Perl 6 - the future is here, just unevenly distributed

IRC log for #moarvm, 2015-12-14

| Channels | #moarvm index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
02:06 llfourn joined #moarvm
02:56 hoelzro joined #moarvm
03:11 vendethiel joined #moarvm
04:38 hoelzro joined #moarvm
07:27 domidumont joined #moarvm
07:32 domidumont joined #moarvm
07:44 dalek joined #moarvm
07:56 FROGGS joined #moarvm
08:29 Ven joined #moarvm
09:13 zakharyas joined #moarvm
09:21 Peter_R joined #moarvm
09:51 dalek MoarVM/even-moar-jit: 58bab11 | brrt++ | src/jit/tile.c:
09:51 dalek MoarVM/even-moar-jit: Factor tiling state out of info node
09:51 dalek MoarVM/even-moar-jit:
09:51 dalek MoarVM/even-moar-jit: Tiling state is temporary data, no need to tag the tree with it.
09:51 dalek MoarVM/even-moar-jit: review: https://github.com/MoarVM/MoarVM/commit/58bab11b4f
09:55 brrt joined #moarvm
09:57 brrt good * #moarvm
10:12 nwc10 good *, brrt
10:13 brrt ohai nwc10
10:35 brrt turns out i think TL was more difficult than i anticipated
10:35 brrt i'm in the process of writing a blog about it :-)
10:54 Ven_ joined #moarvm
10:54 jnthn ohhai o/
10:55 brrt ohhai jnthn
10:55 brrt how is life today?
10:58 jnthn Well, I slept way later than expected, due to staying up rather late finishing my advent post :)
10:58 jnthn And a delivery of curry-cooking ingredients and beer will show up later today.
10:59 nwc10 \o/
10:59 nwc10 #london.pm topic, now out of date, reads:  Only 11 debugging all-nighters until Christmas | http://london.pm.org/
10:59 jnthn And I have stroopwafels for with coffee
10:59 nwc10 so, it should have been "advent post all-nighters" instead?
10:59 jnthn So, life seems OK :)
11:00 nwc10 is your food and important-stuff delivery firm only operatin in Prague?
11:00 * nwc10 is still working out how to get interesting things without air freight
11:00 nwc10 or air-freight-like prices
11:03 brrt stroopwafels++
11:03 jnthn nwc10: It is, believe it or not, Tesco Online :)
11:03 nwc10 aha
11:03 nwc10 yes. OK
11:03 nwc10 I wonder if they deliver from Bratislava :-)
11:04 jnthn I never used the service in Bratislava, 'cus I lived 2 mins walk to 4-story Tesco department store :)
11:05 nwc10 does a non-UK Tesco have a section for "UK products" for expats?
11:05 nwc10 eg cider that is actually cider, and not some evil blend of actual cider bulked up with apple juice and sweetner
11:08 jnthn Um...yes to the section.
11:09 jnthn If Strongbow counts then yes, they have it
11:10 jnthn Then, there's a bar just up the street that also claims to have cider, though I don't know if it's "actually cider" :)
11:10 nwc10 um, yes, I agree with the need for clarification
11:10 nwc10 Strongbow is cider like Bud is beer.
11:10 jnthn http://nakup.itesco.cz/cs-CZ/Search/List?searchQuery=cider&Search=Search
11:10 nwc10 but it is, at least, made from the right ingredients
11:11 jnthn Bud is...beer??!
11:13 jnthn 21 xmas RTs. 7 days left at my comfy dev computer.
11:13 nwc10 Anheuser-Busch make a lot of money from folks who think that it is
11:14 jnthn A beer in the hand is worth two from Anheuser-Busch...
11:14 nwc10 deal!
11:14 nwc10 ("two")
11:15 nwc10 I think your exchange rate is wrong (although the parody doesn't work without it)
11:32 cognominal_ joined #moarvm
12:18 dalek MoarVM: 58b81ac | jnthn++ | src/ (5 files):
12:18 dalek MoarVM: Generalize "is TTY" interface to "introspection".
12:18 dalek MoarVM:
12:18 dalek MoarVM: So we can hang the "get the fileno" off it too.
12:18 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/58b81acea7
12:21 rurban_ joined #moarvm
12:24 brrt jnthn++
12:24 * brrt recalls having plans to do that and never getting around to it
12:27 jnthn :)
12:27 jnthn Will finish it up after lunch :)
12:36 travis-ci joined #moarvm
12:36 travis-ci MoarVM build errored. jnthn 'Generalize "is TTY" interface to "introspection".
12:36 travis-ci https://travis-ci.org/MoarVM/MoarVM/builds/96729522 https://github.com/MoarVM/MoarVM/compare/e7edb06caf41...58b81acea774
12:36 travis-ci left #moarvm
12:36 nwc10 oh, odd, works on "my" machine
13:05 colomon joined #moarvm
13:13 cognominal joined #moarvm
13:14 jnthn What an odd failure
13:14 jnthn "No output has been received in the last 10 minutes, this potentially indicates a stalled build or something wrong with the build itself."
13:15 jnthn The other 7 builds worked out
13:15 jnthn Maybe heisenfail
13:16 moritz is that the gocd on hack?
13:21 leont joined #moarvm
13:24 rurban_ sometimes it's travis fault, sometimes a real hang
13:54 jnthn grmbl
13:55 jnthn So it's easy enough to get the "fileno" from a libuv handle
13:55 jnthn Except on Windows that's a HANDLE, not a C runtime descriptor
13:55 jnthn And there seems to be no way to go from HANDLE to "existing" file descriptor.
13:56 jnthn Perl 5 uses FILE * for its I/O, so gets away with already being in C descriptor emulation land.
13:56 jnthn nqp-m -e "say(nqp::fileno(nqp::getstdout()))"
13:56 jnthn 8
13:56 jnthn :)
13:57 leont ?
13:57 jnthn leont: Well, I'm only going on a grep of the Perl 5 source :)
13:57 jnthn leont: But I'm feeing FILE * being used, which you can _fileno
13:57 leont I'm pretty sure there is such a function in the Windows API
13:58 jnthn I'd love to find it if there is one :)
13:58 jnthn There's this one: https://msdn.microsoft.com/en-US/library/bdts1c9x.aspx
13:58 jnthn But this actually gives a fresh descriptor
13:58 jnthn From the HANDLE
13:59 jnthn So if you do that on, say, stdin, you get...3.
13:59 jnthn And if you do it again on stdin you get 4 :P
14:00 leont First question: why would you want this?
14:00 jnthn I'm not actually sure whether I do
14:00 leont File descriptors on Windows are nothing more than a global table that points to HANDLEs
14:01 jnthn I'm looking into https://rt.perl.org/Ticket/Display.html?id=126541
14:01 jnthn Which wants a way to get the native file descriptor, presumably mostly for passing off to native calls.
14:02 jnthn And then I got curious what Perl 5's fileno does, and discovered that on Windows it does managed to give 0..2 for the standard handles.
14:02 jnthn *manage
14:02 leont https://msdn.microsoft.com/en-us/library/ks2530z6.aspx
14:03 leont No wait, I wasn't paying attention
14:03 jnthn That's the opposite way around.
14:03 leont If you can get away with not exposing either HANDLEs or descriptors, that's a very good thing
14:03 jnthn Well, that works fine until somebody needs it
14:03 leont (well, exposing them as nqp::native_descriptor() or some such is ok, but then it comes without any guarantees
14:04 jnthn Yeah, I'm pondering calling the method .native-descriptor
14:04 jnthn Which dehuffmanizes it somewhat
14:04 jnthn And makes clear it's native, as in "you better care about platform issues"
14:05 leont The main reason why people need a file descriptor is when implementing an event loop and/or asynchronous IO
14:05 FROGGS nqp::fileno?
14:05 jnthn FROGGS: Yes, that's what they op ended up being called
14:05 FROGGS nice that we have it now :o)
14:05 leont Since we already provide both, this should be a rare need
14:05 jnthn Just debating semantics :)
14:05 jnthn leont: Indeed
14:06 jnthn So I'll go with "long method name, and platform specific"
14:06 jnthn I don't really want to shut out people doing interesting I/O stuff we don't provide in core over in module space.
14:07 leont Only piece of p5 I ever wrote that would otherwise need it was a portable sendfile, but given it's actually 5 implementations in one modules that already says a lot about how unportable that is.
14:07 * leont isn't awake enough to type English
14:08 jnthn Close enough :)
14:10 dalek MoarVM: 344113c | jnthn++ | / (10 files):
14:10 dalek MoarVM: Implment fileno_fh op.
14:10 dalek MoarVM:
14:10 dalek MoarVM: Provides access to the native descriptor/handle underlying some OS
14:10 dalek MoarVM: handle. Platform specific.
14:10 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/344113cecb
14:12 Ven joined #moarvm
14:22 Ven joined #moarvm
14:28 travis-ci joined #moarvm
14:28 travis-ci MoarVM build passed. jnthn 'Implment fileno_fh op.
14:28 travis-ci https://travis-ci.org/MoarVM/MoarVM/builds/96747723 https://github.com/MoarVM/MoarVM/compare/58b81acea774...344113cecbf5
14:28 travis-ci left #moarvm
14:39 Ven joined #moarvm
14:45 colomon joined #moarvm
15:14 timotimo o\
15:31 zakharyas joined #moarvm
16:14 retupmoca joined #moarvm
16:21 brrt joined #moarvm
16:22 brrt ohai #moarvm
16:23 diakopter hiiiiiii
16:25 timotimo ohai brrt!
16:40 brrt timotimo: the nativeref/multidimref stuff looks cool
16:40 brrt i hope to add support for that pretty soom
16:41 timotimo multidimref was easy; just reproducing what jnthn did for all other cases
16:41 brrt don't knock it :-)
16:42 timotimo it was necessary and i'm glad i could take it off of jnthn plate, that much is for sure :)
17:25 brrt joined #moarvm
17:49 japhb joined #moarvm
17:57 japhb joined #moarvm
17:57 vendethiel joined #moarvm
18:05 dalek MoarVM: eafdbc0 | jnthn++ | / (20 files):
18:05 dalek MoarVM: Implement async UDP sockets.
18:05 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/eafdbc03ae
18:06 timotimo wait, when did that happen o_O
18:07 jnthn I just did it now :)
18:07 timotimo holy crap, now i can doodle with OSC using perl6
18:07 jnthn Yay, one test bunny recruited \o/
18:23 nwc10 jnthn: second potential test bunny is rabidgravy?
18:24 jnthn Probably :)
18:24 FROGGS joined #moarvm
18:25 jnthn It's a bit of a pity we don't have more IO-hacking folks with copious free time. Implementing this UDP support is probably the first thing I've done in my life with UDP :P
18:28 * leont would ordinarily be an IO-hacker, but free time is in rather limited supply at the moment :-/
18:29 timotimo your work on a pure-perl6 test harness is also very valuable
18:30 leont Well, lots of that was IO hacking :-p
18:31 timotimo heh
18:31 timotimo and hammering our IO implementation
18:31 timotimo from the user space
18:31 timotimo that's good, indeed
19:15 FROGGS wow, gcc 5.2.1 is quite colorful
19:29 mtj_ joined #moarvm
20:12 vendethiel- joined #moarvm
20:28 virtualsue joined #moarvm
21:01 virtualsue joined #moarvm
21:36 dalek MoarVM: df3bbce | FROGGS++ | / (5 files):
21:36 dalek MoarVM: probe for _Bool, and expose bool/size_t to nativecall
21:36 dalek MoarVM:
21:36 dalek MoarVM: Skarsnik++ for the original patch.
21:36 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/df3bbce97f
21:49 leont Should probably file a separate issue for the last issue I hit (non-concrete number in Proc::Async)
22:20 dalek MoarVM: 27f15d9 | FROGGS++ | / (2 files):
22:20 dalek MoarVM: unbreak build on platforms without _Bool
22:20 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/27f15d9156

| Channels | #moarvm index | Today | | Search | Google Search | Plain-Text | summary