github.com/moarvm/moarvm | IRC logs at colabti.org/irclogger/irclogger_logs/moarvm
Set by AlexDaniel on 12 June 2018.
samcv it looks like libtommath is likely to go BSD/MIT license github.com/libtom/libtommath/issues/126 03:54
brrt \o 13:59
yoleaux 22 Nov 2018 02:17Z <kiwi_95> brrt: (and jnthn) I'd love to read your take on "join points" and "basic blocks with arguments" as discussed here www.reddit.com/r/ProgrammingLangua...lift_uses/
brrt .tell kiwi_95 - I'm honestly not personally sufficiently advanced that I could tell you what the difference really is 14:00
yoleaux brrt: I'll pass your message to kiwi_95.
brrt in practical terms. It seems pretty equivalent to me
brrt .ask nine I wonder if you can help me make a test case for nativecallinvoke, and maybe start thinking about deprecating the difference between nativeinvoke, nativeclallinvoke and plain invoke 14:02
yoleaux brrt: I'll pass your message to nine.
Geth MoarVM/pea: 913504df90 | (Jonathan Worthington)++ | 8 files
Preparations for deopt materialization

  * Stub in the data structures we'll use to describe materializations of
   replaced objects and what deopt points need them
  * Add an NYI panic in deopt in the case that we encounter something to
   materialize (should never trigger at the moment)
15:54
brrt ohai jnthn 16:05
jnthn o/ brrt 16:08
This is heavy stuff for a Friday... :) 16:14
Geth MoarVM/pea: 6e28e173bd | (Jonathan Worthington)++ | src/spesh/pea.h
Target register goes with deopt point

Since we might have to materialize *one* object, but put it into *many* registers during deopt, for example because of aliasing.
17:18
MoarVM/pea: d29dd72e77 | (Jonathan Worthington)++ | src/spesh/dump.c
Dump materialization data
MoarVM/pea: ad480447f5 | (Jonathan Worthington)++ | src/spesh/pea.c
Keep a set of tracked registers

These will be needed in order to properly consider deopt points, and see if we need to save materialization information.
17:22
MoarVM/pea: e317922036 | (Jonathan Worthington)++ | 2 files
Sketch out creating deopt/materialization info
jnthn Well, it's some progress... bbl 17:27
nine ~~ 19:18
yoleaux 14:02Z <brrt> nine: I wonder if you can help me make a test case for nativecallinvoke, and maybe start thinking about deprecating the difference between nativeinvoke, nativeclallinvoke and plain invoke
nine .tell brrt I sure hope I can help :)
yoleaux nine: I'll pass your message to brrt.