»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'p6: say 3;' or rakudo:, or /msg camelia p6: ... | irclog: irc.perl6.org or colabti.org/irclogger/irclogger_logs/perl6 | UTF-8 is our friend!
Set by moritz on 22 December 2015.
AlexDaniel m: my %a = 'A' => 1, 'B' => 2, 'cD' => 34, 'Ef' => 56; my %b = %a.map: { .key.lc => .value }; dd %a; dd %b 00:00
camelia Hash %a = {:A(1), :B(2), :Ef(56), :cD(34)}
Hash %b = {:a(1), :b(2), :cd(34), :ef(56)}
AlexDaniel m: my %a = 'A' => 1, 'B' => 2, 'cD' => 34, 'Ef' => 56; my %b = %a.map: { .key.fc => .value }; dd %a; dd %b # probably
camelia Hash %a = {:A(1), :B(2), :Ef(56), :cD(34)}
Hash %b = {:a(1), :b(2), :cd(34), :ef(56)}
ugexe now what about when you have A and a? 00:01
00:01 rba_ joined 00:03 rba__ left
AlexDaniel oh, right 00:05
m: my %a = 'A' => 1, 'B' => 2, 'cD' => 34, 'a' => 56; say %a.categorize({.key.fc}) 00:08
camelia {a => [a => 56 A => 1], b => [B => 2], cd => [cD => 34]}
AlexDaniel almost
m: my %a = 'A' => 1, 'B' => 2, 'cD' => 34, 'a' => 56; say %a.categorize({.key.fc}, as => {.value}) 00:09
camelia {a => [56 1], b => [2], cd => [34]}
AlexDaniel there
m: my %a = 'A' => 1, 'B' => 2, 'cD' => 34, 'a' => 56; say %a.classify({.key.fc}, as => {.value})
camelia {a => [56 1], b => [2], cd => [34]}
AlexDaniel m: my %a = 'A' => 1, 'B' => 2, 'cD' => 34, 'a' => 56; say %a.classify: *.key.fc, as => *.value 00:11
camelia {a => [56 1], b => [2], cd => [34]}
00:13 Rawriful left 00:16 rba__ joined 00:17 Actualeyes left 00:18 rba_ left 00:29 ipatrol joined
ipatrol is the named captures syntax actually an interpolated form of assignment, or does it just look that way? 00:30
00:31 rba_ joined 00:32 Exodist left 00:33 silug left, Exodist joined, rba__ left 00:34 Cabanossi left
raschipi ipatrol: It looks like an asignement on purpose, but it's special syntax not really related to asignments. 00:35
ipatrol raschipi: I'm not sure if I'm a big fan of it, since it puts the name outside of the group instead of as a flag within it 00:37
00:37 Cabanossi joined 00:39 pierre_ joined, ryu0 left, ryu0 joined
ipatrol in other notes, is there anything we could really call a "Perl 6 editor/IDE" available yet? 00:40
00:40 pierre_ left, pierre_ joined 00:41 enheh joined
raschipi There's a plugin for padre and for intelij, I think. Don't know if they are any good. 00:45
00:46 rba__ joined 00:48 rba_ left
ipatrol raschipi: what about one that gives you an extra pallet for unicode operators? 00:49
00:50 nuk9 joined
raschipi you mean an on-screen keyboard? I don't think there's any. 00:51
You don't need to use them anyway.
00:52 pilne joined
ipatrol raschipi: It might be nice to have them anyway. Mathematica does this with a "Basic Math Pallet" to the side of the editor window 00:52
raschipi Most people use xcompose on linux, it's an input method that would work for any ide 00:53
ipatrol raschipi: yes, but then you have to remember the sequences. I think a pallet would be a nice feature 00:54
raschipi Well, I don't think it belongs in the IDE, but go ask them if you think it would be a good feature. 00:55
00:57 astj joined
ipatrol raschipi: as far as I know, Mathematica is the only other language that by default makes use of Unicode operators (it also has an ASCII syntax for them, FYI). We could do a lot worse than cribbing a few ideas from them :-) 00:59
teatime and then there's APL... 01:00
ipatrol but that was pre-Unicode, technically
teatime yar 01:01
ipatrol the symbols later got mapped to Unicode codepoints, but by the time that happened, I'm not sure if anyone was even using APL on a production level anymore.
01:02 rba_ joined
mspo hm 01:03
01:03 rba__ left
MasterDuke ipatrol: i think the closest to an IDE is Atom 01:04
github.com/perl6/Atom-as-a-Perl6-IDE 01:05
01:05 Cabanossi left 01:07 Cabanossi joined
Xliff What is the best way to lowercase all keys in Hash? 01:12
(Or would I need to create a new hash with lowercased keys?)
Juerd my @keys = %hash.keys».lc; 01:13
Xliff m: my %a = ( 'A' => 1, 'B' => 2, 'cD' => 34, 'Ef' => 56 \; my @keys = %hash.keys».lc; dd %a; 01:14
camelia 5===SORRY!5=== Error while compiling <tmp>
Confused
at <tmp>:1
------> 3' => 1, 'B' => 2, 'cD' => 34, 'Ef' => 567⏏5 \; my @keys = %hash.keys».lc; dd %a;
expecting any of:
infix
infix stopper
statement …
AlexDaniel huggable: xy
huggable AlexDaniel, You want to do X, but don't know how. You think you can solve it using Y, but don't know how to do that, either. So, you ask about Y in order to solve X, which doesn't make sense. You should ask about X.
Juerd Oh, in the existing hash? Keys can't be changed afterwards. You'd need to create a new hash.
AlexDaniel Xliff: what are you trying to do?
Xliff m: my %a = ( 'A' => 1, 'B' => 2, 'cD' => 34, 'Ef' => 56 ); my @keys = %hash.keys».lc; dd %a;
camelia 5===SORRY!5=== Error while compiling <tmp>
Variable '%hash' is not declared. Did you mean any of these?
&hash
Hash

at <tmp>:1
------> 032, 'cD' => 34, 'Ef' => 56 ); my @keys = 7⏏5%hash.keys».lc; dd %a;
AlexDaniel Xliff: I mean, why do you need lowercased keys?
Xliff Juerd: Yes, exactly.
AlexDaniel: There is no why. I just wanted to if I could do that in perl6. 01:15
Juerd Xliff: So, if you need lowercased keys, why did you put uppercased letters in there in the first place? :)
Xliff And I think your xy thing is oversimplifying a lot of things.
Sometimes we ask for Y because it is a simplified version of X.
Think golfing.
AlexDaniel :/ 01:16
Xliff Well, here's the whole use case
1) I have a routine with named parameters
01:16 rba__ joined
Xliff But I don't want to create an overly complex signature 01:16
Mainly because I want flexible usage and that means programmatic parameter validation in-sub. 01:17
raschipi Create a new type that's a hash but lowercases keys.
Xliff So in that situation, creating a new hash is a valid option.
But then I got sidetracked and wondered if %hash.keys would give me something I could assign to.
And it doesn't. Question answered
AlexDaniel also, .fc is technically more correct (although you're not going to have unicodey parameters I suppose) 01:18
Xliff AlexDaniel: Now that would be ridiculous. ;)
ipatrol I'd say the only major design difference of opinion I have with perl6, is the fact that subroutine and class identifiers don't generally have sigils
01:18 rba_ left
raschipi They had in Perl4, but it was taken out ages ago. It looks too ugly. 01:19
ipatrol because in my experience (Python), that makes it easy for new versions of the language to break old code
raschipi And remember that in Perl6 there's no difference between built-ins and user-defined. 01:20
01:21 pierre_ left, pierre_ joined
ipatrol raschipi: in theory. in practice some operators are "special cased" by the compiler 01:21
AlexDaniel Xliff: maybe it'd be a better idea to have a hash with default values, and then just set the values in that hash with whatever you got from the args 01:22
Xliff AlexDaniel: Default values is not the issue though.
The names of the arguments are.
raschipi Having the compiler cheats doesn't mean they're any different. Go look at the code, they have just normal definitions. 01:23
Juerd m: class LCHash is Hash { method ASSIGN-KEY(Str:D $key, $val) { nextwith $key.lc, $val }; method STORE_AT_KEY(Str:D $key, $val) { nextwith $key.lc, $val }; method AT-KEY(Str:D $key) { nextwith $key.lc }; method DELETE-KEY (Str:D $key) { nextwith $key.lc } }; my %lch is LCHash; %lch<FOO> = 'bar'; say %lch<Foo>; dd %lch;
camelia bar
{:foo("bar")}
Juerd cringes at the underscores in STORE_AT_KEY...
Xliff Oooh!
Juerd++ # I was already thinking about that approach. 01:24
Juerd Xliff: I think flexibility in named arguments is a bad design decision.
Xliff That's not the question I am trying to answer, though.
Juerd I'd rather have mixed styles between several APIs than within a single API.
[Coke] bah. looks like another node issue blocking the doc build.
Xliff What if I am not the person making the design?
ipatrol raschipi: I don't know how those compiler shortcuts interact with user-defined operators
Xliff But I want to work with said design (read: add on to) 01:25
Juerd Xliff: Then please send the designer my regards and comments. I'll happily discuss the issue with them.
Xliff :P
01:25 mcmillhj joined
Juerd Grepability is important. Try not to have two names for the same thing. 01:25
01:25 pierre_ left
raschipi ipatrol: it's not completely implemented yet. 01:25
Xliff What is design-appropriate, and the reality are two different things. 01:26
mspo I've seen plenty of perl5 code with &sub()
ipatrol If all operators are simply part of the environment, then it shouldn't be an issue if the user defines an operator that is already a builtin; the later user definition simply overwrite's the language's
Juerd You're the programmer, you get to fix your part of reality.
Xliff Juerd: Aaand... now this conversation has devolved into noise. ;) 01:27
Your points are well received, but it is not the question I asked.
Juerd Xliff: It was noise when STORE_AT_KEY came along with its underscores, next to ASSIGN-KEY, AT-KEY, and DELETE-KEY which have hyphens... 01:28
Xliff And whose fault is that??!
Aren't you a P6 dev?
ipatrol hey, play nice children ;-P
Juerd Xliff: I'm not here to answer questions. If I happened to answer your question while trying to have a conversation, then consider yourself lucky :P
Xliff <Juerd> You're the programmer, you get to fix your part of reality.
Juerd: I do. Believe me.
ipatrol raschipi: what parts of user-defined operators are still NYI in Rakudo? 01:29
MasterDuke ipatrol: you can define your own versions of operators and the builtins will get overwritten (in the lexical scope i believe)
raschipi ipatrol: macros
Xliff Should I submit a bug report asking for consistency between STORE_AT_KEY et al. :) 01:30
01:30 mcmillhj left
ipatrol MasterDuke: assuming none of them are special cased. If any of them are, I don't know what happens 01:30
raschipi But traits are already available.
Xliff (that was semi-joking)
(and how much code would I need to rewrite?)
raschipi ipatrol: And it's always cheating: it should be transparent or it's buggy.
MasterDuke Xliff: sure
ipatrol Xliff: s:g/STORE_AT_KEY/STORE-AT-KEY/ 01:31
01:32 rba_ joined
Xliff MasterDuke 01:33
...ask me next week. Pizza is being served!! ^_^
MasterDuke heh
01:33 rba__ left
ipatrol raschipi: one of the problems I've noticed that Python had, historically, was that since it forbids variable names from being the same as language operators, which are AYK typically English words, minor versions that were supposed to be backward compatible, weren't, because given a large enough codebase, some variable would inevitably clash with the new operator 01:35
AlexDaniel Juerd: I think it is from older times when most things had underscores
Juerd: since then all user-facing things were changed to use dashes, except for some of the internal stuff that you shouldn't use anyway 01:36
Juerd: but STORE_AT_KEY is simply a leftover from these times I think
Juerd: maybe ask lizmat about it, or submit a ticket
01:36 Cabanossi left
Juerd m: role TranskeyHash[&transformation] is Hash { method ASSIGN-KEY(Str:D $key, $val) { nextwith transformation($key), $val }; method STORE_AT_KEY(Str:D $key, $val) { nextwith transformation($key), $val }; method AT-KEY(Str:D $key) { nextwith transformation($key) }; method DELETE-KEY (Str:D $key) { nextwith transformation($key) } }; class LCHash does TranskeyHash[*.lc] { }; class UCHash does TranskeyHash[*.uc] 01:36
camelia 5===SORRY!5=== Error while compiling <tmp>
Unable to parse class definition
at <tmp>:1
------> 3 }; class UCHash does TranskeyHash[*.uc]7⏏5<EOL>
ugexe m: sub class { 42 }; say class # ipatrol ? 01:37
camelia 5===SORRY!5=== Error while compiling <tmp>
Unable to parse class definition
at <tmp>:1
------> 3sub class { 42 }; say class # ipatrol ?7⏏5<EOL>
expecting any of:
generic role
Juerd { }; my %lc is LCHash; my %uc is UCHash; %lc<Foo> = "bar"; %uc<fOo> = "bar"; say %lc<FOo>, %uc<fOO>, %lc.keys, %uc.keys;
Aw shit, doesn't fit on a line
ugexe m: sub class { 42 }; say class; # ipatrol ?
camelia 42
01:38 Cabanossi joined, pierre_ joined
ipatrol ugexe: ok, that shows the class operator is relatively well-behaved 01:38
ugexe it shows user defined operators cant be replaced by newer versions magically 01:39
01:39 cdg joined
Juerd m: role TKH[&tr] is Hash { method ASSIGN-KEY(Str:D $k, $v) { nextwith tr($k), $v }; method STORE_AT_KEY(Str:D $k, $v) { nextwith tr($k), $v }; method AT-KEY(Str:D $k) { nextwith tr($k) }; method DELETE-KEY (Str:D $k) { nextwith tr($k) } }; class LCHash does TKH[*.lc] { }; class UCHash does TKH[*.uc] { }; my %lc is LCHash; my %uc is UCHash; %lc<Foo> = "bar"; %uc<fOo> = "bar"; dd %lc, %uc; 01:39
camelia {:foo("bar")}
{:FOO("bar")}
Juerd ^ golfed to fit on a line.
ipatrol right, because code in the script itself takes precedence over the implicit code loaded by the interpreter
Juerd I can think of other useful transformations, such as underscores to kebab case for MIME headers. 01:40
ipatrol Juerd: "kebab case" sounds vaguely racist, for some reason :-P
Juerd ipatrol: What's racist about meat on a skewer?
ipatrol it's a slur for Turks 01:41
Juerd It's also delicious
01:41 mcmillhj joined
ugexe there is a tutorial for http headers 01:42
docs.perl6.org/language/subscripts...pe_example
01:42 enheh left 01:43 cdg left
Juerd Heh, I probably should have added EXISTS-KEY to my example :) 01:43
01:45 ilbot3 left
ipatrol A more minor design disagreement: the way the Cool class and the IO role are used 01:45
01:45 mcmillhj left
Juerd ugexe: I'm not sure HTTP is a good base for such an example, because in HTTP, the same header can be used multiple times (e.g. Set-Cookie) 01:46
ipatrol p6: say 'kitten'.atan;
camelia Cannot convert string to number: base-10 number must begin with valid digits or '.' in '3⏏5kitten' (indicated by ⏏)
in block <unit> at <tmp> line 1
Juerd lizmat: Do you happen to know why STORE_AT_KEY isn't kebab cased?
01:46 rba__ joined
AlexDaniel greppable6: STORE_AT_KEY 01:47
greppable6 AlexDaniel, gist.github.com/a67a16b12e07808765...cc23d2fad6
AlexDaniel greppable6: DELETE-KEY 01:48
01:48 rba_ left
greppable6 AlexDaniel, gist.github.com/4763ee26d33e8f6ab1...e4a609ac79 01:48
ugexe so throw an error if they do that 01:50
01:51 ilbot3 joined, ChanServ sets mode: +v ilbot3
ipatrol I know Perl* historically likes to treat various things as interchangeable with strings, but in practice I've found that coercing values to and from strings is often prone to hard-to-debug errors, and I'm inclined to say that such conversions should never be implicit unless a meaningful result is always possible 01:52
ugexe so write multi methods with type constraints 01:54
ryu0 ipatrol: converting to string is usually non-issue. not so for from so much.
ipatrol ryu0: yes, most things have some kind of string representation, even if it's only useful for printing messages 01:55
ugexe multi method AT-KEY(Str $key) {...} multi method AT-KEY(|c) { ... } 01:56
ipatrol ugexe: that doesn't "fix" the builtin methods. Now, one can enforce a coding style where all conversions must be explicitly laid out, and thankfully Perl6 does, of course, support such a system
p6: '3.14159'.Int.tan;
camelia ( no output )
ipatrol p6: say '3.14159'.Int.tan; 01:57
camelia -0.142546543074278
01:57 mcmillhj joined
ugexe you can store objects as hash keys 01:57
Juerd ipatrol: I disagree, I used Perl 5 a lot and almost never run into this. The only weird thing in this area is "0" being false, really...
raschipi say so "0"
evalable6 True
Juerd ipatrol: The best (unintended) feature in Perl 5 is also a very hard one if you don't know what you're doing: byte strings and text strings are the same thing and can be used interchangably. I really miss that in Perl 6. 01:58
Juerd very regularly uses regexes on binary strings of around 100 MB. 01:59
Writing the same thing without regexes is a lot of work.
ipatrol Juerd: Python made the same change in v3
Juerd: you can't coerce the binary to utf-8? 02:00
Juerd ipatrol: Encoding and decoding would be very slow.
02:01 rba_ joined, mcmillhj left
raschipi ipatrol: If you want to ensure some variable can only hold Numeric data, declare it Numeric. 02:02
ipatrol Juerd: does Perl6 actually need to do any converting to store a byte string as utf-8?
02:03 rba__ left
ipatrol Juerd: only thing to be aware of is that you should use utf8-c8 to avoid hitting problems with invalid byte sequences 02:05
Juerd ipatrol: I don't understand the question. What's "storing as utf-8" exactly? And even then I probably couldn't answer your question because I don't know much about Perl 6 internals. 02:06
raschipi ipatrol: Str in P6 is normalized and stored in strands, so it doesn't have much in common with a sequence of bytes. 02:07
ipatrol Juerd: if they are stored as utf-8, that implies that whatever array in the RAM that the variable ultimately references consists of a sequence of bytes that form the UTF-8 representation of the string 02:08
raschipi ipatrol: Perl6 will also create syntetic codepoints to represent utf-8 chords as one single thing. 02:09
Geth doc: 357e6e2f18 | (Will "Coke" Coleda)++ | doc/Type/Date.pod6
Fix sample output

Closes #1463
ipatrol raschipi: do regexes work on Uni objects?
raschipi Not yet. 02:10
ipatrol but they're planned to be?
raschipi Yes, that's why I said 'yet'. 02:11
ipatrol ok, so Juerd just needs to wait for that
02:11 Actualeyes joined
Geth doc: aeddc020bb | (Douglas L. Schrag)++ | doc/Language/operators.pod6
Add documentation of loose xor operator
 02:12
doc: 3186181df0 | (Douglas L. Schrag)++ | doc/Language/operators.pod6
Expand documentation of loose xor

Add links in loose or / xor to strict sections Summarize chaining behavior
doc: 3fc8e26fb6 | (Will Coleda)++ (committed using GitHub Web editor) | doc/Language/operators.pod6
Merge pull request #1460 from dmaestro/xor-operator

Add documentation of loose xor operator
raschipi No, he will use utf-c8, it's not really difficult.
02:12 mcmillhj joined
ipatrol raschipi: I thought you just said Str does unicode normalization? 02:13
raschipi utf8-c8 doesn't.
ipatrol ah, ok
02:16 rba__ joined 02:17 mcmillhj left 02:19 rba_ left
ipatrol Interestingly, Python 3's `re` module allows regexes to be used on binary data, so long as the pattern is also a binary sequence. 02:19
raschipi: doing so changes the semantics of some of the operators. Perhaps we could implement an adverb to make a blob regex? 02:21
raschipi ipatrol: Yes, that's the idea, but it's not implemented yet. 02:22
ipatrol raschipi: sounds like the setup to a joke 02:23
"How do you know Perl 6 is the language of the future?" "Because that's when all the best features will be implemented!"
raschipi Why should implementers concentrate on something Perl5 does well, since one can always call Perl5 code. 02:25
ipatrol raschipi: I thought what Perl 5 does well is being a shell replacement 02:26
raschipi No, never seen Perl5 work in any way as a shell replacement.
ipatrol I even compiled it for my wireless router so I can make administration scripts in Perl
02:27 pierre_ left
ipatrol raschipi: not interactively, but for shell scripts 02:27
raschipi It's not good as an interactive language.
Perl5 is a multipurpose language.
ipatrol raschipi: that it is, but that doesn't answer what it's best at
raschipi Well, where it's "best" is bioinfomatics, just like Perl6 is specially well suited for commercial analysis. 02:29
ipatrol the backtick inlining, the -n, -p, and -a switches, the diamond operator, the fact that most file system operations are builtins; to me that suggests a language for system administration and gluing programs together 02:30
02:30 mcmillhj joined
raschipi It's also good for that, yes. Perl6 is even better as a glue language in my opinion. 02:31
mspo perl -00 ++ :)
02:31 rba_ joined
ipatrol raschipi: I'd actually dispute Perl5's suitability for bioinfomatics somewhat. I don't think it's OO enough, which is why Python has been eating away at it in that arena 02:31
raschipi It's "eating away" because it's trendy and taught by default. Perl is the exchange format in bioinformatics, just like Tex is the exchange format of math. 02:33
02:34 rba__ left
ipatrol raschipi: the exchange formats are a set of file formats, none of which have anything to do with Perl or Python. 02:34
02:34 mcmillhj left
raschipi Nope, if you publish code in an academic setting, it better be in Perl or otherwise everyone will ignore you. 02:35
02:35 pierre_ joined
raschipi Same thing as math, if it's not in Tex, it might as well not exist. 02:35
02:35 Cabanossi left, mcmillhj joined
ipatrol raschipi: no, I think Perl became prominent because at the time of the Human Genome Project, it was the only really mature language that could efficiently process the text-based formats in use at the time like FASTA 02:36
02:37 Cabanossi joined
Geth doc: 6252e3d6f7 | (Will "Coke" Coleda)++ | doc/Type/Cool.pod6
use OUTPUT: tags
 02:38
doc: 0eb07d333d | (Will "Coke" Coleda)++ | 2 files
fix trim usage
raschipi Right, so you know what it's good for.
02:38 noganex joined
ipatrol raschipi: yes, it's also very good for text processing, though I think Perl 6 is even better for that 02:39
02:40 mcmillhj left 02:41 noganex_ left 02:46 rba__ joined
ipatrol I think Perl6 will really excel in the parsing and linguistic realms 02:48
02:49 rba_ left
b2gills m: class ::('☠') { method state { "DEAD" } }; constant term:<☠> = ::('☠'); ☠.new.state.say 02:49
camelia DEAD
02:51 mcmillhj joined 02:54 john_parr joined, pierre_ left 02:56 mcmillhj left, mcmillhj joined, Actualeyes left 03:00 Actualeyes joined
ipatrol raschipi: hey, I have some data to parse where the order of tokens is generally unimportant, but each token when present only has one meaning, and is not used for anything else. What would be a good way to construct a grammar for this? 03:00
03:01 mcmillhj left 03:02 rba_ joined 03:03 rba__ left
raschipi I don't think a grammar would be the right tool, I would use split to separate the tokens and them put them into a Hash. 03:04
03:04 Cabanossi left
ipatrol raschipi: the data is such that split would not work. That's why it's defied all my attempts to parse it so far, as it doesn't really lex cleanly 03:06
03:06 Cabanossi joined
ipatrol I was thinking just go down the list of tokens, s///'ing each of them out 03:07
03:07 BenGoldberg joined
raschipi right, a grammar would be easier to use, then. 03:07
ipatrol raschipi: using the proto-regexen? 03:08
raschipi define a grammar with a token rule which has all the tokens as an alternation and a rule TOP that calls for a sequence of tokens. 03:09
03:11 char_var[buffer] left 03:12 mcmillhj joined
ipatrol raschipi: would having TOP be a single [<proto><ws>?]* work? 03:12
raschipi The only thing you need there is to capture it, I think. 03:13
ipatrol raschipi: in some cases, I may want to swap the match out with something else 03:14
03:16 rba__ joined, mcmillhj left
raschipi right, the match will return a list of tokens and can substitute whatever then. 03:17
03:18 rba_ left 03:19 Actualeyes left
ipatrol raschipi: since regexes are now considered grammar, does interpolating strings into them still work? 03:24
raschipi There's special syntax for it.
03:24 snarkyboojum joined
ipatrol raschipi: and it is...? 03:25
raschipi don't remember
ipatrol raschipi: does it work with the regex operators too? 03:26
raschipi yes, why wouldn't it?
03:28 mcmillhj joined 03:31 rba_ joined
ipatrol m: my $patt = <spam camel butterfly>.join('|'); 'camelfly spam' ~~ /$patt/ and say ~$/; 03:32
camelia ( no output )
03:33 rba__ left, mcmillhj left
ipatrol m: say <spam camel butterfly>.join('|'); 03:35
camelia spam|camel|butterfly
ipatrol m: 'camelfly spam' ~~ /spam|camel|butterfly/ and say ~$/;
camelia camel
ipatrol raschipi: evidently it does not 03:36
raschipi right, but you're no using the necessary incantations. 03:37
which I don't remember. 03:38
ipatrol abracadabra? :-P
03:39 cdg joined
ipatrol m: my $patt = <spam camel butterfly>.join('|'); 'camelfly spam' ~~ / <$patt> / and say ~$/; 03:44
camelia camel
03:44 cdg left
ipatrol raschipi: found the incantation. It's angle brackets :-P 03:44
03:45 mcmillhj joined 03:46 rba__ joined
ipatrol presumably it forces the string being referenced to be treated as a regex subrule 03:48
03:48 rba_ left
teatime m: my @patt = <spam camel butterfly>; 'camelfly spam' ~~ / @patt / and say ~$/; 03:48
camelia camel
teatime I guess I should have renamed @patt to something else. 03:49
03:49 mcmillhj left
ipatrol teatime: oh that's even more interesting, but where is it documented? 03:49
teatime couple of places 03:50
when you interpolate a string by default it's treated as a literal to match against rather than pattern, as you established 03:51
if it's a @ of strings instead it matches like an alternation
03:52 llfourn joined
ipatrol teatime: how can I negate a portion of a regex? 03:55
teatime can you give more details / an example 03:59
ipatrol teatime: what's the P6 equivalent of (?!pattern) ? 04:00
*from P5
04:00 mcmillhj joined 04:01 rba_ joined
teatime (?!pattern) would be negative lookahead, so <!before pattern> 04:02
ipatrol ah, ok
teatime docs.perl6.org/language/regexes is pretty exhaustive, I think. 04:03
04:03 rba__ left, Cabanoss- joined, Cabanossi left
ipatrol teatime: it didn't explain either of those two interpolation tricks 04:04
04:04 Cabanoss- is now known as Cabanossi 04:05 mcmillhj left
ipatrol teatime: actually, looking at it, it doesn't explain interpolation in regexes at all 04:08
teatime fair enough. 04:09
raschipi time to sleep, bye everyone 04:13
04:13 raschipi left
ipatrol teatime: it is described in design.perl6.org/S05.html#Extensib...%3C...%3E) where it says "A leading $ indicates an indirect subrule call. The variable must contain either a Regex object (really an anonymous method--see above), or a string to be compiled as the regex. The string is never matched literally." 04:13
teatime yes. 04:15
ipatrol teatime: and the array syntax is explained in design.perl6.org/S05.html#Variable...erpolation 04:16
04:17 rba__ joined
teatime so I'm not 100% sure, but I'm thinking the fact that you interpolated your string into the pattern with < > was kindof an accident 04:17
like, it doesn't interpolate into the pattern, it is interpreted as a standalone subpattern 04:18
and perhaps this is how to say what you really meant:
m: my $patt = <spam camel butterfly>.join('|'); 'camelfly spam' ~~ / <{ $patt }> / and say ~$/;
camelia camel
ipatrol teatime: yeah, that's what I said. It coerces it into a subrule
04:19 rba_ left
teatime I think you'll find you don't need/want this behavior as often as you seem to be expecting, though 04:19
instead of building a pattern up out of strings it's more natural to just build with regex objects / subrules 04:20
ipatrol teatime: I wanted to store the information as a hash, so the list interpolation allows me to use the keys method of the hash to build the regex 04:21
because I might later move the information out of the code and into a config file 04:22
04:23 araraloren_fake joined
araraloren_fake good noon 04:23
ipatrol araraloren_fake: you live in East Asia then I take it? 04:25
araraloren_fake yeah, in China
teatime o/
araraloren_fake o/ 04:26
ipatrol \o/
araraloren_fake \o| 04:27
ipatrol At work the other day I had a group of 32 Chinese people show up. They were with some company. For where I work, that's a rather large group, so I jokingly asked the lady in charge if she was bringing the whole country with her 04:29
but anyway, </noise>. So, araraloren_fake, what can we help you with?
04:31 rba_ joined
araraloren_fake no, I'm come here often. 04:31
04:33 rba__ left
araraloren_fake :) And I can help others, I'm not newbie of Perl 6 04:33
teatime lols at the new ⚛ operators
araraloren_fake What is the atomicint for ? 04:37
teatime atomic/safe operations on native ints when multiple threads may be doing the operation 04:38
don't have to acquire a lock first before making the modification
ipatrol araraloren_fake: in that case, since I am a newbie, what can you help me with? :-P 04:39
araraloren_fake oh, so it's aim for safe operator on native int, not for Perl6 object type ?
ipatrol, :) you can ask me, if I know I'll tell you or where it can found in document. 04:40
04:41 mcmillhj joined
araraloren_fake teatime, I see, it's just like a Atomic integer of C/C++ 04:41
ipatrol araraloren_fake: ok, how can I set a hash, or else create a custom subtype thereof, that when a key is not found in the mapping, simply returns the requested key as a value?
04:44 Actualeyes joined, BenGoldberg left
araraloren_fake ipatrol, you can mixin a role with custome AT-KEY method 04:45
04:45 mcmillhj left 04:46 rba__ joined
araraloren_fake m: my %hash; %hash does role :: { method AT-KEY($key, |c) { return $key unless self.{$key}:exists; nextwith(|c); }; }; say %hash{2}; 04:48
camelia 2
araraloren_fake It's just like ^^^
04:48 xtreak joined 04:49 rba_ left
ipatrol araraloren_fake: what does nextwith(|c) do? 04:49
araraloren_fake ipatrol, it'll call next method in inheritance chains 04:50
in there, it'll call same name method AT-KEY of Hash
ipatrol araraloren_fake: what exactly does the |c stand for?
araraloren_fake It is a argument Capture 04:51
m: my %hash = f => 42; %hash does role :: { method AT-KEY($key, |c) { return $key unless self.{$key}:exists; nextwith($key, |c); }; }; say %hash{2}; say %hash<f>; 04:52
camelia 2
42
araraloren_fake I forget pass $key to nextwith, that's the final version 04:53
ipatrol araraloren_fake: so like python's asterisk operator?
araraloren_fake IDK python
ipatrol captures a variable number of arguments and returns them as a list 04:54
04:54 xtreak left
araraloren_fake m: sub foo(|c) { dd c; }; foo(1, 2, "23", :f(3), "z" => "w"); 04:54
camelia \(1, 2, "23", :z("w"), :f(3))
araraloren_fake ipatrol, it's Capture object here
04:54 xtreak joined
araraloren_fake m: sub foo(|c) { say c.WHAT; }; foo(1, 2, "23", :f(3), "z" => "w"); 04:55
camelia (Capture)
ipatrol which docs say gives you a list for positional arguments and a hash for named ones
04:56 xtreak left
araraloren_fake It's *@ and *% 04:56
04:56 xtreak joined
ipatrol or a method-based conversion likewise 04:56
04:57 mcmillhj joined
araraloren_fake m: sub foo(|c, *%nas) { say c; say %nas; }; foo(1, 2, "23", :f(3), "z" => "w"); 04:58
camelia \(1, 2, "23", :z("w"), :f(3))
{f => 3}
araraloren_fake m: sub foo(Int, |c, *%nas) { say c; say %nas; }; foo(1, 2, "23", :f(3), "z" => "w"); 04:59
camelia \(2, "23", :z("w"), :f(3))
{f => 3}
araraloren_fake m: sub foo(Int, *%nas, |c) { say c; say %nas; }; foo(1, 2, "23", :f(3), "z" => "w");
camelia \(2, "23", :z("w"))
{f => 3}
araraloren_fake m: sub foo(Int $z, *%nas, |c) { say c; say %nas; }; foo(1, 2, "23", :f(3), "z" => "w"); 05:00
camelia \(2, "23", :z("w"))
{f => 3}
araraloren_fake m: sub foo(Int $z, *%nas, |c) { say $z; say c; say %nas; }; foo(1, 2, "23", :f(3), "z" => "w");
camelia 1
\(2, "23", :z("w"))
{f => 3}
05:01 Cabanossi left, mcmillhj left
teatime sooo... there is this giant angry hornet nest (maybe 12" around and 14" tall) that I have to attempt to remove this morning 05:01
05:01 rba_ joined
teatime once more unto the breach, dear friends 05:01
05:01 skids left
ipatrol teatime: did you try using smoke? 05:01
teatime like a beekeeper? I don't have any equipment like that, heh 05:02
ipatrol teatime: all you need is a can with some holes in it 05:03
05:03 rba__ left
ipatrol araraloren_fake: why is the . needed between self and {$key}? 05:03
05:04 Cabanossi joined
ipatrol m: my %hash = f => 42; %hash does role :: { method AT-KEY($key, |c) { return $key unless self{$key}:exists; nextwith($key, |c); }; }; say %hash{2}; say %hash<f>; 05:04
camelia 2
42
teatime I wish I had the suit though, I'm pretty scared heh 05:05
ipatrol teatime: the other option is to kill them all with chemistry and then remove the thing
teatime yeah that's the plan 05:06
ipatrol if you're fine being fairly unspecific, burning some sulfur in a can underneath the hive will work quite well. Otherwise there are various insecticides, of course. 05:08
araraloren_fake: I'm not even sure why it worked with the dot. I thought calling obj.{$var} would try to return you a method with the name of whatever $var is once coerced to a strign 05:09
*string
araraloren_fake ipatrol, oh that's like dereference in Perl 5 05:10
docs.perl6.org/language/5to6-nutsh...eferencing
ipatrol araraloren_fake: your version treated self as a reference to a hash, whereas mine treated it as the hash itself 05:12
05:12 mcmillhj joined
araraloren_fake yeah, cause self is sigilless variable, I prefer dereferencing it and call method on it 05:14
sigilless variable is not in any context
05:14 xtreak left, pilne left 05:16 kannan joined, rba__ joined 05:17 mcmillhj left
ipatrol araraloren_fake: whereas I'm inclined to think of self as representing the actual object 05:18
05:19 rba_ left
ipatrol m: say Any.defined; 05:20
camelia False
ipatrol m: say Any // 'pie'; 05:21
camelia pie
ipatrol m: my %hash = f => 42; %hash does role :: { method AT-KEY($key, |c) { nextwith($key, |c) // $key; }; }; say %hash{2}; say %hash<f> 05:22
camelia (Any)
42
ipatrol hm 05:23
05:27 wamba joined
araraloren_fake ipatrol, maybe and the docs said it represent invocant(docs.perl6.org/routine/self). And you can custom the method invocant 05:27
ipatrol m: my %hash = {'f' => 42}; %hash does role :: { method AT-KEY($key, |c) { return callsame // $key; }; }; say %hash{2}; say %hash{'f'}; 05:28
camelia 5===SORRY!5===
Null regex not allowed
at <tmp>:1
------> 3od AT-KEY($key, |c) { return callsame //7⏏5 $key; }; }; say %hash{2}; say %hash{'f'
Other potential difficulties:
Useless use of hash composer on right side of hash assignm…
ipatrol m: my %hash = {'f' => 42}; %hash does role :: { method AT-KEY($key, |c) { return callsame() // $key; }; }; say %hash{2}; say %hash{'f'};
camelia Potential difficulties:
Useless use of hash composer on right side of hash assignment; did you mean := instead?
at <tmp>:1
------> 3my %hash = {'f' => 42}7⏏5; %hash does role :: { method AT-KEY($ke
2
42
araraloren_fake m: my %hash = f => 42; %hash does role :: { method AT-KEY($key, |c) { callwith($key, |c) // $k }; }; say %hash<2>; say %hash<f>;
camelia 5===SORRY!5=== Error while compiling <tmp>
Variable '$k' is not declared
at <tmp>:1
------> 3T-KEY($key, |c) { callwith($key, |c) // 7⏏5$k }; }; say %hash<2>; say %hash<f>;
araraloren_fake m: my %hash = f => 42; %hash does role :: { method AT-KEY($key, |c) { callwith($key, |c) // $key }; }; say %hash<2>; say %hash<f>; 05:29
camelia 2
42
ipatrol m: my %hash = 'f' => 42; %hash does role :: { method AT-KEY($key, |c) { return callsame() // $key; }; }; say %hash{2}; say %hash{'f'};
camelia 2
42
ipatrol I think the braces with a lone pair confused it
araraloren_fake yeah, call* with return to current method when call over
next* will never return to current method 05:30
ipatrol similar to system versus exec 05:31
araraloren_fake yeah
ipatrol from p5
araraloren_fake callsame use same argument of current method, and same like nextsame
05:31 rba_ joined
ipatrol which makes the expression a bit shorter 05:32
I wonder if the return is actually necessary 05:33
araraloren_fake ipatrol, not necessary
ipatrol m: my %hash = 'f' => 42; %hash does role :: { method AT-KEY($key, |c) { callsame() // $key; }; }; say %hash{2}; say %hash{'f'};
camelia 2
42
05:34 rba__ left
ipatrol yay! a birdie! 05:34
araraloren_fake :P 05:35
ipatrol why does it think it's a null regex though?
m: my %hash = 'f' => 42; %hash does role :: { method AT-KEY($key, |c) { callsame // $key; }; }; say %hash{2}; say %hash{'f'};
camelia 5===SORRY!5=== Error while compiling <tmp>
Null regex not allowed
at <tmp>:1
------> 3 { method AT-KEY($key, |c) { callsame //7⏏5 $key; }; }; say %hash{2}; say %hash{'f'
araraloren_fake cause // is like a null regex
m: say // 05:36
camelia 5===SORRY!5=== Error while compiling <tmp>
Null regex not allowed
at <tmp>:1
------> 3say //7⏏5<EOL>
araraloren_fake m: say /123/;
camelia /123/
ipatrol but I'm using it with two operands and not with a matching operator
m: say Mu // 2;
camelia 2
araraloren_fake callsame is a method name, and //, $key is argument 05:37
just like `say // $key'
m: say // $key
camelia 5===SORRY!5=== Error while compiling <tmp>
Null regex not allowed
at <tmp>:1
------> 3say //7⏏5 $key
ipatrol so it's putting commas in there? 05:38
m: say(//,'key');
camelia 5===SORRY!5=== Error while compiling <tmp>
Null regex not allowed
at <tmp>:1
------> 3say(//7⏏5,'key');
ipatrol m: say('key');
camelia key
araraloren_fake so you can use say without parentheses
ipatrol m: say(Mu // 'key');
camelia key
ipatrol m: my %hash = 'f' => 42; %hash does role :: { method AT-KEY($key, |c) { (callsame // $key); }; }; say %hash{2}; say %hash{'f'}; 05:39
camelia 5===SORRY!5=== Error while compiling <tmp>
Null regex not allowed
at <tmp>:1
------> 3{ method AT-KEY($key, |c) { (callsame //7⏏5 $key); }; }; say %hash{2}; say %hash{'f
araraloren_fake m: say( say // $key )); # like this
camelia 5===SORRY!5=== Error while compiling <tmp>
Null regex not allowed
at <tmp>:1
------> 3say( say //7⏏5 $key )); # like this
ipatrol m: say(&say // 'key');
camelia sub say (| is raw) { #`(Sub|36908248) ... }
05:40 pierre_ joined
araraloren_fake yeah, use & get the sub 05:40
ipatrol or use () to call it
araraloren_fake and without & will call it
05:40 cdg joined
ipatrol what does &meth() do? 05:40
araraloren_fake m: say; # this will call say
camelia 5===SORRY!5=== Error while compiling <tmp>
Unsupported use of bare "say"; in Perl 6 please use .say if you meant to call it as a method on $_, or use an explicit invocant or argument, or use &say to refer to the function as a noun
at <tmp>:1
--…
araraloren_fake m: &say; # this will call say 05:41
camelia WARNINGS for <tmp>:
Useless use of &say in sink context (line 1)
araraloren_fake this not
ipatrol nopu
araraloren_fake m: &say(); # this will call say
camelia
05:41 rba_ left
ipatrol m: &say('kitten'); 05:41
camelia kitten
ipatrol I prefer "kitten" to "foo" :-P 05:42
araraloren_fake m: say &say;
camelia sub say (| is raw) { #`(Sub|60094056) ... }
ipatrol or rather, "cat" is foo, and "kitten" is foobar
05:43 mcsnolte left
araraloren_fake haha 05:43
05:43 pierre_ left, pierre_ joined 05:45 cdg left 05:48 nadim joined
ipatrol araraloren_fake: can a hash literal contain a trailing comma? 05:50
m: my %hash = {'cat'=>'kitten','foo'=>'foobar',}; 05:51
camelia Potential difficulties:
Useless use of hash composer on right side of hash assignment; did you mean := instead?
at <tmp>:1
------> 3ash = {'cat'=>'kitten','foo'=>'foobar',}7⏏5;
05:51 nuk9 left
ipatrol m: my %hash = {'cat'=>'kitten','foo'=>'foobar',}; say %hash.gist; 05:51
camelia Potential difficulties:
Useless use of hash composer on right side of hash assignment; did you mean := instead?
at <tmp>:1
------> 3ash = {'cat'=>'kitten','foo'=>'foobar',}7⏏5; say %hash.gist;
{cat => kitten, foo => foobar}
ipatrol m: my %hash = 'cat'=>'kitten','foo'=>'foobar',; say %hash.gist; 05:52
camelia {cat => kitten, foo => foobar}
05:52 brrt joined
ipatrol I don't understand why choosing to use the optional braces is considered a "potential difficulty" 05:52
05:53 araraloren_fake left, brrt left 05:54 Skarsnik joined
Skarsnik Hello 05:55
ipatrol anyway, I need to go to bed. It's like 2AM here. Night all. 05:56
05:56 ipatrol left
Skarsnik AlexDaniel, did you try connecting some of your bot on ipv6 to bypass the 10 connections limits? 05:56
05:58 Skarsnik left 06:02 Cabanossi left 06:04 Cabanossi joined 06:12 kannan left 06:14 espadrine joined 06:33 Cabanossi left 06:34 Cabanossi joined 06:36 andrzejku joined 06:43 espadrine left 06:46 ChoHag left 06:53 TEttinger left 06:57 nadim left 06:59 domidumont joined 07:02 astj_ joined, astj left 07:03 domidumont left 07:05 parv joined, robertle left 07:07 domidumont joined 07:11 snarkyboojum_ joined, snarkyboojum left, snarkyboojum_ is now known as snarkyboojum 07:13 llfourn left, darutoko joined, ChoHag joined 07:15 nadim joined 07:28 domidumont left, domidumont joined 07:36 ChoHag left 07:40 zakharyas joined 07:41 cdg joined 07:42 jferrero left 07:43 wamba left 07:45 cdg left 07:47 dakkar joined 07:48 Dunearhp left, Dunearhp joined 07:59 mcmillhj joined 08:01 Cabanossi left 08:02 jonas1 joined 08:03 ChoHag joined, Cabanossi joined 08:04 mcmillhj left 08:06 rindolf joined 08:08 snarkyboojum left 08:15 mcmillhj joined 08:16 melezhik joined
melezhik Hi Perl6 gurus! 08:16
How can I know the time my script is working at the moment ?
a special variable ... 08:17
nadim docs.perl6.org/routine/now 08:19
08:21 mcmillhj left 08:22 araraloren_fake joined
araraloren_fake :) 08:22
08:24 wigner joined, snarkyboojum joined 08:26 pierre_ left 08:31 melezhik left 08:32 xtreak joined, mcmillhj joined
lizmat Juerd: I see STORE_AT_KEY as an implementation detail that may not survive in the future 08:34
yoleaux 00:43Z <japhb> lizmat: re: rakudo 5db5b1db , couldn't you test for being concrete or being Any? That would only leave the much smaller number of type objects that are Mu but not Any. Not perfect, but an improvement I think.
08:36 pazitiff joined, snarkyboojum left 08:38 wamba joined 08:39 pierre_ joined, mcmillhj left
lizmat clickbaits p6weekly.wordpress.com/2017/08/21/...ng-atomic/ 08:42
araraloren_fake lizmat++ 08:48
08:50 ambs joined 08:51 mcmillhj joined 08:54 samcv left 08:58 robertle joined, mcmillhj left 09:00 samcv joined
pazitiff Hello! 09:02
09:02 parv left
pazitiff Some useing Perl6 in prodcation? 09:02
I am just looking to Perl6. Grammars is realy great idea 09:03
09:05 nuk9 joined 09:06 xtreak left 09:10 nattefrost joined
araraloren_fake pazitiff, hi 09:12
09:12 melezhik joined
araraloren_fake yeah, I like Grammar too :) 09:12
09:13 lowbro joined, lowbro left, lowbro joined
melezhik HI nadim: , thanks. I need indeed time elapsed my script has been running , not current time 09:13
moritz pazitiff: I'm liking grammars so much that I'm writing a book about them :-) 09:14
if you want to be informed when it (or other Perl 6 books) come out, sign up for the mailing list at perl6book.com/ 09:15
</ad type="blatant">
melezhik How to know the time elapsed from the begining till now my script has been running ... ? 09:19
El_Che moritz: how far are you on the book? 09:20
moritz: and aren't you tired from the last one? 09:21
moritz melezhik: now - INIT now 09:22
melezhik: there's also $*INITTIME 09:23
El_Che: I've written about 95 pages in A4
El_Che melezhik: or use "time" from the shell
moritz El_Che: so volume-wise about 80% done
El_Che moritz: you're inspired
nice
moritz El_Che: and I had a break during the time that the production team did all their magic 09:24
El_Che I have been keeping an eye for your book on safari, not there yet
melezhik guys, thanks for options, will take a look
El_Che Think Perl 6 was recently changed from Early Cut to Final
moritz will there be a Director's Cut also? :-) 09:27
09:28 lizmat left
El_Che hehe 09:28
I started browsing the book
(in the hope the author will get paid) 09:29
interesting beginners book
moritz aye; I proof-read the first 100 pages or so 09:31
(and then got bogged down in my own book)
El_Che yes, I saw your name there
and liz
09:32 jmerelo joined
jmerelo Hi :-) 09:33
moritz \o jmerelo
El_Che jota-jota
09:34 wigner left
jmerelo @el_che I loved your meta-analysis of the issues in the TPCiA chat :-) 09:34
moritz speaking of jmerelo -- yesterday I started creating a docker image for the regexes/grammar book, based on the perl6-alpine docker image
El_Che jmerelo: haha, I have an audiance
jmerelo I am having trouble with it now...
El_Che climbs on a box
jmerelo I am trying to update it to the latest version of Rakudo Perl and it fails to build. 09:35
moritz: glad it helps :-) If you need anything, just let me now
El_Che jmerelo: gave alpine a go again, and still ruby-musl problems. Wanted to send a PR for your img to use pkgs (so it builds inmediately). Not there yet 09:36
moritz jmerelo: will do, thanks :-)
jmerelo El_che I don't know if this is related to that. Check out the error:
um 09:37
This is the problem twitter.com/jjmerelo/status/899928919570305026 09:39
It core dumps in stage1
El_Che mmm 09:40
jmerelo: I'll sping the new release for my packages and see if I have it as well
jmerelo Thanks
El_Che I also have a alpine branch were rakudo gets built, but fails at packaging
09:41 wamba left
jmerelo Here's the Dockerfile github.com/JJ/alpine-perl6/blob/ma...Dockerfile 09:41
09:41 cdg joined
jmerelo El_che: how does it fail? Size problems? 09:42
09:43 lizmat joined 09:45 wamba joined 09:46 cdg left
Ulti out of curiosity does anyone here use windows not 10? 09:49
installing R* and the WWW stuff just worked for me apart from some failing tests unrelated to OpenSSL problems 09:50
jmerelo 2017.01 builds correctly in Alpine. There must be some change after that... 09:58
El_Che 2017.07 builds fine on alpine on my side 10:00
(sorry, I was afk)
jmerelo Checking that now. No problem. 10:01
So it must be something in 2017.08 and a completely unrelated thing in v6.c 10:02
It's building now in Docker Hub hub.docker.com/r/jjmerelo/alpine-p...yhxpjl22c/ 10:03
El_Che building 2017.07 with travis: travis-ci.org/nxadm/rakudo-pkg
I'll create pkgs if ok 10:04
jmerelo Great :-)
El_Che I should look into creating release automatically with travis
haven't had the time
jmerelo: I'll run my alpine container manually to here if it build 10:05
jmerelo It does locally. 10:06
Guess there will be no problem in Docker Hub
El_Che I am confused. You fixed the 2017.08 build problem? 10:08
araraloren_fake Is the last release of rakudo rpm released ? 10:09
El_Che Ulti: my ubuntu pkg works on the Linux Subsystem of windows 10. I ran some tests, but don't use win10 much
10:09 nuk9 left
El_Che araraloren_fake: I you refer to 2017.08 and my pkgs, they are being built. If you refer to distro's own pkgs, probably on their "unstable/rawhide/dev" branch. If you mean Rakudo Star, no (the lastest is 2017.07) 10:10
===SORRY!=== 10:12
Found /opt/rakudo/bin/moar version 2017.08, which is too old. Wanted at least 2017.08.1
agh 10:13
again the version mismatch
10:20 Actualeyes left
jmerelo Built new perl6 docker image hub.docker.com/r/jjmerelo/alpine-p...yhxpjl22c/ 10:20
El_Che: I didn't fix it. I checked it worked with 2017.07, as you said. It's still not working with 2017.08, and I have no idea how to fix it. 10:21
araraloren_fake El_Che, I mean rakudo not Rakudo Star
10:24 xinming_ joined 10:27 xinming left 10:28 xtreak joined
stmuk I believe alpine linux has exec stack protection of some sort, there have been some changes to do with that in .08 so maybe related 10:28
does dmesg display anything odd?
10:31 xtreak_ joined 10:34 xtreak left 10:36 xtreak_ left 10:37 xinming_ left 10:39 xinming joined, cdg joined 10:40 leont joined 10:42 sena_kun joined 10:44 cdg left, DarthGandalf joined
sena_kun arnsholt, ping. 10:44
10:51 cog_ joined 10:54 masak left, brimonk left, masak joined, epony left, shadoxx_ joined, shadoxx left, masak is now known as Guest34959, cognominal left, cosimo left 10:55 brimonk joined, itaylor57 left 10:56 cosimo joined 10:57 Voldenet left 11:00 jmerelo left, john_parr left, maettu left, maettu joined, jmerelo joined, El_Che left, incredible left, El_Che joined 11:01 leedo_ left, kubrat left, leedo_ joined, teatime left, pm51 left, cgfbee left, tinita left, teatime joined 11:02 pm51 joined, tinita joined, eythian left, Voldenet joined, Voldenet left, Voldenet joined 11:03 kubrat joined 11:05 incredible joined, cgfbee joined 11:07 eythian joined 11:09 leont left
jmerelo stmuk: I would have to check from the Docker image itself. What kind of things should I look for? 11:14
11:26 leont joined
El_Che jmerelo: rakudo 2018.08 build fails on alpine (this is new), and builds fine on other distors 11:26
distros
11:26 rindolf left
jmerelo Timo is telling me in Twitter that there seems to be a problem with that particular version. 11:27
2018.08.1 should be available soon.
s/2018/2017/
El_Che jmerelo: gist.github.com/nxadm/ea6a2aa2965c...2d1dd7353a
jmerelo Correct, same problem I had. 11:28
El_Che it does build fine however on Ubuntu, Debian, Fedora and Centos: travis-ci.org/nxadm/rakudo-pkg/builds/267123623
jmerelo: I am building again moarvm 2018.08.1 11:29
11:30 pierre_ left, lizmat left
jmerelo I had a problem also with building latest version in ubuntu 14.04 github.com/tadzik/rakudobrew/issues/119 11:30
timotimo jmerelo: 2017.08.1 is already available 11:31
in git
AFAIK rakudobrew doesn't build from tarballs
El_Che timotimo: rakudo 2017.08.1 or moarvm?
timotimo moarvm 11:32
jmerelo does not show up in rakudobrew list-available. Which might be my fault, since that part of rakudobrew is mine...
timotimo rakudo 2017.08 should depend on moarvm 2017.08.1
El_Che timotimo: in that case, it doesn't fix it
timotimo: my output is with that version
11:32 haukex joined, haukex left
timotimo where do i have to look for the output? 11:33
El_Che 13:27 < El_Che> jmerelo: gist.github.com/nxadm/ea6a2aa2965c...2d1dd7353a
timotimo thanks
El_Che I can paste a larger part if needed, shell still open
timotimo can you also get moar --version?
11:35 zakharyas left 11:37 ggoebel left
timotimo what do i do on my local docker to get the same exact setup you have? 11:37
stmuk jmerelo: errors about stack something or other 11:38
jmerelo stmuk: errors are not too informative. Just the line and an error code. But I'll look for that if I can find it. Thanks! 11:39
El_Che timotimo: do you want to build one from a Dockerfile or do you want one from the hub? 11:40
stmuk oh alpine doesn't use glibc either
timotimo whatever lets me go in there and run the command via gdb
so i can figure out why it segfaults
11:41 melezhik left
stmuk " it appears that Alpine Linux (and specifically the musl-libc library it uses) has a very limited stack size by default" 11:42
El_Che timotimo: gist.github.com/anonymous/bd1e3572...54eed8c1e7 11:44
11:44 basket left
timotimo El_Che: need to switch the checkout and the cd :) 11:46
11:46 jmerelo left
El_Che timotimo: if you create the alpine container (skip the create pkg command) from the gist, you can just do: docker run -ti --rm --entrypoint=/bin/bash rakudo/pkgrakudo-alpine-amd64 11:46
from there you can do the needed steps manually 11:47
timotimo the "create pkg command" is "pkg_rakudo.pl"?
El_Che that's a wrapper for passing the version information to the packager script in the image 11:48
timotimo mhm
El_Che github.com/nxadm/rakudo-pkg/blob/3...64-3.5#L21
that copy there puts "pkg_rakudo" in the / of the image
with the download and build instructions
github.com/nxadm/rakudo-pkg/blob/a...pkg_rakudo 11:49
11:49 ggoebel joined
El_Che it's a just simple distro-agnostic script that download the sources, compiles them and packages them 11:50
timotimo i can't do the docker run command
"Unable to find image 'rakudo/pkgrakudo-alpine-amd64:latest' locally"
El_Che yes, you need to build it first
timotimo: let us do something else
I'll prepare an image with the sources in place and gdb
timotimo i ran create_container.pl, shouldn't that do it? 11:51
El_Che upload it to the docker hub and let you run that
timotimo: yes, it should
timotimo Successfully built 3dc7ded9a64f 11:52
Successfully tagged rakudo/pkgrakudo-alpine-amd64:3.5
i probably have to put the version at the end to be able to ues it?
El_Che yes
timotimo that was easy
the tarball for moarvm's point release is already up, btw 11:54
i can't run it under gdb: warning: Error disabling address space randomization: Operation not permitted 11:56
During startup program terminated with signal SIGSEGV, Segmentation fault.
El_Che yes
musl 11:57
I think stmuk commented about that a few weeks ago
(the randomization stuff, not gdb)
timotimo oh, docker threw everything i did away, okay 12:00
is that what --rm odes?
El_Che timotimo: I created a sandbox for you: 12:02
docker run -ti nxadm/timoalpine bash
in the container you can run /pkgrakudo with no params
timotimo too late, i'm already in gdb now :)
El_Che it will download the rakudo, nqp and moarvm dists and unpack them on / 12:03
then you can build them however you like it
12:04 leont left
timotimo i wonder if rr works inside there 12:04
El_Che apk is your friend if you need extra packages 12:05
12:05 ggoebel left
timotimo yup 12:05
apk doesn't have rr though
El_Che and --rm, yes ;)
timotimo #0 0x00007ffff7d95dc6 in malloc () from /lib/ld-musl-x86_64.so.1 12:06
#1 0x0000000000000000 in ?? ()
fantastic backtrace
El_Che timotimo: while you're at it, fix the fpm crash so I can create rakudo packages on alpine :) 12:07
(to illustrate that a lot of soft seems to have trouble with musl)
timotimo i'm running it through valgrind right now 12:09
that'll potentially take a long while
12:10 wigner joined
timotimo ==3532== Bad permissions for mapped region at address 0x4089F68 12:11
==3532== at 0x54DD1C6: build_cfg (graph.c:102)
the what now?
12:16 raschipi joined 12:18 P6steve joined
timotimo well, gdb recording is out 12:18
araraloren_fake Why I can use `is export` on variable declared with my, but can apply it on `my sub` ?
12:18 wigner left, ggoebel joined 12:19 colomon joined
moritz araraloren_fake: are you missing a negation somewhere in there? 12:19
timotimo [Inferior 1 (process 3541) exited normally] 12:20
graaahhh
araraloren_fake m: module Foo { my $x is export; };
camelia 5===SORRY!5=== Error while compiling <tmp>
Can't apply trait 'is export' on a my scoped variable. Only our scoped variables are supported.
at <tmp>:1
------> 3module Foo { my $x is export7⏏5; };
expecting any of:
constrain…
araraloren_fake m: module Foo { my sub x() of Int is export { 42 } };
camelia ( no output )
araraloren_fake moritz, I'm not get you mean 12:21
colomon is there a way to get the process’s exit code from Proc::Async? I need to figure out how to detect it when the process segfaults… 12:22
jnthn colomon: the Promise returned by .start will be kept with an object that has a .exitcode method
araraloren_fake or maybe can someone explain something about symbol table of PACKAGE ? 12:23
I notice this question talk about some about public symbol table of PACKAGE: stackoverflow.com/questions/457701...-namespace 12:24
Is there a private symbol table ,or it called lexical symbol table ?
timotimo "lexical pad" usually
araraloren_fake oh lexical pad, I got it 12:25
so we can too table or pad. But I can not get why we can export a sub declared with my? 12:26
timotimo El_Che: a quick google search shows how the alpine people patched mpd to set the stack size to 1024*1024 for threads created with pthread_create
El_Che: they used to get the exact same error, now they don't any more
araraloren_fake s/too/two/ :) 12:27
moritz araraloren_fake: in first approximation, scoping and exporting are orthogonal
12:27 wigner joined
araraloren_fake yeah, I see 12:28
El_Che timotimo: I can bump the image to 3.6
should be out by now
12:29 domidumont left
colomon jnthn: is there an example for it somewhere? I am having a hard time figuring out how to make it work. 12:32
timotimo libuv has a piece of code to set the stack size to RLIMT_STACK when creating a new thread
maybe this is necessary on alpine linux as well for whatever reason? 12:33
12:33 zakharyas joined, Cabanossi left 12:34 Cabanossi joined
jnthn colomon: say (await $proc.start).exitcode 12:34
timotimo El_Che: how do you feel about patching moarvm for the alpine image? 12:35
12:35 pazitiff left
El_Che timotimo: docker run -ti nxadm/pkgrakudo-alpine-amd64:3.6 bash 12:35
timotimo: as long as I cant get fpm to create packages, it won't do much 12:36
12:36 OtherNick joined
timotimo ? 12:36
El_Che timotimo: I think jj merelo is closer to the issue
12:36 OtherNick left
El_Che timotimo: jjmerelo's image == rakudo on alpine 12:36
mine == an alpine image for creating alpine packages. The package part does not work yet as fpm as trouble as well 12:37
in short, my img is a short-live build environment for native packages
timotimo ok i have the image, where's th sources? 12:38
El_Che timotimo: but If you want me to patch that image, sure
timotimo i mean literally include a piece of C code in there
El_Che This are my images: github.com/nxadm/rakudo-pkg/tree/master/docker , this is jj's: github.com/JJ/alpine-perl6/blob/ma...Dockerfile 12:39
timotimo i mean i'm inside the image, where are moar, nqp, and rakudo extracted to?
El_Che root
/
timotimo i only see the three scripts there for zef and pkg_rakudo 12:40
El_Che ok
in the regular image you need to run pkg_rakudo with env variables
timotimo oh
El_Che on the one I built from you I hardcoded those
so it downloaded and extracted them automatically 12:41
but that was 3.5
do you want me to create one like that for 3.6?
colomon jnthn++
timotimo what do i need REVISION for?
colomon will figure this out sooner or later.
El_Che put 01
it's a pkg thing
timotimo ok 12:42
El_Che you can repkg the same version of the software with changes on the pkg
12:43 AnotherNick joined 12:44 AnotherNick left 12:45 Ven`` joined 12:46 lizmat joined, AnotherNick joined
timotimo i don't think libuv lets us tell them a stack size 12:47
Ven`` twitter.com/coetry/status/896169863684792320 if coetry comes around, he might want to check out #6macros :).
timotimo so we'd have to patch 3rdparty/libuv inside the alpine package
12:47 domidumont joined, leont joined
colomon jnthn: so it’s the .result of the Promise which has an exitcode, yes? 12:48
jnthn Yes
colomon may be starting to undertand
\o/
El_Che timotimo: that's ok for a self contained pkg as mine (if I fix the ruby prob), but distro's won't like tha
t
12:48 AnotherNick left, AnotherNick joined
timotimo "distros"? 12:49
oh
El_Che alpine, debian, centos
timotimo well, alpine will have to Deal With It
El_Che they don't like the include-3rd-party-libs
timotimo or offer us a magical stack compression library or something 12:50
El_Che yeah, my enthousiasm for alpine (see, 5mb images!) is somewhat cooled
timotimo i don't even think reducing default stack size for new threads helps with making small images 12:51
El_Che no, that's unrelated
timotimo well i guess it makes programs a little bit smaller? if they only ever use a kilobyte of the stack for every new tread
Ven`` well if they're OS threads, then they're a least a page big
El_Che I mean the idea of small images is very nice until you meet real world apps
Ven``
.oO( Always detect EVAL )
 12:52
timotimo with MVM_SPESH_DISABLE=yes i can build nqp, but it won't test because "prove" is not installed
that's only a matter of a single apk add, though
how do i ask apk what package provides a given binary? 12:53
oh, second google search result
12:53 mcmillhj joined
timotimo apkfind() {wget -q --no-check-certificate -O - pkgs.alpinelinux.org/contents\?file=$1 | sed -e '1,119d' -e '/panel-footer/,$ d' -e 's/]*>//g' -e 's/\&\#x2F\;/\//g' -e '/^ *$/d' | xargs -n 5 | awk '{ printf("%-20s %-7s %-11s %-13s %s\n",$2,$3,$4,$5,$1)}' } 12:53
Operations
cool.
El_Che oh crap
Ven`` beautiful, glorious sed. 12:54
timotimo that output is disgusting 12:55
12:55 nuk9 joined
timotimo it's perl-utils 12:56
12:58 DarthGandalf left
timotimo El_Che: looks like spesh is the only thing that blows the stack limit, and you can just disable that and still build your packages 12:58
El_Che timotimo: I am stuck at the packaging level. Until this release, rakudo built fine on alpine. 12:59
timotimo: jjmerelo is stuck on his rakudo img
the question is why this release needs the workaround and previous don't 13:00
timotimo spesh changed significantly between the previous and the current release
it got pushed to its own thread
i'll tweet our results at him 13:01
13:02 cosimo left 13:03 Cabanossi left, Cabanossi joined, cdg joined
jnthn Also, basic block walking recurses on the C stack, and spesh graphs split up into more basic blocks, so it can end up deeper. There's an upper limit on the code size we'll specialize. 13:06
But it assumes a stack size of around a meg
lizmat seems like someone referred to the latest P6W on HN: news.ycombinator.com/item?id=15071037
no comments yet, though 13:07
:-)
raschipi Reddit has comments: www.reddit.com/r/perl6/comments/6v...nd_perl_6/
13:08 cdg_ joined 13:09 cosimo joined 13:10 cdg left
ilmari nobody tell tektektektektek that all ASCII characters are UTF-8... 13:13
nor that "UTF-8 character" isn't a thing, for that matter 13:14
timotimo jnthn: it already crashes inside build_cfg :) 13:15
at least it looked like it from valgrind's output
jnthn timotimo: haha
Wow
I wonder what recurses there 13:16
timotimo i wonder if the interp_run stackframe is a big part of the blame
why is that even on there for just the spesh worker? :P
colomon has rebuilt his crash testing script to recognize the difference between crashes, hangs, returned no output, and success. o/
timotimo initial_invoke, surely
jnthn timotimo: Just so we can have it set up as a normal thread
timotimo: So it plays the same way with GC as everything else 13:17
It's hard enough to get threads and GC right without two kinds of thread :)
It's possible we could still have that and avoid the run stack frame though
Just was expedient to do it this way
timotimo right 13:19
i'm still not sure how big the stackframe for interp_run even is 13:20
probably not terribly bug
big*
13:23 gfldexwork joined
jnthn Dunno 13:23
timotimo hey google, how do i measure stack frame size? - here's a whole bunch of info about how to find the right sized bike for you
jnthn Anyway, time to start heading towards Switzerland...
o/
13:23 gfldexwork left
timotimo have a good trip! 13:24
AlexDaniel .tell Skarsnik ipv6 or not, how would I choose the ip unless the ticket is resolved?
yoleaux AlexDaniel: I'll pass your message to Skarsnik.
timotimo oooh -fstack-usage 13:28
13:29 wamba left, DarthGandalf joined 13:34 pierre_ joined
timotimo interp_run: 1312 dynamic,bounded 13:35
bounded means that number is a reliable upper limit 13:37
leont I should have finished my PR Before 2017.8 came out :-/ 13:40
lizmat leont: there's always 2017.09 ?
leont Can't make TAP::Harness parse subtests until Test.pm outputs them correctly, there's a chain effect :-/ 13:42
fg 13:44
Test::_diag looks rather optimized (I mean, it's using nqp directly), and I'm not sure how to fix the issue without unoptimizing it 13:50
The easiest way is just to remove the empty-line-removing line entirely 13:51
Geth doc: 36251c6ecf | (Zoffix Znet)++ (committed using GitHub Web editor) | doc/Language/unicode_texas.pod6
List Texas atomics

  - Create a separate table (following format of the first one
   with all the ops got kinda messy)
P.S.: Some of the subs appear to be undocumented P.P.S: the docs for Texas/Unicode versions are duplicated; Wouldn't
   be easier to just list both variants with one description?
P.P.S.: Shouldn't ⚛+= and ⚛-= be called atomic-add-assign and atomic-sub-assign
   instead of atomic-add-fetch and atomic-sub-fetch, since ⚛= is atomic-assign,
   not atomic-fetch, which is just prefix:<⚛>		 13:54
13:54 domidumont left 13:56 raiph joined 13:57 skids joined 13:58 lizmat left, domidumont joined 14:00 epony joined
Ven`` I love it. 14:01
araraloren_fake :) 14:02
Geth doc: e5cadaedaa | (Zoffix Znet)++ (committed using GitHub Web editor) | doc/Language/unicode_texas.pod6
Remove accidentally duped line
 14:03
14:03 ShalokShalom joined
nadim What's the format of the rakudo release tags that are considered stable? yyyy.mm(.v)? 14:03
moritz that is the format of release tags 14:04
we don't have a separate syntax for those we consider "stable"
nadim release is good enough
moritz that's something that we typically can only tell after the fact
nadim :)
El_Che ha 14:05
14:06 Skarsnik joined
Skarsnik Hello 14:06
yoleaux 13:24Z <AlexDaniel> Skarsnik: ipv6 or not, how would I choose the ip unless the ticket is resolved?
Skarsnik AlexDaniel, if you put a ipv6 addr as dest it will be forced to use an interface with an ipv6 addr on it 14:07
araraloren_fake Skarsnik, Hello
timotimo can you also supply an interface name that way?
Skarsnik I am not 100% sure. but I am pretty sure that will work
timotimo or is that for outgoing addresses?
AlexDaniel Skarsnik: oh, so I can double the amount of bots by simply connecting to different hosts?
Skarsnik I don't think you can bind your socket on a an interface 14:08
14:08 rindolf joined 14:09 cosimo left
Skarsnik AlexDaniel, it's my guess since I can't bind a socket on a ipv4 address if I try to connect to a ipv6 address 14:09
14:09 newb2 joined
AlexDaniel I'll try it now 14:09
El_Che new linux packages: github.com/nxadm/rakudo-pkg/releas...ag/2017.08
Xliff I'm bored. Can someone suggest a Perl6 project to work on? 14:10
14:10 cosimo joined
Skarsnik (I have to go somewhere in 5 min, I will check more later) 14:10
timotimo Xliff: how do you like Terminal::Print?
Xliff I've worked with Terminal::Print /some/
moritz Xliff: you could implement <commit> in regexes. That would be very helpful :-) 14:11
Xliff moritz: Linky?
I think that might be a bit more than I can chew, atm
Plus, I have not found a good way to test nqp changes in rakudo.
THAT would be good to know, TBH 14:12
moritz Xliff: design.perl6.org/S05.html search for <commit>
timotimo bailador is getting a bunch of attention & traction, perhaps that's interesting to you?
with <commit> you can alreday test it in nqp without building rakudo
Xliff timotimo: That also is on my roadmap.
timotimo: Ah. I will look into how to do that. Any hints?
14:14 quotable6 joined, ChanServ sets mode: +v quotable6
newb2 I want to import Readline.pm from drforr/perl6-readline into the perl6 RPL but 14:14
14:14 quotable6 left, quotable6 joined, statisfiable6 joined, ChanServ sets mode: +v statisfiable6
AlexDaniel Skarsnik: it works XDDD 14:15
Xliff Re: <commit> -- Ah... such a small description implies such.... complexity. :P
So a <commit> assertion accepts either a Code block or a string argument? 14:16
AlexDaniel anyway, time to update rakudo…
AlexDaniel hides
14:16 Actualeyes joined, committable6 left, evalable6 left, releasable6 left, quotable6 left, statisfiable6 left, greppable6 left
Xliff AlexDaniel: I just did that. LOL! 14:16
timotimo Xliff: i don't know :< 14:17
Xliff Are there test cases written for <commit> ?
AlexDaniel Skarsnik: (‘chat.freenode.net’, ‘185.30.166.38’).pick # :D
Xliff S05 isn't anywhere near as clear on it.
Ven`` wasnt <commit> spelt :: ?
I think we had :, :: and :::. I remember FROGGS talking about it
AlexDaniel Skarsnik: now the limit is effectively 20 bots, so I have some room for 8 more or something 14:18
Xliff I am working on the "<commit>" assertion, as written between quotes.
newb2 I want to import Readline.pm from drforr/perl6-readline into the perl6 REPL for testing
14:19 silug joined
newb2 but "use lib '/home/cvs/.zef/tmp/perl6-readline.git/lib'; use Readline" hangs up on the 2nd use pragma 14:19
So obviously there's something here Cover looked
.. here I overlooked 14:20
timotimo you are using "zef look"?
newb2 PS it doesn't work as a perl6 script either
OK will try ..
timotimo no, don't
newb2 OK 14:21
timotimo the REPL will try to load readline itself to provide line editing if it's installed
14:21 bloatable6 joined, ChanServ sets mode: +v bloatable6, quotable6 joined, evalable6 joined, coverable6 joined, committable6 joined, ChanServ sets mode: +v evalable6, ChanServ sets mode: +v coverable6, ChanServ sets mode: +v committable6, bisectable6 joined, releasable6 joined, greppable6 joined, unicodable6 joined, benchable6 joined, ChanServ sets mode: +v benchable6, statisfiable6 joined, pierre_ left
newb2 yes but Iwant to do testing 14:21
timotimo i'm not sure if it'll work to load readline inside the repl, because you'll be using a readline inside a readline
newb2 such as adding or changing bindings or setting vars 14:22
Xliff Ven: If I am reading this correctly: "::", "::>" and ":::" have to do with alternation.
14:22 wamba joined
timotimo the REPL is quite wobbly in general. if at all possible, you should prefer a script somehow 14:22
newb2 Right - but it also doesn't work with a regular p6 script
timotimo gotta go AFK for a bit
newb2 check
AlexDaniel Skarsnik: thanks :)
newb2 So my question is "if the module is in the directory from the use lib statement 14:23
14:23 lizmat joined
newb2 which returns nil - what does that mean? 14:23
14:23 pierre_ joined, pierre_ left
newb2 then why does it hang up on 'use Readline'? 14:23
14:23 Kyo91 joined
newb2 INSIDE a script 14:23
timotimo you can set the RAKUDO_MODULE_DEBUG (or something?) environment variable and see what it tries at the time it starts hanging 14:24
newb2 SHOULD "use lib '/home/cvs/.zef/tmp/perl6-readline.git/lib'; use Readline" work unless I have made a typo? 14:25
From inside a script? 14:26
OK time will try ...
s/time/tim/
moritz uhm, did you install it? 14:27
newb2 "zef install'
moritz then you shouldn't need a "use lib..."
newb2 but it's placed in a ~/.zef directory NOT in the expected path 14:28
moritz that's where the sources are
14:28 andrzejku_ joined
moritz installing it should place it in other places too, potentially with file names you wouldn't expect 14:28
newb2 OK so when I do 'find / | grep Readline.pm" that's the only thing that is found 14:29
AHA so how would I find it and what difference does it make if I do 'use lib .." 14:30
moritz why would you want to find it? isn't it enough if rakudo finds it?
14:31 araraloren_fake left
ugexe zef locate lib/Readline.pm 14:31
newb2 I have to import the Module in order to use it if its not in @INC or whatever in perl 6
14:31 wigner left
newb2 ugexe will try .. 14:31
14:31 araraloren_fake joined
ugexe but you wouldnt use that result of that in a `use lib "..."` 14:31
14:31 wigner joined
moritz newb2: if you installed it with 'zef install Readline', there is no need to do anything manually with paths to be able to use it 14:32
just "use Readline;"
newb2 : "Readline => /opt/share/perl6/site/sources/D8BAC826F02BBAA2CCDEFC8B60D90C2AF8713C3F" Thanks! 14:33
14:34 newb2 left 14:36 andrzejku_ left 14:49 raschipi left 14:56 smls joined
smls m: role R [::T] { has T $.a }; say R[Int].new.a; 15:01
camelia No such method 'gist' for invocant of type 'T'
in block <unit> at <tmp> line 1
smls ^^ Is this supposed to work?
Ven`` no, AFAIK, because T is not available at that time. There's no "instantiation" type 15:04
s/type/time/
smls m: role R [::T] { has T $.a = T}; say R[Int].new.a;
camelia (Int)
smls Hm, like that it works... 15:05
Ven`` m: role R [::T] { my constant T $.a = T}; say R[Int].new.a;
camelia 5===SORRY!5=== Error while compiling <tmp>
Missing initializer on constant declaration
at <tmp>:1
------> 3role R [::T] { my constant T7⏏5 $.a = T}; say R[Int].new.a;
Ven`` m: role R [::T] { my constant $.a = T}; say R[Int].new.a;
camelia 5===SORRY!5=== Error while compiling <tmp>
Variable $.a used where no 'self' is available
at <tmp>:1
------> 3role R [::T] { my constant $.a7⏏5 = T}; say R[Int].new.a;
Ven`` whoops
m: role R [::T] { my constant $a = T}; say R[Int].new; 15:06
camelia R[Int].new
Ven`` not sure then :)
araraloren_fake m: role R [::T] { has T $.a }; say R[Int].new(a => 42).a;
camelia 42
15:07 araujo joined, araujo left, araujo joined
araraloren_fake Of course it works 15:07
Ven`` m: role R [::T] { has T $.a }; say WHAT R[Int].new.a;
camelia No such method 'gist' for invocant of type 'T'
in block <unit> at <tmp> line 1
Ven`` right 15:08
araraloren_fake That error is not about generic role
Ven`` m: role R [::T] { has T $.a }; # well
camelia ( no output )
araraloren_fake m: m: role R [::T] { has T $.a; }; say R[Int].new.a.WHAT; 15:09
camelia No such method 'gist' for invocant of type 'T'
in block <unit> at <tmp> line 1
araraloren_fake Hmm, but that's strange here..
Ven`` funnily enough, removing the `.a` gives a slightly different error message: 15:10
m: role R [::T] { has T $.a; }; say R[Int].new
camelia No such method 'perl' for invocant of type 'T'
in block <unit> at <tmp> line 1
Ven`` it's just recursively calling `.perl`, surely?
smls m: role R [::T] { has T $.a }; say R[Int].new.a.HOW; 15:11
camelia Perl6::Metamodel::GenericHOW.new
smls will RT. 15:12
araraloren_fake m: role R [::T] { has T $.a }; say R[Int].new.HOW;
camelia Perl6::Metamodel::ClassHOW.new
Ven`` smls: I'm really not sure it's not how it's supposed to be. It's just that a GenericHOW has no .gist? 15:14
I know I had similar questions 1-2 years back, and jnthn++ explained that it was meant to be that way 15:15
smls Oh?
Couldn't `has T $.a;` be made to do the same thing as `has T $.a = T;` ? 15:16
araraloren_fake m: m: role R [::T] { has T $.a = T; }; say R[Int].new.a;
camelia (Int)
araraloren_fake m: role R [::T] { has $.a }; say R[Int].new.HOW; 15:17
camelia Perl6::Metamodel::ClassHOW.new
araraloren_fake m: role R [::T] { has $.a }; say R[Int].new.a;
camelia (Any)
araraloren_fake I think it should be
15:20 kannan joined 15:21 domidumont left 15:22 domidumont joined, zakharyas left 15:29 devmikey joined 15:30 devmikey left 15:31 Cabanossi left 15:33 Cabanossi joined 15:36 P6steve left
timotimo damnit! i forgot about plug adapters for going to switzerland 15:38
i had one and i knew exactly where it was ... then i moved to a different place
15:44 lizmat_ joined, ufobat joined
Skarsnik btw does IO::Socket::Async::SSL is installable? it fails test with rakud 17.08 15:45
15:45 lizmat left
jnthn What does it fail? I installed it fine on my laptop last night 15:48
timotimo worked for me just now, too 15:49
15:50 cdg_ left
Skarsnik (btw the binding does provide an abi version so the dev package is needed x) 15:50
+does not
ilmari openssl version difference?
jnthn Skarsnik: Patches welcome :) 15:51
kannan my laptop failed due to hardware probelm. now i am re-installing on a debian 9 VM. i get the following error. Killed
15:51 P6steve joined
kannan Makefile:492: recipe for target 'CORE.setting.moarvm' failed 15:51
make[1]: *** [CORE.setting.moarvm] Error 137
make[1]: Leaving directory '/usr/src/perl6/rakudo-star-2017.07/rakudo'
Makefile:43: recipe for target 'rakudo/perl6-m' failed
make: *** [rakudo/perl6-m] Error 2
15:52 lowbro left
Skarsnik libssl1.0.0: /usr/lib/x86_64-linux-gnu/libssl.so.1.0.0 15:52
kannan ah ok, thanks
i will try
15:53 cdg joined
Skarsnik ilmari, ABI version, not just version :) 15:53
ugexe install works on osx here as well
araraloren_fake good night! 15:54
15:54 araraloren_fake left
kannan oop, i thought maybe i am missing the libssl-dev package, but i read wrongly. not sure why install failed. 15:54
Skarsnik jnthn, I don't patch nc lib finding stuff anymore. I think github.com/rakudo/rakudo/pull/716 need to be discussed/pushed. I am actually not sure how to handle libxml2 without that x)
ilmari tries on debian stable (which has libssl 1.1.0f)
Skarsnik Original exception: 15:55
Could not find symbol '&SSL_ctrl'
in block at /root/.zef/store/p6-io-socket-async-ssl.git/62212cdbb1940ea5d3dd4b8824aa10c83aa4e1d9/lib/IO/Socket/Async/SSL.pm6 (IO::Socket::Async::SSL) line 176
should have use gist maybe
ilmari Cannot locate symbol 'SSL_load_error_strings' in native library 'libssl.so' 15:56
is my error
Skarsnik yeah libssl.so is in libssl-devel
ilmari yes, I know
libssl-dev:amd64: /usr/lib/x86_64-linux-gnu/libssl.so
Skarsnik that why the abi version should be here when it's possible (not like libmysqlclient)
ugexe but OpenSSL installs? thats where the bindings are 15:57
ilmari yes
15:57 cdg left, Ven`` left
ugexe OpenSSL has one spot where it checks for some interface change by trying one method and if that fails, some other 15:57
ilmari it's only available if OPENSSL_API_COMPAT < 0x10100000L 15:58
and deprecated and a no-op in 1.1.0
Skarsnik I am not sure if the error come from the module or just my perl6 install
ugexe github.com/sergot/openssl/commit/7...0ca1c67R38
Skarsnik I already had a 2016 rakudo an /opt and I installed the last release on top of it 15:59
ilmari ugexe: but on libssl 1.1.0 SL_load_error_strings isn't a symbol, it's a conditional macro 16:00
#if OPENSSL_API_COMPAT < 0x10100000L
# define SSL_load_error_strings() \
OPENSSL_init_ssl(OPENSSL_INIT_LOAD_SSL_STRINGS \
| OPENSSL_INIT_LOAD_CRYPTO_STRINGS, NULL)
#endif
Skarsnik well we can't really call MACRO from NC ^^
16:01 robertle left, nhywyll joined
Xliff I thought there was a provided fix in the comments. I don't know if that was ever excepted. 16:01
I remember because I ran into that macro problem myself.
16:02 bdmatatu joined
ilmari OPENSSL_init_ssl() is only necessary if you need non-default settings 16:03
16:03 Cabanossi left, Cabanossi joined 16:05 lizmat_ left 16:06 brimonk left, brimonk joined
timotimo kannan: that sounds like you don't have enough ram to build the core setting 16:08
it takes quite a bit of resources, since it's a multiple hundred thousand lines of code file
or is it just multiple tens of thousands?
sorry, it's just 60 thousand lines of code 16:09
kannan timotimo , makes sense, i spin up the cheapest thing at 4 $ / mo has 0.6 mb ram,
timotimo i hope you mean 0.6 GB ram :)
kannan i did
0.6 gb ram
Skarsnik that expensive for 0.6GB
kannan really? , its in google cloud... 16:10
i will wait for my laptop to get repaired i guess. 16:11
i had one day adevnture now in perl6, lol.
Skarsnik Dunno I pay a vps for like 3€/month and has 2Gbytes
kannan but successfully ran helloworld... written in atom editor.
El_Che kannan: use pkgs 16:12
mspo use swap
El_Che kannan: github.com/nxadm/rakudo-pkg/releases
kannan: fresh from today
kannan El_Che : thank you, i will trythis 16:14
16:15 domidumont left, mcmillhj left 16:17 leont left 16:18 andrzejku_ joined
stmuk hmm don't read reddit 16:20
16:21 lizmat joined
mspo yeah ⚛ renders a little weird for me too 16:22
I mean it's cute and all 16:23
El_Che stmuk: that sounds like good advice even without context :)
ilmari hasn't reddit 16:26
kannan El_Che : gr8 , it worked 16:27
El_Che people complaining that the ascci variant of atom is a lot less expressive and more typing ($var ⚛= $value (aka atomic-assign($var,$value))) do have a point though 16:29
kannan: great to hear
ugexe no they dont
mspo yeah are there other places where "texas" is not in the same position?
ugexe how often do you think this gets used?
mspo ⚛ also renders wide 16:30
16:30 mcsnolte joined
El_Che raiph's answer was pretty useful, showing .A++ 16:30
ugexe it people looking for a reason to complain without thinking about the use case
timotimo i just wrote a little reddit 16:31
El_Che I am just at the top, but that comment didn't struck to me as just-commplainig-because
timotimo www.reddit.com/r/perl6/comments/6v...6/dlyyfgs/ - this is what i wrote 16:32
kannan well, zef will not install and says memory not enough. i guess i will wait a day or so to resume 16:33
El_Che kannan: really?
16:33 domidumont joined
kannan yes, the vm is pretty costly for me, and my laptop is in service 16:34
El_Che kannan: give it a swapfile in the meanwhile? 16:35
timotimo: good point
kannan ok i can try that also.
AlexDaniel propdump: a⚛
16:35 itaylor57 joined
unicodable6 AlexDaniel, gist.github.com/57e1344fe924fa6786...c3efea6a23 16:35
El_Che I am not a mac user, so I haven't done much configuration on this machine, but the atom operator looks awful in vim 16:36
mspo like why isn't the texas of "÷" divide(num,dem)? 16:37
El_Che: oh osx it renders as a purple icon for me
El_Che double wide (partly on top of =) and it looks like a pink emoji
mspo El_Che: and eats the next character
El_Che agh
16:37 P6steve left
AlexDaniel “Strictly speaking, it makes no sense to talk of narrow and wide for neutral characters, but because for all practical purposes they behave like Na, they are treated as narrow characters (the same as Na) under the recommendations below.” 16:37
mspo yep yep
AlexDaniel “Narrow (and neutral) Unicode characters always map to halfwidth characters.” 16:38
so if I'm reading this correctly, ⚛ should be halfwidth
“⚛ also renders wide” is likely a problem with your system, or font, or editor, or something… 16:39
El_Che ok, it's not a mac thing. Same results on vim on Ubuntu
AlexDaniel: you mean fixed width fonts? What kind of heretic would other something else in the terminal? 16:40
geekosaur I suspect it's a font substitution
AlexDaniel ah, actually the last quote was from the wrong place
geekosaur your current font doesn't have that glyph so a fallback font is used to render it, and the character metrics don't match up 16:41
AlexDaniel El_Che: how can you possibly render 本 as a halfwidth character?
El_Che: it's fullwidth both in my terminal and emacs
El_Che AlexDaniel: that's kind of my point
not a big fan of cute-but-a-lot-readable 16:42
16:42 dakkar left
El_Che I takes effort to get someone to try perl6 (and most of the time they like it) 16:44
AlexDaniel actually, it doesn't really say what should be done with neutral characters
so ⚛ could be anything? I don't know…
El_Che guess what is the chance people in my surrounding will try it if they need to change their font
AlexDaniel samcv: ↑ ? Maybe you do?
geekosaur terminals can't generally deal with this, sadly 16:45
El_Che I am not trying to troll, I think it's an issue
geekosaur: we have zero IDE support
and some editors
geekosaur this is a big part of why you don't see much unicode in practice
El_Che so a lot of people will use a terminal
AlexDaniel El_Che: typically it's not about changing their font, but about fixing their system 16:46
El_Che: if it can't render ⚛ then chances are it can't render emoji and a bunch of other stuff
El_Che AlexDaniel: sure, but that won't happen if tehy don't see it as broken
AlexDaniel 🤷 16:47
El_Che I am a linux user and have years of experience of tweaking stuff, it's a second nature
how many mac and win users will do that?
16:47 mcmillhj joined
AlexDaniel .tell moritz speaking of unicode… I know that I am probably being very annoying, sorry, I don't know how to help… but want it to be fixed anyway… Any news regarding unicode in irc log? Or some info or something so that I can be more useful? :/ 16:49
yoleaux AlexDaniel: I'll pass your message to moritz.
moritz AlexDaniel: Unicode in IRC logs seems fine, for the BMP 16:50
yoleaux 16:49Z <AlexDaniel> moritz: speaking of unicode… I know that I am probably being very annoying, sorry, I don't know how to help… but want it to be fixed anyway… Any news regarding unicode in irc log? Or some info or something so that I can be more useful? :/
16:50 nattefrost left
El_Che besides the emoji discussion, great addition for having thread-safe libs 16:50
16:51 robertle joined
moritz AlexDaniel: and due to the mysql backings, getting non-BMP Unicode is probably quite a journey... 16:51
AlexDaniel moritz: that I know, but even the output from camelia is broken
moritz AlexDaniel: irclog.perlgeek.de/perl6/2017-08-22#i_15052699 looks fine here
AlexDaniel (because every single message has at least ␤ in it)
moritz huh, the N/L is broken, the «..» look fine 16:52
wtf?
AlexDaniel propdump: «»␤
unicodable6 AlexDaniel, gist.github.com/4cacf070d5962466ef...632d12e16f
AlexDaniel m: ‘␤’.ords.say 16:54
camelia (10)
moritz AlexDaniel: if you want, I can give you shell access to the box so that you do your own debugging
AlexDaniel camelia: that's not what I meant
moritz: I can try, but do you have any clues?
moritz: like where should I look approximately 16:55
github.com/AlexDaniel.keys ← first one
moritz AlexDaniel: probably in the logger, since it comes out as a question mark in the direct DB output
AlexDaniel: ssh [email@hidden.address]
AlexDaniel worked 16:56
moritz AlexDaniel: 'mysql moritz5' gives you the DB prompt
and the sources are nextgen-ilbot, systemd service file in /etc/systemd/system/ilbot@.service
config in ~/nextgen-ilbot/config/*.conf 16:57
I tend to test by creating a copy of ilbot2.pl and give it a config that joins #bottest 16:58
16:58 mr_ron joined
geekosaur sometimes my irc mapping confuses things. this discussion would probably make more sense if I weren't rendering U+0010 as ␤ + newline 17:00
17:01 Cabanossi left 17:04 Cabanossi joined
mr_ron m: say [so "\c[CR]\c[LF]" ~~ / \x[0D] /, so "\c[CR]\c[LF]" ~~ / \x[0A] /, so "\c[CR]" ~~ / \x[0D] /, so "\c[LF]" ~~ / \x[0A] /]; say "\c[CR]\c[LF]".encode 17:04
camelia [False False True True]
utf8:0x<0d 0a>
timotimo m: say "\n".encode("utf8", :!translate-nl); say "\n".encode("utf8", :translate-nl) 17:05
camelia utf8:0x<0a>
utf8:0x<0a>
mr_ron Is there a good document someplace on p6 on dos newlines? 17:08
17:08 cdg joined 17:09 cdg_ joined
timotimo m: say "\r\n".encode("utf8", :!translate-nl); say "\r\n".encode("utf8", :translate-nl) 17:09
camelia utf8:0x<0d 0a>
utf8:0x<0d 0a>
timotimo maybe only on windows
mr_ron I am running on ubuntu 17:10
kannan El_Che : everything is good. i upgraded the VM, and re-installed rakudo-star 2017.07 from source. 17:11
17:13 cdg left
El_Che kannan: great to hear 17:15
mr_ron m: say so "\c[CR]\c[LF]" ~~ / \x[0D] /; say so "\c[CR]\c[LF]" ~~ / \x[0A] / # new RT ?? 17:18
camelia False
False
17:26 mr-fooba_ joined, stmuk_ joined 17:28 mr-foobar left 17:29 stmuk left 17:32 Cabanossi left 17:34 Cabanossi joined 17:38 itaipu joined, devmikey joined 17:40 pierre_ joined 17:45 pierre_ left 17:48 kannan left 17:56 MilkmanDan left 18:03 MilkmanDan joined 18:10 setty1 joined
jnthn mr_ron: CRLF is a single grapheme, regexes match at grapheme level, so no, wouldn't expect those to match 18:11
Working as designed.
18:12 wigner left
Geth doc: 489cc2d6c7 | (Zoffix Znet)++ (committed using GitHub Web editor) | doc/Language/unicode_texas.pod6
"Fix" display of atomic ops

The POD parser chokes due to ops having `=` in them, which are things to draw the table with.
Fixed by wrapping them in C<>, but the parser doesn't parse those in tables :(		 18:12
18:12 zakharyas joined
mr_ron jnthn: ABNF Core ( tools.ietf.org/html/rfc5234 ) does matching for CR, LF and CRLF. Any way for CRLF to match single ASCII CR / LF ? 18:14
jnthn No 18:15
mr_ron OK - will just document in US-ASCII module. Thanks. 18:16
18:19 geekosaur left 18:20 geekosaur joined 18:22 Kyo91 left 18:23 zakharyas left 18:24 DerAlex joined 18:29 lostinfog joined, konsolebox left 18:30 Kyo91 joined 18:32 Cabanossi left 18:34 Cabanossi joined, nuk9 left 18:36 konsolebox joined 18:37 cdg joined 18:40 cdg_ left
ryu0 grapheme? 18:41
18:42 cdg left 18:47 itaipu left
timotimo "something the user would call 'a single thing'" 18:47
18:48 Kyo91 left
moritz or a base character + combining characters also lumped into a single character 18:48
mr_ron maybe: stackoverflow.com/questions/273318...a-grapheme 18:49
timotimo also series of emoji joined with the zero-width-joiner, like the family emoji
18:49 Kyo91 joined 18:50 devmikey left 18:51 devmikey joined
ryu0 I see. 18:51
18:51 darutoko left
ryu0 a grapheme is a single letter or so, as a human would see it. 18:51
i've only really dealt with English/ASCII so the distinction is largely irrelevant to what i write... 18:52
timotimo english has words like "naive" which is supposed to have two dots on the i instead of just 1 :) 18:53
sena_kun arnsholt, re-ping.
18:53 itaipu joined 18:57 smls left, smls joined 18:59 domidumont left
AlexDaniel m: say ‘i’ ~ ‘̈’ 18:59
camelia 5===SORRY!5=== Error while compiling <tmp>
Missing required term after infix
at <tmp>:1
------> 3say ‘i’ ~7⏏5 ‘̈’
expecting any of:
prefix
term
AlexDaniel m: say ‘i􏿽xCC􏿽x88􏿽xE2􏿽x80􏿽x99 19:00
camelia ï
AlexDaniel timotimo: ↑ this does not feel right 19:01
it normalizes to ï, but shouldn't it have three dots in the end? :)
timotimo got no clue :) 19:02
[Coke] AlexDaniel: why? 19:03
oh. a funny.
AlexDaniel m: say ‘ı̈’ 19:04
camelia ı̈
geekosaur that may be up to font metrics 19:05
Geth doc: e610ddb097 | (Zoffix Znet)++ (committed using GitHub Web editor) | doc/Language/unicode_texas.pod6
Clarify Texas atomics are just subs
AlexDaniel so dotless i (ı) + diaeresis does not normalize into ï, but normal i does :)
Geth doc: 1ee4ec3a2c | (Zoffix Znet)++ (committed using GitHub Web editor) | doc/Language/unicode_texas.pod6
Fix typo in sub name
 19:06
19:06 domidumont joined 19:12 domidumont left 19:13 espadrine joined
smls m: say foo List; sub foo (::T --> T) { T.new: |<a b c> } 19:16
camelia Died with X::TypeCheck::Return
in sub foo at <tmp> line 1
in block <unit> at <tmp> line 1
smls ^^ Should this succeed ? 19:17
El_Che fyi, on my home ubuntu machine, the atom is not wide and noet emoji like. On the other hand, it's not very readable, more like a badly printed asterisk 19:19
19:22 jonas1 left
El_Che it looks like this: claudio.ulyssis.be/var/tmp/atom_vim_ubuntu.png 19:23
the size of it should be dependant of the font then
19:30 |oLa| left 19:31 |oLa| joined
[Coke] I think they're subbing in the other symbol, looks like. 19:33
19:35 brrt joined
[Coke] ... or maybe I'm crazy. nevermind. :) 19:35
Geth ecosystem: 2936c63286 | (Brian Duggan)++ | META.list
Add p6-jupyter-kernel to the ecosystem
 19:37
19:39 ccntrq left 19:40 ccntrq joined 19:44 Kyo91 left 19:55 bdmatatu left
timotimo yes!!! 20:06
finally <3 20:07
20:07 cdg joined
Skarsnik Good night people ^^ 20:08
lizmat Skarsnik: good night
20:08 Skarsnik left, itaipu left 20:09 cdg_ joined 20:10 Kyo91 joined, andrzejku_ left 20:12 cdg left 20:29 mr_ron left, AnotherNick left, khw joined
Xliff timotimo: why so excited over... what looks to be python bindings (?) for other languages? 20:30
Well... other than it makes them available to P6 (which is really cool now that I think about it) 20:31
20:31 char_var[buffer] joined
Xliff Ooor... that I miss the fact that there are notebooks using this thing. 20:34
Xliff googles further.
lizmat m: my %h = :1a, :2b; for %h -> (:$key,:$value) { dd $key, $value } # isn't that almost like magic ? 20:35
camelia Str $key = "a"
Int $value = 1
Str $key = "b"
Int $value = 2
lizmat just something I realized when looking at a very old P6Weekly report
20:36 Sgeo joined
Xliff lizmat: Yes, that is cool, once you learn it. 20:36
However I've always found that initialization format to be confusing.
:<Value><Key> 20:37
Breaks my haid.
From: try.jupyter.org/: "ts = ts.cumsum()" 20:39
Which proves that with a little thought, sometimes the obvious shortening is not a good idea. 20:40
];-)
20:42 mr-foobar joined
Xliff Probably better to use this link: tmp59.tmpnb.org/user/08yVC3UsZc5j/...thon.ipynb 20:42
Which may not work after a bit.
20:44 mr-fooba_ left 20:45 cdg_ left 20:47 Cabanossi left
Xliff extemporizes: "Where notebooks == Cool google doc like pages that can run code like an online MatLab" 20:47
20:49 Cabanossi joined 20:51 colomon left 20:54 cog_ left
timotimo Xliff: .sumcumlaude 20:56
223002 Xliff │ :<Value><Key> 20:57
oops
Xliff timotimo: .sumcumlaude.graduate 21:03
But I don't think ".cumsum.graduate" will do what you expect. :P 21:04
Mainly because I have a gutter brain. That's my story and I'm sticking to it!!
21:05 mcmillhj left
Xliff is still waiting for "zef install ." to complete. 21:05
21:05 cdg joined
Xliff I've since gone and: smoked cigarette, and diagnosed a wireless problem. It's "still going" 21:06
21:07 TEttinger joined 21:08 cdg left, cdg joined 21:20 skids left 21:23 ufobat left 21:27 wamba left 21:31 Guest29805 left 21:32 giraffe joined, giraffe is now known as Guest54652 21:33 lostinfog left 21:37 brrt left 21:38 cpage_ left 21:41 espadrine left 21:46 Cabanossi left 21:48 cdg_ joined 21:49 Cabanossi joined 21:51 cdg left 21:52 rindolf left 21:56 setty1 left 22:04 DerAlex left, Kyo91 left 22:05 cpage_ joined 22:06 nadim left 22:10 Peter_R joined 22:18 ggoebel left 22:19 nhywyll left 22:23 snarkyboojum joined
Xliff "zef install Net::ZMQ" - hangs rakudo 22:37
22:37 smls left
sena_kun Xliff, do you use arnsholt's master? 22:38
Xliff I use whatever is in META6.json
22:38 snarkyboojum left
sena_kun Seems like it then, well. 22:39
Xliff Using panda now, and at least it installs the dependencies.
sena_kun Xliff, you can try my fork out: github.com/Altai-man/Net-ZMQ - git clone, git checkout revitalize, zef --verbose install .
Xliff Still, no response from the Net::ZMQ tests.
sena_kun it works good with zmq 4.1, but 4.2 is... Yet to work, yes. 22:40
Xliff Crud. I think I am on zmq 5
sena_kun I am trying to contact arnsholt to merge update into master.
Xliff No. 4.2.1
sena_kun But still no luck.
Xliff, can you downgrade to 4.1 somwhow? Stable work for 4.2 is, ugh, a hard thing to talk about. 22:41
Xliff, for what package do you need this library?
Xliff Trying to install Net::ZMQ from clone
Worked. 22:42
p6-jupyter-kernel
sena_kun ah, okay.
the thing is.
22:43 mcmillhj joined
Xliff # Could not find Net::ZMQ::Constants at line 6 22:43
sena_kun Net::ZMQ was started on 2013 when zmq version was 2. And since then many things were changed. Some time ago I've updated it, wrote some docs, made api easier, etc. And broke API, of course.
Xliff, ^
Xliff Oh. Shuks. 22:44
sena_kun I can help you to port to my version.
It should be SO HARD.
I mean.
Xliff p6-jupyter-kernel doesn't realize that Net::ZMQ is now Net::ZMQ4
sena_kun s/Net::ZMQ::Constants/Net::ZMQ4::Constants/
Yeah.
Only in my fork though. :)
And some obsolete constants may be used too. 22:45
Xliff :(
This sounds bad.
sena_kun Actually, it is possible to write a compatibility layer for older zmq realizations, but I thought it is more important to support actual version in conditions of manpower lack.
Xliff, not SO bad, I think. I can try to run this module and write a update patch tomorrow or on weekends. Will it be merged? :) 22:46
Xliff So downgrading is also not an option then?
sena_kun Downgrading what?
Xliff zmq to 4.1
ugexe sena_kun: your fork can be installed via `zef install github.com/Altai-man/Net-ZMQ.git` 22:47
Xliff I have not analyzed p6-jupyter-kernel yet. I don't know how the changes in your Net::ZMQ branch will react to it.
timotimo i found it a bit difficult to understand what version and what api version of zmq i'd be on
sena_kun downgrade to 4.1 will make you safe from very persistent errors I'm getting on 4.2. But constatns are constants.
timotimo from any given package version
22:47 Cabanossi left
Xliff I blame timotimo 22:47
He started it.
22:47 mcmillhj left
sena_kun timotimo, well, Net::ZMQ used a mixture of versions API as I looked on it. 22:48
ugexe, thanks, it's fine, but I anyway want this changes be in the main repo.
22:48 Cabanossi joined
sena_kun If arnsholt is against... The fork will get a different name or something. :/ 22:49
and zef won't change installation branch of the repo for me. :P 22:50
Xliff This blows the docs for p6-jupyter-kernel and it was just added to the ecosystem, today. 22:51
Methinks they tested with an older version of Net::ZMQ
sena_kun Sure they were.
Looking on p6-jupyter-kernel sources right now... 22:53
22:54 mcmillhj joined
sena_kun I see some obsolete pieces, but it is not really a disaster, just to easy-fix 5-10 lines of code. 22:56
ugexe you can give it a commit id then, zef install github.com/Altai-man/Net-ZMQ@a317944
er zef install github.com/Altai-man/Net-ZMQ.git@a317944
sena_kun ugexe, oh, cool. didn't know about it. Still doesn't look enterprisey enough to just write a "Net::ZMQ" for everyone in meta file dependencies and go. :) 22:58
but I will use it next time.
ugexe no, but it lets other people test your fork with less steps
sena_kun sure.
22:59 mcmillhj left
Xliff LOL 23:00
Undeclared name:
ZMQ_FORWARDER used at line 113
So yes. This is now becoming an effort in futility.
Not to denegrate your efforts, sena_kun. 23:01
But now I would need to look in to what ZMQ_FORWARDER was, its intended purposes, and a workaround.
And I was so hoping to play with this tonight.
ugexe do you have an old version of Net::ZMQ installed? 23:02
sena_kun Xliff, forwarder is a piece of zmq_device, which was killed years ago. Now you just use zmq_proxy. 23:04
Xliff, should I write you a guide about an update? I'd more like to: 1)merge my fork with master; 2)create an issue for p6-jupyter-kernel with an explanations about how to port it to modern means. But it can easily take more than one night. :/ 23:06
Xliff Only used in one line: "try zmq_device(ZMQ_FORWARDER, $!socket, $!socket);"
I don't have the strength tonight for "more than one night"
Not even: "More than 20 minutes tbh."
Xliff looks up zmq_proxy
sena_kun use Net::ZMQ4::Proxy; 23:07
Then zmq_proxy(First_socket, Second_socket, Pointer[void]).
But I didn't test it. :)
23:07 mcmillhj joined
Xliff What does the poiner do? 23:08
Looks like a callback.
k... looks like I can do a NULL equiv.
sena_kun It's the "intermediate" you can stick in if you sant to sniff all packages that are forwarded from both sides. 23:09
See api.zeromq.org/4-2:zmq-proxy
s/sant/want/
Xliff Undeclared routine: 23:10
zmq_proxy used at line 115
sena_kun did you add an import? 23:11
Xliff Nope. You didn't imply one. :P
sena_kun :)
Try it. iirc it should work with `our` even without explicit "is export". If now I'll commit a patch quickly. 23:12
23:12 mcmillhj left
sena_kun *If not 23:12
I probably need to go to sleep already...
Xliff Shouldn't you be able to selectively import symbols even if not marked "is export" 23:16
23:17 skids joined, devmikey left
sena_kun There is more than one way to do it... :) I yet to remember everything about importing/exporting. 23:17
ugexe with `our` you use the full name: My::Module::sub_name(...)
Xliff kk
sena_kun Patches are welcome, btw. :) I am sure my perl 6 skills have a long way to go. 23:18
Xliff t/02-sandbox.t .. ok 23:20
Cannot resolve caller AUTOGEN(Net::ZMQ4::Socket: Net::ZMQ4::Message, Int); none of these
signatures match:
kk... consider that. I got past the first test with changes to Jupyter::Kernel::Service.pm6
23:20 mcmillhj joined
sena_kun What I dislike about this message is that it doesn't say you the actual subroutine name... 23:20
anyway, it seems like `send`-related trouble. 23:21
Xliff Yeah. And that is the limit of my strength ATM.
sena_kun or not. :/
Xliff Gotta crash. Or at least take a break.
sena_kun Xliff, while you're still here.
Xliff nooooooo
:)
sena_kun: ga 23:22
sena_kun I'll go to sleep now and be back in 10-12 hours or so. Then ping me with yoleaux or something, I will help you further with everything I can.
Regarding my fork, of course. :)
Xliff You ping ME with yoleaux. If I am active, I will respond. 23:23
I don't know what I will be doing in 10-12 hours.
Heck... I don't know what I am doing in the next 20 minutes..... maybe sleep.
Mmmm......sleeep
sena_kun Ah, okay.
Xliff Have a good night!
sena_kun o/ 23:25
timotimo i'll go to sleep
have a good one!
23:25 mcmillhj left 23:29 BenGoldberg joined
BenGoldberg m: say // ~~ Method; 23:34
camelia 5===SORRY!5=== Error while compiling <tmp>
Null regex not allowed
at <tmp>:1
------> 3say //7⏏5 ~~ Method;
BenGoldberg m: say /./ ~~ Method; 23:35
camelia True
23:35 ggherdov joined 23:38 cdg joined 23:40 mcmillhj joined 23:41 cdg_ left 23:44 sena_kun left 23:45 mcmillhj left 23:47 cdg left 23:50 cdg joined 23:55 cdg left, mcmillhj joined
Xliff .tell sena_kun You're right. It's .send related. 23:59
yoleaux Xliff: I'll pass your message to sena_kun.

Powered by Cro and the Raku® Programming Language.

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

Thank you!