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.
MasterDuke caught that inline_cache spesh oops in rr. can i interest anyone in the recording? 13:03
MasterDuke hm, something seems wrong with this recording 13:22
MasterDuke nine: unrelated question, is is possible to create a raku hash in c code and return it via nativecall? 14:03
nine MasterDuke: you mean from some user library? No, dabbling in internals like that is explicitly unwanted. 14:09
MasterDuke yeah. rf over in #raku was asking recently because they were creating a module binding to simdjson and wanted to return a hash 14:10
jnthn One would return, say, an array of values and build the hash back in Raku space. 14:12
That way would work on other backends too. 14:13
MasterDuke thanks, i'll pass the suggestion along 14:16
jnthn Although it seems the API of simdjson is more iterator based, so it may make more sense to consume that directly in Raku (that is, iterate to produce the hash) 14:20
MasterDuke arg, something seems borked with my rr 14:42
nine Loaded the zen workaround? 14:43
MasterDuke yep. at first it wouldn't run because of a libcapnproto mismatch, so i tried rebuilding it 14:46
that failed at first because of complaints about flexible array members not at the end of structs
so i applied a patch from a recent commit to their git repo 14:47
now it builds and installs again, but my replays just all end in syscall_traced()
maybe i'll just try building the latest from their git repo, the last 5.6.0 release was sometime last year 14:50
MasterDuke hm, might be related to using the `--chaos` flag... 15:02
[Coke] heh. 15:27
"why is there so much chaos!?" "oh."
MasterDuke huh. rr built from git, no `--chaos`, but the replay still doesn't work right 18:51
well, it repros easily enough, hopefully one of you with a working rr can catch it 18:52
github.com/MoarVM/MoarVM/pull/1740 seems to be in a somewhat decent state. the only problem now that i think is unique to/cause by the PR is the windows fails 19:48
but it happens when building rakudo, not moarvm or nqp, and only on the new windows server 2022 images 19:49
Stage start      :   0.000 19:51
NMAKE : fatal error U1077: 'C:\Strawberry\perl\bin\perl.exe' : return code '0xff'
Stop.
Command failed (status 512): nmake
 at D:/a/1/rakudo/3rdparty/nqp-configure/lib/NQP/Config.pm line 34.
    NQP::Config::__ANON__("Command failed (status 512): nmake\x{a}") called at D:/a/1/rakudo/3rdparty/nqp-configure/lib/NQP/Config.pm line 1502
    NQP::Config::system_or_die("nmake") called at Configure.pl line 124
some random googling suggests maybe "Microsoft Visual Studio is not correctly installed, or that the paths to it are set incorrectly" 19:52
but then i don't understand how moarvm and nqp built
[Coke] MasterDuke: does it work OK on windows 10, but not 2022? 20:28

Powered by Cro and the Raku® Programming Language.

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

Thank you!