»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'p6: say 3;' or rakudo:, std:, or /msg camelia p6: ... | irclog: irc.perl6.org | UTF-8 is our friend!
Set by masak on 12 May 2015.
00:01 bpmedley left 00:07 rangerprice joined
rangerprice Hi 00:09
i need some help
i'm on a forum, and the subject is general discussion. Someone tell me "What i should learn Perl instead of Python ?" and wants concrete a answer. 00:12
and wants a concrete answer* 00:13
psch doesn't know an answer to that question
konobi there's nothing concrete
_but_ i would say that perl tooling for authors is much better in my experience 00:14
in terms of deployment too
00:15 firstdayonthejob left
raoulvdberge and perl is more fun! altough that is subjective 00:15
konobi things like overuse of things like decorators in python are also a turn off 00:17
psch .oO( maybe the best reason is that pythonistas want to convince you their language is better... ) 00:24
00:27 Skarsnik left 00:28 rangerprice left 00:31 robinsmidsrod left 00:35 BenGoldberg joined 00:36 Juerd left, Juerd joined 00:37 vendethiel joined 00:40 aborazmeh joined, aborazmeh left, aborazmeh joined, adu joined 00:42 Juerd left
vendethiel o/ 00:43
00:44 Juerd joined 00:45 tokuhiro_ joined
dalek p: 19a7c2e | peschwa++ | src/vm/jvm/runtime/org/perl6/nqp/tools/EvalServer.java:
Print the StackTrace to the client.

Other solutions like e.g. printing only the classname to the client or somesuch end up printing out of order and even duplicate information when running spectest, so this seems least weird.		 00:45
psch now when does camelia rebuild..? 00:46
dalek kudo/nom: ff9e6de | peschwa++ | src/ (4 files):
Teach JVM about the BOOTSTRAP Binder.

Also use it for its trial_bind implementation and remove org.perl6.rakudo.Binder.trialBind.		 00:48
00:49 tokuhiro_ left 00:51 sprocket joined 00:53 BenGoldberg left 00:55 BenGoldberg joined 00:57 robinsmidsrod joined 01:07 Ven joined 01:08 Juerd left, vendethiel left 01:10 Juerd joined 01:19 bpmedley joined, Ben_Goldberg joined 01:21 raiph left, Ven left 01:22 BenGoldberg left 01:24 Ven joined 01:26 Juerd left
colomon I have a moar process running using up 9.43 GB of memory. I don’t think it’s attached to any terminal window? 01:27
01:28 Ven left
colomon 99% CPU, too. 01:28
01:30 raiph joined
TimToady howdy from Hackers 01:42
raoulvdberge hi! 01:45
01:53 BenGoldberg_ joined 01:56 Ben_Goldberg left
ugexe perl6 -e 'my $x = shell("ls", :out); $x.out.close;' 02:00
The spawned process exited unsuccessfully (exit code: 141)
im aware of the recent change, but doesn't that seem wrong?
the jvm does not do that fwiw
ah it does... so it must have to do with how soon .out is closed 02:01
or is intended behavior now that .out will have to be read entirely before closing it? 02:02
sprocket question about slurp 02:03
i’m trying to read in a file using slurp (/proc/cpuinfo) but it doesn’t return any lines
however, when i use $FILENAME.IO.lines, it gives the correct result 02:04
gist.github.com/cspencer/5b7ec50a4917c0e2a2f6
there’s a gist of the code
am i slurp()’ing incorrectly?
ugexe or how do i launch a proc i expect to fail but want to examine the exitcode? 02:14
try/catch? 02:17
AlexDaniel ugexe: $proc.exitcode ? 02:18
02:18 b2gills joined
ugexe hmm calling exitcode before the .out.close makes it work, so screw it 02:19
AlexDaniel ugexe: exitcode lies if you call it before .out.close 02:20
ugexe i *think* that might have changed the other day with a Proc related pr
AlexDaniel well, two bug reports are still there #126561 #126037 02:21
ugexe ah yeah a bogus command gives exitcode 0 while its still open
02:22 raiph left
ugexe looks like on a non-zero exit code it throws an exception and the Proc 02:22
02:24 nys left 02:36 BenGoldberg_ left, BenGoldberg_ joined
timotimo huh, how can slurp be broken and our test suites don't catch it :| 02:39
ugexe i guess im not sure how you are now supposed to launch a process you expect to fail where you want to also capture its output and exitcode once its finished 02:41
but the process might also succeed 02:42
perl6 -e 'my $proc; try { my $x = shell("ls", :out); $x.out.close; CATCH { default { $proc = $_.proc } } }; say $proc.exitcode' # exitcode of 141 now
02:43 sprocket left 02:44 Juerd joined, ilbot3 left
AlexDaniel ugexe: my $proc = run(‘xxx’, :out); my $out = $proc.out.slurp-rest; $proc.out.close; my $exitcode = $proc.exitcode; 02:44
ugexe: and what about this? 02:45
ugexe The spawned process exited unsuccessfully (exit code: 127) 02:46
02:46 ilbot3 joined, tokuhiro_ joined 02:48 sufrostico left, sufrosti1o left
ugexe and i'd want it to not print anything to the terminal, just capture the exitcode 02:48
AlexDaniel ugexe: then you have to do something with stderr perhaps 02:49
e.g. capture it or redirect to /dev/null
ugexe doesnt seem to help
i already redirect it
this is a change from the last day or two
AlexDaniel hmmm I don't have the latest version 02:50
02:51 tokuhiro_ left
ugexe i have to wonder if Proc shoudl really decide what is a failure... i thought some whacky programs use non-0 exit codes for success 02:52
AlexDaniel only shell stuff but you are not going to use these anyway 02:54
ugexe even then if i CATCH it, $promise.keep($proc) just throws it when it re-sinks it
AlexDaniel I have not seen any normal program that would do that
ugexe so you cant close .out before its done or it will be an exception. and if you try to get the exit code before closing it its wrong. and if you catch the proc in an exception it will just rethrow if you sink it again. i dont see how you are supposed to handle a non-zero exitcode process given this 02:57
maybe it should only throw an exception the first time its sunk? 02:58
AlexDaniel well, on my setup slurp-rest blocks until it ends 03:01
and then you can get exitcode. What happens if stdout is closed earlier? I have no idea 03:02
(I mean what if it is closed by the process itself. Not sure if that's even possible) 03:03
ugexe nickl@li685-90:~/perl6/http-useragent$ perl6 -e 'my $proc = run("xxx", :out); my $out = $proc.out.slurp-rest; $proc.out.close; my $exitcode = $proc.exitcode;' 03:04
The spawned process exited unsuccessfully (exit code: 127)
in block <unit> at -e:1
but i'd expect what you said to happen 03:08
AlexDaniel ugexe: indeed, it seems to be broken if you run weird stuff 03:10
for example “xxx”
say run(‘xxx’).exitcode;
prints -1 03:11
which means that the program did not even start
and it's ok, but why is it 127 when you attempt to get the output? 03:12
sounds like a bug
ugexe but say i do something else like git clone repo exiting-folder/, it would do the same thing
existing 03:13
AlexDaniel ugexe: what is the command that you are trying to run? 03:14
ugexe: because that's weird
I get 128 :) 03:15
which is correct, I think
ugexe before i could do `git clone $repo $exiting-directory` which would clone the repo, but if it existed it would exit 127. because i expect this to happen a lot, i use .out to hide the output of the git command, and then look at the exit code and decide if i should actually fetch/pull or throw a real exception 03:16
128 yea probably
AlexDaniel ugexe: say run('git', 'clone', '-q', 'github.com/TiMBuS/Net--IRC.git', 'repo2', :err('/dev/null')).exitcode; 03:19
for me it is 128 if the directory already exists or 0 if everything is OK
03:20 liufeng left
ugexe but i need to capture out as well 03:22
which means i need to close it
so this throws the exception: 03:23
'my $proc = run("git", "clone", "-q", "github.com/TiMBuS/Net--IRC.git", "repo2", :out, :err("/dev/null")); say $proc.out.lines; $proc.out.close; say $proc.exitcode;'
after its been cloned already
03:24 kid51 left
AlexDaniel ugexe: with or without -q, it seems like git clone prints only to stderr 03:26
and stdout is empty in any case
ugexe right, but i do use it for more than that one command 03:27
i seem to be able to catch the proc and still use it if i wrap all the stuff in a try instead of just the run or the .out.close i see. so i'll be able to figure it out now yay 03:29
03:32 cognominal left 03:33 rurban joined 03:36 yqt left 03:46 telex left 03:50 telex joined, liufeng joined
zengargoyle sproket: i would guess linux's /proc filesystem is special with regards to slurp. i've seen mention in places of ways you can/can't read from /proc files. not sure of slurp's implementation. would have expected .open.slurp-rest to work but it doesn't... 03:52
03:53 telex left
zengargoyle you can probably only do the equivalent of getline on /proc files or something like that. 03:53
03:53 BenGoldberg_ is now known as BenGoldberg, uruwi joined 03:54 telex joined
BenGoldberg m: say + .lines; 03:55
camelia rakudo-moar ff9e6d: OUTPUT«Method 'lines' not found for invocant of class 'Any'␤ in block <unit> at /tmp/YLRehYqJyt:1␤␤»
BenGoldberg m: say +lines;
camelia rakudo-moar ff9e6d: OUTPUT«20␤»
04:15 advwp left 04:32 raoulvdberge left 04:37 MadcapJake_ joined 04:38 MadcapJake left, MadcapJake_ left 04:39 MadcapJake_ joined 04:40 MadcapJake_ left 04:41 MadcapJake_ joined, MadcapJake_ left 04:42 MadcapJake joined, MadcapJake is now known as MadcapJake_, MadcapJake_ left 04:43 MadcapJake joined, MadcapJake is now known as MadcapJake_, MadcapJake_ left 04:44 MadcapJake joined, MadcapJake is now known as MadcapJake_ 04:45 MadcapJake_ is now known as MadcapJake, advwp joined 04:48 tokuhiro_ joined
ugexe the secret to getting the old Proc behavior back straight up was just `$proc does role { method sink(|) { } }`... now i feel dumb 04:52
04:53 tokuhiro_ left 05:07 uruwi left 05:13 liufeng` joined 05:16 liufeng left 05:22 AlexDaniel left 05:34 zengargoyle left 05:35 zengargoyle joined
dalek kudo/nom: 9e4902f | TimToady++ | src/Perl6/Actions.nqp:
failed S/// should return original string

fixes RT #126476		 05:40
ast: ffe19e0 | TimToady++ | S05-substitution/subst.t:
RT #126476 fixed, untodo
05:43 adu left 05:53 ribasushi left 05:55 ribasushi joined 05:57 sprocket joined
Timbus so when using \r\n delimited lines, somesing in zee .lines Supply method causes it to buffer the last line until a new line comes through. 06:06
06:08 khw left
Timbus also, .chars-supply does not ever seem to run my taps and I'm not sure if that's my fault? But if I put an empty .do supply on it, stuff works. one sec while I paste an example.. 06:09
gist.github.com/TiMBuS/3d3ca10471e175c8358e 06:17
06:19 sprocket left 06:31 aborazmeh left 06:32 skids left
dalek kudo/nom: e1e03e6 | TimToady++ | src/Perl6/Grammar.nqp:
don't typocheck types that actually exist
 06:37
06:38 tokuhiro_ joined 06:39 liufeng` left 06:41 rurban left 06:43 tokuhiro_ left 07:14 kmel joined
kmel hello 07:42
07:44 tokuhiro_ joined 07:48 tokuhiro_ left 08:01 rindolf joined
moritz \o 08:09
kmel \o 08:10
moritz www.mikeash.com/pyblog/friday-qa-2...-hard.html "Swift has settled on the grapheme cluster as its idea of a "character."" 08:14
08:16 kmel is now known as hankache, darutoko joined 08:18 pmurias joined
pmurias hi 08:18
08:19 hankache left 08:20 kmel joined, hankache joined, kmel left, hankache left 08:21 hankache joined
moritz hi pmurias 08:24
moritz is glad that Task::Star builds again with current rakudo 08:25
08:27 virtualsue joined 08:30 leont joined 08:31 firstdayonthejob joined 08:36 spider-mario joined
dalek p: 02f0f5e | (Pawel Murias)++ | src/vm/js/nqp-runtime/ (3 files):
[js] Support associative access on type objects.
 08:36
p: ac51c16 | (Pawel Murias)++ | src/vm/js/ (2 files):
[js] Hack around the lack of typed attributes support to unbreak nqp-js-on-js.
konobi pmurias: how goes? 08:37
pmurias konobi: hi 08:43
planning to work this week on the getting the failing nqp-js tests passing 08:44
08:45 RabidGravy joined
konobi cool 08:45
i should look again... was trying to get moarvm working on illumos of late
08:47 _28_ria left
pmurias haven't really worked on the packaging the generated code up for npm yet, as trying to build up monentum first so that I don't get bogged down and discouraged by Makefiles ;) 08:47
08:47 _28_ria joined
pmurias konobi: if you have any questions about the nqp-js code base, feel free to ask as I should likely document/clean up any confusing stuff 08:48
konobi pmurias: i suppose a lot of it comes down to the split of responsibilities for each thing generated 08:49
stmuk www.youtube.com/watch?v=R4_rBlodftU 08:52
in Spanish oddly enough
pmurias I'm thinking that the way we have a mechanism that resolves NQP/Perl 6 modules to file paths, something like that should resolve to npm packages 08:53
I should look into the CURLI etc. stuff more 08:54
08:59 _28_ria left 09:00 _28_ria joined 09:07 Ven joined 09:17 vendethiel joined 09:28 xfix joined
RabidGravy This is about Java, but it should probably be read by everyone foxglovesecurity.com/2015/11/06/wha...erability/ 09:29
09:31 Psyche^ joined 09:35 Psyche^_ left 09:37 Ven left 09:38 Begi1115 joined 09:42 Ven joined 09:45 tokuhiro_ joined
Woodi hi #perl6 :) 09:47
RabidGravy marning!
Woodi RabidGravy: "Nearly two years ago, we decided we wanted 0-day in WebSphere application server." ;)
moritz I tried to skim the article for the actual vulnerability, but didn't found it very clear 09:48
something with unsafe deserialization
RabidGravy yeah
moritz but why do they give the user access to serialized objects? 09:49
09:50 tokuhiro_ left
RabidGravy because "dumb" 09:51
09:52 Ven left
RabidGravy www.masteringperl.org/2012/12/the-s...-problem/, slides from a talk I gave on this *moree than ten years ago* gellyfish.co.uk/old/yapc_talk/ 09:52
09:52 Ven joined 09:54 BenGoldberg left
moritz one could also sign the data, and verify the signature to ensure that it hasn't been tampered with 09:55
as Mojolicious does for session cookies
Woodi becouse "Java LOVES sending serialized objects all over the place." :) 09:59
RabidGravy however in an RPC or data sharing context that may give a false sense of security
09:59 _mg_ joined
moritz aye 09:59
nine "Every application server comes with its own bundle of libraries, even worse, every application you deploy on the server often comes with its own set as well" 10:05
That's why after all the hype of local::lib, containers and whatnot, I'm still a fan of using system perl and system libraries. 10:06
moritz it's a different situation if you make in-house applications 10:08
if you want to update a library, you just rebuild the application and deploy it 10:09
RabidGravy quite a while ago I worked at an ISP where the provisioning system use three different versions of Perl :(
moritz (provided build and deployment are somewhat automated)
RabidGravy: hey, they had a provisioning system!
pink_mist my alma mater just announced they're switching away from Java as their primary language for teaching ... I kindof wish they'd switch to perl6, but I don't think there'd be enough competence yet to teach that :/
10:10 rurban joined
pink_mist seems they've chosen Scala 10:10
nine moritz: that assumes that I actually know that I need to update a library. Which again means that I'd have to follow changelogs of 100s of modules we use. I'd much rather have our Linux distribution do that for us.
moritz nine: agreed, if you have that option; otherwise www.versioneye.com/ exists, at least 10:11
RabidGravy though for a lot of even quite large applications there are only a relatively few direct dependencies and the rest are dependencies of those dependencies 10:13
10:14 espadrine joined
[Tux] T::C seriously broken after recent commit. 10:26
10:26 hankache left
lizmat [Tux] ah? perhaps my :k/:v etc work ? 10:27
moritz or the CRLF grapheme?
[Tux] I think *my* :k work committing in the wrong branch
lizmat :-)
[Tux] yup, much better 10:28
RabidGravy yay!
[Tux] test 50000 35.260 35.148 10:33
test-t 50000 18.171 18.059
tadzik :o
moritz wow 10:35
RabidGravy half 10:37
Woodi ok, I see numbers second time but what generate them ?? ;) 10:38
arnsholt That's a really nice speedup! What's changed between the two versions of the code? 10:40
RabidGravy harder whipping of the hamsters 10:41
lizmat afk for a few hours 10:42
Woodi btw. today I finished very interesting article about history of SmallTalk. one of the features (early 70s or earlier) was safe serialization ability :) and there is so many interesting things to note that I do not put them becouse someone/something will kick me from the chanell :) 10:46
link is: gagne.homedns.org/~tgagne/contrib/E...oryST.html
10:47 Ven left
RabidGravy :) 10:48
Woodi just one note: at the end of article is chapter "Coda", just random things from the past and about future... *THAT* is why I am realy sad IT "oldtimers" do not do blogs or more such perspective view talks... 10:49
[Tux] Woodi, the test suite for Text::CSV 11:00
arnsholt, split (@needles, ...) instead of split (rx{...}, ...)
11:01 diakopter joined
pmurias nine: re follow the changelogs, why not just update to the latest version? 11:02
nine pmurias: that's the exact opposite of what people using local::lib or containers want. They use exactly specified versions because they tested those. New versions break things far too often. 11:04
pmurias using the version the distribution provides will also break things 11:05
nine Only on upgrades. Security fixes are mostly backported 11:08
pmurias nine: I really dislike distributions doing things like patching perl 11:09
when debian perl prints out that a few hundred patches are applied I'm all "Leave my perl alone you bastard"
the way debian handled node.js was really horrible 11:11
nine Maybe debian is just not the prime example :) 11:13
arnsholt The thing that annoys me the most is Ubuntu splitting core Perl into several packages, so that perldoc isn't installed by default, for example 11:14
RabidGravy a lot of LInux packagers do that 11:17
waaaay annoying
11:20 _mg_ left 11:28 TEttinger left, MARTIMM joined
spider-mario luckily, arch linux doesn’t 11:31
MARTIMM reading through S10 section Autoloading; is the use of CANDO() already implemented in rakudo?
spider-mario even most of Qt4 is just in one big “qt4” package 11:32
including the development tools :-°
11:33 Skarsnik joined
stmuk but embedded systems! blah blah 11:36
11:36 Sqirrel left 11:37 iH2O joined
stmuk in an "enterprise" vendor backported security patches do add "value" 11:38
11:39 iH2O left 11:40 Ven joined, ^elyse^ joined
moritz MARTIMM: no 11:44
MARTIMM: and I don't think we will for christmas; we have FALLBACK though
11:48 Ven left, Zoffix joined, Zoffix left, Zoffix joined
MARTIMM moritz: Can I find that in the documents 11:48
timotimo o/ 11:49
MARTIMM moritz: found something in S12, thanks 11:51
timotimo moritz: i don't think signatures on the serialization blobs would help this issue at all 11:52
moritz: because you use serialized blobs for unauthorized commands against the server and the general setup requires everyone to have access to the programs that generate these blobs as commands 11:53
what would help is a whitelist of classes that any given serialized blob is allowed to generate 11:54
because for a command like "hey what's your version?" why accept anything but List, Map, String, Int, Float? 11:56
(and at that point you have a complicated, binary version of JSON)
if there was some way to only allow deserialization of things that don't have a "post-deserialize" method, that'd also be A Thing 11:59
moritz timotimo: yes, I was commenting on the use case of session storage, not the RPC use case 12:02
timotimo ah, yes
for that, signature is a good idea
12:16 MARTIMM left 12:23 weihan joined 12:24 virtualsue left 12:26 Ven joined 12:27 weihan left 12:28 weihan joined 12:34 weihan left 12:35 kid51 joined 12:37 Ven left, weihan joined 12:39 Ven joined 12:48 ChoHag left 12:50 raoulvdberge joined, ChoHag joined 12:51 weihan left 12:53 Ven left, Ven joined 12:54 weihan joined 12:55 weihan left 12:59 weihan joined 13:04 hankache joined 13:08 dayangkun joined 13:09 dayangkun left, dayangkun joined, Sqirrel joined 13:10 dayangkun left, dayangkun joined
RabidGravy If one had a thing that indicated either "Optional" or "Mandatory" what would you call it 13:13
psch neccessity marker? 13:14
13:14 Ven left
RabidGravy Hmm, bit wordy 13:15
psch need-it bit? :P
RabidGravy specifically if I had "enum Something <Mandatory Optional>" what would work for Something ;-)
psch m: A: for 1 { A.Str.say } 13:16
camelia rakudo-moar e1e03e: OUTPUT«Label<139845112624704>␤»
13:16 Ven joined
psch m: A: for 1 { A.chars.say } 13:16
camelia rakudo-moar e1e03e: OUTPUT«Method 'chars' not found for invocant of class 'Label'␤ in block <unit> at /tmp/pUT7Y3acaN:1␤␤»
psch m: A: for 1 { A.perl.say } 13:17
camelia rakudo-moar e1e03e: OUTPUT«Label.new␤»
13:17 raiph joined
psch RabidGravy: english.stackexchange.com/questions/110394/ 13:18
RabidGravy hahaha, Optionality 13:19
Ordinality is a weird metaphor really
RabidGravy goes with Optionality as it's better than sitting here worrying about it
cheers 13:20
lizmat Mandatum
lizmat likes Optionality
works with Whipuptitude and Manuplexity
RabidGravy :)
13:21 colomon left
psch m: A: for 1 { for 1 { A.last }; CONTROL { default { .perl.say } } } # this used to segfault afair 13:21
camelia rakudo-moar e1e03e: OUTPUT«chars requires a concrete string, but got null␤ in block at /tmp/Mtjls394hH:1␤ in block at /tmp/Mtjls394hH:1␤ in block <unit> at /tmp/Mtjls394hH:1␤␤»
psch now it just doesn't know how to .chars the Label
RabidGravy psch, because of the above I just discovered one can apply a role to a Label like that, haven't a clue what you could do with it
timotimo could very well be; .chars on a null thingie used to just segfault
psch m: A: for 1 { for 1 { A.last }; CONTROL { default { .Str.say } } } 13:22
camelia rakudo-moar e1e03e: OUTPUT«concatenate requires a concrete string, but got null␤ in block at /tmp/QX0OWZbKlk:1␤ in block at /tmp/QX0OWZbKlk:1␤ in block <unit> at /tmp/QX0OWZbKlk:1␤␤»
psch m: A: for 1 { for 1 { A.last }; CONTROL { default { .WHAT.say } } }
camelia rakudo-moar e1e03e: OUTPUT«(X::AdHoc)␤»
psch m: A: for 1 { for 1 { A.last }; CONTROL { default { .message.say } } }
camelia rakudo-moar e1e03e: OUTPUT«concatenate requires a concrete string, but got null␤ in block at /tmp/r0enFXXACy:1␤ in block at /tmp/r0enFXXACy:1␤ in block <unit> at /tmp/r0enFXXACy:1␤␤»
psch vOv
it's clearly a miswrapped CX::Last
really though, something about exception handling and labels is just utterly opaque to me
13:23 dayangkun left
psch as in, the P6-level seems to bypass the NQP layer and is handled directly in the vm 13:23
timotimo i wonder if the CONTROL block wraps it for you?
psch m: use nqp; A: for 1 { for 1 { A.last }; CONTROL { default { say nqp::getextype(nqp::decont($_)) } } } 13:24
camelia rakudo-moar e1e03e: OUTPUT«getexcategory needs a VMException␤ in block at /tmp/zdboewrxj1:1␤ in block at /tmp/zdboewrxj1:1␤ in block <unit> at /tmp/zdboewrxj1:1␤␤»
psch m: use nqp; A: for 1 { for 1 { A.last }; CONTROL { default { say nqp::getextype(nqp::getattr(nqp::decont($_), '$!ex', Exception)) } } } # uhh 13:25
camelia rakudo-moar e1e03e: OUTPUT«Cannot unbox a type object␤ in block at /tmp/cmaHQEfveO:1␤ in block at /tmp/cmaHQEfveO:1␤ in block <unit> at /tmp/cmaHQEfveO:1␤␤»
psch timotimo: i think the X::AdHoc isn't even the CX::Last i want, maybe..?
timotimo you mean something else could potentially be thrown in there? 13:26
13:26 dayangkun joined
psch yeah, something like that 13:27
maybe it breaks differently too, as mentioned i don't see through that at all :/
m: for 1 { last; CONTROL { default { .perl.say } } }
camelia rakudo-moar e1e03e: OUTPUT«CX::Last.new␤»
psch labels man
RabidGravy I don't suppose that anyone has compiled a list of "customary" extensions to the S22 META spec as might be used by various tools?
psch m: A: for 1 { A.last; CONTROL { default { .perl.say } } }
camelia rakudo-moar e1e03e: OUTPUT«chars requires a concrete string, but got null␤ in block at /tmp/zCF0XdToYG:1␤ in block <unit> at /tmp/zCF0XdToYG:1␤␤»
13:30 AlexDaniel joined
psch j: A: for 1 { for 1 { last A }; CONTROL { default { .perl.say } } } 13:32
camelia rakudo-jvm e1e03e: OUTPUT«X::AdHoc.new(payload => "Died")␤»
13:32 _mg_ joined
psch m: say X::AdHoc.payload 13:32
camelia rakudo-moar e1e03e: OUTPUT«Invocant requires an instance of type X::AdHoc, but a type object was passed. Did you forget a .new?␤ in block <unit> at /tmp/arMb4vz58J:1␤␤»
psch m: say X::AdHoc.new.payload
camelia rakudo-moar e1e03e: OUTPUT«Unexplained error␤»
psch j: say X::AdHoc.new.payload
camelia rakudo-jvm e1e03e: OUTPUT«Unexplained error␤»
13:35 ilbelkyr left 13:36 ilbelkyr joined 13:37 dayangkun left
psch r: use nqp; A: for 1 { for 1 { last A; }; CONTROL { default { say nqp::getextype(nqp::getattr($_, Exception, '$!ex')) } } } 13:37
camelia rakudo-{moar,jvm} e1e03e: OUTPUT«4112␤»
psch getattr is hard :l
13:37 pecastro_ left
psch at least they agree its a labeled last 13:37
13:38 Ven left 13:39 Ven joined
Skarsnik Hello, is there a way to get a counter in a for loop without setting/inc a var for it? 13:39
psch Skarsnik: do you have an example (maybe in another language) that shows what you mean? 13:40
lizmat m: my @a = <a b c d e>; for @a.kv => $index, $letter { dd :$index, :$letter } 13:41
camelia rakudo-moar e1e03e: OUTPUT«5===SORRY!5=== Error while compiling /tmp/GklweIJCls␤Variable '$index' is not declared. Did you mean any of these?␤ &index␤ &rindex␤␤at /tmp/GklweIJCls:1␤------> 3my @a = <a b c d e>; for @a.kv => 7⏏5$index, $letter { dd :$index,…»
lizmat m: my @a = <a b c d e>; for @a.kv -> $index, $letter { dd :$index, :$letter }
camelia rakudo-moar e1e03e: OUTPUT«Str letter = "a"␤Int index = 0␤Str letter = "b"␤Int index = 1␤Str letter = "c"␤Int index = 2␤Str letter = "d"␤Int index = 3␤Str letter = "e"␤Int index = 4␤»
lizmat m: my @a = <a b c d e>; for @a.kv -> $index, $letter { say "$index: $letter" } # clearer ?
camelia rakudo-moar e1e03e: OUTPUT«0: a␤1: b␤2: c␤3: d␤4: e␤»
lizmat Skarsnik: ^^^ that what you mean ? 13:42
Skarsnik I mean I have ,for example, for @tab -> $t {}; and I want to count each loop; like if I write my $cpt = 0; for @tab -> $t {$cpt++}
Ah yes
or can I loop through 2 array at the same time? like for @t1, @t2 -> $t1, $t2 ? 13:44
lizmat m: my @a = <a b c d e>; my @b = ^5; dd @a Z @b # something with Z maybe ? 13:46
camelia rakudo-moar e1e03e: OUTPUT«Seq $var = (("a", 0), ("b", 1), ("c", 2), ("d", 3), ("e", 4)).Seq␤»
RabidGravy or 13:47
m: my @a = <1 2 3 4>; my @b = <a b c d>; for (@a Z @b).flat -> $a, $b { say $a, $b } 13:48
camelia rakudo-moar e1e03e: OUTPUT«1a␤2b␤3c␤4d␤»
RabidGravy depending on your requirement
lizmat m: my @a = <a b c d e>; my @b = ^5; for @a Z=> @b { .say } # another alternative
camelia rakudo-moar e1e03e: OUTPUT«a => 0␤b => 1␤c => 2␤d => 3␤e => 4␤»
Skarsnik Thx :) 13:49
AlexDaniel m: my @a = <a b c>; my @b = <x y z>; for (flat @a Z @b) -> $x, $y { say $x ~ $y, $++ }
camelia rakudo-moar e1e03e: OUTPUT«ax0␤by1␤cz2␤»
AlexDaniel another way to get counter
dalek kudo/nom: 4aa63f0 | lizmat++ | src/core/Str.pm:
Implement :v/k/kv/p also for Str.split(Regex:D)
 13:50
lizmat next step, documentation, then more tests 13:51
AlexDaniel oh wow 13:52
m: my @array1 = <a b c>; my @array2 = <x y z>; for @array1 Z @array2 -> $one, $two { say $one, $two }
camelia rakudo-moar e1e03e: OUTPUT«(a x)(b y)␤Too few positionals passed; expected 2 arguments but got 1␤ in block <unit> at /tmp/Qvu3QWsFNO:1␤␤»
AlexDaniel example taken from perl6advent.wordpress.com/2009/12/...and-profi/
perhaps it should be corrected? 13:53
13:53 Ven left
AlexDaniel m: my @array = <a b c>; for ^Inf Z @array -> $index, $item { say $index, $item } 13:54
camelia rakudo-moar e1e03e: OUTPUT«(0 a)(1 b)␤Too few positionals passed; expected 2 arguments but got 1␤ in block <unit> at /tmp/sdUOLoQ1Uf:1␤␤»
AlexDaniel that's another example from this post
psch single-arg rule /o\ 13:57
timotimo you would want @array.kv instead, no?
oh, wait
m: say <a b c> Z <x y z>
camelia rakudo-moar e1e03e: OUTPUT«((a x) (b y) (c z))␤»
psch m: my @array = <a b c>; for (^Inf Z @array).flat -> $index, $item { say $index, $item }
camelia rakudo-moar e1e03e: OUTPUT«0a␤1b␤2c␤»
timotimo right, that wants a .flat
psch AlexDaniel: not that $index is (0 a) and $item is (1 b) in your example 13:58
so, yeah, semantics changed
timotimo yeah
psch s/not/note/
timotimo for a post from 2009, that's ... fine?
psch well, the question is whether the advent posts should be kept up to date or at least have errata
13:58 Ven joined
AlexDaniel timotimo: I'm not sure. If it is outdated, then why keep it at all? 13:59
psch they are kind of a documentation precursor
AlexDaniel timotimo: or we can just fix it
psch fwiw, i know from experience that any author for perl6advent can edit any post 14:00
(because i got my post edited... :) )
i think moritz++ has "add author" authority
Skarsnik m: use NativeCall; my $a; $a.^set_name("int32"); say nativesizeof($a.WHAT); 14:01
camelia rakudo-moar e1e03e: OUTPUT«NativeCall op sizeof expected type with CPointer, CStruct, CArray, P6int or P6num representation, but got a P6opaque␤ in sub nativesizeof at lib/NativeCall.pm:335␤ in block <unit> at /tmp/79RatZRdx1:1␤␤»
psch i don't know who Matthew Walton is, is why i bring this up
Skarsnik: .WHAT isn't .HOW.name
Skarsnik m: use NativeCall; my $a; $a.^set_name("int32"); say $a.WHAT; 14:02
camelia rakudo-moar e1e03e: OUTPUT«(int32)␤»
14:03 hankache left
psch oh, it is..? 14:03
that's curious :P
well, i guess that means that isn't the problem, but the repr is
Skarsnik m: use NativeCall; my $a; $a.^set_name("int32"); say nativesizeof(int32);
camelia rakudo-moar e1e03e: OUTPUT«4␤»
psch m: use NativeCall; my $a; $a.^set_name("int32"); say $a ~~ int32
camelia rakudo-moar e1e03e: OUTPUT«False␤»
psch m: use NativeCall; my $a; $a.^set_name("int32"); say $a.WHAT === int32 14:04
camelia rakudo-moar e1e03e: OUTPUT«True␤»
RabidGravy that's odd
psch m: use NativeCall; say int32.REPR; my $a; say $a.REPR
camelia rakudo-moar e1e03e: OUTPUT«P6int␤P6opaque␤»
psch m: use NativeCall; my int32 $a; say $a.REPR 14:05
camelia rakudo-moar e1e03e: OUTPUT«P6opaque␤»
psch m: use NativeCall; my int32 $a; say nqp::decont($a).REPR
camelia rakudo-moar e1e03e: OUTPUT«P6opaque␤»
psch m: use NativeCall; my int32 $a; say nativesizeof($a.WHAT)
camelia rakudo-moar e1e03e: OUTPUT«NativeCall op sizeof expected type with CPointer, CStruct, CArray, P6int or P6num representation, but got a P6opaque␤ in sub nativesizeof at lib/NativeCall.pm:335␤ in block <unit> at /tmp/sq3kV4eWpI:1␤␤»
psch m: use NativeCall; my int32 \a; say nativesizeof(a.WHAT) 14:06
camelia rakudo-moar e1e03e: OUTPUT«5===SORRY!5=== Error while compiling /tmp/wZ7eMuJwIJ␤Term definition requires an initializer␤at /tmp/wZ7eMuJwIJ:1␤------> 3use NativeCall; my int32 \a7⏏5; say nativesizeof(a.WHAT)␤»
psch m: use NativeCall; my int32 \a = 5; say nativesizeof(a.WHAT)
camelia rakudo-moar e1e03e: OUTPUT«Type check failed in binding; expected int32 but got Int␤ in block <unit> at /tmp/dkPl5zDbMo:1␤␤»
psch m: use NativeCall; my int32 $a; say nativesizeof($a.VAR.of)
camelia rakudo-moar e1e03e: OUTPUT«Method 'of' not found for invocant of class 'Int'␤ in block <unit> at /tmp/59qMYdvzCW:1␤␤»
Skarsnik m: use NativeCall; my int32 $a; say nativesizeof($a.WHAT); 14:08
camelia rakudo-moar e1e03e: OUTPUT«NativeCall op sizeof expected type with CPointer, CStruct, CArray, P6int or P6num representation, but got a P6opaque␤ in sub nativesizeof at lib/NativeCall.pm:335␤ in block <unit> at /tmp/yKLLjaOzdp:1␤␤»
14:08 MARTIMM joined, Ven left 14:09 Ven joined
Skarsnik m: use NativeCall; my $a = "int32"; say EVAL("nativesizeof($a)"); 14:11
camelia rakudo-moar 4aa63f: OUTPUT«4␤»
psch well, that's just < nativesizeof(int32) > 14:12
Skarsnik I know x)
mrf ugexe: you about?
psch Skarsnik: ooc, what's the use case? or are you just looking for WATs? 14:13
Skarsnik I am not sure to understand why nativesizeof($var.WHAT) does not work still x)
arnsholt m: my int32 $x; say $x.WHAT; # What about this?
camelia rakudo-moar 4aa63f: OUTPUT«(Int)␤»
arnsholt Huh. 's odd 14:14
psch arnsholt: method calls box
arnsholt: because natives don't know methods
Skarsnik I am writing something that compare C struct size and their defined size in perl6
dalek kudo/nom: f4266f2 | lizmat++ | src/core/Cool.pm:
Make sure sub split() passes all nameds as well
 14:15
lizmat oddly enough, we seem to be low on test on split()
psch m: say int32.WHAT
camelia rakudo-moar 4aa63f: OUTPUT«(int32)␤»
psch well, .WHAT isn't a method, but it somehow still promotes... :/
timotimo "promotes"?
oh, i see 14:16
yeah, that's right
arnsholt Skarsnik: "C struct size" and "defined size in Perl 6"?
14:16 sprocket joined
lizmat github.com/rakudo/rakudo/pull/578 # leont++ does dogfooding 14:17
leont :-)
arnsholt Yeah, leont++ 14:18
timotimo m: say nativesizeof(int32)
camelia rakudo-moar 4aa63f: OUTPUT«5===SORRY!5=== Error while compiling /tmp/gUIQ8cDxSc␤Undeclared routine:␤ nativesizeof used at line 1␤␤»
timotimo m: say nqp::nativecallsizeof(int32)
camelia rakudo-moar 4aa63f: OUTPUT«4␤»
timotimo m: say nqp::nativecallsizeof(int64)
camelia rakudo-moar 4aa63f: OUTPUT«8␤»
arnsholt That's a seriously important task, I think
timotimo m: use NativeCall; say nativesizeof(int32) 14:19
camelia rakudo-moar 4aa63f: OUTPUT«4␤»
arnsholt I've been meaning to look into a Perl 6 test harness for ages, but never getting around to it
timotimo m: use NativeCall; say nativesizeof(int64)
camelia rakudo-moar 4aa63f: OUTPUT«8␤»
Skarsnik Well you want to bind a C lib that has struct ex { int a; char *ag} to perl6 code: class ex is repr('CStruct') { has int $a; has Str *ag} it's a good thing to check if their size match
timotimo m: use NativeCall; my int32 $foo = 0; say nativesizeof($foo)
camelia rakudo-moar 4aa63f: OUTPUT«NativeCall op sizeof expected type with CPointer, CStruct, CArray, P6int or P6num representation, but got a P6opaque␤ in sub nativesizeof at lib/NativeCall.pm:335␤ in block <unit> at /tmp/rHjdIhJDoO:1␤␤»
timotimo m: use NativeCall; my int32 $foo = 0; say nativesizeof($foo.VAR)
camelia rakudo-moar 4aa63f: OUTPUT«NativeCall op sizeof expected type with CPointer, CStruct, CArray, P6int or P6num representation, but got a NativeRef␤ in sub nativesizeof at lib/NativeCall.pm:335␤ in block <unit> at /tmp/s0xpapD01k:1␤␤»
timotimo m: use NativeCall; my int32 $foo = 0; say nativesizeof(nqp::decont($foo))
camelia rakudo-moar 4aa63f: OUTPUT«NativeCall op sizeof expected type with CPointer, CStruct, CArray, P6int or P6num representation, but got a P6opaque␤ in sub nativesizeof at lib/NativeCall.pm:335␤ in block <unit> at /tmp/sLf2cXWWG2:1␤␤»
timotimo mhm, mhm.
m: use NativeCall; my int32 $foo = 0; say nqp::nativecallsizeof(nqp::decont($foo))
camelia rakudo-moar 4aa63f: OUTPUT«NativeCall op sizeof expected type with CPointer, CStruct, CArray, P6int or P6num representation, but got a P6opaque␤ in block <unit> at /tmp/uRSjeadre1:1␤␤»
arnsholt m: my int32 $x = 3; say $x.WHAT 14:20
camelia rakudo-moar 4aa63f: OUTPUT«(Int)␤»
leont The fudge tools are other targets to dogfood, IMO
arnsholt It's still not clear to me why the .WHAT of an int32 is Int though
nine arnsholt: because my native int32 is repr('P6int') is Int is nativesize(32) { } 14:21
lizmat leont: if they live in roast, they should not be 14:22
or at least, not remove the perl 5 ones
leont Yeah, I figured we can't remove the perl 5 ones
But that's no reason to use them if we have a working perl 6 14:23
lizmat cycling&
agree :-)
really afk&
Skarsnik hm EVAL does not work for me because it can't find the type defined in a module even with EVAL("require $modulename; nativesizeof($ncname);")
arnsholt nine: But how does that make the .WHAT be Int?
14:24 pmurias_ joined
Skarsnik m: my str $x; say $x.WHAT; 14:24
camelia rakudo-moar 4aa63f: OUTPUT«(Str)␤»
nine arnsholt: int32 is just an Int with an added trait
timotimo welllll
if you look at the --target=ast, the WHAT will be called on a lexicalref
it might just be it has no clue to resolve the actual lexical behind that
m: my int32 $foo = 0; say $foo.VAR.WHAT 14:25
camelia rakudo-moar 4aa63f: OUTPUT«(IntLexRef)␤»
Skarsnik m: my str $x = "a"; my $p := $x; say $p.WHAT;
camelia rakudo-moar 4aa63f: OUTPUT«(Str)␤»
arnsholt Yeah, that sounds more reasonable. `is Int` means it's a subclass of Int, after all
psch m: my int32 $foo = 0; say $foo.VAR.of
camelia rakudo-moar 4aa63f: OUTPUT«Method 'of' not found for invocant of class 'Int'␤ in block <unit> at /tmp/zmq_1b3C8_:1␤␤»
psch m: my int32 $foo = 0; say $foo.VAR.^native_type 14:27
camelia rakudo-moar 4aa63f: OUTPUT«Cannot find method 'gist': no method cache and no .^find_method␤ in block <unit> at /tmp/BFHixWxCza:1␤␤»
psch m: my int32 $foo = 0; say $foo.VAR.^native_type === int32
camelia rakudo-moar 4aa63f: OUTPUT«Cannot call infix:<===>(int, int32); none of these signatures match:␤ ($?)␤ ($a, $b)␤ (Int:D \a, Int:D \b)␤ (int $a, int $b)␤ (Num:D \a, Num:D \b)␤ (Num $ where { ... }, Num $ where { ... })␤ (num $a, num $b --> Bool:D)…»
Skarsnik m: use NativeCall; class T is repr('CStruct'){has int32 $.a}; my T $b; say nativesizeof($b.WHAT);
camelia rakudo-moar 4aa63f: OUTPUT«4␤»
14:27 pmurias left
arnsholt Skarsnik: That sounds reasonable. But how are you going to query for the C size of the struct? 14:27
Skarsnik m: use NativeCall; class T is repr('CStruct'){has int32 $.a}; my $b; $b.^set_name("T"); say $b.WHAT; nativesizeof($b.WHAT); 14:28
timotimo arnsholt: nativesizeof can handle CStruct
camelia rakudo-moar 4aa63f: OUTPUT«(T)␤NativeCall op sizeof expected type with CPointer, CStruct, CArray, P6int or P6num representation, but got a P6opaque␤ in sub nativesizeof at lib/NativeCall.pm:335␤ in block <unit> at /tmp/TOWtuQb8na:1␤␤»
psch hm, can i pull a PR from github?
timotimo m: use NativeCall; class T is repr('CStruct'){has int32 $.a}; say nativesizeof(T)
psch i'd like to look at what the EvalServer isn't doing for leont++'s work
camelia rakudo-moar 4aa63f: OUTPUT«4␤»
arnsholt Well yeah. But that's the Perl 6 side size. How are you going to find out what size the C side operates with?
leont psch: awesome 14:29
Skarsnik I compile a small C file that just printf sizeof(cstruct)
arnsholt Without resorting to on-the-fly C compilation
Right =)
RabidGravy psch, pull the branch the PR comes from
leont You need to uncomment the jvm specific code at the end of harness6 to re-enable that
psch RabidGravy: yeah, i found the CLI instructions on github... (reading before asking)++ 14:30
Skarsnik m: use NativeCall; class T is repr('CStruct'){has int32 $.a}; my $b; $b.^set_name("T"); say $b.WHAT;
camelia rakudo-moar 4aa63f: OUTPUT«(T)␤»
RabidGravy :)
Skarsnik but why I can't nativesizeof this WHAT?
timotimo what do you .^set_name for?
just curious
m: use NativeCall; class T is repr('CStruct'){has int32 $.a}; my $b; $b.^set_name("T"); say $b.WHAT; say nativesizeof($b); say nativesizeof($b.WHAT) 14:31
camelia rakudo-moar 4aa63f: OUTPUT«(T)␤NativeCall op sizeof expected type with CPointer, CStruct, CArray, P6int or P6num representation, but got a P6opaque␤ in sub nativesizeof at lib/NativeCall.pm:335␤ in block <unit> at /tmp/PCaZAX17EW:1␤␤»
timotimo m: use NativeCall; class T is repr('CStruct'){has int32 $.a}; my $b; $b.^set_name("T"); say $b.WHAT; try say nativesizeof($b); try say nativesizeof($b.WHAT)
camelia rakudo-moar 4aa63f: OUTPUT«(T)␤»
timotimo er, wait ...
arnsholt Why the .^set_name to the name it already has?
timotimo what the hell are you doing :)
Skarsnik to construct something with the name of the type.
timotimo "my $b" does NOT give you a T-typed thing 14:32
even if you set its name to T
just like if you call yourself "timo", you're not me
Skarsnik well .WHAT give T
psch m: sub f(Int $x) { }; my $x; $x.^set_name('Int'); f $x
arnsholt Well it says that it's *name* is T
camelia rakudo-moar f4266f: OUTPUT«Type check failed in binding $x; expected Int but got Int␤ in sub f at /tmp/0IguzjNYY0:1␤ in block <unit> at /tmp/0IguzjNYY0:1␤␤»
psch DIHWIDT
arnsholt Which makes sense, since you just renamed its type to T =)
m: use NativeCall; class T is repr('CStruct') { has int32 $x; }; my T $x; say $x.WHAT # Here you go 14:33
camelia rakudo-moar f4266f: OUTPUT«(T)␤»
14:33 lizmat left
Skarsnik yes but WHAT give the exact same thing x) 14:33
arnsholt m: my $x; say $x.WHAT; $x.^set_name("T"); say $x.WHAT
camelia rakudo-moar f4266f: OUTPUT«(Any)␤(T)␤»
psch m: my $x; $x.^set_name("int32"); say $x.WHAT =:= int32
camelia rakudo-moar f4266f: OUTPUT«False␤»
psch it doesn't give the exact same name, infix:<===> is just the wrong tool for the job 14:34
arnsholt You just reconfigured Any to say its name is T
psch s/name/thing/
RabidGravy m: my Int $a; $a.^set_name("Str"); say $a.WHAT; say $a.^mro;
camelia rakudo-moar f4266f: OUTPUT«(Str)␤((Str) (Cool) (Any) (Mu))␤»
Skarsnik Ok, so how I create something with T type (given I only have his name)? 14:35
RabidGravy weird
psch m: my $type = "Int"; my ::($type) $x; say $x.WHAT
camelia rakudo-moar f4266f: OUTPUT«Method 'gist' not found for invocant of class '($type)'␤ in block <unit> at /tmp/brUDD27eD6:1␤␤»
psch well, that's how i thought it would work vOv
arnsholt m: my $x = ::("Str"); say $x 14:36
camelia rakudo-moar f4266f: OUTPUT«(Str)␤»
RabidGravy m: class T { }; my ::("T") $a; say $a.WHAT
camelia rakudo-moar f4266f: OUTPUT«Method 'gist' not found for invocant of class '("T")'␤ in block <unit> at /tmp/xz3ZwPJcT2:1␤␤»
arnsholt Skarsnik: But remember that types can be passed around just like any other object
psch well, indirect lookup for container constraints is apparently different than type object lookup... 14:37
not sure if that's even supposed to work, anyway
leont: can you expand on your "SourceHandler" comment?
Skarsnik m: use NativeCall; sub foo ($t) { say nativesizeof($t)}; foo(int32); 14:38
camelia rakudo-moar f4266f: OUTPUT«4␤»
14:38 mr-foobar joined
Skarsnik Ah nice x) 14:38
leont eval-client is a small p5 script that makes a connection with the eval-server, tells it what to run, and then copies anything it gets back to the harness 14:39
14:39 ^elyse^ left
timotimo m: my ::Foo $test = 1; say Foo 14:39
camelia rakudo-moar f4266f: OUTPUT«5===SORRY!5=== Error while compiling /tmp/GmDBhhmytb␤Undeclared name:␤ Foo used at line 1␤␤»
leont This gluing together could/should be done without shelling out, IMO
timotimo i would expect that'd be parsed as a type capture, TBH
leont SourceHandlers are the objects in TAP::Harness that turn an input (currently always a filename) into a Source object (e.g. something that runs that file as a perl 6 script) 14:40
So we'd need a SourceHandler that sets up that connection with the eval server instead of running it directly 14:41
arnsholt Skarsnik: Yeah, that's what I meant with passing around types. The problem, AFAICT is that .WHAT on some things doesn't return what you want 14:42
jnthn: The .WHAT on a scalar declared as int32 is apparently Int. Bug or feature?
Skarsnik it's the int32 is bind from a clib func call I think it remain int32 14:43
*if
psch leont: the "it" that's ran directly is the eval client? 14:45
leont Yes
See nqp/tools/jvm/eval-client.pl
It does very little, really 14:46
Skarsnik does %hash.keys and %hash.values give the contents in the same orders?
psch yeah, that does sound generally useful, to have a "feed this into a socket" kind of handler for a harness
...apparently i can't build panda on r-j 14:47
14:47 pmurias_ is now known as pmurias
psch "java.lang.RuntimeException: Cannot access a native attribute as a reference attribute4~ 14:48
grml
BooK in perl6, there is no need for try because CATCH is "attached" to the enclosing scope. did I get that right? 14:50
moritz aye 14:51
leont Usually
moritz though of course, 'try' is still a very handy shortcut
RabidGravy in that sense yes, but try is still useful to "try and get a value or undef"
psch our try is notably different than e.g. javas
arnsholt Skarsnik: If you're iterating a hash, for %hash will give you the pairs in it though. So no need to zip together the keys and values separately 14:52
BooK ok, I should have actually looked for a try keyword :-)
psch in that the latter uses it to denote a "we care about the Exception" scope, while we use it for a "we don't care about the Exception, just don't die" scope
arnsholt In case that's what you're doing =)
BooK but the doc for Exception pointed to CATCH only
leont try without a catch swallows the exception
BooK and design.perl6.org/S04.html#Phasers said "A CATCH block is just a trait of the closure containing it" 14:53
leont I tend to use try with catch, but that's probably more out of habit from other languages
BooK so I assumed I didn't even need try, and leont comment makes me understand why it's still needed
Skarsnik arnsholt, I am putting the keys and the values in separate arrays
psch leont: uhm, how do i run harness6? the make invocations i tried use harness5
arnsholt Skarsnik: Right, in that case it's not what you want =) 14:54
leont I hadn't enabled it because of said issues
psch oh, so manually
leont Just run ./perl6 -Ilib t/harness6 t/<whatever>
./perl6-j -Ilib t/harness6 t/01-sanity t/04-nativecall
timotimo m: my $b; $b.^set_name('MyClass'); my $c; say $c.WHAT 14:55
camelia rakudo-moar f4266f: OUTPUT«(MyClass)␤»
BooK I should probably have read that whole section
timotimo Skarsnik: that's how you change every untyped variable into MyClass instead!
14:56 ^elyse^ joined
leont I should have added that incantation as a comment to the jvm Makefile 14:56
Skarsnik What?! $c get the last type? 14:57
timotimo no
$b's type is Any. you set Any's name to MyClass
arnsholt Skarsnik: It changes the *name* of the Any type
Skarsnik oohh
timotimo now every Any is actually called MyClass
leont Dogfooding TAP::Harness in its travis configuration, it works :-) 14:58
14:58 isBEKaml joined
RabidGravy haha, that's evil 14:58
arnsholt m: class T {}; Any.^set_name("T"); my $a; my T $b; say $a.WHAT, $a.WHAT.WHERE; say $b.WHAT, $b.WHAT.WHERE # Observe
camelia rakudo-moar f4266f: OUTPUT«(T)33942928␤(T)140221499061984␤»
arnsholt The two objects report types with the same names, but the object identities, crucially, are different
Declaring the type of a variable is done between the "my" and the variable name, not MOP shenanigans (which is what anything using .^ is) 14:59
14:59 hankache joined
BooK interestingly, it's try { CATCH { ... } } and not try { ... } catch { ... } 15:02
psch leont: fwiw, running the eval-server separately (with commenting it out in harness6) gives me "All 24 subtests passed" for 01-sanity/01-literals.t, but still shows "Dubious, test returned 255"
leont Hmmmm
timotimo BooK: yeah, the good thing about that is you have all your lexicals still valid inside the CATCH block like that 15:03
15:03 _mg_ left
timotimo psch: does that mean it doesn't have a plan in it? 15:03
or doesn't have a done()?
Skarsnik m: use NativeCall; my %h = "I32" => int32, "I8" => int8; for %h.kv -> $k, $v {say $k, nativesizeof($v)}
arnsholt Oh, also relevant to the .^set_name shenanigans is that it's only the *reported* name of Any that's changed. It's still spelled Any in the code =)
camelia rakudo-moar f4266f: OUTPUT«I324␤I81␤»
BooK timotimo: the other nice thing is you can put it at the beginning, or wherever you want 15:04
timotimo that's true, too
psch timotimo: i guess, it manually outputs "1..24" at the start
no &plan, no &done-testing
timotimo hm, if the 1..24 gets to the tap parser, that should work
15:04 zakharyas joined 15:05 ponpon is now known as ponbiki
RabidGravy BooK. also you can CATCH { } specific exceptions in the scope, but leave unexpected ones to the try { } 15:05
15:05 weihan left
BooK RabidGravy: I'm trying to work that out 15:06
a bare CATCH {} just lets it go through
psch same for a file with plan, though
BooK: CATCH { } is similar to given { }, in that regard
BooK so I need to mark it as handled, I guess
timotimo so something makes it crash in the end?
psch BooK: a given { } without anything inside also doesn't do anything 15:07
BooK the doc says the exception is in $_, but that's because of the similarity to given {} you just mentionned
Skarsnik hm why I get an undeclared routine error writing use Gumbo::Binding; test-cstruct-size(:cheaders(@headers), :types(("GumboOutput" => gumbo_ouput_s, "GumboNode" => gumbo_node_s)), :clibs(@libs), :modulename("Gumbo::Binding")); for the gumbo_* stuff
BooK I'm not familiart yet with .perl or .gist, and say @!.perl dies in a weird way 15:08
hankache m: say 'abc' ~~ m/ a?c /;
camelia rakudo-moar f4266f: OUTPUT«｢c｣␤»
BooK m: say @!.perl
camelia rakudo-moar f4266f: OUTPUT«5===SORRY!5=== Error while compiling /tmp/TwCrboHyTs␤Cannot negate . because dotty infix operators are too fiddly␤at /tmp/TwCrboHyTs:1␤------> 3say @!.7⏏5perl␤»
RabidGravy m : class X::Foo is Exception { }; { CATCH { when X::Foo { say "woo" } }; Bool.pick && X::Foo.new.throw }
hankache shouldn't this return Nil?
RabidGravy m: class X::Foo is Exception { }; { CATCH { when X::Foo { say "woo" } }; Bool.pick && X::Foo.new.throw }
camelia ( no output )
RabidGravy m: class X::Foo is Exception { }; { CATCH { when X::Foo { say "woo" } }; Bool.pick && X::Foo.new.throw }
camelia rakudo-moar f4266f: OUTPUT«woo␤»
psch hankache: you didn't anchor anything 15:09
RabidGravy m: class X::Foo is Exception { }; { CATCH { when X::Foo { say "woo" } }; Bool.pick && X::Foo.new.throw; die "bah" }
camelia rakudo-moar f4266f: OUTPUT«bah␤ in block <unit> at /tmp/_be_QY_o2A:1␤␤»
RabidGravy m: class X::Foo is Exception { }; { CATCH { when X::Foo { say "woo" } }; Bool.pick && X::Foo.new.throw; die "bah" }
camelia rakudo-moar f4266f: OUTPUT«woo␤»
BooK ok so when implicitely handles the exception
leont psch: what happens if you add exit 0 to the end of the test?
RabidGravy that is
BooK: yep
the "default" case can be used for any others 15:10
hankache psch: i am new to regex
leont Also, it worked for me, but I did it with another test, could you try t/02-op-match.t or some such?
hankache psch can you kindly tell me how to "anchor" 15:11
psch hankache: doc.perl6.org/language/regexes
hankache coming from SQL ? and * are wildcards
timotimo ah, yeah, ? and * work differently in regex
i think ? means "any single character"? 15:12
in SQL, i mean
hankache yes timotimo
moritz % is the wildcard in LIKE queries
corresponding to .* in a regex
timotimo the equivalent of that is the dot in regex
BooK and m: CATCH { $_.handled = True }; die "aaahhh";
m: CATCH { $_.handled = True }; die "aaahhh";
camelia rakudo-moar f4266f: OUTPUT«Method 'handled' not found for invocant of class 'X::AdHoc'␤ in block <unit> at /tmp/5klJXEGBW9:1␤␤»
psch leont: i tried t/spec/S05-operators/match.t, cause that exists, and still get the dubious 255
BooK m: CATCH { default {} }; die "aaahhh";
camelia ( no output )
hankache psch thanks i'll read it
BooK and m: CATCH { say $_.handled; default { say $_.handled } }; die "aaahhh"; 15:13
psch leont: note though it still passes every test
15:13 Ven left, yqt joined
BooK m: CATCH { say $_.handled; default { say $_.handled } }; die "aaahhh"; 15:13
camelia rakudo-moar f4266f: OUTPUT«Method 'handled' not found for invocant of class 'X::AdHoc'␤ in block <unit> at /tmp/ZSNf7YzHBo:1␤␤»
BooK ok, there's something I don't know how to do 15:14
timotimo i think handled may be a private attribute
psch it is
accessed by Booling
(amongst others afair)
leont What do they return when run by hand?
BooK but S04 says : $!.handled = 1;
psch leont: when ran through eval-client.pl, $? is 0 15:15
(bash $? that is)
and all tests pass
same EvalServer instance as the harness6 invocation uses
hankache is there a wildcard other than . 15:16
. is only one character 15:17
something that allows many characters
psch hankache: that's what quantifiers are for
timotimo instead of multi-character-matching wildcards, regex has "quantifiers"
psch m: say "aaaa" ~~ / a+ / 15:18
camelia rakudo-moar f4266f: OUTPUT«｢aaaa｣␤»
timotimo every kind of wildcard matches exactly once
pmurias nqp-m: say(nqp::escape("line2\r\n"))
camelia nqp-moarvm: OUTPUT«line2
15:18 MARTIMM left
timotimo ? is actually also a quantifier; it means "match the previous thing either once or not at all" 15:18
pmurias nqp-p: say(nqp::escape("line2\r\n"))
15:18 camelia left
pmurias is nqp-m behaving correctly? it seems like some unicody refactor broke nqp::escape 15:19
timotimo could be
15:19 camelia joined
timotimo it does the opposite of what it's supposed to do %) 15:19
ah, no, it just does nothing to it 15:20
hankache ok but say you have "abcdef" and "aef" and you want to check if it starts with "a" and ends with "f"
camelia nqp-parrot: OUTPUT«Can't exec "./rakudo-inst/bin/nqp-p": No such file or directory at lib/EvalbotExecuter.pm line 193.␤exec (./rakudo-inst/bin/nqp-p /tmp/J_UNGEpZZl) failed: No such file or directory␤Server error occurred! Closing Link: ns1.niner.name (Quit: camelia)␤Lost conn…»
15:20 ChanServ sets mode: +v camelia
timotimo right, then you do / ^ a .* f $ / 15:20
pmurias it seems like \r\n was turned into one grapheme and nqp::escape wasn't updated 15:21
timotimo alternatively $foo.startswith("a") && $foo.endswith("f")
hankache ok
timotimo m: say "abcdef".startswith("a")
camelia rakudo-moar f4266f: OUTPUT«Method 'startswith' not found for invocant of class 'Str'␤ in block <unit> at /tmp/taCufjiHor:1␤␤»
pmurias puts on his MoarVM hat and starts assembling a pull request
timotimo oops
disregard that!
psch timotimo: ENOKEBAB
timotimo those methods don't exist!
m: say "abcdef".starts-with("a") 15:22
camelia rakudo-moar f4266f: OUTPUT«True␤»
timotimo m: say "abcdef".ends-with("f")
camelia rakudo-moar f4266f: OUTPUT«True␤»
RabidGravy chilli SAUCE?
hankache i can also quantify \D
right?
timotimo of course
what is \D?
psch \D is probably our \d
RabidGravy anything that is a literal or stands for a literal can be quantified
timotimo oh, "not digit"
RabidGravy: what do you call what % and %% do? 15:23
psch at least i think SQL escapes are case insensitive as well, but i might be misremembering
timotimo because they modify a quantifier
well, \d is "a digit" and \D is "anything but a digit"
BooK so a failed parse returns Nil, but where do I find details about the failure? 15:24
RabidGravy "quantifier modifying thingies"
timotimo BooK: it's very hard to generally figure out things about a failure to match 15:25
leont is observing the return value of MAIN not being the return value of the program; is that deliberate?
timotimo because matches usually do backtracking and scanning
leont: i think it's very easy to accidentally return something you don't mean to return in a sub MAIN
BooK timotimo: Perl6 manages to give nice error messages about itself, doesn't it? 15:26
timotimo but if you throw an exception, you get a failing exit code, right?
that's right; the perl6 grammar has a "high water mark" system
it remembers what's the farthest that it had matched and what possibilities there were to continue matching 15:27
BooK timotimo: I'm trying to give a more useful error than this one: github.com/moritz/json/blob/master...iny.pm#L59
timotimo that's what you get "expected: postfix, semicolon, ..." from
15:29 spider-mario left
BooK at the stage the exception is thrown, it's too late to have access to the useful info, I assume 15:29
timotimo "the exception"? 15:30
oh
when i said "the perl6 grammar", i meant the one that rakudo has internally to parse perl6 code
sadly, you don't get that for free in any grammar
so in this case, the "useful info" doesn't actually exist :( 15:31
RabidGravy I suppose that you would have to keep track of where you got to in the actions 15:32
BooK I was wondering 15:33
so the info would be in the action object handed to the grammar
15:33 spider-mario joined
RabidGravy could be 15:33
BooK for a first contribution to that module, I'd like to be able to say something like "parsing stopped at character 1234" 15:34
pmurias the high mark thing should be made pluggable
timotimo i think it'd be better to put code for that in the grammar rather than the actions
BooK how do you get access to the data?
attributes of the grammar? 15:35
which need to be reset when entering TOP?
timotimo the current position is a property of the match object, or rather, the cursor
BooK the doc on Cursor was very sparse, kinda "nothing to see here, move along"
timotimo yeah
BooK doc.perl6.org/type/Cursor 15:36
15:36 isBEKaml left
timotimo m: "foo bar baz quux 123" ~~ / (\w+ { say $/.CURSOR.pos } ) + % \s+ 15:36
camelia rakudo-moar f4266f: OUTPUT«5===SORRY!5===␤Regex not terminated.␤at /tmp/8M6BNDcEhg:1␤------> 3~ / (\w+ { say $/.CURSOR.pos } ) + % \s+7⏏5<EOL>␤Unable to parse regex; couldn't find final '/'␤at /tmp/8M6BNDcEhg:1␤------> 3~ / (\w+ { say $/.CURSOR.pos } ) + % \s+…»
timotimo m: "foo bar baz quux 123" ~~ / (\w+ { say $/.CURSOR.pos } ) + % \s+ /
camelia rakudo-moar f4266f: OUTPUT«3␤7␤11␤16␤20␤»
timotimo m: "foo bar baz quux 123".substr(20) 15:37
camelia ( no output )
timotimo m: "foo bar baz quux 123".substr(20).say
camelia rakudo-moar f4266f: OUTPUT«␤»
timotimo m: say "1" ~~ / \w /
camelia rakudo-moar f4266f: OUTPUT«｢1｣␤»
timotimo oh, duh :)
m: "foo bar baz quux 123" ~~ / (\a+ { say $/.CURSOR.pos } ) + % \s+ /
camelia rakudo-moar f4266f: OUTPUT«5===SORRY!5=== Error while compiling /tmp/txeLOrfZs1␤Unrecognized backslash sequence: '\a'␤at /tmp/txeLOrfZs1:1␤------> 3"foo bar baz quux 123" ~~ / (\7⏏5a+ { say $/.CURSOR.pos } ) + % \s+ /␤ expecting any of:␤ term␤»
timotimo how do i "alpha"?
m: "foo bar baz quux 123" ~~ / (<.alpha>+ { say $/.CURSOR.pos } ) + % \s+ /
camelia rakudo-moar f4266f: OUTPUT«3␤7␤11␤16␤»
timotimo m: my $highest_so_far = 0; (my $source = "foo bar baz quux 123") ~~ / (<.alpha>+ { $highest_so_far max= $/.CURSOR.pos } ) + % \s+ /; say $source.substr($highest_so_far) 15:38
camelia rakudo-moar f4266f: OUTPUT« 123␤»
timotimo do note that JSON::Tiny is already woefully slow; adding those closures in there won't make it any faster :(
15:39 tokuhiro_ joined, rangerprice joined
timotimo JSON::Fast is only 2x faster, but due to its implementation it's able to give you the last position it tried to match, and also what it was trying to match at that point 15:40
BooK is it using grammars too? 15:42
timotimo nah
github.com/timo/json_fast/blob/mas...ON/Fast.pm
pmurias are the grapheme numbers for composed characters constant?
RabidGravy I found something that I considered perfectly valid that JSON::Fast borked on the other day, but forgot what it was
timotimo pmurias: you mean the generated numbers? don't rely on them 15:43
RabidGravy: that's potentially easy to do :) 15:44
15:45 _28_ria left, _mg_ joined, kaare_ joined
BooK timotimo: so perl6 grammars are awesome, but don't use them because they're slow? 15:45
timotimo json doesn't need backtracking at all 15:46
so a big part of what makes grammars awesome is wasted on them
BooK but it's possible to not backtrack in grammar, I just read that
timotimo right, "ratchet" mode
i'm not sure if we generate much better code from that yet
15:46 mr_ron joined
timotimo it's been a while since i last touched that part of the code 15:46
15:48 tokuhiro_ left
timotimo on top of ratchet mode there's also more precise backtracking controls, but at least parts of that aren't implemented yet 15:49
15:50 _28_ria joined
dalek c: bbc2e39 | (Philippe Bruhat (BooK))++ | doc/Language/grammars.pod:
Fix typo
 15:54
15:54 khw joined 15:56 rurban left 15:57 skids joined
raiph .tell pmurias nwc10 on NFG attacks: irclog.perlgeek.de/perl6/2015-04-07#i_10400549 15:59
yoleaux raiph: I'll pass your message to pmurias.
16:02 zakharyas left
pmurias I made a pull request on MoarVM to fix nqp::escape 16:04
yoleaux 15:59Z <raiph> pmurias: nwc10 on NFG attacks: irclog.perlgeek.de/perl6/2015-04-07#i_10400549
timotimo pmurias: looks good to me 16:06
i'll just wait for the travis checks to complete and i'll merge it 16:07
pmurias timotimo: travis runs tests for rakudo-m? 16:09
16:10 Ven joined
timotimo no, it only runs the nqp tests 16:11
god damn it, clang is slow
16:20 yeahnoob joined
grondilu pmurias: what's the status of the js backend? I've just learnt about Emscripten (C/C++ compiler to javascript) and I was wondering if someone has ever tried to port MoarVM to it, or if that would make sense. 16:23
awwaiid contemplates the concept of "Perl6 Axioms" -- the smallest possible part of the language upon which the rest could be built
16:23 Ven left
grondilu awwaiid: that's almost certainly arbitrary. In truth, any turing complete language would do. 16:24
awwaiid grondilu: pmurias and I talked about that a bit a while back, and they said that the emscripten performance/integration is good enough for demos of languages but not as much else -- I tried to counter with the two OCaml compilers, but then realized that neither of them work as just "run ocaml via enscripten" like I initially thought 16:25
grondilu ok 16:26
awwaiid grondilu: mmm. I don't mean what are the axioms of a perl6 _implementaiton_, I mean of the language itself. So like some LISP languages have a pretty small core language and the reset can be implemented (possibly inefficiently) in libraries.
re emscripten -- that doesn't mean you shouldn't try!
16:26 Ven joined
grondilu wouldn't that be STD? 16:27
16:27 habamax joined
awwaiid well my hypothesis is that the STD is more than axioms -- it has shortcuts/optimizations and such, and is invented like before our future macro system. 16:28
Skarsnik m: class T {}; my %h = "T" => T; say %h; 16:33
camelia rakudo-moar f4266f: OUTPUT«T => (T)␤»
16:33 ^elyse^ left
Skarsnik hm When I write 16:34
use Gumbo::Binding# contening the gumbo_ type; my gumbo_output_s $m;
my %typ = ("GumboOutput" => gumbo_ouput_s, "GumboNode" => gumbo_node_s);
16:35 Oatmeal left
Skarsnik the hash affection fail, saying it can't find gumbo_ouput_s routine 16:35
16:37 rindolf left
geekosaur right, that would be looking for a function, not a type. not sure how you write the type there but suspect it involves :: 16:38
sprocket hello, perl6! 16:39
Skarsnik m: use NativeCall; class T is repr('CStruct'){}; my %h = "T" => T; say %h;
camelia rakudo-moar f4266f: OUTPUT«T => (T)␤»
Skarsnik hm
16:40 hankache left
geekosaur maybe not. so... exporting of types? hm 16:40
that seems wrong too. 16:41
geekosaur doesn't know :(
ugexe gumbo_output_s = ::($m) ? 16:43
moritz what's wrong with exporting types?
Skarsnik pastebin.com/MZ4yrz3p 16:45
Types can be find in github.com/Skarsnik/perl6-gumbo/bl...inding.pm6 16:48
16:48 Oatmeal joined
ugexe did you happen to try GumboOutput => ::("gumbo_output_s")? 16:48
the error doesn't happen until line 14, so im guessing the type applied fine to $m 16:49
Skarsnik yes
it work with ::(""), but why an imported type can't be used directly in this case? 16:51
ugexe how does it know its not a function?
moritz you need ::('&foo') or &::('foo') to dynamically lookup a function 16:52
Skarsnik well why my previous example work? m: use NativeCall; class T is repr('CStruct'){}; my %h = "T" => T; say %h;
16:55 ZoffixW joined
pmurias awwaiid: I haven't actually measured the performance of the enscripten approach 16:58
grondilu: re status, nqp-js-running-on-moar fails 4 t/nqp tests, nqp-js-compiled-by-nqp-js-running-on-node.js fails 10 t/nqp tests 17:00
grondilu: performance will be moar-with-jit-disabled-slightly-slower 17:01
17:01 Begi1115 left 17:02 raoulvdberge left
pmurias grondilu: with slightly being an unknown factor 17:02
dalek href="https://modules.perl6.org:">modules.perl6.org: 31d6c3d | (Zoffix Znet)++ | web/build-project-list.pl:
Clean up useless code
 17:06
href="https://modules.perl6.org:">modules.perl6.org: 9a1ed6e | (Zoffix Znet)++ | web/build-project-list.pl:
Use system() safely
17:06 _mg_ left, yeahnoob left, Begi1115 joined 17:07 ^elyse^ joined
awwaiid pmurias: you keeping some repo up to date for us to stare at / mess with? 17:08
pmurias awwaiid: perl6/nqp on github 17:11
it's in the main nqp repo, in the src/vm/js 17:12
awwaiid oh right, I think you told me that. no branch or anything?
17:12 spider-mario is now known as spidermario, spidermario is now known as spider-mario
pmurias awwaiid: it's merged in 17:13
keeps bitrot from piling in
awwaiid awesomesauce
that's the way to go
RabidGravy I'm sure I've asked this before but how does one extract the version from a module i.e. "class F:ver<v0.0.1> { }" <- the value of ver 17:17
17:18 skids left
moritz m: class F:ver<v0.0.1> { }; say F.^ver 17:18
camelia rakudo-moar f4266f: OUTPUT«Method 'gist' not found for invocant of class 'NQPMu'␤ in block <unit> at /tmp/bZPqTLZB0y:1␤␤»
RabidGravy ah yes
moritz :(
doesn't seem to work, though
m: class F:ver<1> { }; say F.^ver 17:19
camelia rakudo-moar f4266f: OUTPUT«Method 'gist' not found for invocant of class 'NQPMu'␤ in block <unit> at /tmp/18FlOiHyVM:1␤␤»
RabidGravy I think I worked it out myself, found it didn't work and wandered off on another problem
sprocket moritz: what does the “^” imean for that method call?
RabidGravy "on the HOW"
sprocket aah, ok
RabidGravy i.e. the meta-object
sprocket gotcha
17:21 habamax left 17:23 rindolf joined 17:27 zacts left 17:29 Ven left 17:32 grondilu left 17:34 zakharyas joined
RabidGravy I RT'd the "ver not working properly" with RT #126583 for information 17:35
it may have already been in there but it was difficult to search for 17:36
17:42 grondilu joined
ugexe yea i kinda dread using rt 17:43
17:45 tokuhiro_ joined 17:49 Actualeyes left, tokuhiro_ left 17:51 zacts joined 17:53 eliasr joined 18:04 Begi1115 left 18:07 Begi1115 joined, lizmat joined 18:08 colomon joined 18:14 zakharyas left 18:17 Ven_ joined 18:21 cognominal joined 18:24 Ven_ left 18:25 yqt left 18:26 Ven_ joined
zengargoyle hrm, i think NativeCall is beyond me at the moment. 18:28
is CArray[uint32].new( "string".comb».ord ) a reasonable thing to do? 18:30
when the C version is (uint32 *)L"strint" 18:32
psch m: use NativeCall; my $arr = CArray[uint32].new( "string".comb>>.ord ); say $arr[0]; say "s".ord # seems to work fine..?
camelia rakudo-moar f4266f: OUTPUT«115␤115␤»
moritz "string".comb».ord # shorter: "string".ords
zengargoyle ah, i guess my Segfault is maybe something else... :P 18:34
18:36 Ven_ left 18:37 Ven_ joined 18:38 telex left 18:39 Ven_ left, brrt joined 18:40 telex joined, brrt left
llfourn does p6 have something like __DATA__? 18:40
psch zengargoyle: that might be NativeCall free()-ing before the lib expects it
zengargoyle: i'm not really into that, though, so take that as a "there might be something spooky somewhere", nothing else :) 18:41
llfourn: q:to//
18:41 Ven_ joined
psch llfourn: the difference there is that __DATA__ (afair) can be wherever in the file, but q:to// expects it directly after the statement 18:41
well, almost, __DATA__ still has to be at the end of the file i think? 18:42
llfourn psch: q:to// is like HEREDOC then?
psch: yeah data has to be at the end
psch llfourn: right, q:to// is probably more like HEREDOC
llfourn psch: kk thanks :)
psch llfourn: design.perl6.org/S02.html#Double-un...core_forms for reference 18:44
zengargoyle psych: found my bug, a rogue '-' snuck in somewhere and my lack of adding || fail bit me. 18:45
do submethod DESTROY {…} get called when you do $obj = Type, or how do the Perl 5 like $obj=undef to get DESTROY to free something? 18:47
i'm not seeing my debug note in the DESTROY happening.
psch zengargoyle: you can't really rely on DESTROY being called if you don't call it yourself, due to GC-specifics i don't understand 18:50
zengargoyle: note that it was added primarily to make Inline::Perl5 not leak things, and Inline::Perl5 calls it manually 18:51
zengargoyle: it gets called whenever an object gets GC collected, but this might not happen on program exit
Skarsnik nativecall has a explicitly-manage to avoid the GC to do work on stuff 18:52
*not
AlexDaniel m: say ^1000.pick 18:58
camelia rakudo-moar f4266f: OUTPUT«Potential difficulties:␤ Precedence of ^ is looser than method call; please parenthesize␤ at /tmp/6GzxSAlgWN:1␤ ------> 3say ^10007⏏5.pick␤^1000␤»
AlexDaniel m: say (^1000).pick
camelia rakudo-moar f4266f: OUTPUT«615␤»
18:58 uruwi joined
dalek href="https://modules.perl6.org:">modules.perl6.org: 1877371 | (Zoffix Znet)++ | web/ (2 files):
Add P6Project::SpriteMaker module
 18:59
psch AlexDaniel: we got auto-unspace for dotty ops 19:03
m: say ^1000 .pick
camelia rakudo-moar f4266f: OUTPUT«931␤»
psch (actually i think it's precedence lowering relative to before..?) 19:04
AlexDaniel psch: oh nice! Thanks!
19:12 yqt joined 19:17 nys joined 19:18 rindolf left 19:19 pmurias left
arnsholt Skarsnik, zengargoyle: explicitly-manage is only for strings, and also a terrible terrible hack 19:19
And the reason you can't rely on DESTROY being fired is that when a process is exiting, it's cheaper to just free all of the memory you've allocated (or just exit without freeing) than doing a GC run 19:20
After all, it's all garbage at that point 19:21
19:21 pmurias joined
arnsholt And if your process is short enough to not trigger a GC during execution of the mainline, no DESTROYS will be fired 19:21
19:24 Ven_ left 19:25 Ven joined
psch there is nqp::force_gc, but it still relies on some sort of internals knowledge 19:26
m: use nqp; class A { submethod DESTROY { say "destroyed" } }; for ^1000 { A.new }; nqp::force_gc
camelia ( no output )
psch m: use nqp; class A { submethod DESTROY { say "destroyed" } }; for ^10000 { A.new }; nqp::force_gc
camelia rakudo-moar f4266f: OUTPUT«destroyed␤destroyed␤destroyed␤destroyed␤destroyed␤destroyed␤destroyed␤destroyed␤destroyed␤destroyed␤destroyed␤destroyed␤destroyed␤destroyed␤destroyed␤destroyed␤destroyed␤destroyed␤destroyed␤destroyed␤destroyed␤dest…»
psch as in, the different gc gens have sizes and whatnot
as i said, i don't really know anything about the moar gc 19:27
zengargoyle i'm more concerned that the library will have done something (lockfile, other permanent thing) that doesn't get cleaned up unless the free() is called at some point. 19:37
probably not in this case...
psch zengargoyle: that's probably best served by an END block 19:39
19:39 Ven left 19:40 Ven joined, Praise left 19:43 Praise joined, Praise left, Praise joined 19:47 tokuhiro_ joined 19:51 tokuhiro_ left
dalek kudo/nom: 4b8937f | lizmat++ | src/core/Str.pm:
Fix some off-by-one and skip-empty issues
 19:53
19:55 Ven left 19:56 Ven joined 20:02 Begi1115 left
lizmat m: dd "abcde".split("",:all) # historically, this would *not* put any delimiters in the result 20:02
camelia rakudo-moar f4266f: OUTPUT«List $var = $("", "a", "b", "c", "d", "e", "")␤»
lizmat in the light of the new :v,:k,:kv,:p on split, I think this is inconsistent 20:03
TimToady: do you agree ?? ^^
20:03 Begi1115 joined 20:06 khw left 20:10 Ven left 20:11 Ven joined 20:14 ^elyse^ left
dalek kudo/nom: 8b834cc | lizmat++ | src/core/Str.pm:
Fix an off-by-one just introduced :-(
 20:18
moritz did anybody mention libcello.org/learn/garbage-collection yet? 20:20
20:24 Ven left
dalek c: fa86569 | lizmat++ | doc/Type/Str.pod:
Document all old/new features of Str.split
 20:25
20:27 darutoko left 20:31 Ven joined
Ulti looks like done is now a reserved word? what do you use to end tests? 20:37
lizmat done-testing 20:38
done is now part of the supply { } sugar
as is emit()
Ulti okedoke, thanks 20:39
lizmat yw :-)
Ulti: of course, using a plan is always better!
Ulti ahh yeah I remember changing from done_testing with my other tests
:P
20:40 Ven left, Ven joined
lizmat m: &&::{}[];; 20:40
camelia ( no output )
lizmat m: say &&::{}[];;
camelia rakudo-moar 8b834c: OUTPUT«Nil␤»
dalek p: fb0aad7 | (Pawel Murias)++ | src/vm/js/QAST/Compiler.nqp:
[js] Implement QAST::Stmt.
 20:41
p: 61f36ca | (Pawel Murias)++ | tools/build/MOAR_REVISION:
Bump Moar for fix with nqp::escape
p: c4c2b73 | (Pawel Murias)++ | t/nqp/59-nqpop.t:
Test that nqp::escape works on \r\n.
ast: 73f6740 | lizmat++ | integration/weird-errors.t:
Fudge what seems to be a bogus test now
 20:48
20:54 kaare_ left
lizmat m: sub a($x is rw) { $x; }; a 42 # dies as expected 20:55
camelia rakudo-moar 8b834c: OUTPUT«Parameter '$x' expected a writable container, but got Int value␤ in sub a at /tmp/2NK1Hbv1Ub:1␤ in block <unit> at /tmp/2NK1Hbv1Ub:1␤␤»
lizmat m: sub a($x is rw) { $x; }; { a 42 }() # but not if within a block ? 20:56
camelia ( no output )
lizmat m: sub a($x is rw) { $x; }; { a 42 } # seems to be optimizer related 20:58
camelia ( no output )
lizmat $ perl6 --optimize=off -e 'sub a($x is rw) { $x }; { a 42 }'
Parameter '$x' expected a writable container, but got Int value
RT #126585 21:01
dalek ast: 4ab8381 | lizmat++ | S06-traits/misc.t:
Fudge test for RT #126585
 21:02
21:04 fjk joined, fjk left
lizmat All tests successful. 21:08
Files=1071, Tests=49769, 225 wallclock secs (12.20 usr 3.41 sys + 1383.10 cusr 125.43 csys = 1524.14 CPU)
.oO( just a nice sight :-)
psch lizmat: that "is rw" thing is weird. before my patch relating that "sub f (int $x is rw) { 'int' }" would dispatch successfully, but it's clearly wrong 21:09
lizmat psch: well, it's an optimizer issue, as it fails correctly if switched off or the body of the sub is more than just $x
psch lizmat: i have no clue how exactly my patch interacts with what we had before to make it work as-is currently, though...
21:10 Ven left
psch lizmat: yeah, my patch was indirectly Optimizer related as well 21:10
21:10 ^elyse^ joined
lizmat m: sub a($x is rw) { $x; say "huh" }; { a 42 } 21:10
camelia rakudo-moar 8b834c: OUTPUT«Parameter '$x' expected a writable container, but got Int value␤ in sub a at /tmp/AkQtodCfhj:1␤ in block <unit> at /tmp/AkQtodCfhj:1␤␤»
lizmat so as soon as the body is more than just $x, it fails correctly
psch star-m: sub a($x is rw) { "foo" }; say a 5 21:11
camelia star-m 2015.09: OUTPUT«foo␤»
psch star-m: sub a($x is rw) { $x }; say a 5
camelia star-m 2015.09: OUTPUT«5␤»
psch star-m: sub a($x is rw) { $x = $x + 1 }; say a 5
camelia star-m 2015.09: OUTPUT«Cannot assign to an immutable value␤ in sub a at /tmp/lJnKvlKbm5:1␤ in block <unit> at /tmp/lJnKvlKbm5:1␤␤»
21:11 Ven joined
psch note that's not Parameter::RW 21:11
so yeah, my patch is definitely just a side-grade... :)
lizmat well, at least it's not changing the value of the constant 5 :-) 21:12
like in Perl 5, you could change the value of undef
dalek href="https://modules.perl6.org:">modules.perl6.org: cf95cb4 | (Zoffix Znet)++ | / (23 files):
Use a sprite instead of multiple images (Closes #21)
 21:15
21:16 advwp left, eliasr left 21:17 raoulvdberge joined 21:18 eliasr joined
psch lizmat: yeah, it's weird. i'll look at it again tomorrow, jvm ExceptionHandling is a bit too tough atm and xmas kinda matters more anyway 21:22
dalek osystem: 72a76b1 | (Zoffix Znet)++ | README.md:
Suggest to check for common errors in META file
 21:23
lizmat yeah, and it seems to fail when it really matters
Timbus could I get a very quick rundown on how newline parsing now works in perl6, and if theres any way to redefine the delimiter, or what have you? I'm specifically pointing at this CCLASS_NEWLINE thingy 21:24
its now matching \n but I guess it used to match \r sometimes?
21:25 Ven left
Timbus is there a uhh $INPUT_RECORD_SEPARATOR that I can use now 21:25
21:26 Ven joined
lizmat $?NL ? 21:26
m: say ?NL
camelia rakudo-moar 8b834c: OUTPUT«5===SORRY!5=== Error while compiling /tmp/7zdDPEAfTP␤Undeclared name:␤ NL used at line 1␤␤»
lizmat m: say $?NL 21:27
camelia rakudo-moar 8b834c: OUTPUT«␤␤»
lizmat m: print $?NL
camelia rakudo-moar 8b834c: OUTPUT«␤»
RabidGravy Zoffix, strangely I am just making a thing to test META files
Timbus lizmat, does this change the behaviour of matching CCLASS_NEWLINE in nqp?
lizmat afaik, it doesn't 21:28
Timbus oh
then Supply.lines needs a fix
lizmat the synthetic takes the properties of the base char, which is CR, which is CCLASS_NEWLINE
21:28 raoulvdberge left
lizmat Timbus: it probably does... :-) 21:28
busy with Str.split testing now... will look at it after that unless someone beats me to it :-) 21:29
21:29 raoulvdberge joined
ZoffixW RabidGravy, I'm secretly watching you. 21:30
:)
RabidGravy, I've also made this checker github.com/zoffixznet/p6-ecosystem-meta-checker and the modules.perl6.org builder also has a checker that outputs any errors at the end of the log file: modules.perl6.org/log/update.log
moritz, the log tells me Mojolicious::Plugin::AssetPack module is not installed 21:32
RabidGravy my plan is to just have a "meta-ok" sub which just laughs if there is no hope of it working
ZoffixW moritz, and the sprite stuff would also need Imager::File::PNG installed
21:36 khw joined
dalek osystem: 6c47654 | (Zoffix Znet)++ | README.md:
Fix typo
 21:36
21:37 xfix left 21:44 espadrine left 21:46 advwp joined, pmurias_ joined
dalek oblem_solver_tutorial: 8ad43f9 | lichtkind++ | chapter/text0.md:
about demo program
 21:46
oblem_solver_tutorial: d2ab7e2 | lichtkind++ | chapter/text0.md:
Merge branch 'master' of github.com/perl6/problem_solver_tutorial
21:48 tokuhiro_ joined 21:50 pmurias left 21:51 zakharyas joined 21:53 tokuhiro_ left 21:55 Ven left, Ven joined 21:56 cognominal left
ugexe and thats 1 years of commits. what a terrible idea that was... 21:57
21:57 cognominal joined
zengargoyle can you do casting in NativeCall? i.e. sub foo(Pointer[void]) is native(…); my CArray[uint32] $a.=new; foo(Pointer[void]($a)); # or something like that 21:58
21:58 TEttinger joined
pmurias_ ugexe: ? 21:59
21:59 pmurias_ is now known as pmurias
zengargoyle or would it be better to have multi sub foo(<each type>) or sub foo(type1|type2|type3) 21:59
the library i'm trying to wrap returns a few things that it expects me to free() 22:01
dalek href="https://modules.perl6.org:">modules.perl6.org: c50c8b3 | (Zoffix Znet)++ | Build.PL:
Add new modules needed by spriter to the prereqs
ugexe pmurias: i did at least 1 perl/perl6 github commit every day for the last year
22:02 Begi1116 joined
AlexDaniel ugexe: and why is that a terrible idea? 22:04
22:04 Begi1115 left
pink_mist AlexDaniel: sounds like a terrible schedule to haveto stick to 22:04
ugexe ^
AlexDaniel actually, I did the same thing. But somehow it slipped simply because I had a really busy week 22:05
and then I decided not to keep it
only 179 days though 22:06
pink_mist it's the 'decided not to keep it' bit that's different I think...
TEttinger ugexe: congrats. keep it up, I'm at 511 days (512 if/when I commit something today) 22:09
vendethiel has us both beat on longest streak 22:10
ugexe heh i at this point im not sure i could stop without actively trying
22:10 Ven left
TEttinger same here 22:10
vendethiel *g* 22:11
TEttinger my commits haven't been to one project, and they've been in a number of languages
22:12 Ven joined
ugexe hmm so what kind of mental problems should i look forward to developing? 22:14
ZoffixW OCD :) 22:15
ugexe that got me this far already
ZoffixW Then nothing to worry about :P
ugexe so much for scoring some drugs out of it 22:16
22:17 virtualsue joined
moritz ZoffixW: re modules.perl6.org, I installed the modules, but fucked up the perlbrew environment 22:18
ZoffixW :)
moritz the updating script uses "source ~/perl5/perlbrew/etc/bashrc"
but from cron I start it with "sh update-modules.perl6.org"
and sh != bash
should run again in one minute 22:19
dalek href="https://modules.perl6.org:">modules.perl6.org: 3dae26d | (Zoffix Znet)++ | web/ (8 files):
Use pure CSS instead of images for travis stuff
 22:20
22:22 zakharyas left
dalek href="https://modules.perl6.org:">modules.perl6.org: cd6a3bc | (Zoffix Znet)++ | web/assets/css/style.css:
Make travis buttons fill the cell and change on hover
 22:24
22:25 Ven left 22:28 Ven joined 22:30 AlexDaniel is now known as alexine`, alexine` is now known as alexine
ZoffixW moritz++ sweet :D modules.perl6.org works now 22:32
ZoffixW leaves for a beer
22:32 ZoffixW left 22:33 raiph left
dalek kudo-star-daily: 8b7e4eb | coke++ | log/ (9 files):
today (automated commit)
 22:35
kudo-star-daily: a5a33d2 | coke++ | log/ (9 files):
today (automated commit)
rl6-roast-data: bb2876a | coke++ | / (9 files):
today (automated commit)
rl6-roast-data: eb9022c | coke++ | / (9 files):
today (automated commit)
22:37 raiph joined, eliasr left
dalek kudo/nom: 4988c70 | lizmat++ | src/core/Str.pm:
Fix :skip-empty handling on empty splitter
 22:38
alexine m: say ^2**128 .pick.base(36); # hm, this kinda works as a password generator 22:40
camelia rakudo-moar 8b834c: OUTPUT«8TK82I6IS9BMZ25SZI2RAHSY4␤»
22:40 Ven left
pink_mist nice .. except most places require mixed-case passwords :/ 22:41
22:43 Ven joined 22:55 ^elyse^ left 22:56 BenGoldberg joined
dalek osystem: 31c458d | RabidGravy++ | META.list:
Add META6

  github.com/jonathanstowe/META6		 22:57
RabidGravy Zoffix, there you go ;-)
If anyone can think of any "customary" usage in META files that I have missed please PR me ;-) 22:58
I'll do the Test::META tomorrow 22:59
22:59 Begi1116 left
RabidGravy It's interesting that the "numbers of modules per author" goes almost exactly as you would predict 23:08
Skarsnik ?
RabidGravy number -> number of modules per author: 23:10
37 1
16 2
10 3
8 4
7 5
5 6
5 7
4 8
1 10
2 11
1 13
1 14
1 18
1 22
1 24
1 25
1 27
:)
cat META.list | sed -e 's/https:\/\/raw.githubusercontent.com\///' -e 's/\/.*$//' | sort | uniq -c | sort -n | awk ' { print $1 }' | uniq -c
for the interested
23:10 Ven left
Skarsnik I am still not sure to understand how to read this, I am probably too tired x) 23:11
RabidGravy There are lots of authors with one module, and the number of authors with more modules goes down in a curve 23:12
until it gets to 10 where it sort of asymptotes out
Skarsnik Oh I see 23:13
who has 27 modules x)
RabidGravy retupmoca
23:14 Ven joined
RabidGravy the strange thing is that the people in the top ten hasn't changed much just the order slightly but I guess the more prolific people are, er, more prolific 23:15
23:19 virtualsue left 23:20 pmurias left
RabidGravy Skarsnik, BTW "Invalid json found at: raw.githubusercontent.com/Skarsnik...META.info" from the modules.perl6.org/log/update.log 23:24
23:26 Ven left 23:30 Ven joined
RabidGravy tony-o # Module::Does utter crack but fab 23:30
23:30 virtualsue joined
RabidGravy tony-o++ # Module::Does utter crack but fab even 23:30
pink_mist might need a (tony-o)++ ... not sure how sensitive the karma bots are about - in the identifier 23:32
RabidGravy :) 23:34
who knows
lizmat good night, #perl6!
RabidGravy goonight
Zoffix Skarsnik, I've sent a PR to fix that couple of hours ago 23:36
yoleaux 4 Nov 2015 09:38Z <Ven> Zoffix: yes, thanks for your PR. I'll try to take a look at the noncrossed ones soon-ish
RabidGravy Zoffix++ # star 23:37
23:37 N joined, N is now known as Guest53701
Guest53701 is :kv argument to grep function recent I can not seem to get it to work 23:37
RabidGravy yes 23:38
Guest53701 thanks I will try to download latest thing
RabidGravy m: say <a b c d>.grep(/d/, :kv) 23:39
camelia rakudo-moar 4988c7: OUTPUT«(3 d)␤»
Guest53701 it must be recent I have 2015.06 23:41
23:41 Ven left
RabidGravy oh yes waaaay old, lots and lots of changes since then 23:41
23:41 Ven_ joined 23:44 alexine left
RabidGravy right now, it's probably a good move to keep as up to date as possible so the small changes don't build up :) 23:44
23:45 virtualsue left
Guest53701 m: my $max = [max] ("asdas","qweqwe","gfdgdfg").map({.chars}); say @arr.grep({.chars==$max},:kv); 23:48
camelia rakudo-moar 4988c7: OUTPUT«5===SORRY!5=== Error while compiling /tmp/6Jx7judgn_␤Variable '@arr' is not declared␤at /tmp/6Jx7judgn_:1␤------> 3,"qweqwe","gfdgdfg").map({.chars}); say 7⏏5@arr.grep({.chars==$max},:kv);␤»
Guest53701 m: my @arr = "asdas","qweqwe","gfdgdfg"; my $max = [max] @arr.map({.chars}); say @arr.grep({.chars==$max},:kv); 23:49
camelia rakudo-moar 4988c7: OUTPUT«(2 gfdgdfg)␤»
Guest53701 hm it does not output that on my machine
23:50 tokuhiro_ joined
Guest53701 it only gives me (gfdgdfg) 23:50
I upgraded
perl6.bat -v says This is perl6 version 2015.09 built on MoarVM version 2015.09 23:51
llfourn Guest53701: that's still old :P 23:52
RabidGravy that's still nearly two months old
23:52 raoulvdberge left
RabidGravy camelia is updated very frequently 23:52
Guest53701 there seems to be no windows download for latest thing I guess I better find linux 23:54
23:54 tokuhiro_ left
RabidGravy don't know what the status of packaged build is, sorry 23:55
Guest53701 thanks
23:55 Ven_ left 23:56 Guest53701 left
Zoffix Guest22572, there's a Sept. build of Rakudo Star: rakudo.org/downloads/star/rakudo-st...0(JIT).msi 23:56
Oh, they left.
llfourn isn't that what he had? 23:57
23:57 Ven joined
Zoffix Ah, right. I only saw the "I have 2015.06" above and not the "I upgraded" bit 23:58

Powered by Cro and the Raku® Programming Language.

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

Thank you!