github.com/moarvm/moarvm | IRC logs at irclog.perlgeek.de/moarvm/today
Set by moderator on 8 October 2013.
00:33 retupmoca joined 00:52 colomon joined 01:07 colomon joined 01:18 FROGGS_ joined 01:31 benabik joined 01:47 _ilbot joined
moderator github.com/moarvm/moarvm | IRC logs at irclog.perlgeek.de/moarvm/today
03:19 woolfy left
TimToady what a boring channel :) 05:58
moritz moar fun! 05:59
JimmyZ :-) 06:00
FROGGS_ TimToady: well, you can always entertain us :o) 06:11
06:58 lizmat joined 09:43 ssutch joined, lizmat joined 10:09 FROGGS_ joined
diakopter moritz=+ 10:55
er, moritz++
oh btw www.iloveopensource.io/projects/525...497d000bb4 11:07
FROGGS_ diakopter: we should put libuv and others as a dep 11:08
diakopter heh, why :)
FROGGS_ because I think libuv is great :o) 11:09
diakopter oh, right... :)
forgot for a minute we used it. :|
FROGGS_ hehe 11:10
(and then I spend some money for libuv and force a libuv-hacker to look at our problem) 11:11
diakopter from the "people named Perl" department: en.wikipedia.org/wiki/Emily_Kingsley 11:26
and this guy: www.newschool.edu/lang/faculty.aspx?id=10372 11:27
and this guy: www.imdb.com/name/nm0674144/ 11:29
12:07 FROGGS joined 13:36 japhb joined 14:00 grondilu joined 14:25 jnap joined 15:50 FROGGS joined
rurban did you already change the new fs_events api? 15:58
diakopter for what? libuv?
how new is it?
15:58 raiph joined
rurban I had to do the libuv 3rd rewrite now. first the error not the loop anymore (huge) and now this 15:58
2 days or so 15:59
diakopter you mean rewrite of stuff that uses libuv in p2?
rurban rewrite my aio.c wrapper
diakopter in p2?
rurban they advance like rust
FROGGS do we use fs_events at all?
rurban in potion
I export the whole libuv API
I'm now on 0.11.14, and was on 0.10.12 before 16:04
diakopter rurban: why potion and not p2?
rurban because they use the samme vm. ps is just sugar and a parser on top of potion
p2 is
FROGGS moarvm runs fine with libuv HEAD fwiw... though we use 0.11.4 or so atm 16:05
rurban potion uses scheme like predicate names, like defined?. p2 uses perl like names: defined
great
so you are not using the tricky, fast moving parts 16:06
I still have to keep up with the process class
FROGGS there is only one issue that I know: if you print to stdout/-err, close that handle, and the delete the loop, it segfaults 16:07
and I have no idea to get around that
diakopter FROGGS: we shouldn't be closing stdout/err
I thought I disabled that
FROGGS diakopter: I mean uv_close(handle), not nqp::closefh($stderr) 16:08
diakopter right
I thought I disabled uv_close on handles representing stdout/stderr
FROGGS and if you dont uv_close it, you should not delete the loop because this will assert when there are open handles
diakopter: right, my local copy closes these handles, without any change in behaviour 16:09
diakopter FROGGS: we should save flags to close stdout/stderr upon global destruction then
if they've been openeded
opened 16:10
commute really &
FROGGS did that, changes nothing
rurban shouldn't you use uv_fs_close()?
FROGGS for file descriptors yes, by stderr is a tty
rurban see test/test-fs.c
I see
FROGGS so uv_close takes care of it
raiph Am I right in saying the elf format makes little sense for p6 because you have to have a runtime (eg moarvm) which means elf has no advantages over bytecode? 16:11
FROGGS so, if I close it when we shut down, then it segfaults in uv__stream_destroy or so
rurban their testcase avoids tty_err
because it wouldn't work on windows, i guess
FROGGS rurban: that makes me think I should create a super small test case and create a bug report... 16:12
rurban just add it to test/test-tty.c
FROGGS yeah, will do, thank you!
raiph: well, I would like to see something for p6 what staticperl is for p5... so, having one binary (elf/windows/whatever) would be nice 16:14
diakopter FROGGS: well..... 16:38
FROGGS: fakecuteables with .moarvm bytecode embedded as binary blobs would be just fine, imho 16:41
(esp if the executable had an option to write out the binary blob to .moarvm back out again) 16:42
FROGGS: why does it need to be one executable?
16:51 colomon joined 17:03 benabik joined 17:05 lizmat joined 17:14 lizmat_ joined 17:16 woolfy joined
FROGGS diakopter: it would have an virtual filesystem inside, for games etc 17:16
diakopter: so you an just copy it around 17:17
diakopter: that means: modules it depends on, gfx and sounds files are included in the binary like in a zip file 17:22
diakopter eh 17:27
unsigned executables aren't treated very well
17:29 raiph joined
FROGGS diakopter: "treated"? what do you mean? 17:30
diakopter they are looked down upon 17:31
FROGGS from what?
diakopter windows?
FROGGS well, our binaries are not signed as well 17:32
diakopter they need to be
(if they're downloaded)
FROGGS so, having an option to distribute a game beats no option I'd say
raiph .oO ( looks down from second story window and sees elf; frowns ) 17:33
rurban but more important for modules and dll's
like par
FROGGS diakopter: "they need to be" ... in future, or "they need to be" ... already? 17:34
raiph FROGGS, diakopter, rurban: thanks for discussing this
(still listening)
FROGGS :o)
rurban on windows you simply add resources and mmap it 17:35
FROGGS I enjoyed staticperl because it lets me bundle up my little games on an old debian 3, and then it will run on most linux boxes
with a 3meg executable (== single file) 17:36
which is pretty awesome
rurban the problam mwith par is the temp. unpacking. it should really mmap it. but you cannot fool the loader to mmap dll's in memory
FROGGS no system perl involved, no deps to install... it is all in one file
rurban: this would be the perfect thing if we get this to work with moarvm (or p2 for that matter) 17:37
rurban and on windows you hardly get any library static only
I have now an explicit static target in p2, but the problems are the libraries, like pcre, libuv, libtomath, ... 17:38
FROGGS yeah, staticperl is/was unix-only... for windows I used sfx-archives, which are slooooooooow on first startup 17:39
rurban with games you need sdl, and many more
FROGGS rurban: moarvm uses static builds of libuv and libtomath
rurban interesting. I needed a special trick to move away static libs when building dynamic 17:40
FROGGS there is a static build for SDL too of course, but the default of the perl5 SDL module is to build dlls
rurban maybe a special path for static libs is better (slib?)
FROGGS all deps moarvm has (the ones in 3rdparty folder) are built statically
rurban hmm 17:41
diakopter rurban: how long does potion/p2 take to build? :)
FROGGS if the lib naming would be right there were no confusion... libname.a for static, libname.dll and libname.dll.a for dynamic
IIRC nwc10 can configure and build moarvm (-j26) in less than two seconds... 17:44
that is my box:
real\t0m4.639s
user\t0m13.444s
but it doesnt build the deps in that case I think
17:45 ssutch joined 17:47 ssutch joined
raiph Having a single binary packaging option would be a feature of a given backend, not something that NQP cares about, right? 17:51
s/packaging/packaging (eg ELF)/ 17:52
FROGGS I think a module an totally figure out what is needed, and do the packaging 17:53
but I guess for some backends it would be simpler and for other impossiblier :o)
18:12 cognominal joined
rurban potion-0.1 released: github.com/perl11/potion/releases/tag/v0.1 18:33
18:57 cognominal joined
rurban diakopter: 1m - 1m30 (dynamic + static) 20:26
jnthn Grr. I simply don't have the energy to look at extops wiring tonight. 20:31
If all else fails, I'll have a look at the weekend
One task I did realize needs doing: passing along our C/link stuff so we have the info handy to build the extops in the Rakudo repo. Suspect some moarvm --dump-config thing will suffice, if somebody wants to work on it. 20:33
Plus add an op to get at it, and then we can use it to supply NQP's -show-config too. In fact, maybe that obviates a --dump-config for Moar. 20:34
So if anybody wants to grab that task, go for it :) 20:35
20:48 lizmat joined
diakopter just did a health assessment at work 20:52
... apparently my blood pressure is in the "omg call an ambulance" range 20:53
jnthn o.O
Ouch
timotimo :(
japhb diakopter: I suggest a change of workplace. You know, just sayin'. 20:58
diakopter o_O 21:01
21:01 woolfy joined
PerlJam diakopter: perhaps you were clenching your fists when they took your blood pressure :) 21:01
diakopter yeah or micro hyperventilating like always 21:03
timotimo that's not a concept that is known to me 21:07
diakopter eh just hyperventilating very small breaths 21:08
timotimo oh
doesn't sound pleasant
21:13 grondilu_ joined
japhb timotimo, It's not. 21:22
Gives you a headache pretty quickly.
Does to me, at least. 21:23
timotimo i remember regular hyperventilating. it hurts in the hands 21:24
grondilu_ diakopter: don't pay too much attention to what doctors say. They're often wrong. 21:30
21:32 grondilu joined
diakopter grondilu: yeah, that's the problem 21:32
the doctors' clinic I attend measure blood pressure by hand with a gauge
and it varies wildly
the readins
when I have it measured by the machine at the pharmacy or grocery store, it shows I'm in a hypertensive crisis, always. 21:33
grondilu maybe you have a phobia of the measuring apparatus or somehting.
diakopter grondilu: so I go to the Dr and they measure it and say, well, lgtm
no, the nurses at the clinic are just clueless buffoons 21:34
grondilu noticed that the less he sees doctors, the better he feels.
21:55 BenGoldberg joined 22:14 Ben_Goldberg joined
[Coke] raises your hypertension with high ocular pressure! 22:59
23:10 ssutch joined 23:53 BenGoldberg joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!