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.
00:02 reportable6 left 01:02 bloatable6 left, quotable6 left, shareable6 left, releasable6 left, bisectable6 left, nativecallable6 left, greppable6 left, notable6 left, committable6 left, sourceable6 left, squashable6 left, statisfiable6 left, coverable6 left, evalable6 left, unicodable6 left, tellable6 left, benchable6 left, linkable6 left 01:03 sourceable6 joined, squashable6 joined, shareable6 joined 01:04 bisectable6 joined, tellable6 joined 01:05 evalable6 joined, benchable6 joined, releasable6 joined, reportable6 joined, statisfiable6 joined, greppable6 joined 02:04 nativecallable6 joined 02:05 unicodable6 joined 03:04 quotable6 joined, committable6 joined 03:05 coverable6 joined 04:04 linkable6 joined 04:50 frost joined 05:05 bloatable6 joined 06:02 reportable6 left 06:03 notable6 joined 07:03 reportable6 joined 07:59 linkable6 left 09:01 linkable6 joined 09:22 frost left 10:24 lizmat left 10:25 lizmat joined 12:03 reportable6 left 12:06 reportable6 joined 14:08 committable6 left, benchable6 left, bisectable6 left, sourceable6 left, evalable6 left, nativecallable6 left, reportable6 left, unicodable6 left, bloatable6 left, quotable6 left, squashable6 left, shareable6 left, releasable6 left, greppable6 left, tellable6 left, notable6 left, coverable6 left, statisfiable6 left, linkable6 left 14:09 shareable6 joined 14:10 evalable6 joined, reportable6 joined, squashable6 joined, bloatable6 joined, sourceable6 joined, notable6 joined 14:11 tellable6 joined 14:12 releasable6 joined, benchable6 joined
dogbert17 MaterDuke++, nice work yesterday 14:20
that should ofc be MasterDuke++ 14:21
there are two juicy new-disp bugs waiting for the rr experts :) 14:32
15:10 statisfiable6 joined 15:27 leont left 15:31 leont joined 16:09 quotable6 joined 16:10 coverable6 joined, linkable6 joined 16:11 nativecallable6 joined, unicodable6 joined, bisectable6 joined
nine there are_ 16:15
dogbert17 nine: indeed 16:20
t/spec/S32-io/utf16.t hangs with MVM_SPESH_NODELAY=1 MVM_SPESH_BLOCKING=1 and a small nursery (12k) 16:21
t/spec/S04-statements/with.t fails to comipile given the same settings
17:09 greppable6 joined 17:10 committable6 joined 18:02 reportable6 left 19:03 reportable6 joined
vrurg How hard would it be to add atomics support to NQP? 19:56
lizmat isn't it already in nqp as nqp:: ops ? 20:07
$ nqp -e 'my int $a; nqp::atomicstore($a,42)' 20:09
Cannot perform atomic store to a non-container value of type BOOTInt
vrurg lizmat: that's it. There is no atomicint.
lizmat but isn't that just a matter of definition?
afaik, the "atomicint" is reallt just an int64 ? 20:10
the magic being in the ops, *not* in the variable ?
going afk&
vrurg In the core it is 'my native atomicint is repr('P6int') is Int is ctype('atomic') { } ' 20:11
ctype is what's missing for NQP.
lizmat I see :-(
20:14 evalable6 left, linkable6 left 20:15 evalable6 joined 20:16 linkable6 joined
vrurg Ok, correct me if I'm wrong, but atomics wouldn't work without containers. So, no go for NQP. 20:44
21:04 squashable6 left 21:07 squashable6 joined
timo atomics can work on locals by way of NativeRef, right? 21:10
maybe we can only sensibly emit that at the qast level at the moment in nqp 21:11
"is ctype('atomic')" is only for getting the right size, isn't it?
but yeah, nqp doesn't build containers by itself, and if we can't "just" get a nativeref or similar with "is rw" or so, that will need ome extra work 21:36
vrurg timo: but that'd require more work than it worth, perhaps. 22:23
timo do you have anything in particular in mind? 22:27
vrurg No. I just played with adding atomicint to NQP (required a grammar fix) only to realize that nqp::cas results in a "not a container" kind of error. 22:41
timo yeah, nqp also doesn't have "is rw" 22:42
vrurg That's easy to understand. But with atomic I needed the error message to realize the fact. :) 22:44
timo have you tried adding something to t/nqp/104-refs.t? there are reference types there which may support atomic ops 22:45
vrurg No. I didn't even know that exists. But looks promising. Unfortunately, I have discarded all changes in NQP already. Would have to try it later. 22:48
23:21 linkable6 left, evalable6 left, evalable6 joined
timo the nqp test suite is a little treasure trove in general 23:25