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.
drakonis how much longer until it gets merged? 01:18
Nicholas good *, * 06:28
nine drakonis: there's quite a bit of work ahead still 07:31
Nicholas something like: 07:36
1162 red spectests with a status fail
and if one red spectest should unexpectedly pass
there'd by 1161 red spectests with a status fail
(repeat for another 1161 verses)
also, "unexpectedly pass" is not a good thing 07:37
"it's not easy being green"
nine My gut feeling says we're probably somewhere in the middle. That may sound disappointing considering the 2 years or so since the start. But then so far it was pretty much only jnthnwrthngtn working on it and he did new-disp in the same time frame. The latter is done and I have just joined the effort. And I think most of the difficult design questions are answered. 07:40
jnthnwrthngtn nine: I think there's still quite a few tricky design questions to go :) 09:24
nine: btw, it was 320 passing not so long ago, so back to 277 is still quite the regression 09:25
MasterDuke `Cannot have attributes on an array representation`. i just attempted to add some methods to MAST::Bytecode that would do a write_at at an incrementing position (to clean up github.com/Raku/nqp/pull/768), but the idea may not work nicely 10:44
i.e., i tried to add `has $!pos` and then the new method would do `write_at(..., $!pos); $!pos := $!pos + x;` 10:45
guess i can just implement via MoarVM::BytecodeWriter in NQP instead 10:48
nine jnthnwrthngtn: I have never managed to get 320 passing, even on the commit before my very first one 10:52
So 277 was my base line
MasterDuke: if it becomes too troublesome, I'd just leave your PR as it is. I don't like it, but sometimes we just have to accept a little uglyness 10:53
MasterDuke i'll see if doing in on MoarVM::BytecodeWriter is any cleaner, it's not difficult to give it a try. if not, i'll revert to as is + a variable rename for clarity 10:55
jnthnwrthngtn nine: Will check what I get. t/12-rakuast/var.t fails on latest `rakuast`, btw 13:03
But package seems OK
nine: Back up to 320 for me! \o/ Trying to guess the difference: do you do a `make spectest` without RAKUDO_RAKUAST=1 and then one with? 13:07
The first one pre-compiles all the various utility modules 13:08
And then the RakuAST-based compiler can load them
I guess just doing `make test` followed by the spectest with RAKUDO_RAKUAST=1 would pass rather less (for example, all that use is-run) 13:09
Ah, and I guess the var.t failure is the one we talked about :)
nine Yes, that's the one somewhat fishy test 13:10
And yes, I'm pretty sure that explains it :) I have noticed that I get only failures until I run one test without RAKUDO_RAKUAST=1. But of course that one test will not load all the modules
jnthnwrthngtn Aha, OK, then it makes sense. 13:11
MasterDuke hm, i don't think adding anything to MoarVM::BytecodeWriter is any cleaner. i could add an `$!mbc_pos` along its `$!mbc` and some methods which manipulate the two, but it/they would only be used in `write_frame` 20:12
dunno. all those `$!mbc.write_uint32_at($foo, $idx); $idx := $idx + 4;` could become `self.add_uint32($foo);`, with implementation `method add_uint32($data) { $!mbc.write_uint32_at($data, $!mbc_pos); $!mbc_pos := $!mbc_pos + 4; }` 20:18
maybe just two lexical subs in `write_frame` is better
then if the same kind of change is made elsewhere they could be pulled out into methods 20:19
nine 320 it is! 20:28
MasterDuke nice 20:29
jnthnwrthngtn High time for me to get to work and get 100 more passing... 20:49
MasterDuke ~1352 is the final target, correct? 20:53
jnthnwrthngtn Yup 20:57
Now it's 321 21:20
[Coke] woot 21:26
jnthnwrthngtn Will try and get a few more tomorrow. My brain is still mushy today. 21:35
MasterDuke nice. after the excitement of putting an offer on a house, going to go save earth from (another) alien invasion 21:42
jnthnwrthngtn Cool, best of luck with the house. :) 21:49
japhb ... and the alien invasion 21:54
jnthnwrthngtn
.oO( Would they be worse than humans? )
 21:57
MasterDuke thanks (for both) 22:07

Powered by Cro and the Raku® Programming Language.

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

Thank you!