6 left 23:12 evalable6 joined, linkable6 joined