github.com/moarvm/moarvm | IRC logs at colabti.org/irclogger/irclogger_logs/moarvm
Set by AlexDaniel on 12 June 2018.
moon-child discoD: thank you!! 00:11
discoD happy to help. I've got one more for arrays (needs more testing), then all the nativecall bugs I've experienced will be squashed 00:20
moon-child Perhaps I'll take another crack at a raku embedding api once that's merged 00:25
discoD That'd be nice. I've wished for it a few times. 00:36
moon-child the main difficulty is that the global namespace isn't dynamic, so you have to decide ahead of time all of what you want to expose 00:38
or use weird hacks like the repl does 00:39
(also, I think there's a bug in mvm with shared state; when I tried to start up multiple mvm instances, I got a crash. But the overhead of an instance is high enough that you'd probably want everything on one instance anyway) 00:42
discoD you mean if you link libmoar and start 2 instances it chokes? 00:43
moon-child yeah
that might also be resolved, though; it was a while ago
discoD i see. I've never tried it. I do have dreams about standalone executables, but I haven't tried that yet either 00:44
moon-child standalone executables would be cool. Can't you kinda get that with the jvm backend, though? 00:49
discoD dunno, I've never used it, but maybe I should take a look. I want it because no antivirus would catch it, so I could write undetectable payloads in raku. 00:51
moon-child ...oh 00:52
discoD for pentesting
moon-child I mean, you can do that with something like lua
statically link it, embed source code in binary
discoD sure, there's always a way around av, but I don't know lua very well. 00:55
my only experience with it involves nmap, wireshark, and the syntax highlighter in my text editor. life without good regular expressions is hard 00:57
nwc10 good *able6, #moarvm 06:28
MasterDuke discoD: there was a GSoC project last year to enable creating standalone executables 08:01
i think they got to the point of being able to do so for scripts, but not if it used any modules
6ring.github.io/?pamplemousse 08:03
ggoebel discoD: I'm pretty sure pamplemouse got to the point of creating a standalone executable for a script with 1 module. 11:44
I think when they stopped they were butting up against the right way to include >1 module in the standalone executable. Nine recently shared his thoughts on that... colabti.org/irclogger/irclogger_lo...-04-29#l42 11:48
discoD Thanks for the links. I'll take a look. 20:09
I'm fuzzing string decoders because I'm going stir-crazy. Is there a way to get at the chinese decoders(gb2312,gb18030) from raku, or is that NYI?
discoD nevermind 20:21
timotimo if you can, i'd really recommend doing fuzzing with the function directly, completely without rakudo or even nqp 21:27
discoD i'd prefer that too. I tried fuzzing in memory with PIN a few weeks ago, but I could never wrangle all the background threads, and restoring state always crashed. 22:00
i'll look at starting a vm instance manually and go from there. i want the gc to run, because that's where most bugs would manifest 22:02
timotimo you can force a GC run, too 22:11