github.com/moarvm/moarvm | IRC logs at colabti.org/irclogger/irclogger_logs/moarvm
Set by AlexDaniel on 12 June 2018.
jnthn Wonder what keeps us from deleting them entirely... 00:20
timotimo "block with cuid BLAH failed to show up" 00:23
i guess i can just toss that out and see what happens :P 00:24
lizmat in the past, I've seen that message every now and then 00:25
brrt \o 13:04
nwc10 o/ 13:05
jnthn |o 13:07
brrt fullish house? 13:09
(I still miss timotimo for that)
timotimo o/ 13:11
brrt ohai timotimo
lizmat unlurks
brrt :-) 13:17
I looked into it... it seems that we have 4 classes of labels: 13:21
- for graph start and end
- for basic blocks
- for individual instructions 13:22
- internal to expression trees 13:23
within all of these classes, labels all increase
jnthn So we can use going to a smaller label in order to detect a back-branch? 13:35
Geth MoarVM: 9709537d90 | (Jonathan Worthington)++ | src/6model/serialization.c
Serialize the names of closures

The name is held per code object, meaning different closures may carry a different name. This is notably true of auto-generated `proto` methods. However, we did not preserve this during serialization, meaning that Raku language users could see things like AUTOGEN showing up in both backtraces and introspection. Fix this by serializing the name along with the closure. This, one a spectest is added, will let us resolve
  github.com/rakudo/rakudo/issues/3323.
15:52
timotimo oooooh that is why that happens! 16:05
Guest47772 what, is jnthn at it again, hooray :) 16:06
jnthn Doing some easy things to get back into the code :) 16:19
Guest47772 very cool
so you're cherry picking the bugs then :) 16:20
jnthn Yeah. :) 16:21