»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg camelia perl6: ... | irclog: irc.perl6.org | UTF-8 is our friend!
Set by sorear on 25 June 2013.
00:01 araujo left 00:14 alc joined, alc left 00:17 ajr_ left 00:22 Guest4386 is now known as ponbiki
timotimo t.h8.lv/p6bench/2014-08-21-recent_stuff.html <- some benchmarks -> t.h8.lv/p6bench/2014-08-21-recent_rakudo.html 00:27
TimToady: it may be of special interest to you that in the array_set_xx benchmark the improvement is about 18x 00:29
TimToady \o/ 00:31
timotimo i wish everyone traveling to YAPC::EU now or in the next few hours or tomorrow before i get on irc safe and comfortable travels :) 00:32
TimToady firshir 00:33
00:37 Timbus left, Timbus joined
timotimo i'll probably be following the perl6 related streams closely 00:38
maybe even more :3
colomon timotimo: hmmm, the jansson thing works great in parrot 00:42
timotimo colomon: it'd be nice if you could upload a bit of code that reproduces the problem 00:43
with an example json file as well if possible :)
japhb \o/ For massive xx performance improvement
timotimo still kind of frustrating that in our bigger benchmarks, improvements are much, much weaker 00:44
especially the tiny bump the jit does is ... sub-optimal
but ... yeah, the param_* ops will have their jit impls soon-ish and that'll give us many, many more frames attempted to jit
TimToady is anyone else having trouble downloading those graphs? 00:46
00:47 thou joined
colomon timotimo: working on it. ;) 00:48
00:49 Tassilo joined
Tassilo hello 00:49
timotimo thank you
japhb Can you get the page at all? If so, do you have a script blocker running?
timotimo i'll probably have my nightly dose of sleep before i dive into your code
TimToady I haven't had trouble before 00:50
japhb Hmmm, they came up just fine for me.
And it's not like the JS code changes all that often. 00:51
colomon timotimo: github.com/colomon/p6-json-jansson -- the tests I've added in t/01-basics.t show the problem nicely.
TimToady oh, I'll bet I just have to restart ff
japhb (JQPlot appears to be essentially "stable" at this point.)
TimToady istr that it goes wonky when it upgrades under you 00:52
japhb Same with Chrome, sigh
00:52 thou left 00:55 telex left
TimToady nope, same deal 00:56
00:56 telex joined, muraiki left
TimToady just sits there flashing Transferring data from t.h8.lv... 00:56
00:58 woosley joined, muraiki joined
japhb Reload seemed to work, so it's not like the endpoint is dead. 00:59
00:59 rurban1 joined
timotimo colomon: all i can think of is that perhaps we're not doing the reference counting thing properly? 01:01
and after we get the one list, the other list gets get'd, but jansson's like "you didn't increase the ref count of the first list, so i'll overwrite it with the first" 01:02
not quite sure why it'd end up the same which, though ...
colomon timotimo: or why it would work in parrot? 01:03
01:03 rurban1 left 01:04 Tassilo left
timotimo well, it also doesn't help 01:04
i mean when i incref it
timotimo goes to bed 01:06
gnite
colomon \o
timotimo it's a very good find, i must say
01:10 BenGoldberg joined 01:24 dayangkun joined 01:28 FROGGS_ joined 01:31 FROGGS left 01:35 rindolf joined 01:40 klapperl_ joined 01:41 klapperl left 01:44 ilbot3 left 01:46 ilbot3 joined 01:52 xenoterracide joined 01:54 hypolin joined 01:55 hypolin left, Akagi201 joined 02:11 nbrown_ joined 02:14 araujo joined 02:34 BenGoldberg left 02:35 thou joined
dalek ast: e2a83ff | (David Warring [email@hidden.address] | S02-literals/misc-interpolation.t:
tests for RT #122583 list comprehension + $(...) interpolation
02:36
synopsebot Link: rt.perl.org/rt3//Public/Bug/Displa...?id=122583
02:37 nbrown_ left 02:40 thou left 02:45 noganex joined 02:47 noganex_ left 02:53 ivanshmakov left 02:56 ivanshmakov joined 03:11 raiph left 03:16 kaare_ joined 03:55 ivanshmakov left 03:58 ivanshmakov joined 04:07 kaare_ left 04:15 xenoterracide left 04:19 mattp___ left, [Sno] left 04:21 xenoterracide joined, kaleem joined 04:24 thou joined 04:29 thou left 04:36 kaare_ joined 04:38 SamuraiJack joined
lizmat commute& 04:44
04:44 lizmat left 04:55 xenoterracide left 04:58 anaeem1 joined 05:37 gfldex joined 05:38 Mouq joined 05:45 hoelzro left, alc joined, hoelzro joined 05:48 jtpalmer_ left, jtpalmer joined 05:51 SamuraiJack left 05:54 SamuraiJack joined 06:08 [Sno] joined, gfldex left
sergot morning o/ 06:10
06:12 thou joined 06:16 denis_boyun_ joined 06:17 thou left
kurahaupo evening \o 06:17
06:19 mattp___ joined 06:23 PotatoGim left
Mouq middle of the night \o 06:26
06:40 kaleem left
dalek kudo/nom: 47d315a | TimToady++ | src/core/LoL.pm:
optimize X operator, esp for ranges
06:47
kudo/nom: 1260360 | Mouq++ | docs/ChangeLog:
Add some ChangLog entires

I just did this from a quick perusal of the git logs; please add to this if you think something is missing.
06:48
kudo/nom: 7e23526 | Mouq++ | src/core/LoL.pm:
Merge branch 'nom' of github.com/rakudo/rakudo into nom
TimToady ^200 X ^200 now runs about 3 times faster
Mouq Woot! 06:50
TimToady though it's still slower than nested loops 06:51
since we have to allocate the entries
so at some point we'll want to optimize 'for @a X @b -> $a, $b {}' into 'for @a -> $a { for @b -> $b {...} }, only without the extra lexical scope, if we can get away with it 06:52
07:02 darutoko joined 07:04 zakharyas joined
moritz when Range { eager @result } 07:05
what does that do for infinite ranges?
TimToady that wants to generalize someday
07:05 takesako left
TimToady eager will just batch things up in that case, I think 07:06
it's only "mostly eager"
in any case, we probably want to generalize that test with more input from the compiler 07:07
but adding the same thing to X+ makes it maybe 30% faster 07:08
not as much gain since it's doing a lot of other work to call +
still, beats a kick in the head
if it turns out to be a problem, we can ask the range for it's max 07:09
moritz for 1..$n -> $i { # might be even faster to write a while loop with a native int 07:10
TimToady that's already optimized, I think
though perhaps not to native 07:11
but yeah, if you can prove max value, you can use natives
07:12 gtodd left 07:13 chenryn joined 07:14 gtodd joined
moritz $n is already a native int :-) 07:15
07:16 PotatoGim_ joined 07:19 PotatoGim joined 07:22 PotatoGim_ left 07:23 FROGGS_ is now known as FROGGS 07:24 kaleem joined 07:44 smls joined 08:00 thou joined 08:05 thou left 08:07 alc left 08:10 chenryn left, rindolf left 08:11 rindolf joined 08:12 bjz left 08:16 klaas-janstol joined 08:27 smls left 08:32 bjz joined 08:34 smls joined 08:38 jercos left 08:39 jercos joined 08:40 chenryn joined 08:45 fhelmberger joined 08:46 chenryn left 09:04 Mouq left
dalek kudo/nom: c0d8643 | TimToady++ | src/core/ (2 files):
pessimize .infinite X args more consistently
09:07
TimToady moritz++ for making me take a second look 09:09
timotimo o/
sergot hi timotimo
TimToady timotimo: your ^SCALE X ^SCALE benchmarks should run faster now 09:13
09:14 SamuraiJack left
nwc10 TimToady: are you experimenting with the xkcd 28 hour schedule? :-) 09:14
TimToady no, just the flipside of SAD :) 09:15
and what with all this staying at home, have more energy than usual
and since my wife is up visiting in Seattle, nobody to tell me to go to bed :) 09:16
timotimo i wonder how hard it'll be to build an optimization in code-gen or optimizer that turns an ^$foo X ^$bar -> $a, $b into a nested while loop internally while still properly handling redo/next/last 09:18
on the other hand, making it a non-nested loop wouldn't be terribly hard either, just code-gen an increment operation that's a tiny bit more complicated than just $i++ 09:19
TimToady surely only last is a problem
and you can solve that with loop labels 09:20
timotimo ah, hm.
TimToady redo and next go to INNER: while last goes to OUTER:
timotimo but if the outer loop itself has a label, we can't properly turn it into a nested loop, right? 09:21
or would it be enough to transfer that loop to the inner loop?
hm, not quite; you can pass labels to functions you call, right? 09:22
in that case, a "last" could appear with the then-inner loop's label and mess up our pretty scheme
TimToady you just split the label, if the loop as a whole is FOO:, then you end up with OUTERFOO: and INNERFOO: or some such 09:25
but yeah, dynamic can mess that up, unless you install a FOO handler that is smart
timotimo sounds like work :) 09:27
nwc10 TimToady: PASS (other than t/spec/S32-io/IO-Socket-Async.rakudo.moar and t/spec/S32-list/uniq.t) 09:28
TimToady well, and lock still flaps
nwc10 several of them SEGV on me if I build optimised. Timing issues. I believe jnthn is aware of this
TimToady there was a test that required an infinite list on the right side of X*, and metaop handled that, but X didn't (though there's no test for it) 09:29
should handle it now by just not eagerizing anything 09:30
GIGO, just slowly :)
moritz m: say (5 X* 1..*)[^10]
camelia rakudo-moar 7e2352: OUTPUT«5 10 15 20 25 30 35 40 45 50␤»
TimToady m: say (5 X 1..*)[^10] 09:31
moritz I'm sure that for a Haskellian, that's a legitimate way to generate this series
camelia rakudo-moar 7e2352: OUTPUT«This type cannot unbox to a native integer␤ in sub infix:<X> at src/gen/m-CORE.setting:8806␤ in block at /tmp/InmOavjVJV:1␤␤»
TimToady eh?
09:31 klaas-janstol left
timotimo moritz: could just Z or <<*<< it 09:31
TimToady oh, notmine
timotimo well, would have to use a range or repetition on the left
TimToady hypers are eager
timotimo (for Z)
oh!
of course they are
TimToady > perl6 09:32
> say (5 X 1..*)[^10]
5 1 5 2 5 3 5 4 5 5 5 6 5 7 5 8 5 9 5 10
so I fixed the X case, just hasn't recompiled yet
moritz aye 09:34
TimToady I suppose X should look for infinities after the first argument, and if the second value of the first argument can never be reached, warn
or basically, if there is more than one .infinity, you've written something sorta bogus 09:35
09:35 Ven joined
TimToady or anything that's not a single value left of a .infinite, for that matter 09:36
but coming up with a message that doesn't confuse the (obviously already confused) programmer is gonna be a trick 09:37
well, the message is probably not worth the memory it would consume 09:40
09:43 chenryn joined 09:48 chenryn left 09:49 thou joined 09:53 klaas-janstol joined, thou left 10:02 kurahaupo left 10:03 chenryn joined 10:17 jaffa4 joined 10:22 chenryn left 10:26 spider-mario joined 10:29 jaffa4 left 10:36 raiph joined 10:37 rindolf left 10:40 raiph left, raiph joined 10:46 lizmat joined 10:48 spider-mario left 10:54 dayangkun left
dalek kudo/nom: f1edae0 | (Elizabeth Mattijsen)++ | docs/ChangeLog:
Change of lives/dies_ok did not exceed a release

So we don't need to mention this in the ChangeLog
10:54
lizmat is sitting rather numbed behind the keyboard
10:55 spider-mario joined 10:56 Akagi201 left 11:02 telex left 11:03 Ven left 11:04 telex joined 11:07 dayangkun joined 11:13 woolfy joined
timotimo what's wrong? 11:17
lizmat our cat Whoopi died ~ 2 hours ago 11:21
she was with us for 12+ years
we also missed our flight and are now considering our options, as there is no identical flight available tomorrow 11:22
and on Saturday, we probably would only be able to make it for the lightning talks, maybe a bit earier 11:23
but, truth be told, right now I don't feel like doing anything much 11:24
11:25 SamuraiJack joined
timotimo oh no! :( 11:25
you have my sympathies 11:26
losing a beloved pet is terrible
only recently i've been reminded of our dog Nicki who passed away about 6 years ago 11:27
11:35 raiph left, raiph joined 11:36 dwarring left 11:37 thou joined
lizmat timotimo: thank you 11:41
11:42 thou left
moritz lizmat: my sympathies 11:48
lizmat moritz: thank you 11:51
11:56 raiph left
sergot lizmat: my sympathies as well 11:59
12:01 chenryn joined 12:04 bjz left 12:07 kurahaupo joined 12:14 kaleem left 12:24 raiph joined 12:28 bjz joined 12:32 Bucciarati left, betterworld left 12:33 Bucciarati joined, betterworld joined 12:37 kaare_ left
tadzik ohh 12:41
lizmat: hugs to both of you :( I lost my 13yo dog this summer too
lizmat sergot tadzik thank you
colomon lizmat: hugs 12:42
lizmat colomon thank you 12:43
12:48 kaleem joined 12:51 raiph left, chenryn left, kurahaupo left, raiph joined 12:57 bjz left 12:58 bjz joined 13:04 smls left, Akagi201 joined 13:08 prevost joined 13:13 ggoebel11111115 joined, anaeem1 left 13:15 anaeem1 joined 13:16 ggoebel11111114 left 13:19 donaldh joined, anaeem1 left 13:25 cooper left, thou joined 13:30 thou left 13:31 cooper joined 13:37 guru joined 13:38 guru is now known as Guest60219, jnap joined, Guest60219 is now known as ajr_ 13:44 kaare_ joined 13:51 jimmyz_ joined 13:52 jimmyz_ left
hoelzro lizmat: I'm really sorry to hear that 13:57
(sorry, a little late) 13:59
14:05 RandalSchwartz joined, thou joined
RandalSchwartz when I try to make rakudo with moar, I always get this error: 14:06
... perl -MExtUtils::Command -e cp perl6-m perl6
... gmake: write error
am I doing something wrong? It seems to stop the make
I never got that building the parrot backend
14:07 dakkar joined
moritz RandalSchwartz: disc full? or is anythinig wrong with your local 'perl6' (dangling symlink?) 14:13
jnthn lizmat: Really sorry to hear about Whoopi. And the flight, though that seems somehow insignificant in comparison... :( 14:14
masak lizmat: what jnthn said :/
RandalSchwartz nope
where is it trying to copy to?
jnthn RandalSchwartz: You may also have the misfortune of being on a version of OSX that ships with a busted make (yes, really). If you "make" again it actually completes successfully.
RandalSchwartz nope.
using gmake
built from ports 14:15
jnthn OK. The error matches exactly.
Does typing "make" again make it work?
RandalSchwartz and no, it repeats precisely the same way, although faster since things are already made.
jnthn Ouch. That is different.
I don't have a platform where it happens, but from the digging I've heard about, the error actually is spat out by make, rather than anything of ours. :/ 14:16
moritz and to answer it the question, it just tries to perl6-m to perl6 14:17
all in the current directory
rurban I also had the same error as Randal recently. Didn't investigate
timotimo it's also a thoroughly unhelpful error message :\ 14:20
14:21 pyrimidine joined, molaf joined
pyrimidine I had that error as well, but ‘make install’ worked fine 14:22
(on Mac OS X)
btyler_ likewise, also using system make, OSX 10.9 14:23
pyrimidine Also: irclog.perlgeek.de/perl6/2014-04-28#i_8647503 14:25
betterworld maybe 'make -s' helps
14:25 molaf left 14:26 dakkar left, dakkar joined
rurban I also got it on the darwin make only 14:27
pyrimidine testing ‘make -s’ 14:28
14:28 klaas-janstol left 14:29 tgt joined
tgt Has anyone managed to golf the OS X make error down to anything I could usefully submit to Apple? 14:30
moritz git clone [email@hidden.address] cd rakudo; perl Configure.pl --gen-moar --gen-nqp; make 14:31
pyrimidine hasn’t
moritz just 4 commands! :-)
even better, use the https:// URL, so that they don't need a github account 14:32
14:33 tassilo joined
pyrimidine ugh 14:34
RandalSchwartz repeats... "using gmake" 14:35
pyrimidine after git pull and make: Stage parse : moar(81319,0x7fff7c7e6310) malloc: *** error for object 0x7fcaf38cd0b0: pointer being freed was not allocated
tgt I think they'd prefer us to do all the work :( 14:36
14:36 dakkar left
moritz if both make and gmake exhibit the problem (and they aren't the same), maybe it's really a problem in our Makefile 14:36
pyrimidine moritz: seeing compilation error with a git pull now ^^^ 14:38
moritz pyrimidine: :(
14:38 tgt left
pyrimidine I’ll try bisecting 14:40
betterworld according to the sources, "write error" is printed when (g)make gets an error closing standard output, so I guess it is a problem with the terminal, and "make -s" might be a way around that 14:41
daxim_ code in p5, but question is grammar-centric: now that I have my AST <paste.scsys.co.uk/417408>, how do I go about transpiling it to a different programming language? any tutorials I should read? 14:42
14:43 treehug88 joined
moritz daxim_: depends on the target language 14:43
daxim_: there's tons of literator on code generation, though most assumes it's assembly or something similar 14:44
daxim_: if it's more like a high level language, I approach it more like I'd approach any other text document generation (think HTML...)
daxim_ no, also high-level like the dynamic P-languages
HTML is vague - do you mean use a templating system? 14:45
moritz yes, at least for the broad structure
and then you can have objects that know how to codegen-ify themselves 14:46
jnthn daxim_: Sometimes I take the approach of building up a tree structure that describes the target language, and that stringifies to the code.
moritz or you can have template functions that know how to codegen-ify a certain type of operation, and use some form of genericity to pick the right function for each operation 14:47
jnthn daxim_: In general, the goal is to push dealing with strings to the edges of the system.
daxim_ ooh, then I need to study the target languages in detail first
jnthn So that as little as possible remains to be decided by the time you turn it into a string.
timotimo oh hey jnthn, you've arrived near sofia? 14:50
jnthn Yeah, I'm at my hotel 14:51
Wasn't too bad a journey. Ended up doing a slightly bumpy propellor plane ride for the last bit. And then the taxi desk gave us a hotel to the wrong place! But made it :)
timotimo good to know 14:53
jnthn Nice to see some Cyrillic again :) 14:54
gtodd nice beaches? 14:55
:-) ... gotta head east for that 14:56
or south
felher + 14:57
jnthn yeah, red line 14:58
oops, ww
gtodd I have a mechanical question about the rakudo build ... in what "order" are the backends built? if I do --backends=moar,parrot --gen-moar --gen-parrot do moarvm.parrot get built then the nqp specific portions of perl6-m perl6-p ... ? 14:59
actaully better Q: I'd like to try to build and install everything with say a cron job and have failures skipped ... can I do that somehow?
[Coke] lizmat: very sorry about your cat. :| 15:00
gtodd (was trying to write a quick script that would remove a failed build from the configure stage and rerun) 15:01
[Coke] RandalSchwartz's problem happens every single time on os x, and has been doing for MONTHS.
lee_ i've seen that on OS X for a long time, but it didn't seem to prevent installing of perl6 15:02
15:02 chenryn joined
[Coke] I typically have to then "chmod a+x ./perl6", and things can continue; but if you're doing it as part of a bigger process, it kills you dead. 15:03
jnthn wonders if just tinkering with the Makefile a bit would make it somehow go away... 15:04
[Coke] I tried debugging the issue, but if you tell the os x make to use debug mode... it segfaults. (make)
jnthn o.O
[Coke] If you're bisecting, try before and after --backends change. Though I'm not sure how much Moar there was before that change. 15:05
jnthn: yah, it's sneaky. :|
15:05 tassilo left
[Coke] is doing the release on a linux box to avoid the issue. :| 15:06
[Coke] will get the release out this evening local time.
pyrimidine moritz: the bad commit that was causing the malloc: gist.github.com/cjfields/4ced0ba97eabbf82dfce 15:07
github.com/rakudo/rakudo/commit/c0...abe0920b02 15:08
BTW, it is not related to —enable-jit (get the error with and w/o it) 15:09
jnthn [Coke]: Remember to bump to latest MOAR_REVISION and NQP_REVISION, if that's not done already. :) 15:11
[Coke] need to do nqp release, so I'll just use the release versions, ayes?
jnthn Correct 15:12
I did the Moar release already
15:12 denis_boyun_ left
[Coke] jnthn++ 15:12
jnthn Tarball link is in the backlog of #moarvm 15:13
dalek kudo-star-daily: 064986e | coke++ | log/ (14 files):
today (automated commit)
[Coke] I bumped the parrot release to 670, as it had one or two things that sounded worth getting. 15:15
testing a moarvm run now, expecting no issues..
15:19 dayangkun left 15:23 chenryn left 15:25 chenryn joined 15:26 cognome left, cognome joined 15:31 cognome left
dalek p: cc6e693 | coke++ | tools/build/MOAR_REVISION:
bump MOAR to latest (recent) release
15:36
15:40 fhelmberger left, chenryn left 15:41 chenryn joined
jnthn pre-conf meat & 15:41
nwc10 om nom nom
15:41 ajr_ left, zakharyas left 15:45 guru joined, guru is now known as Guest69825 15:46 Guest69825 is now known as ajr_ 15:57 donaldh left, ajr_ left 16:07 ajr joined 16:08 ajr is now known as Guest85205, Guest85205 is now known as ajr_ 16:11 chenryn left 16:13 kaleem left
TimToady wonders if OS X comes with a default umask that prevents creating an executable, hence the need for an explicit chmod +x 16:14
though that wouldn't explain why a second make would work sometimes
16:15 chenryn joined
TimToady wonders how his last patch could possibly be causing a malloc problem 16:15
16:16 Ven joined
TimToady pyrimidine: what arch and hardware is the malloc problem? and can anyone else reproduce this? 16:17
.oO(they're probably all out partying...)
er, I meant OS--arch and hardware are sort of the same thing 16:18
pyrimidine TimToady: Mac OS X 10.9.4
TimToady ah...hmm...haven't there been malloc issues on OS X before? 16:19
anyway, that patch is all high-level code, so it must be tickling a bug somewhere else 16:20
colomon just got back from the podiatrist.
TimToady did you walk?
colomon to and from the car. ;)
pyrimidine TimToady: not sure. I bisected to that specific commit, but I agree re: bug tickling
colomon TimToady: painkillers haven't worn off yet 16:21
TimToady better living through chemistry
colomon so how can I try to duplicate this on OS X 10.8?
gtodd [Coke] something still gives this error on FreeBSD with clang ... paste.scsys.co.uk/416737?tx=on&...rmat+it%21 (I think it is when nqp-p is getting built) ... I recall some kind of fix being discussed that was possibly Makefile-ish (??) 16:23
TimToady colomon: see irclog.perlgeek.de/perl6/2014-08-21#i_9222686
gtodd [Coke]: all I could find was irclog.perlgeek.de/perl6/2014-06-10 ChoHag
TimToady or are you asking about the version skew?
colomon TimToady: trying to build f1edae0be91da7b9faaee0e247be5db276ac770c here now. 16:24
TimToady also some discussion back at irclog.perlgeek.de/perl6/2014-08-18#i_9205478 16:26
which btw predates my latest patch, so it's not about the patch at all, seems 16:27
sounds like building with LOLLY=1 fixed it for lizmat++, but that could be coinkydink
and it might just be moving a fencepost somewhere 16:28
gtodd so probably just wishful thinking on my part ... anyway it was something to do with parrot's version of nqp's dyncall so hopefully that bit of src/build strangeness will just get straightened out and the problem will go away magically in some future release :-) ... 16:30
TimToady can anyone on OS X confirm that the umask is something like 0022 and not something like 0044 or 0066? 16:32
btyler_ 0022, OSX 10.9.4 16:36
pyrimidine TimToady: on my MBP it’s 0022
same as btyler_
16:37 cognome joined
colomon TimToady: Building f1edae0be91da7b9faaee0e247be5db276ac770c on OS X 10.8.5 worked fine 16:37
0022 here as well
TimToady well, and that wouldn't explain it, since tools/build/create-moar-runner.pl does a chmod 0755 anyway 16:38
16:41 cognome left
TimToady oh, and it's the 'perl -MExtUtils::Command -e cp perl6-m perl6' that fails, anyway 16:41
16:43 pmurias joined
pmurias daxim_: which language are you transpiling? 16:43
16:45 [Sno] left, chenryn left
pmurias daxim_: what do you want to turn your toy language into? 16:46
TimToady now here's the funny thing: my bulid doesn't actually run that command, presumably because I already have a symlink from perl6 -> perl6-m 16:47
*build
could a previous build of perl6-p or perl6-j be putting a perl6 out there that can't be overwritten, if there's no symlink? 16:50
(my symlink dates from, like, February or so...) 16:51
16:56 takan joined, takan left 17:01 klaas-janstol joined 17:07 Rotwang joined 17:12 tgt joined, Ven left, klaas-janstol left 17:20 zakharyas joined 17:29 raiph left 17:32 raiph joined 17:34 ajr_ left
[Coke] TimToady: second make never works any times. (Os x) 17:35
gtodd: I don't know that one. is there a ticket?
TimToady is there already a perl6 file sitting there when it fails? 17:43
but then why does it fail that way, if it's just a file collision? 17:44
17:47 Ven joined 17:48 telex left 17:50 telex joined
TimToady okay, it doesn't do the m-runner-default because I said "make m-install", and m-runner-default is only called from "all" 17:51
I note that the m-runner-default rule doesn't do a CHMOD on the target; could the copy on OS X not be carrying over the executable permissions? 17:52
17:52 [j4jackj] is now known as j4jackj
TimToady lemme just add a CHMOD there, maybe it will solve the problem 17:52
17:52 kurahaupo joined
dalek p-js: 52c8e38 | (Pawel Murias)++ | src/vm/js/ (3 files):
Pass test 14 - while loops.

Implement nqp ops: islt_n, isgt_n, add_i, sub_i, preinc, predec, postinc, postdec. And partially the ops: while, until, repeat_while, repeat_until.
17:59
[Coke] to be clear, you getting a "make file: write error" ? 18:01
TimToady yes, but you said a chmod then fixes it, I thought
18:01 treehug88 left
TimToady can't hurt to have the chmod there, anyway 18:02
also told it to ignore errors on the cp, so maybe that will suppress the write error, but probably not 18:03
maybe it should try to rm what's there already 18:04
we don't know exactly what is giving make indigestion there
could even be something rotten in a variable
[Coke] "fix" is strong. I'm not saying it's dying on the chmod, I'm saying that after I do that, I can use the perl6 that's there. 18:05
(don't know exactly) exactly! ;)
TimToady a write error could be caused by an invisible char in the filename that OS X gets upset at, I suppose, but that's a long shot 18:07
18:07 Ven left
dalek p-js: bbd25fc | (Pawel Murias)++ | src/vm/js/QAST/Compiler.nqp:
Implement mod_n, isge_n, bitor_i, bitand_i, bitxor_i.
18:10
p-js: 001ca80 | (Pawel Murias)++ | src/vm/js/bin/run_tests:
Pass test 13.

Implement the ops bitor_i, bitand_i, bitxor_i, mod_n, isge_n.
TimToady or what if, for some reason, someone has a symlink from perl6 to something unwritable? that would produce these symptoms 18:11
an rm would fix that
FROGGS hmmm, I don't see any invisible char in my Makefile 18:12
TimToady you on OS X? 18:13
FROGGS no
18:13 guru joined
FROGGS I looked at the templates and my generated Makefile on linux 18:13
18:14 guru is now known as Guest64656
FROGGS from #moarvm: <rurban> ExtUtils::Command cp is usually the culprit, dropping the -x perm 18:14
TimToady in the old days, we had to make our own symlink to perl6, so if someone remade in an old directory that had a symlink to a root-owned directory, it could cause this
18:14 Guest64656 is now known as ajr_
TimToady I will put in a rm as well 18:15
dalek kudo/nom: 79fc622 | coke++ | docs/ChangeLog:
minor updates
18:16
kudo/nom: 3bb45b7 | TimToady++ | tools/build/Makefile-Moar.in:
make sure we don't have perl6 symlink, chmod 755
18:20
TimToady anyone on OS X care to try that?
unfortunately, requires a reconfig
18:20 prevost left
TimToady then a make all 18:20
Randa 18:23
RandalSchwartz: I may have fixe it in HEAD
*ed
huf for a second there i was really confused... didnt think you spoke that language :D
18:24 cognome joined
TimToady it's amazing what happens when your corpus colossum misfires and you type Return instead of Tab 18:24
rurban I'll try 18:25
TimToady thanks 18:26
dalek p: 4a8fe9f | coke++ | VERSION:
bump VERSION to 2014.08
tgt The write error still happens here. I ended up with an executable perl6 this time though. 18:31
TimToady huh
perhaps an educational message is in order 18:32
tgt The write error also happens when just running m-all.
TimToady which command does it happen on? 18:33
tgt After M_RUNNER
colomon wonders if this is the same write error he's been getting seemingly forever…. 18:34
RandalSchwartz TimToady - cool! 18:35
running fetch-n-make now 18:36
btyler_ still have a write error here as well
TimToady maybe M-RUNNER needs the chmod too
rurban me too 18:37
RandalSchwartz different write error now
colomon is this the write error in make install?
RandalSchwartz perl -MExtUtils::Command -e chmod 755 perl6
colomon after bin/perl6 has been installed? 18:38
RandalSchwartz happens because of that or shortly after that. hard to tell
tgt The write error happens after the last thing run.
RandalSchwartz that's the line just before the error
perl6 had just been updated 18:39
rurban I get it with osx make and a new gmake from macports
colomon When I spent time trying to debug this months ago, I concluded it was happeneing after the last thing run was done.
RandalSchwartz yeah - I'm using gmake from ports
colomon and yes, it's still happening even with the latest
(latest rakudo, that is)
rurban I have no idea how to repro it easily (other than make clean) 18:40
colomon apologizes to TimToady++ for not reporting this earlier -- this bug has been happening forever and a day, and so didn't think it could be what TimToady was looking for. 18:41
PerlJam greetings #perl6
dalek kudo/nom: bc6a32a | TimToady++ | tools/build/Makefile-Moar.in:
insurance on mode of perl6-m as well
rurban yes, maybe that
18:42 Sqirrel joined
rurban repro with rm perl6.moarvm; gmake ... gmake: write error 18:43
[Coke] the system make is gnu make also, same problem, RandalSchwartz.
TimToady let's keep our eyes peeled for other things like this that we've just put up with, but we wouldn't want to inflict on Real Users 18:44
18:44 SamuraiJack left
TimToady we need to start minimizing the cultural hurdles to adoption 18:44
well, s/start/continue/ 18:45
colomon TimToady: to be fair, if I could have figured out how to fix that, I would have months ago. It's a major PITA for me. 18:47
And it's still happening with bc6a32acb7afbcfa32b92f868609ac9e4814c3e9
18:48 [Sno] joined
colomon when I did look into it, I think the conclusion was it was most likely a bug in OS X's make. 18:48
PerlJam Is someone building R* for macports or fink or whatever is commonly used these days? 18:49
colomon Hmmm…. irclog.perlgeek.de/perl6/2014-01-20
sorry, irclog.perlgeek.de/perl6/2014-01-20#i_8144630
FROGGS I think this is relevant: openbsd.7691.n7.nabble.com/gmake-qu...81450.html 18:50
colomon also irclog.perlgeek.de/perl6/2014-04-28#i_8647503
FROGGS: this, then? www.mail-archive.com/commits%40crat...02534.html 18:53
colomon has no idea how to go from that (which seems relevant) to something useful 18:55