github.com/moarvm/moarvm | IRC logs at colabti.org/irclogger/irclogger_logs/moarvm
Set by AlexDaniel on 12 June 2018.
MasterDuke timotimo: is moarperf integrated into comma at all? 00:08
timotimo no 00:12
comma has its own profiler stuff :)
MasterDuke huh, didn't realize that
timotimo well, only its own profiler UI 00:13
it uses the sql profiler stuff
MasterDuke nice 00:14
timotimo it's a bit simpler than what moarperf has, but i'm on the comma team as well :) 00:16
jnthn timotimo: That wasn't actually merged yet, though probably will be next week and in the next release. :)
timotimo oh!
jnthn But yeah, there'll be loads of stuff we can do with presenting the performance data alongside the code :) 00:18
timotimo yep! 00:19
exciting times ahead
MasterDuke well, just bought comma. don't think i've ever used any jetbrains products before, but i'm looking forward to it 00:24
timotimo it'll hopefully be a good time :) 00:28
especially looking forward to what other magic stuff comma's gonna get. like integration with Cro 00:29
MasterDuke first question, how do i point it to a non-system perl6? 00:30
timotimo it has a concept of "SDKs" 00:31
jnthn Project Structure dialog, and then New SDK in there
timotimo oh, i gave the "correct, but completely useless" answer 00:32
MasterDuke well, i tried creating a new project and figured it out (and saw "SDK" in the text, so timotimo's hint did help) 00:33
jnthn Ah, I figured you'd just sailed past that while making a project already :) 00:35
MasterDuke have you guys been using for hacking on rakudo itself? 00:36
timotimo nah, i'm still using vim for that mostly
comma doesn't have (a reason to have) support for nqp or nqp-heavy perl6 code 00:37
MasterDuke i would think at least the syntax highlighting would pretty much be the same 00:38
timotimo and i'm still experiencing an oddity in syntax highlighter/parser performance on bigger files
even though jnthn had that at one point and fixed it two or more releases ago
00:42 manchicken joined
manchicken Hello. I just encountered (and _DEFEATED_) a build problem brought on by having mixed Xcode + GNU toolchain on OSX Mojave. As such, I thought it might be a good idea to add a safety to the `Configure.pl` for moarvm to throw out a big warning. 00:44
That seems like the quick fix. Long-term, I'm wondering if it might be a good idea to try to infer (based on the path or the `--version`) of the compiler and the ar/ranlib to add a self-correcting safety. 00:46
I know that this is a problem that exists sometimes, but I wanted to make sure I'm not duplicating efforts.
jnthn A "you probably are going to have a bad time" warning would probably be friendly, if you're able to patch that in. 00:48
timotimo maybe even reference the error message that you're likely to get in that instance
so that when you search for it you may stumble upon it, too
unfortunately, the error will come at the most amount of output that moarvm's "make" step is capable of 00:49
so it'd be as far away from Configure.pl's warning message as possible
manchicken That seems like it'd be a pretty straight-forward patch. 00:53
Well, I was thinking of having a prompt to continue. 00:54
jnthn timotimo: BTW, will you be at the German Perl Workshop this year?
manchicken "This isn't going to go well. Would you like to continue anyway? [y|n]"
jnthn manchicken: Hmm, maybe better would be to exit with non-zero exit code unless there's some --force flag passed or similar 00:55
A prompt isn't so friendly to anyone scripting stuff 00:56
manchicken Yeah, that's true.
Righto.
That seems pretty straight-forward.
timotimo jnthn: oh! i still need to get all of that set up! 01:03
jnthn :)
I did my hotel a while back, but booked trains there today
Should probably write my talks soon also :)
timotimo wait, munich?
jnthn Yeah
timotimo ... oh yeah
somehow i was confusing munich with cologne?!? 01:04
jnthn On the upside, there's a comfy ICE between them if you went to the wrong one :P
timotimo oh, it'd have been karneval time in kƶln 01:05
that's a thing up there
jnthn I heard about that...in the days before the workshop I'm going to be in Železna Ruda / Bayerisch Eisensteinen for a few days relaxing, and the Localbahnmuseum is open more than usual thanks to karneval. I've no idea if that means it'll also be packed. :) 01:08
timotimo schedule's not yet ready to be seen; not sure if the evening before the 6th would have a meetup 01:16
jnthn Dunno; I'll be arriving sometime in the afternoon of the 5th
timotimo archinect.imgix.net/uploads/ai/aiy...s%2Cformat - why don't they have ose of those in munich? :D 01:26
01:32 manchicken left 01:34 manchicken joined
manchicken Apple's making this easy. `grep -b gnu $(which ar)` yields `$? != 0`. Yay. 01:35
timotimo .o( was fĆ¼r ein vergnĆ¼gen )
MasterDuke i'm going to be in the uk for roughly the next three years, maybe i'll be able to make it to some european perl6 conferences 01:38
jnthn MasterDuke: PerlCon in Riga in August is probably a good one to go to :) 01:46
Sleep time for me; 'night o/ 01:47
MasterDuke well, not arriving until mid-june, so that might be too soon, but it's certainly attractive
later...
timotimo gnite jnthn 01:49
if the schedule isn't ready to be seen yet, does that mean i can still consider giving a talk? :D 01:54
manchicken @jnthn: I'm not sure I need to implement a `--force`, we already have a `--toolchain=gnu` in the configure. If I see that the basics are GNU-derived, and we didn't get that arg, it seems like we should die then.
01:58 manchicken left 02:01 manchicken joined
manchicken Weeeee! Submitted PR. 02:22
I haven't submitted a PR for a few months. Too much work stuff going on.
Sweet! Passed my checks. I'd be interested in any feedback anybody had. github.com/MoarVM/MoarVM/pull/1055 02:38
03:47 manchicken left 03:53 manchicken joined
manchicken There are a bunch of PRs which are failing tests. 03:58
Are any of them worth digging in to? 03:59
MasterDuke manchicken: you'd want to check why exactly they're failing. sometimes travis/appveyor just time out downloading/installing required things 04:00
manchicken Yeah; I see one of them (branch `reloc`) is failing due to some non-macOS friendly linker flags. 04:02
04:18 Ven`` joined 04:36 manchicken left 04:37 Ven`` left 08:05 robertle joined 08:54 MasterDuke left 10:04 synopsebot left, dalek joined, synopsebot joined, p6lert joined, Geth joined 12:16 MasterDuke joined, MasterDuke left, MasterDuke joined
MasterDuke timotimo: do we really need this mp_clear? github.com/libtom/libtommath/blob/...v_d.c#L103 12:44
yoleaux 10:50Z <timotimo> MasterDuke: this seems to be the big integer to string algorithm cpython uses: github.com/python/cpython/blob/mas...ct.c#L1679
timotimo that's necessary to free the buffer inside the mp_int, isn't it? 12:48
MasterDuke i saw some commentary that suggested those were for security in crypt applications, but maybe we don't care? 12:49
timotimo doesn't mp_init_size have some kind of malloc in it? or is it a macro that uses alloca? 12:50
MasterDuke `XMALLOC(sizeof(mp_digit) * (size_t)size)` 12:51
timotimo what does XMALLOC do, then?
is that what we're free to redefine to supply our own memory management? 12:52
MasterDuke github.com/libtom/libtommath/blob/....h#L39-L52
timotimo ah, ok
so it seems like if you don't mp_clear the q, it'll be leaking that buffer 12:53
MasterDuke mp_clear sets it to zeroes before freeing, we could just free directly instead
github.com/libtom/libtommath/blob/...mp_clear.c 12:54
or just remove the setting to zero from mp_clear if we *never* need it 12:58
hm, what do i have to do for edits to 3rdparts/libtommath to be picked up? 13:04
timotimo oh, i see 13:05
will probably have to commit it inside the submodule and then "git add" and commit the submodule itself inside the moar repo? 13:06
MasterDuke thanks. probably won't be able to experiment right now, but will try later 13:07
timotimo though it'd surprise me if "make" would immediately revert any changes to 3rdparty things 13:23
14:16 scovit left 14:19 lucasb joined
MasterDuke hm, commenting out the setting to zero in mp_clear doesn't seem to have much effect speedwise 14:33
oh, and we're using 1.0 of libtommath, but 1.1 was just released last month github.com/libtom/libtommath/blob/...hanges.txt 14:38
timotimo ah, yeah 14:39
MasterDuke how do i specify/upgrade to 1.1 to test?
timotimo you'd go into the libtommath submodule in moarvm's 3rdparty folder, "git pull" or whatever it takes to bring it to the right commit, then go back out and "git add 3rdparty/libtommath" and commit that 14:56
15:22 manchicken joined 15:45 Ven`` joined 16:12 scovit joined 16:14 Ven`` left 16:24 manchicken left 16:25 manchicken joined 16:44 manchicken left 16:50 zakharyas joined 17:39 Ven`` joined 18:32 Geth left, Geth joined
MasterDuke well, i think i updated to libtommath v1.1.0, and nqp built ok and passed `make m-test` and rakudo built ok and passed `make m-test m-spectest` 18:40
but i didn't rebase, so i assume any custom commits we had on top of v1.0.0 weren't included? in that case maybe they aren't required anymore?
timotimo some of them probably got replaced by upstream changes 18:41
18:47 Geth left, Geth joined
MasterDuke i have no idea how to submit this as a PR 18:50
timotimo just pretend that the folder with the changes in it is a single file 18:54
MasterDuke or...........blow away all my changes and create a new issue saying i tested it already... 18:57
18:58 Geth left, Geth joined 19:09 Geth left, Geth joined 19:28 lucasb left 19:30 Kaiepi joined 19:49 Ven`` left 19:56 Kaypie joined, Kaiepi left 21:04 Kaypie left 21:29 zakharyas left 21:59 Ven`` joined 22:55 Geth left 22:56 Geth joined 23:09 Ven`` left