00:18 TEttinger joined 01:13 vendethiel joined 01:40 TEttinger joined 01:47 ilbot3 joined 02:18 vendethiel joined 03:07 vendethiel joined 05:04 harrow joined 05:10 vendethiel joined 07:05 zakharyas joined
dalek arVM: f7f201f | (Stéphane Payrard)++ | src/core/args.c:
typo
07:43
arVM: f0dde50 | lizmat++ | src/core/args.c:
Merge pull request #227 from cognominal/patch-1

typo
07:57 brrt joined
brrt \o 08:34
JimmyZ o/
brrt the 'hard parts' of my little toy are finished :-) 08:35
JimmyZ congratulations
brrt it's not the compiler yet
unfortunately
although
masak brrt: did you blog about progress recently? inquisitive minds are so curious!
:) 08:36
brrt as a means of demonstration, i aim to write a very simple codegenerator, that should be almost as good as you can do for a RISC machine
not yet masak :-) i was going to do that
but then my brain got stuck on this subproblem 08:37
masak I know the feeling 08:39
just remember that there's a market demand :> 08:40
08:46 zakharyas joined
dalek arVM/even-moar-jit: 131d018 | brrt++ | / (2 files):
Add initial tree expression template compiler.

This is a tool to prepare the templates from which the new JIT expression tree IR can be generated. Tool is not done, but parser and expression compiler are.
08:48
08:58 vendethiel joined
brrt Getopt::Long is a standard module, isn't it? 09:02
lizmat in P5 yes, in P6 that would be called multi MAIN 09:04
brrt ok, awesome
lizmat: could you perchance test the libuv-1.6.1-update branch 09:08
on a mac :-)
lizmat brrt: gist.github.com/lizmat/cfa3c247f5b7928cef66 09:24
brrt that's... weird 09:25
the 'reference is not a tree' bug
lizmat same after make clean / distclean 09:26
also after nuking the install dir 09:27
timotimo hm, does "git submodule update" give an error? 09:28
is that the right command?
lizmat $ git submodule update
fatal: reference is not a tree: 30c8be07bb78a66fdee5141626bf53a49a17094a
Unable to checkout '30c8be07bb78a66fdee5141626bf53a49a17094a' in submodule path '3rdparty/libuv'
brrt it is *really* weird. doesn't happen when I'm using a clean source directory
but it is very much a commit
timotimo isn't update supposed to do a fetch first? 09:29
that gives no output whatsoever?
nwc10 lizmat: at a guess, you need to edit 3rdparty/libuv/.git/config
timotimo in any case, cd 3rdparty/libuv; git fetch - should help?
nwc10 url = github.com/libuv/libuv.git
because that's what I needed to do yesterday.
timotimo oh, i thought we had our own repository on github for libuv 09:30
brrt isn't that handled by the branch
timotimo maybe "git submodule sync" helps?
brrt oh, yes
git submodule sync
timotimo that could perhaps go into the configure.pl script
brrt should *probably* be aprt of Configure.pl
brrt nods
lizmat Configuration SUCCESS
so yes
make is successful 09:31
brrt \o/ 09:33
dalek arVM/libuv-1.6.1-update: cdc6d03 | brrt++ | Configure.pl:
Add submodule sync before update in Configure.pl

This is needed to deal correctly with submodule repo url changes.
09:38
brrt i'm thinking of making the sizeof / offsetoff thing a more generic construct 09:52
maybe something like (&foo 3 4) always translating into foo(1,2)
then we can add in whatever macros we like (as long as they can be calculated at compile time :-))
does \w match _ 09:57
yes, it does
09:58 lizmat_ joined
brrt bbiab 09:58
09:59 prammer joined
jnthn Anyone here seen or understand gist.github.com/virtualsue/0b757b139f3946c2515e ? 10:18
Something to do with the recent dyncall/libffi thing?
timotimo huh? i think that's stuff froggs recently introduced 10:20
the leading underscores are probably put in by clang?
jnthn Looks like lack of re-configure... 10:25
timotimo aye 10:27
i re-configure regularly to mae sure nqp's and rakudo's configure scripts are happy with the versions
11:24 brrt joined
brrt no, haven't seen it before 11:24
timotimo brrt: it was because new files were added 11:29
when froggs split nativecall into dyncall and libffi versions
brrt hmm, on master i guess? 11:30
i'm a bit behind :-)
oh, i see
yes 11:31
y'all need to reconfigure more often :-)
timotimo not me, and not jnthn ;)
brrt also so hot anywhere else? 12:12
jnthn brrt: It's on its way to being here 12:16
Today is plenty hot enough...tomorrow will be worse.
brrt remind me to plan a *winter* of code next time 12:21
saturday will be the hottest day in nl
12:33 colomon joined
jnthn urgh 12:34
It seems the hot is moving west to east
So I didn't get the full force of it yet
I'm flying east a bit on Saturday, but it seems it'll only buy we a little delay :)
*me 12:35
12:37 colomon joined
brrt :-) i understood that it be over in about a week 12:40
jnthn phew
Should have airco for a week starting on Sat 12:41
brrt i have fan and a garden spray :-) 12:46
unfortunately, that doesn't work very well for laptops 12:47
jnthn Uh, no, it won't 12:48
If you make your compile cycle long enough you can go cool off while compiling :P 12:49
brrt my compile cycle is too short :-(
14:09 vendethiel joined 14:11 colomon joined
dalek arVM/even-moar-jit: 5ee272c | brrt++ | tools/tree-expr-compiler.pl:
Remove dependency on Test::Exception

This tool should run on every perl we ever encounter, so relying on nonstandard modules is not acceptable.
14:12
arVM/even-moar-jit: e87903c | brrt++ | tools/tree-expr-compiler.pl:
Write header containing expression template tables

The generated file includes a function to get the template information belonging to an opcode.
brrt so, that's good news, because the tool is done
14:13 JimmyZ_ joined
brrt i shall now blog about it 14:13
bbl 14:17
15:12 sivoais joined 16:25 vendethiel joined 16:29 colomon joined 16:49 colomon joined 17:09 mj41 joined 17:20 colomon joined 17:37 vendethiel joined 17:48 brrt joined 18:38 Ven joined
masak looks out for a blog post 18:40
nwc10 I don't think that valgrind tells us anything more than ASAN: paste.scsys.co.uk/491789 18:54
well, I guess other than "there was no undefined behaviour prior to the read from freed memory that ASAN barfed over"
18:57 brrt joined
timotimo you know, i think i really ought to build a tool that you can just paste some text into that contains file:lineno and whenever it sees interp.c:$n it figures out what the op is 19:02
check this out, nwc10, jnthn: paste.scsys.co.uk/491792 19:19
i'm also positively surprised by how quick that script runs 19:21
perl6 performance is quite hard to predict sometimes
nwc10 nice 19:22
dalek arVM: 25dccb2 | timotimo++ | tools/interp_backtrace_enhancer.p6:
a tool that scans text for interp.c:123 and annotates op names

useful for backtraces and such
19:25
timotimo isn't the prettiest code but it seems to work well enough :) 19:26
brrt translate.google.com/translate?sl=...mp;act=url i just had to share this 19:27
timotimo "it went up again and the signals and signals." 19:28
someone drove into some water? 19:29
i don't understand most of the text %)
brrt yes
bridge was open
so cars were waiting before it
19:30 Peter_R joined
brrt so, bridge is closing, but hangs open 19:30
timotimo it failed to close properly?
brrt yes
timotimo damn
and the signals failed to notice
brrt and the motorist just ... drives into it
how does that even happen?
timotimo that's the kind of bridge that just has one static part in the middle that goes up and down? 19:31
jnthn fail
brrt yes, that kind of bridge 19:32
a suspension bridge is i think the english name for it 19:35
jnthn Yeah, you're supposed to suspend your driving before reaching it, when it's open... :P 19:36
brrt fortunately, all ended well for the driver and the passenger 19:41
ooh thunder 19:52
brrt afk 19:54
20:06 colomon joined 20:13 zakharyas joined 20:54 xaduha joined 22:46 TEttinger joined 22:49 japhb joined 23:58 vendethiel joined