github.com/moarvm/moarvm | IRC logs at colabti.org/irclogger/irclogger_logs/moarvm
Set by AlexDaniel on 12 June 2018.
01:21 lucasb left 02:30 gdonald joined 04:47 evalable6 left 04:49 evalable6 joined 06:10 sena_kun joined 07:07 AlexDaniel left, AlexDaniel joined, AlexDaniel left, AlexDaniel joined 07:25 sena_kun is now known as Altai-man, Altai-man is now known as sena_kun 07:58 domidumont joined
MasterDuke timotimo: src/profiler/heapsnapshot.c:73:9: error: ISO C90 forbids mixed declarations and code [-Werror=declaration-after-statement] 73 | MVMuint8 i; 09:19
oh, that line looks unused. but if you remove it then you get the same error about `MVMHeapDumpTableOfContents *toc = MVM_calloc(1, sizeof(MVMHeapDumpTableOfContents));` 09:27
dogbert17 hmm, I wonder if 'Zeroed owner in item added to GC worklist' could be some kind of fromspace error in disguise 10:17
I get it when running t/04-nativecall/13-union.t with a small nursery (768 bytes) and it looks like this: gist.github.com/dogbert17/f396ed9a...7f26227232 10:20
11:19 Altai-man_ joined 11:20 sena_kun left 11:57 Altai-man_ is now known as sena_kun
timotimo thanks MasterDuke i'll go fix that 12:12
my compiler was supposed to explode on that
Geth MoarVM: 289e9dfcec | (Timo Paulssen)++ | src/profiler/heapsnapshot.c
fix "mixed code/declaration" error
 12:13
MasterDuke huh, mine did 12:17
but doesn't now
timotimo glad to hear
MasterDuke gcc (GCC) 9.1.0
dogbert17 timotimo: are you less distracted today, i.e. are you working on your grant? 12:35
MasterDuke any reason not to merge github.com/MoarVM/MoarVM/pull/1164 ?
or github.com/MoarVM/MoarVM/pull/1145 ? 12:36
or github.com/MoarVM/MoarVM/pull/1167 ? 12:39
timotimo dogbert17: i didn't start yet, because i was distracted :| 12:46
but i want to do grant work today if possible
did you have some burning questions? :)
haha
the internet connection broke
so i connected to the router's web interface
while i was waiting for it to appear, i switched over to irc to read and answer 12:47
not realizing that that means the internet connection is back
MasterDuke heh. did you simultaneously ask where your glasses were while wearing them? 12:48
timotimo quite
but kind of more like someone put the glasses on my head while i was looking for them
MasterDuke timotimo: can i use `MVM_exception_throw_adhoc_free` to free `MVMGrapheme32 *buffer`? 12:55
dogbert17 timotimo: I have no burning questions atm :) 12:56
timotimo MasterDuke: sure, it just calls MVM_free on the pointers you pass 12:57
MasterDuke cool. i sort of assumed so, but all the uses i've seen in the code are with `char *` 12:58
timotimo right, because the reason the function exists is because you can't free those before you call the exception throw function, because they are used in the format string, but you also can't free them afterwards because the exception throw function doesn't return
MasterDuke i think i'm going to go the _free route to address github.com/MoarVM/MoarVM/pull/1139...304626105, but i guess i'm surprised nobody else has ever used anything other than chars 13:00
13:01 AlexDaniel left 13:02 AlexDaniel joined, AlexDaniel left, AlexDaniel joined
timotimo using the _free variant will allow you to have ("%02x %02x %02x", buffer[pos - 2], buffer[pos - 1], buffer[pos]) 13:03
BBL 13:06
MasterDuke .tell brrt if you don't get github notifications, i replied to your comments in github.com/MoarVM/MoarVM/pull/1139 13:08
tellable6 MasterDuke, I'll pass your message to brrt
13:14 lucasb joined
MasterDuke whoops, `MVMGrapheme32 *waste[] = { buffer, NULL };` does not make the compiler happy 13:16
interesting. the conflicting types isn't the main problem 13:21
src/strings/utf8.c:387:21: error: a label can only be part of a statement and a declaration is not a statement 387 | MVMGrapheme32 *waste[] = { buffer, NULL };
13:24 AlexDaniel left 13:28 AlexDaniel joined, AlexDaniel left, AlexDaniel joined
MasterDuke oh, that was just because of the missing newly required braces for the case 13:29
13:30 AlexDaniel left, AlexDaniel joined, AlexDaniel left, AlexDaniel joined 13:32 brrt joined, AlexDaniel left 13:33 AlexDaniel joined, AlexDaniel left, AlexDaniel joined 13:43 domidumont1 joined, domidumont left
MasterDuke hm. now that i've replied to github.com/MoarVM/MoarVM/pull/1139 and github.com/libtom/libtommath/pull/330 is just about done, i don't remember what of the 5-10 things i was working on comes next (or even what they were) 14:14
14:24 brrt left 15:22 zakharyas joined
MasterDuke how do you get a perl6 backtrace if you ctrl-c in gdb? 16:00
oh, i think i just had an old alias, now it works if i just start perl6 manually with gdb 16:02
timotimo call MVM_dump_backtrace(tc)
16:16 MasterDuke left 16:39 brrt joined 17:00 brrt left 17:12 zakharyas left 17:15 brrt joined
brrt \o 17:15
tellable6 2019-09-07T22:25:17Z #moarvm <jnthn> brrt I've looked at deopt a few times and figured there's a simplification to be had. Probably also a memory saving since we only use a fraction of the deopt points we potentially might need.
2019-09-08T13:08:51Z #moarvm <MasterDuke> brrt if you don't get github notifications, i replied to your comments in github.com/MoarVM/MoarVM/pull/1139
brrt jnthn: I'm working on making the deopt use the index... that will simplify the implementation of guards as expression templates
which becomes entirely nonspecial, at that point 17:16
but... at least one place is messing with my assumptions 17:17
also, there's a bunch of places in which we could save a bunch of memory by using bitmaps, I think 17:18
17:34 MasterDuke joined
brrt I tweaked all known places where we insert sp_guardconc ops into using the index, yet I get one that's using the code offset like before 17:55
18:45 domidumont1 left 18:55 zakharyas joined 19:19 brrt left
timotimo i bet it's uninlining :) 19:43
20:03 Ven`` joined, Kaiepi left, Kaiepi joined 20:09 sena_kun left 20:19 domidumont joined, domidumont left 20:21 domidumont joined 20:22 domidumont left
timotimo i'd love it it rakudo could ping moarperf when it has written a profile file 20:35
like, have a unix pipe in /tmp or something that it could just write the path to?
21:28 zakharyas left 22:12 Ven`` left

Powered by Cro and the Raku® Programming Language.

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

Thank you!