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.
00:08 reportable6 left 00:11 reportable6 joined 00:21 raiph left 03:49 squashable6 left 04:51 squashable6 joined 06:02 squashable6 left, committable6 left, bloatable6 left, coverable6 left, notable6 left, shareable6 left, sourceable6 left, bisectable6 left, unicodable6 left, greppable6 left, quotable6 left, evalable6 left, linkable6 left, benchable6 left, nativecallable6 left, statisfiable6 left, releasable6 left, reportable6 left 06:03 unicodable6 joined, releasable6 joined, statisfiable6 joined 06:04 linkable6 joined, evalable6 joined, reportable6 joined, coverable6 joined, sourceable6 joined, notable6 joined 06:05 shareable6 joined, greppable6 joined 07:04 squashable6 joined 07:05 quotable6 joined 08:03 bisectable6 joined 08:04 nativecallable6 joined 08:06 committable6 joined 08:11 discord-raku-bot left, discord-raku-bot joined 09:04 benchable6 joined 10:05 bloatable6 joined
jnthnwrthngtn Hah, that StackOverflow issue that got linked mentions *Parrot*. How does one come accross a Rakudo that old these days? :) 10:47
lizmat uname -v #1-Microsoft Wed Dec 31 14:42:53 PST 2014
Nicholas not sure when debian stopped packaging Rakudo with parrot. 10:49
jnthnwrthngtn Wow
Nicholas But that will be one answer. but it might be "your debian is ancient"
jnthnwrthngtn "stable" :)
Nicholas stable, old stable, old old stable, fell-off-LTS-and-is-definately-special-biologist-word 10:50
Voldenet wsl flavour of ubuntu: `rakudo is already the newest version (2015.11-2build1).` 10:56
lizmat that's even before the first official release 11:09
MasterDuke wonder if there's a way to get wsl to include something newer 11:14
btw, Nicholas, jnthnwrthngtn, nine, anybody who knows more about linking than i do, any thoughts about github.com/MoarVM/MoarVM/issues/1691 ? 11:21
jnthnwrthngtn I know linking on Mac tends to be weird...now, let's see if this is about mac :)
hah, no :)
Wait, what does --static do? 11:22
I thought that was the thing that didn't give you a moar.so, for the Raku ext ops to link against
Nicholas "um, not what one thinks it should" IIRC when I was trying to use it to work around something
jnthnwrthngtn In which case, I'm surprised you even got a Raku that compiles
Or at least it linked against a previously left-behind .so in extops? 11:23
Nicholas oh, completele guess - it might cause symbols in the main program not to be exposed, and hence the dynamic load of the .so won't link properly (and maybe get a NULL instead of failing hard and early)
MasterDuke well, after using --static, `ldd moar` no longer shows `libmoar.so => //home/dan/Source/perl6/install/lib/libmoar.so (0x00007f9fd9e09000)`
Nicholas or the "previosuly left behind"
I'm a bit snowed under by new crazy work code
MasterDuke --static using the libmoar.a instead of the .so, right? 11:24
jnthnwrthngtn I think so, but my understanding is that extops would not work out this way, because they'd end up with an entire separate MoarVM linked in to them, rather than sharing the dynamic library? 11:25
MasterDuke hm. i'd have expected some other/earlier failure then, but this is pretty far outside my experience/knowledge 11:26
github.com/MoarVM/MoarVM/blob/mast...che.c#L260 this is what happens and then the oops hits up at line 234 11:28
lists.fedoraproject.org/archives/l...L5C5SJXB2/ is what inspired me to experiment with --static. it is faster, stage parse drops by ~1s 11:30
also experimenting with -flto and -fno-semantic-interposition 11:32
nine FWIW I haven't seen a difference with lto. Except for the extraordinarily long link time of course
MasterDuke yeah, it didn't seem to help all that much 11:33
though i'm hoping that once gcc 12 arrives and supports mold, the linking time shouldn't be a problem 11:34
nine It can multi thread the linking, which kind of fixes the runtime issue. But since I couldn't measure any advantage, I didn't persue it further 11:37
jnthnwrthngtn Once we are done with getting rid of extops, we can probably do --static without a problem 11:56
MasterDuke ok, so no reason to pursue this until then? 11:57
12:09 reportable6 left 12:12 linkable6 left 12:15 linkable6 joined
jnthnwrthngtn MasterDuke: I don't think it'll work out until then, no 12:27
By this point we can't have many extops left, a lot went away during new-disp
nine p6captureouters, p6stateinit, p6setfirstflag, p6takefirstflag, p6setpre, p6clearpre, p6inpre, p6staticouter and p6invokeunder 12:33
There's also p6init, but that does just nothing :) 12:34
lizmat so it can go from e.g. src/main.nqp ? 12:36
jnthnwrthngtn Probably not as it does soemthing on other backends 12:37
Probably on MoarVM we can map it to a noop though and then delete the extop
p6staticouter could move into core as a syscall 12:38
And compile into that
Nicholas I'd like extops to go, because then AIX is easier 13:12
13:12 reportable6 joined
MasterDuke oh, using clang to build with `--telemeh` and `--c11-atomics` is broken 14:36
15:24 [Coke] left 15:28 [Coke] joined
dogbert11 m: use nqp; say nqp::mod_i(-1,8); say nqp::mod_I(-1,8,Int) 15:41
camelia -1
dogbert11 bug ?
16:28 ismustachio joined 18:10 reportable6 left 19:56 linkable6 left, evalable6 left 19:57 evalable6 joined 20:12 reportable6 joined 20:56 linkable6 joined 20:57 linkable6 joined 23:41 Kaipi left, Kaiepi joined