🦋 Welcome to Raku! raku.org/ | evalbot usage: 'p6: say 3;' or /msg camelia p6: ... | irclog: colabti.org/irclogger/irclogger_log/raku
Set by ChanServ on 14 October 2019.
00:12 wamba left 00:25 aborazmeh joined, aborazmeh left, aborazmeh joined 00:38 cpan-raku left 00:39 cpan-raku joined, cpan-raku left, cpan-raku joined 00:40 markoong left
jdv79 tony-o: your addendum doesn't say how cpan does not support ver/auth/api. 00:53
also, i still don't get the point of api - feels like a misfeature 01:01
01:07 maggotbrain left 01:08 maggotbrain joined 01:12 maggotbrain left 01:13 Altai-man_ joined 01:14 maggotbrain joined 01:15 sena_kun left
guifa2 .tell jmerelo I have the Superposition article posted. Scheduled it for the 8th, but feel free to change it to someone else, didn't see if anyone else was scheduled 01:18
tellable6 guifa2, I'll pass your message to JJMerelo
01:36 guifa2 left 01:37 guifa2 joined 01:42 aborazmeh left 01:47 molaf left 02:00 molaf joined
codesections I can see Raku being hit by the Lisp Curse pretty badly. Just while working on a simple CLI, I ran into some limitations of the built-in arg-parsing (not providing helpful error messages for incorrect options). Since Raku is so powerful, I wrote my own argument handling code which, in a dozen or so lines, is 60% of the way to being an arg parsing library 02:22
And there's a decent chance that it will *stay* 60% of a library, and never grow into a real one 02:23
02:27 gnufr33dom joined 02:33 clarjon1 joined 02:36 lucasb left 03:36 quotable6 left, nativecallable6 left, coverable6 left, tellable6 left, shareable6 left, linkable6 left, bisectable6 left, releasable6 left, greppable6 left, benchable6 left, unicodable6 left, evalable6 left, bloatable6 left, statisfiable6 left, reportable6 left, committable6 left, sourceable6 left, notable6 left, squashable6 left, evalable6 joined, benchable6 joined, unicodable6 joined 03:37 bisectable6 joined, statisfiable6 joined, greppable6 joined, squashable6 joined, sourceable6 joined, reportable6 joined 03:38 bloatable6 joined, quotable6 joined, committable6 joined, nativecallable6 joined, coverable6 joined, linkable6 joined, tellable6 joined, notable6 joined, releasable6 joined 03:39 shareable6 joined 03:59 marcusr left 04:00 marcusr joined 04:01 Sgeo left 04:03 Sgeo joined 04:07 Sgeo left 04:11 Sgeo joined 04:14 sena_kun joined 04:16 Altai-man_ left 04:22 dataangel left 04:24 xinming joined 04:27 dataangel joined
tony-o jdv79: cpan won't index File::Temp from two different pause users with the same version 04:29
you can test that by downloading a File::Temp and uploading it to the appropriate directory and waiting to get an email letting you know it couldn't index 04:30
the "api" is basically handling the features of pause, indexing/filing into appropriate places a package manager should look/authenticating so it's not just a trash heap of tar files 04:31
04:33 Kaeipi left, Kaeipi joined 04:39 ab5tract left 04:43 zacts left 05:14 bocaneri joined 05:15 ab5tract joined 05:41 zacts joined
jdv79 tony-o: I don't think that used to be true but in any case it shouldn't be true. for instance, there are 2 JSON::Tiny dists from 2 different users in the p6-dists index. if its broken, it just needs to be fixed. 05:43
06:09 Kaeipi left 06:10 Kaiepi joined, ab5tract_ joined 06:11 ab5tract_ left 06:12 ab5tract_ joined 06:14 Altai-man_ joined 06:16 sena_kun left 06:30 skids left 06:40 ufobat joined 06:50 kensanata joined 06:54 andrzejku joined 07:01 grumble left 07:02 grumble joined 07:07 leont joined 07:23 patrickb joined 07:35 ufobat left 07:40 fluca1978 joined
fluca1978 is there a way to get operators from an object using MOP? Something like .^methods ? 07:42
moritz if by operators you mean methods, then yes :D 07:43
ab5tract_ fluca1978, you can use `$obj.^can($method)` to do duck typing 07:55
as moritz++ points out, operators are "subroutines with funny names" so they don't really fit into a "get them from an object" kind of concept 07:56
fluca1978 moritz: this is not working Array.^methods.grep: { .name ~~ '[]' } 07:57
I suspect that, as ab5tract says, the method has a different name like `postfix` something 07:58
ab5tract_ fluca1978, you are calling it on the class,
try on an instance
fluca1978 ab5tract: Array.new.^methods.grep: { .name ~~ '[]' } is not working too, there is something different with the names 07:59
ab5tract_ I don't know that Array has any method of that name. if it is implemented in a manner that matches userspace, the [] syntax is provided by defining `GET-POS/SET-POS` 08:00
m: say Array.new.^methods.grep: *.name ~~ 'GET-POS' 08:01
camelia WhateverCode object coerced to string (please use .gist or .raku to do that)
in block <unit> at <tmp> line 1
Cannot use Bool as Matcher with '.grep'. Did you mean to use $_ inside a block?
in block <unit> at <tmp> line 1
ab5tract_ m: say Array.new.^methods
camelia (iterator from-iterator from-list new STORE reification-target Method+{is-nodal}.new Method+{is-nodal}.new shape Method+{is-nodal}.new Method+{is-nodal}.new grab name of default dynamic clone Method+{is-nodal}.new Method+{is-nodal}.new Method+{is-noda…
ab5tract_ so we aren't giving it the right question
m: say Array.^methods ~~ Array.new.^methods 08:02
camelia False
ab5tract_ m: say Array.^methods
camelia (iterator from-iterator from-list new STORE reification-target Method+{is-nodal}.new Method+{is-nodal}.new shape Method+{is-nodal}.new Method+{is-nodal}.new grab name of default dynamic clone Method+{is-nodal}.new Method+{is-nodal}.new Method+{is-noda…
ab5tract_ so the instance isn't necessary but there cane be differences between the available methods between the class object an object instance of the class 08:03
m: say (Array.^methods (-) Any.^methods)
camelia Set(ACCEPTS ASSIGN-POS BIND-POS BIND-POS BUILDALL Bool CALL-ME Capture FLATTENABLE_HASH FLATTENABLE_LIST Int Method+{is-nodal}.new Method+{is-nodal}.new Method+{is-nodal}.new Method+{is-nodal}.new Method+{is-nodal}.new Method+{is-nodal}.new Method+{is…
ab5tract_ m: say (Array.^methods (-) Any.^methods (-) Iterable.^methods) 08:04
camelia Set(ACCEPTS ASSIGN-POS BIND-POS BIND-POS BUILDALL Bool CALL-ME Capture FLATTENABLE_HASH FLATTENABLE_LIST Int Method+{is-nodal}.new Method+{is-nodal}.new Method+{is-nodal}.new Method+{is-nodal}.new Method+{is-nodal}.new Method+{is-nodal}.new Method+{is…
ab5tract_ fluca1978, does that help explain it a bit? 08:05
m: say (Array.^methods (^) Iterable.^methods)
camelia Set(ACCEPTS ASSIGN-POS BIND-POS BIND-POS BUILDALL Bool CALL-ME Capture FLATTENABLE_HASH FLATTENABLE_LIST Int Method+{is-nodal}.new Method+{is-nodal}.new Method+{is-nodal}.new Method+{is-nodal}.new Method+{is-nodal}.new Method+{is-nodal}.new Method+{is…
ab5tract_ m: say (Array.^methods (^) Nodal.^methods)
camelia 5===SORRY!5=== Error while compiling <tmp>
Undeclared name:
Nodal used at line 1
ab5tract_ m: say (Array.^methods (^) Node.^methods) 08:06
camelia 5===SORRY!5=== Error while compiling <tmp>
Undeclared name:
Node used at line 1. Did you mean 'Code'?
ab5tract_ anyway, enough of me poking at poor camelia :)
fluca1978 ab5tract: thanks, I need to reason about it 08:07
ab5tract_ m: Array.new.^can('BIND-POS').say; say Array.new.^methods.grep: { .name ~~ 'BIND-POS' } 08:10
camelia (BIND-POS BIND-POS BIND-POS)
(BIND-POS BIND-POS)
ab5tract_ ^^ fluca1978 I can't explain that one :( ... maybe moritz or another core dev has an idea though 08:11
Ahh, are those just multi method targets? so the can also includes the proto but the grep doesn't? or..? 08:12
Wasn't expecting a list of methods in either case, to be honest
fluca1978 ab5tract: a multi method, may be? 08:13
ab5tract_ m: my @a = ^10; my &look = @a.^can('GET-POS'); dd &look; @a.&look(4).say 08:14
camelia Type check failed in assignment to &look; expected Callable but got List (())
in block <unit> at <tmp> line 1
ab5tract_ well that's a weird little corner I haven't gotten myself into before :( 08:15
I guess one is meant to go digging into signatures to find the right one?
08:18 JJMerelo joined 08:28 molaf left 08:34 Black_Ribbon left 08:40 JJMerelo left 08:42 wamba joined
fluca1978 ab5tract: what I discovered is that the method to access and element seems to be AT-POS, by Positional. The fact is that it is not printed out (seems to be hidden): .say for Array.^methods.sort: *.name cmp *.name; 09:07
while this is working fine: @array.AT-POS( 2 );
El_Che lo #rqku 09:11
lo #raku
09:13 JJMerelo joined
ab5tract_ ah, of course. silly brain fart had me thinking GET-POS. 09:13
fluca1978, what I can say with some certainty is that there are a few corners of the MOP where things are still not at the capabilities that we eventually expect 09:14
09:14 sena_kun joined
ab5tract_ IIRC there is no way for monkey patching a parent class to have an impact on descendant classes 09:14
(as an example) 09:15
09:15 Sgeo left
ab5tract_ o/ El_Che 09:15
El_Che OO is overrated :) 09:16
09:16 Altai-man_ left 09:18 rindolf joined
Geth_ advent: 780226cb38 | L'Alabameñu++ (committed using GitHub Web editor) | 20th/articles/rfc223.md
Create RFC233
09:31
advent: 1540ff5f07 | (Juan Julián Merelo Guervós)++ (committed using GitHub Web editor) | 20th/articles/rfc223.md
Merge pull request #57 from alabamenhu/patch-3

Create RFC233
09:47 clarjon1 left
tbrowder codesections: ref arg parsing have you looked at leont's Getopt::Long? 09:50
09:55 markoong joined 09:56 clarjon1 joined 09:59 JJMerelo left 10:00 markong joined, markoong left 10:01 clarjon1 left
leont I don't know the question, but I'd always recommend Getopt:Long ;-) 10:02
10:02 clarjon1 joined 10:10 marcusr left, marcusr joined 10:26 oddp joined 10:27 shlomif joined 10:28 rindolf left, shlomif is now known as rindolf
oddp Late to the party, but thanks again for yet another weekly! 10:35
Oh, yeah, wanted to ask: do we have something more idiomatic than grep in conjunction with the flip-flop operator for something like {drop,skip}_while in other langs? 10:36
fluca1978 I'm curious: why hash has some adverbs instead of methods? %hash<foo>:exists instead of %hash.exists( <foo> )? I think because it could be a little easier for slicing, but I cannot see the real advantage.
oddp fluca1978: does that help? perl6advent.wordpress.com/2013/12/...-only-way/ 10:38
Also check the comments below the article.\ 10:39
fluca1978 oddp: I'm reading 10:40
jnthn fluca1978: Various reasons, including array and hash subscripts being parsed differently from argument lists, but also because the indexers are operators, and the slicing semantics are part of the current language rather than part of the object's semantics (an indexable object only has to provide the semantics for getting/checking for existence of/deleting a single element). 10:43
fluca1978 jnthn: thanks, I was just realizing it while reading your comment. 10:54
jnthn: anyway I still don't understand why %hash.AT-KEY( <username>, exists => True ) is returning the value and not a boolean, being the adverb a "method argument" to an accessor I would expect it to perform as %hash<username>:exists 10:58
jnthn fluca1978: You have to call .EXISTS-KEY; the exists adverb there is just being ignored 10:59
fluca1978 jnthn: thanks 11:00
11:11 pecastro joined 11:14 Altai-man_ joined 11:16 sena_kun left
tbrowder leont: codesections mentioned problems with param parsing and rolling his/her own 11:21
yours looks like something to start with instead 11:22
11:24 ab5tract left, ab5tract_ left 11:28 kensanata left
codesections tbrowder, that looks like a good module :) It's not a good fit for what I'm doing right now for a couple of reasons. First, I misspoke – I'm not really augmenting Raku's argument *parsing* so much help text generation, which is slightly outside Getopt::Long's purview (I think) 11:31
Second, and more importantly, for this project I'm not using external modules (for … reasons) 11:32
tbrowder codesections: i'm working on a wrapper for that to auto generate stuff for my CLI use. that will use my Abbreviations module as a helper 11:33
11:35 pecastro left
tbrowder my idea is to provide a hash keyed by leont's option specifier whose value is the help string. the options get autoalised with Abbreviations 11:36
*aliased
El_Che tbrowder: yeah, that's what I do, but I need to fix a few bugsd 11:37
tbrowder wrapper name at the moment is Opt::Handler
El_Che: is it published? 11:38
codesections That seems a bit fancier than I'm being. I'm just grouping CLI commands into ARGS, FLAGS, and OPTIONS, printing their WHY, and printing their default value (if any)
El_Che github.com/nxadm/SuperMAIN
was on holiday, so I didn't had the time to fix stuff 11:39
codesections Unrelated question: there isn't any way to introspect into a subset, is there?
El_Che codesections sent great bug reports
the '='bug is probably a 1 line change :)
tbrowder oh yeah, my only prob with yrs is it uses MAIN which i stay away from 11:40
codesections What do you dislike about MAIN?
El_Che lol, MAIN was what I wanted to keep specifically
requirements vary of course 11:41
tbrowder i can't get any support for line wrapping for one thing
codesections You mean based on terminal width?
tbrowder and my style of CLI is really baked in
yes
El_Che I love the idea of a built-in capable MAIN 11:42
but the capabilities are just very slightly lacking for my use case 11:43
for something more advanced, you can always reach for sonething more advanced, like leont's modules
codesections tbrowder, how are you detecting the terminal width, by the way?
tbrowder i've played with Raku's a bit but i like mine better. after i figured out leont's capture handling i think my opt::handler will be a viable alternative
El_Che tbrowder: it could be interesting indeed 11:44
tbrowder i don't. i use gnu's guidance on 70 or so characters and either manually wrap or use my Text::Utils line wrapper. 11:45
codesections Oh, well in that case, what about MAIN prevents line wrapping? 11:46
tbrowder as an example of what bugs me, look at Documentable's help
codesections I'm using MAIN and wrapping text in USAGE
tbrowder well that's one of the strengths of raku isn't it. something like: do it my way or take the highway! ;-D 11:49
codesections :D
Yep! Though it can also be a weakness – which is what got me thinking about the Lisp Curse angle 11:50
tbrowder i'm too set in my ways to learn everything all over again
yep.
codesections Oh, yeah, I wouldn't like lines like `documentable --help` has. But that's easy to fix with USAGE
tbrowder my prob is the rabbit hole effect. my current rabbit started out as a pdf check writing project....and there i go hopping down the bunny trail 11:52
how 11:53
i''k
codesections Ha. That reminds me of the judge I used to clerk for. He was always telling us to stay on the trail of the big 'coon, not to go down rabbit trails 11:54
tbrowder i'll have to look at it again, but leont's option specs are really powerful for my needs
El_Che yak shaving to the recue 11:56
tbrowder does anyone here use mi6. i just started and can't get my second release.
because it doesn't like my Changes or version or something. 11:57
it's very useful but lite on help 11:58
got to go. bye
12:07 zacts left 12:29 wamba left
tbrowder back. can anyone help me with an mi6 release problem? 12:37
codesections Wish I could, but I haven't yet used mi6 12:56
tbrowder the first release i did went pretty well, but the second is wedged. i'm trying to take good notes and file issues, but the overall process does ease a release (just not yet well documented). 13:06
13:19 skids joined 13:36 kensanata joined 13:42 MasterDuke joined
rypervenche What is wrong with this regex? It is taking forever to run and I'm sure it should be a pretty quick thing. 13:56
$_ = 'This is my test string, for testing purposes.'; say m/ test <!before ing> /;
evalable6 「test」
rypervenche Hmmm.. 13:58
evalable6: say $*RAKU.compiler.version; 14:00
evalable6 v2020.07.54.gd.4.bac.47.ff
rypervenche Oh....my bad. Ignore me.
14:01 lucasb joined
rypervenche I somehow left off the "raku" after "#!/usr/bin/env" >< 14:01
codesections That does seem like it'd be a problem for Raku :D I'm surprised it even started
how does `OUTER.keys` work, again? It's not really documented, and I keep unexpectedly getting `()` 14:02
14:03 wamba joined
codesections Oh, because I was supposed to be spelling it OUTER::.keys 14:04
14:05 Kaiepi left
codesections nevermind, then :) 14:05
14:05 aluaces left, Kaeipi joined 14:10 squashable6 left 14:11 Kaiepi joined, squashable6 joined 14:12 Merfont joined 14:13 Kaiepi left 14:14 sena_kun joined 14:15 Merfont left, Kaeipi left 14:17 Altai-man_ left 14:22 Rawwar joined
codesections m: my $first = 1,; my $second = 2,; $first.set_why('1st'); $second.set_why('2nd'); say "first: {$first.WHY}"; 14:23
camelia first: 2nd
14:26 andrzejku left
codesections .tell guifa You probably already know this, but I didn't: it's not only not currently possible to document a variable with a Pod declarator block, you can't currently set the WHY manually – trying to do so will set the WHY for the entire *type*, not just the variable. (which can be seen with this code: my $first = 1,; my $second = 2,; $first.set_why('1st'); $second.set_why('2nd'); say "first: 14:28
tellable6 codesections, I'll pass your message to guifa
codesections {$first.WHY}";
14:28 andrzejku joined
guifa rypervenche: indeed, before/after means “immediately before/after”. That said…. it’s possible to program a new token that would be anywhere-before and anywhere-after 14:30
tellable6 2020-08-06T08:20:38Z #raku-dev <JJMerelo> guifa2 OK, great, thanks. That date is probably OK, I'll check.
14:40 Rawwar left 14:53 Kaiepi joined
codesections I can't decide if this hack is so horrible that I need to cut it, or so amazing that I need to leave it in: 14:54
m: my $total = 97; my $subtotal = 84; sub fmt-with-percent-of($num, $name) {sprintf("%4d (%d%% of $name)", $num, 100 × $num/OUTER::("\$$name")) }; say 42.&fmt-with-percent-of('total'); say 42.&fmt-with-percent-of('subtotal');
camelia 42 (43% of total)
42 (50% of subtotal)
codesections (It really ought to be a macro, once Raku AST is fully implemented) 14:55
14:55 Kaiepi left
guifa lol but at that point, why not just pass in the variable itself? 14:58
42.&fmt-with-percent-of($total) saves you a character ;-)
codesections But then it can't print "(XX% of $NAME)" :D 14:59
guifa fair 15:01
15:01 gnufr33dom left
guifa Also I’m reminded now of why the internatinoal formatters are going to be so awesome 15:02
codesections oh?
guifa you’ll be able to do local-number $foo, :type<percent> and it’ll autmagically format a number as a percent for you
I’ll probably add the binding myself but you could also do 15:03
codesections what's it a percent of? 1?
guifa my \local-percent = &local-number.assuming(:type<percent>) 15:04
Yeah, that aligns with how most other formatters handle it
15:04 orinthe left
guifa (e.g. ICU) 15:04
15:05 orinthe joined
codesections tangent: why do you write that as `my \local-percent` instead of `my &local-percent`? 15:05
guifa no morning coffee yet 15:06
codesections (I'm still trying to get my head around when it's good style to use sigil-less variables)
guifa it should be &
15:13 Kaiepi joined 15:17 MilkmanDan left, MilkmanDan joined 15:29 domidumont joined 15:31 ssm left
tony-o jdv79: JSON::Tiny, are you referring to a search done on modules.rakudo.org? 15:32
15:32 JJMerelo joined, MilkmanDan left
jdv79 nope. what is on cpan and in the, I believe unused, p6-dists index 15:35
15:35 Murilo joined, Murilo left
jdv79 also, how are you going to fight off the bots in terms of signups to this new "ecosystem"? afaik that's the only reason pause is still manual approval. 15:37
El_Che signup like it's 1999 15:38
15:40 ssm joined
jdv79 yeah, its annoying, but its simple and effective and one only has to go through it once... 15:41
15:41 Sgeo joined
El_Che I disagree 15:41
jdv79 on what points?
El_Che it makes no sense whatsoever in 2020 15:42
cpan, registering, the pain
it has been working for perl for decases, no sense chaging it there
but shoehorn that model for a newer language? crazy
*changing 15:43
npm tried it
it turned out like sh*t
codesections "it turned out like sh*t" That can be taken two ways – and, for each meaning, I think you'll find *lots* of people who agree! 15:45
jdv79 what exactly about npm are you referring? the crazy dep graphs or the lack of security or...?
El_Che ask the devops people to deploy perl, python, ruby, JS and they will cry 15:46
jdv79 its impossible to discuss something without facts. cool. 15:47
El_Che I am saying that the CPAN model only worked for Perl because they had the talent and the manpower at a time where public git repos weren't a thing
15:48 demostanis joined
codesections El_Che, I'm not following what side your on in this discussion. Are you saying that manual approval is good or bad in your opinion? 15:48
moritz one option would be to allow login with github, and only require manual approval for those who don't
codesections s/your/you're/ gosh
El_Che codesections: I prefer no login, no registration, just a repo + git tag
you put a repo + tag as a dependency 15:49
jdv79 El_Che: that's p6c
El_Che jdv79: that's closer, yes
jdv79 which goes back to my original idea - why can't the existing solutions be used?
El_Che besides the need to maintain a central list
jdv79 cpan and/or p6c - ok. but why a third.
El_Che jdv79: in my eyes p6c > cpan 15:50
codesections so how does that fit in with the stuff you said that's critical of npm? Isn't that basically the non-approval free-for-all approach that npm takes?
jdv79 well, the client has to have a list somehow
demostanis Hello everybody! Is there an easier way to check if a list of string isn't in a hash's keys? Currently I am doing it like that: for <first second> { .callback if $_ ∉ hash.keys.cache }
El_Che jdv79: I would put that in the META6.json
15:50 MilkmanDan joined
jdv79 it is in the METAs 15:50
El_Che repo + tag instead of name
jdv79 the "index" is really just a list of all metas
El_Che so a client knows where to get the dep en what version
jdv79: the problem with p6c is that nog tagging is requiring 15:51
-ed
Grinnz I haven't kept up on this discussion but for a data point, Perl used to have something called a module list which you could register for and which was manually maintained. It got abandoned 15:52
jdv79 no, we're just re-hashing basic dist repo stuff
Grinnz I just saw something about manual approval, sorry 15:53
codesections demostanis, what don't you like about that form? I.e., are you looking for something shorter, or what? (That reads pretty well to me)
jdv79 tag as in the git thing, not metadata
moritz demostanis: if you don't want to check individually, you could use a junction
15:54 zacts joined
moritz demostanis: like, %hash{none(@list)}:exists 15:54
jdv79 well, one of the "pain points" of using cpan for raku dists is the manual approval process
demostanis About what you guys are saying: can't we have a part free-for-all like NPM which requires something like an --insecure flag with Zef and another part manually checked not requiring a special flag? So that everyone is happy?
codesections: Yeah, I find it not short enough and I was wondering if there was another way
moritz in general, using %hash{$key}:exists is way faster than checking %hash.keys
demostanis moritz: I'm gonna try
El_Che demostanis: the security check is an illusion
demostanis Isn't the code read before putting it into p6c ecosystem? 15:55
Grinnz ah right. Well the reason for manual approval is to avoid obvious spam and malware mainly
El_Che demostanis: code is not set on stone
Grinnz Those things get uploaded by people who have no intention of contributing
or bots.
codesections There's been a fair bit of action on code review for security at github.com/crev-dev/crev/ 15:57
Grinnz If you base it on github accounts that would be something.
You can at least then report bad actors to github
jdv79 Grinnz: you're talking about pause or something else?
Grinnz yes
jdv79 that could be neat
codesections One thing I'd like to see related to Grinnz point is to have a highly-visible count of transitive dependencies per module 15:58
El_Che raku has a small ecosystem. Base it on CPAN and it will stay smaller (or smaller, as several people are not fond of using cpan)
Grinnz For perl it wouldn't work, I couldn't see relinquishing authentication control to github. But it's possible
jdv79 that's easy to do - dep graphs
El_Che want to create go module? Make the repo public, tag it using semantic version, done
codesections it's one thing to trust an author to not be malicious; it's entierly something else to trust that they've reviewed 500 transative deps for malice 15:59
Grinnz If you have an api that can return the deps for any given module you can graph it like on cpandeps.grinnz.com
codesections jdv79, yeah I know it's easy to do. But I'm talking about making it visible automatically. It's more about keeping that number front-and-center
jdv79 El_Che: so your only issue with p6c is that it doesn't support git tags?
El_Che my issue is mostly with a cpan flow 16:00
jdv79 the version is in the meta so seems a non--issue with me
El_Che for p6c, yes tags would make it more stable
how it would be implemented is less important
jdv79 is it unstable now?
El_Che if someone points to master, instead of a specific version (tag or commit), it can be shaky, yes 16:01
moritz JJMerelo, lizmat: are we still looking for articles for the 20th advent? if yes, I'd consider doing one about RFC 64
jdv79 its an interesting discussion to have - maybe we should draft something so its not just oral/typing/ephemeral
El_Che jdv79: most people are sane, so it's not a huge problem now
jdv79 some would argue that's a feature. and if someone did cause a mess i bet it would be better to address the root cause than try to box them in with more restrictive flows 16:03
sounds kinda like one of the argument used for java - "harder to make a mess" so good for any idiots to code in
Grinnz a dependency system definitely should be based on tags to be reliable, i agree
El_Che I think that it's important to combine a low bar for new contribution + reproducibility 16:04
Grinnz it's how docker and npm type stuff all work
El_Che so, github/gitlag without a central cpan is the easy way to public code
jdv79 tags are not immutable, history is rewriteable
demostanis moritz: Why does this return False? so %(:working, :good){none(<working good>)}:exists
jdv79 i fail to see how that factors well
El_Che tag is for making build repoducible
what google added to golang, for example, is a read-only proxy for tags 16:05
Grinnz it is, but you have to intentionally rewrite them
jdv79 a tag is just an alias to a sha and as long as the sha doesn't get gc'ed or otherwise disappers its repro'able...
16:05 xinming left
Grinnz whereas branches become unstable as soon as you work on them 16:05
El_Che (which you are free not to use)
jdv79: commits disapper (or are rewritten), tag don't 16:06
of at least people should know you're not suppose to reuse them
Grinnz a tag also keeps said commit from getting gc doesn't it?
jdv79 tags can move/disappear
yeah
El_Che jdv79: yes, and people will hae you for that
hate
and the blame will be clearly on you :) 16:07
codesections I hope that the "github" comments are just a for instance and that any system would work with any git host (not just the Microsoft-owned one)
jdv79 if its my repo i'll git how i like and if i cause a mess then do something about it...?
16:07 xinming joined
Grinnz modifying any tag on a repo is just as problematic as force pushing a branch other people are using 16:07
El_Che codesections: yes, thatś why I metioned gitlab
but I also think gitea
codesections source hut, too
El_Che the git protocol basically
moritz I'd like to point out that for some some people, a tag in a git repo is a perfectly fine release. Others require a build process 16:08
jdv79 anyway, there's nothing stopping anyone from using a tag in p6c
guifa demostanis: well, both working and good exist
jdv79 its just not the only ref you can use:)
codesections yeah, exactly — pure git protocol, without relying on any vendor-specific addons
El_Che moritz: build from a tag should be reproducible
moritz El_Che: the emphasis is on *should*
guifa none would only return true if they *didn’t* exist 16:09
m: say so %(:bar, :foo){none(<working good>)}:exists
camelia True
moritz demostanis: it's false that none of these keys exists
El_Che moritz: if it doesn't something is broken
demostanis Ohhhh my mistake I understand now
El_Che that brings me on my 2nd stolen idea, but for apps
moritz El_Che: most software builds in the wild are not reproducible
El_Che they should vendor their deps in the repo 16:10
ie including the code of the deps
moritz you can call that broken all you want, it's just reality
El_Che so the build is identical eacht time
moritz then build tool versions, time-dependent things, OS versions etc. can still make a difference
El_Che I agree 16:11
but you are not offering build artifacts
the user's CI is building stuff. And nothing changes on their part they shouldn't expect surprises 16:12
We can't avoid it completely, but we can limit the surprise factor
moritz > but you are not offering build artifacts 16:13
jdv79 iirc there was some linux distro or soemthing that managed actual repro'able builds a few years ago
moritz well, that's the point, maybe we should
El_Che we're offering source code
jdv79 seemed extreme but ok
tony-o shoe horning an ecosystem into a stack it wasn't designed for is just a bad idea. there isn't much more to say about it. 16:14
moritz nearly all of the package repos I know offer build artifacts, not (just) source code
cpan, npm, pypi, ruby gems etc
they all have tarballs or other archives that are built from a source 16:15
El_Che tony-o: I don't know if you're agreeing with me or or giving a counter argument. But's that exactly how I feel about the raku+cpan combination
tony-o git is already being used as the protocol. that doesn't mean the current ecosystem is a good design and doesn't solve the problem of it being a PITA to contribute modules
El_Che "cpan, npm, pypi, ruby gems" <== that's exactly what devops and ops hate to deploy
16:15 dolmen joined
tony-o El_Che: i agree with you in the sense that the current arrangement is becoming tedious and can be majorly improved 16:16
jdv79 i personally believe the current cpan AND p6c options can be much improved to the point where a third option is not necessary
tony-o jdv79: what is your plan to do that?
jdv79 there is no plan:) 16:17
tony-o then how can you confidently say it's less work to "fix" those than do a third option? 16:18
jdv79 because the pain points are known and the solutions are not terribly complicated. as opposed to a from scratch effort? seems like a good bet;) 16:19
tony-o if that were true then someone would surely have taken credit for such low hanging fruit 16:20
jdv79 i'll stop now. i've said my bit. i like your energy and effort and don't want to argue anymore.
tony-o jdv79++ 16:21
cpan-raku New module released to CPAN! Abbreviations (0.1.0) by 03TBROWDER 16:26
16:29 ccube joined
ccube where can I find the public key for rakudo (to verify detached signature) 16:30
p6: say time 16:34
camelia 1596731659
ccube p6: say time
camelia 1596731670
Geth_ advent: 1e061ddd2d | (Moritz Lenz)++ | 2 files
Claim RFC 64, and add a draft for it
16:37
advent: ed152047d8 | (Moritz Lenz)++ | 20th/articles/rfc64.md
Wording improvement
16:46
16:46 ccube left
moritz feedback on this article would be very welcome 16:46
JJMerelo moritz++ I'll try to have a look, probably tomorrow. 16:55
Thanks a lot.
OK, I finally did it now... 16:57
ccube I think you can get it from raku.org/downloads 16:58
And yes, moritz, we're still looking for articles. We've still got 6 slots to fill. 17:00
So thanks.
guifa JJMerelo: lemme know if you need another artile and I’ll try to get.......
oh nm
there we go :-)
JJMerelo guifa we definitely need them... 6 unfilled slots yet. If you want, a slot is yours... 17:01
jnthn JJMerelo: How soon do they need writing? I may have some post-vacation writing energy... :) 17:05
17:05 JJMerelo left
guifa jjmerelo: I’ll work on it as soon as I get done fighting with this stupid localtime/gmttime algorithm. I’ve got… something… off even though I’ve sworn I’ve translated the code exactly 17:06
tellable6 guifa, I'll pass your message to JJMerelo
guifa jnthn: feeling refreshed? 17:07
moritz jnthn: welcome back :-) 17:11
jnthn guifa: Yes; I turned off github email notifications and didn't touch a computer (well, my smartphone aside) for about 2 weeks, which was quite nice :) 17:13
17:14 Altai-man_ joined
guifa ….until you turned it back on and got a flood of INT_MAX notifications ;-) 17:14
17:16 sena_kun left
jnthn :D 17:18
No, it doesn't resend everything :) I can find those via the web UI though :)
Home time, bbl 17:22
17:22 andrzejku left 17:30 dolmen left 17:45 domidumont left 17:55 cpan-raku left 17:56 cpan-raku joined, cpan-raku left, cpan-raku joined 18:17 MilkmanDan left 18:21 bocaneri left
Geth_ advent: 3400415e07 | (Moritz Lenz)++ | 20th/articles/rfc64.md
Apply feedback by JJ++
18:33
codesections I'm having a bit of trouble with a syntax issue that seems like it should be easy: destructuring a hash into %sigiled variables. This works (with `$`): 18:38
m: my %h = a => 0, b => 1; my ($a, $b) = %h; say $a; say $b
camelia a => 0
b => 1
codesections how do I do the same thing, but with %a and %b? 18:39
guifa my %h = a => %(x => 1, y => 2), b => %(x => 1, y => 2); my (:%a, :%b) := %h; say %a 18:41
evalable6 {x => 1, y => 2}
guifa I don’t think there’s actually a guarantee that in your scalar example that $a and $b will necessarily be a and b from the hash.
m: my %h = a => 0, b => 1; my ($b, $a) = %h; say $b; 18:42
camelia b => 1
guifa hrm 18:44
actually even when using :%a and :%b you’re still not guaranteed it it seems.