🦋 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.
00:27 Noisytoot left 00:30 Noisytoot joined 01:10 approaching236 left, vgrato_ joined 01:34 zakame joined 01:59 molaf left 02:12 molaf joined 02:15 marcusr left 02:16 marcusr joined 02:41 mahafyi joined
mahafyi With respect to building rakudo star from latest src, a small suggestion for an addition in rakudo.org/star/source : Build instructions can include the README.md instructions so 'cd <to pwd>' ; git init ; ./bin/rstar install' . 03:19
03:30 fooist left 04:30 sourceable6 left, statisfiable6 left, coverable6 left, releasable6 left, unicodable6 left, linkable6 left, quotable6 left, shareable6 left, tellable6 left, greppable6 left, reportable6 left, bloatable6 left, committable6 left, notable6 left, evalable6 left, nativecallable6 left, bisectable6 left, squashable6 left, benchable6 left, notable6 joined, reportable6 joined 04:31 nativecallable6 joined, squashable6 joined 04:32 tellable6 joined, benchable6 joined, coverable6 joined, linkable6 joined, quotable6 joined, bloatable6 joined, sourceable6 joined, unicodable6 joined, committable6 joined, shareable6 joined, releasable6 joined 04:33 statisfiable6 joined, greppable6 joined, bisectable6 joined, evalable6 joined 05:14 molaf left 05:27 aborazmeh left, zacts joined 05:33 bocaneri joined 05:37 mahafyi left 05:40 andrzejku joined 05:41 andrzejku left 05:45 squashable6 left
SmokeMachine guifa: cool module (Test::Inline) it reminds me of github.com/FCO/Test-Fuzz (different intent, but using modules as “test files”) 05:46
05:48 squashable6 joined 06:03 hungryd19 left 06:04 domidumont joined 06:27 billN1VUX left 06:31 zacts left, domidumont left 06:33 billN1VUX joined 06:51 domidumont joined 06:54 aluaces joined 06:58 mahafyi joined 07:01 finanalyst joined 07:05 dolmen joined 07:13 MasterDuke joined 07:32 pecastro joined 07:33 tejr left 07:40 tejr joined 07:46 Sgeo left 07:54 Altai-man joined 08:00 hungrydonkey joined 08:04 mahafyi left
El_Che ok, I'll relqse linux packages today 08:22
let's see if everything builds ok
Altai-man and everyone: thx for the releaase
Altai-man El_Che, hi! There were some issues, ahem, but 2020.08.2 point is the sane one. 08:23
I am not sure about building previous ones.
Do we package broken releases or skip them?
El_Che ah, good coicidence I reappeared after the latest dot release then
Altai-man Yes.
El_Che I skip them if I know they are broken 08:24
Altai-man Roger. So go straight for 2020.08.2 then.
timotimo an annoyance i just stumbled upon is that i read 52 bits from a 64 bit big buffer, and i got back a signed integer
El_Che I leave them in the repos if already packaged because they are updates by the system on the user's machine
08:25 rir_ joined, rir left
El_Che Altai-man: I'll check, but did everything get a dot release or only rakudo? 08:25
Altai-man El_Che, Moar is 2020.08, no points, nqp and rakudo are at 2020.08.2 08:26
El_Che thx
08:31 hungryd79 joined
MasterDuke timotimo: that's what i'd expect right now. rakudo/moarvm support for unsigned stuff is still a bit lacking, so i assume everything is signed 08:31
08:32 hungrydonkey left
MasterDuke i had a moarvm branch that added a bunch of *_u versions of a lot of ops, but i either ran out of steam or hit a sticking point, because i don't think i ever submitted a pr 08:33
08:43 mahafyi joined 08:46 dotdotdot joined
timotimo hm, so reading n bits where n < 64 you'll get sign-extension 08:51
i guess that's fair
08:55 aluaces left
timotimo i'm definitely too dumb to take apart a num after writing it to a buf and then mathematically putting it back together 09:03
MasterDuke github.com/dankogai/p6-num-hexfloat might have something to steal 09:13
09:13 hungryd79 left, hungrydonkey joined
timotimo i wanted to double-check the num from the recent moarvm issue on github by using Rat arithmetic 09:22
MasterDuke github.com/MoarVM/MoarVM/issues/1289 ? 09:30
09:34 aluaces joined
timotimo exactly 09:34
09:35 ufobat joined
MasterDuke there's got to be some online tool for that 09:36
timotimo for sure 09:50
10:05 vgrato_ left 10:15 mniip joined 11:00 dolmen left 11:07 gnufr33dom joined
finanalyst HI. I just noticed that there is no reference on the raku.org home page of the license. There is a reference to Artistic 2.0 on docs.raku.org, so I assume its the same for Raku. 11:15
11:38 aborazmeh joined, aborazmeh left, aborazmeh joined
lizmat finanalyst: could you make an issue for that at github.com/raku/raku.org ? 11:39
11:41 domidumont left 11:42 domidumont joined
lizmat clickbaits rakudoweekly.blog/2020/08/31/2020-...ndidacies/ 11:43
finanalyst lizmat: done. I have to go out now, so can't do a PR. It should be a simple fix
11:46 sena_kun joined, Xliff left 11:48 Altai-man left 12:00 molaf joined 12:15 mniip left 12:17 mniip joined 12:20 kent\n left 12:23 domidumont left 12:24 kent\n joined 12:25 domidumont joined 12:33 molaf left 12:45 squashable6 left 12:46 KotH left 12:47 KotH joined 12:48 squashable6 joined 12:49 finanalyst left 12:54 aluaces left 12:55 mahafyi left 13:09 john_parr_ joined, john_parr_ is now known as st_elmo, st_elmo left, __jrjsmrtn__ joined, _jrjsmrtn left
[Coke] lizmat: Want me to come up with a template for the voting that is supposed to start in 5 days? 13:12
tellable6 [Coke], I'll pass your message to lizmat
[Coke] lizmat: It says "vote 1 to 5" but doesn't say what a winning vote looks like. 13:15
tellable6 [Coke], I'll pass your message to lizmat
[Coke] lizmat: is it "the top X vote recipients" (if so, what's X?) 13:16
tellable6 [Coke], I'll pass your message to lizmat
13:18 lizmat joined
[Coke] lizmat: opened github.com/Raku/Raku-Steering-Council/issues/23 13:36
13:36 aluaces joined
[Coke] anyone else knows the answer or has a comment, please chime in there. 13:36
13:48 finsternis left
[Coke] vrurg++ 13:59
AlexDaniel`: can you add your employer to the RSC nomination doc? 14:03
14:03 maggotbrain left
AlexDaniel` sure 14:03
I'm going to change the nomination a bit anyway
[Coke] jnthn: can you add your employer to the RSC nomination doc? 14:04
tellable6 [Coke], I'll pass your message to jnthn
[Coke] lizmat: can you add your employer (self, I assume) to the RSC nomination doc?
14:04 maggotbrain joined
[Coke] nine: can you add your employer to the RSC nomination doc? 14:04
tony-o, ugexe: can you add ugexe's employer to the RSC nomination doc? 14:05
.ask ugexe do you accept tony-o's nomination, btw?
tellable6 [Coke], I'll pass your message to ugexe
[Coke] if anyone knows how to reach the others who have nominations, please do so and have them update the pages. 14:06
(JJ already had his) 14:07
14:07 aluaces left
AlexDaniel` [Coke]: is this good enough? 14:08
ah, it's not markdown 14:09
OK I'll fix it up and change it later
[Coke] (everyone) assuming you're OK sharing it publicly. vrurg points out that maybe it's not public info. (but it may impact voting results) 14:10
14:10 finsternis joined 14:12 mowcat joined
lizmat I'd leave it up to the candidates themselves whether they want to include their employer or not 14:12
tellable6 2020-09-01T13:12:24Z #raku <[Coke]> lizmat: Want me to come up with a template for the voting that is supposed to start in 5 days?
2020-09-01T13:15:30Z #raku <[Coke]> lizmat: It says "vote 1 to 5" but doesn't say what a winning vote looks like.
2020-09-01T13:16:37Z #raku <[Coke]> lizmat: is it "the top X vote recipients" (if so, what's X?)
[Coke] lizmat: ok, but if we are supposed to reject candidates based on employer, I don't see how it can be optional. 14:13
lizmat I don't think, at this stage, can reject anybody based on emplyer
[Coke] based on my reading of github.com/Raku/Raku-Steering-Coun...f-interest
Then let's add something that it's not relevant to the initial round of voting? 14:14
lizmat well, I *do* find it relevant
codesections [Coke]: Couldn't candidates state that they do not share an employer with other candidates without disclosing their employer (if that's what they prefer?) 14:15
lizmat but I also deem it an honour question at this stage
that would work for me
[Coke] lizmat: the rules say people get disqualified. Are you saying that they will not get disqualified at this poitn?
Also: there's no guarantee for many employers that you know someone also works at your company. 14:16
lizmat hmmm... perhaps stating that they're not employed by anybody else of the candidates would work best
AlexDaniel` [Coke]: yeah, the process is a little bit of a hodgepodge :S Thanks for trying to bring clarity to it!
lizmat and *if* that work for the same employer, state which people they share employers with
*they
[Coke] ok. I'll just release the votes "as is" and let the RSC sort it out once votes are tallied. 14:17
but -1 from me on having that language in the rules and then not really enforcing it.
codesections [Coke]: It's *theoretically* true that two people could work for the same employer, both not feel comfortable revealing their employer, and *neither* know that the other works there. But that's a *lot* of things that have to go wrong
I still think that, so long as every candidate reveals their employer or avows that they don't share an employer, then you *are* enforcing it 14:18
AlexDaniel` yeah, absolutely weird to have a protection for getting a council that is too biased, and then to let the council itself to figure it out
lizmat codesections: indeed, and reversely, to give an example:
El_Che the idea is to prevent a takeover by a single company?
lizmat I am employed by WenZPerl BV in the NL, but I could be doing work for Edument 14:19
technically, not the same employer
still, I could have significant interest in lining up with opinions of Edument employees on the council, or lose the work 14:20
14:20 MasterDuke left, finanalyst joined
AlexDaniel` correct! And I feel like this is what that section is supposed to prevent 14:20
lizmat indeed 14:21
[Coke] the text in the document only mentions employer, not contract work. If you're also concerned about that, I think that would require more language
lizmat indeed...
and I think that is something for the first voted RSC to figure out
[Coke] but I'm fine. I'll just report the results "as is"
lizmat ++[Coke]
in the end it is all about trust of the community in the functioning of the RSC 14:22
if there is no trust, there is no way that enforcing any rules will be able to fix that 14:23
[Coke] My take: don't have rules if you
're not going to follow them.
s/follow/enforce/, I mean.
lizmat well, probably, but let's just work with what we have now, and let the RSC adjust the rules with a mandate from the community 14:25
[Coke] jnthn, ugexe, tony-o, ugexe: please feel free to consider your employer private information. I think it will not be required for the voting process.
lizmat or state the candidates that you share an employer with, without needing to mention the employer 14:26
[Coke]: to give another example 14:27
in 2000, Wendy was a member of a Dutch foundation for Internet providers
codesections [Coke]++ that said, for anyone who does _not_ consider their employer to be private info, listing it could be helpful because it would allow others to know whether they share an employer with you without realizing it
lizmat she could be that, because she worked for a company that was a member of that foundation
then we both were fired from that company 14:28
14:28 rindolf joined
lizmat the people of the foundation wanted her on the board so badly that they offered her a token job at their companies, just so she could stay on the board 14:28
she luckily declined
neither the company nor the foundation survived the next year 14:29
afk& 14:34
14:37 aborazmeh left 14:38 Sgeo joined 14:47 skyl4rk left
El_Che hi liz 14:49
14:52 gnufr33dom left
nine [Coke]: done 14:55
guifa SmokeMachine: Nice. I hadn’t seen that one. It could be interesting to see how well the two combine
[Coke] adds github.com/Raku/Raku-Steering-Coun...ng-form.md 15:02
nine++
Please feel free to add any new people that show up before voting starts. I won't hold people directly to the form, but it'll definitely make it easier for me to process votes. 15:03
Also: we only have nicks for 2 people. Won't impact vote tallying, but not sure if we need "real"/"legal" names.
15:16 skyl4rk joined 15:20 eseyman left 15:32 MasterDuke joined 15:34 approaching236 joined
tbrowder ok, i'm confused: 5 or 7 on the council? how many to vote for? 15:37
El_Che Altai-man AlexDaniel`: I am getting test fails about the profiler: gist.github.com/nxadm/1635b36be894...a5314baa47 15:40
codesections tbrowder: you vote for 5. 7 go on the council. Voting doesn't start until next week and candidates still have time to sign up 15:41
[Coke] tbrowder: you get 5 votes. there are 7 winners
guifa &a cmp &b # <— what should determine More Less Same?
(speaking hypothetically, since right now that will always return Same and produce a warning) 15:42
tbrowder ok, thnx. i saw the convo where the tpc voting method was being confused with ours and that confused me. how about adding info that to the ballot for fuzzy thinkers like me :-D 15:43
[Coke] guifa: that doesn't return same to me.
m: sub a { ... }; sub b { ... }; dd &a cmp &b; 15:44
camelia Sub object coerced to string (please use .gist or .raku to do that)
in block <unit> at <tmp> line 1
Sub object coerced to string (please use .gist or .raku to do that)
in block <unit> at <tmp> line 1
Order::Less
guifa Coke: hmm
Geth doc: codesections++ created pull request #3608:
Add xt/check-signatures
guifa I was testing with regex and they return Same always 15:45
15:45 Altai-man joined
Altai-man El_Che, hmm, our CI thinks it's fine. 15:45
guifa m: my &a = { $^b + 1}; my &b = { 'a' }; dd &a cmp &b 15:46
camelia Block object coerced to string (please use .gist or .raku to do that)
in block <unit> at <tmp> line 1
Block object coerced to string (please use .gist or .raku to do that)
in block <unit> at <tmp> line 1
Order::Same
El_Che I had 2 so far
restarted them
is the test time based?
Altai-man El_Che, can you log into an environment where it happens and try to manually invoke something with --profile?
codesections guifa: that's because regex all coerce to the empty string
which isn't great
El_Che no, it's travis
the container is gone
codesections m: say /a/.Str eq '' 15:47
camelia Regex object coerced to string (please use .gist or .raku to do that)
True
in block <unit> at <tmp> line 1
Altai-man It is not a dockerfile-driven?
guifa ahaaaaa I found it
Altai-man El_Che, travis allows you to log for debugging, by the way
s/log/log in/
guifa subs will stringify to their names
El_Che Altai-man: it is 15:48
guifa regularly blocks stringify to empty strings like regex
15:48 sena_kun left
guifa I think 15:48
Altai-man El_Che, I know it would take time and probably not the most interesting thing in the world, but I don't think there is a neat way to see why it fails otherwise.
And I don't think it is time based. :( 15:49
guifa Yup, that’s it
m: sub a { … }; sub b { … }; my &c = { … }; my &d = { … }; say [~] &a, &b, &c, &c, ‘*’ 15:50
camelia Sub object coerced to string (please use .gist or .raku to do that)
ab*
in block <unit> at <tmp> line 1
Sub object coerced to string (please use .gist or .raku to do that)
in block <unit> at <tmp> line 1
Block object coerced to string (p…
El_Che Altai-man: this one failed twice in a row: travis-ci.org/github/nxadm/rakudo-.../723130833 (just restarting it again, so it's live)
got to be a little afk
codesections m: my regex Foo { a }; say &Foo.Str 15:51
camelia Regex object coerced to string (please use .gist or .raku to do that)
Foo
in block <unit> at <tmp> line 1
codesections guifa: ^^^ looks like Regex stringify to their names too
They're just usually not named
guifa yeah. Hmm
I wondered if labeled blocks stringify to their label 15:52
codesections (yay consistency, I guess)
Geth doc: a4f343a9a2 | (Daniel Sockwell)++ (committed by Juan Julián Merelo Guervós) | xt/check-signatures.t
Add xt/check-signatures

This commit adds a new xtest, check-signatures.t. Here is the description of the test from it's documentation:
  > For each method documented in a .pod6 file in doc/doc/Type/, this
  > test compares the documented signature against the signature in the
... (24 more lines)
15:53
guifa Okay, so we stringify using the name of a Code object, or empty string if there’s not one. Why produce the warning then? Is it just for people who think doing “The result of sub a is &a” will run it? 15:57
codesections I think it's because we want to guide people towards .gist and .raku, right? 15:58
guifa …which leads to the question of why? .gist and .Str produce the same output lol. 15:59
Right now I’m thinking the best solution for github.com/rakudo/rakudo/issues/3885 16:00
And basically, I think the two best solutions are either (a) add a few more multis so that Code types use .gist instead of .Str, or (b) just get rid of the warning altogether for Code.Str 16:01
codesections Yeah, but it's a code clarity/intent sort of thing. `.gist` signals the reader "I want a human-readable representation of this thing" whereas `.Str` signals the reader "I want to convert this to a Str value" 16:02
guifa FWIW, the only two things that produce warnings on stringifying is Code and Nil 16:03
Threads feel roughly on the same level as Code, but they don’t produce an error 16:04
16:04 cpan-raku left
guifa ForeignCode doesn’t either, per the docs (though I’ve not tested it) 16:04
16:04 cpan-raku joined, cpan-raku left, cpan-raku joined
Altai-man El_Che, raised github.com/rakudo/rakudo/issues/38...-684962540 again, need more details of why it doesn't work in this particular env, will see how it goes. 16:04
codesections m: sub greet { 'Hey '}; say &greet ~ 'Bob'
camelia Sub object coerced to string (please use .gist or .raku to do that)
greetBob
in block <unit> at <tmp> line 1
codesections ^^ that feels like something that should produce a warning 16:05
guifa Fair
I could definitely see a beginner using that
El_Che Altai-man: I can try to recreate (where do I add the profile? at test time?). So far i think about 30% fails 16:06
Altai-man El_Che, if it's broken on some env the same way it was until we "fix" it, mere `raku --profile -e 'say "foo"'` would trigger error. 16:07
codesections guifa: I haven't looked into 3885, but could it be as simple as wrapping a few things in `quietly` blocks?
guifa That would work, but I think getting a general solution is better: everything else can pass through cmp without a warning (although they might get a weird ordering) 16:09
16:09 veesh left
El_Che (going to prepare food, bbl) 16:09
codesections No, I meant as a fix: putting the relevant parts of cmp inside quietly blocks
(instead of using a different multi with .gist)
16:10 veesh joined
guifa That just feels sort of….bandaid-y 16:11
codesections Fair :)
16:14 finanalyst left 16:15 finanalyst joined 16:17 finanalyst_ joined
guifa Final question, I suppose, is where to ptu it. in core.c/Order.pm6 or core.c/Code.pm6 16:18
lizmat: vrurg: y’all have most of the commits in Order.pm6, thoughts?
lizmat guifa: I guess in Order, as Code doesn't know about Order yet, looking at tools/templates/6.c/core_sources 16:23
16:23 mowcat left
codesections Is there such a thing as an `Any:D`? 16:24
Or is Any inherently :U ?
guifa Sure. Anything defined is also Any:D 16:25
guifa . o O ( unless it’s direct subclass of Mu, I suppose )
But Any:D === :D 16:26
codesections guifa: it *matches* Any:D, but it isn't Any:D, is it?
16:26 aluaces joined
guifa I mean 16:26
codesections Like, there's nothing where .WHAT returns Any and .defined returns True ? 16:27
guifa m: my $a = Any.new; say $a.WHAT; say $a.defined
camelia (Any)
True
guifa Not really sure what you’d use it for though lol
codesections Aha, interesting!
Not a thing in the world! Just ran into a type signature and got curious 16:28
m: say Any.^lookup('prepend').candidates.map(*.signature)[0] 16:29
camelia (Any:U \SELF: |values)
codesections (that type signature, in fact)
SmokeMachine guifa: yes, that would be interesting 16:31
16:46 ufobat_ joined 16:48 aluaces left 16:49 ufobat left
guifa okay weird, when I tried using .name … it ended up producing the same output as .raku instead of .name normally 16:54
17:06 eseyman joined 17:14 dotdotdot left, aluaces joined 17:15 billN1VUX left
tbrowder ref upcoming vote: does voting for more than five invalidate the ballot? i assume so. if true, that needs to be highlighted near the top of the ballot. 17:16
i would also put the note about the actual number being seated near the top. 17:17
tony-o [Coke]: i talked to ugexe before i nominated him. i'll ask him about his employer 17:19
(i know who it is, just want to make sure he's fine with it)
17:27 jmerelo joined
jmerelo Hey 17:27
17:27 hungrydonkey left
codesections o/ 17:27
17:27 hungrydonkey joined
tbrowder \o 17:28
i 17:29
i'm looking for a good, non-wordpress blogging platform that permits editing after the original post. suggestions? 17:30
codesections Do you want a dynamic site rather than one built by a static site generator? 17:31
17:43 molaf joined 17:44 finanalyst_ left, vrurg left, Util left, shadowpaste left, Celelibi left, gfldex left, agentzh left, stux|RC-only left 17:46 finanalyst_ joined, vrurg joined, Util joined, shadowpaste joined, Celelibi joined, gfldex joined, agentzh joined, stux|RC-only joined, stux|RC-only left
codesections Well, here's a LTA error msg: 17:46
m: class A { method file(IO:Path:D $f) {}}
camelia 5===SORRY!5=== Error while compiling <tmp>
Multiple type smileys cannot be used
at <tmp>:1
------> 3class A { method file(IO:Path:D7⏏5 $f) {}}
lizmat :Foo:Bar 17:47
hmmm
17:47 stux|RC-only joined
lizmat perhaps "did you forget a ":" somewhere" should be added to that message ? 17:47
[Coke] tony-o: (employer) followup: lizmat says not to require it, so no worries. 17:48
tbrowder: Rapi::Blog, based on Rapi::App, which is what news.perlfoundation.org uses? 17:49
[Coke] hurls metacpan.org/pod/Rapi::Blog
codesections lizmat: Yeah, or maybe even not say anything about multiple type smileys when one of them is ≥ 2 characters long?
it seems unlikely that the user tried to have 2 type smileys _and_ an incorrectly long smiley -- much more likely that they forgot a ':' 17:50
tbrowder [Coke]: thnx 17:51
17:51 wamba joined
tbrowder codesections: static sites imho are great for blogging 17:52
codesections Agreed. That's what I use (specifically, Zola)
17:53 wamba left 17:56 wamba joined
Geth doc: codesections++ created pull request #3609:
Fix typos detected by xt/check-signatures.t
17:57
guifa codesections: don’t forget that IO:Path could very well be a valid identifier. I wonder…. 17:58
ora t least, it can be with subs
[Coke] codesections++ 17:59
17:59 bocaneri left
[Coke] codesections: I may patch that xt file so it races, if you don't mind. 17:59
guifa Only :ver and :auth on classes
but
m: class A:ver<1> { }; sub foo (A:ver<1>:D $bar) { } 18:00
camelia 5===SORRY!5=== Error while compiling <tmp>
Multiple type smileys cannot be used
at <tmp>:1
------> 3class A:ver<1> { }; sub foo (A:ver<1>:D7⏏5 $bar) { }
codesections [Coke]: Please feel free. I tried to do so naively, but it caused tests to fail and I didn't think it worth investigating right now
(where naively === "I just added `race`") :D
guifa Aparently :ver can’t be specified in the signature? That actually feels like a great thing to use sometimes
codesections So, if you do, I'll be interested to see what needs to be done 18:01
[Coke] needs race, a lock around the hash you're saving results in, and then a loop at the end to dump results in alpha order.
18:01 aborazmeh joined, aborazmeh left, aborazmeh joined
[Coke] lizmat: it doesn't say this in the repo, but I'm assuming invididual votes are "secret", and I'll just be reporting on the aggregrate results. 18:02
codesections [Coke]: but what hash? The test pass/fail within the loop
[Coke] but if you race, they will be reported out of order.
so you save all the results, then dump them in order when the race is done. 18:03
it's not mandatory, but is nice to have.
lizmat [Coke] yes, I think voting should be secret in the sense that only you will know who voted for what
codesections Right, and I expected that. But, instead, results were wrong – even though I didn't see any shared state
[Coke] if you don't care about order, yes, emit the test result asap
lizmat: ok, good, same page.
codesections: maybe Test has some thread safety issues? 18:04
guifa Test not thread safe
codesections Yeah, that's what the Test docs say. (And that's what I didn't feel like debugging :D)
[Coke]: Why would I use race and then sort the results? My understaning is that that's what hyper does -- is that wrong? 18:05
[Coke] You're not returning results, you're doing stuff in the loop. If you return them, sure. 18:09
codesections Oh, right, I'm with you 18:10
[Coke] could probably stick a .map() onto the end of the .race.map we have to handle that. not a bad idea. 18:11
(then we could start getting output asap, not "when all chunks are done'
I'll see about updating aspell to take that approach.
codesections Well, anyway, if you make it work with race, I'll be interested in seeing what it takes to overcome the lack of thread safety. And if you don't, I may try my hand at it later
[Coke] er, hyper is sorted, race is not. 18:12
rypervenche tbrowder: I use Hugo for my personal stuff. I much prefer static site generators now. Much easier syntax and can pretty easily move it to another platform if need be. 18:13
codesections Well, hyper's *output* is similar, but it runs/can run out of order
s/similar/sorted/
[Coke] "Unlike hyper, race does not preserve the order of elements.
"
^^
codesections right, but hyper gives you the same output order as input order but doesn't perform the operations in order 18:14
[Coke] ah. looks tricky to update xt/aspell.t to use hyper because the map doesn't have a clearly defined "exit" point, it's inside a react block where we capture the output.
guifa Well, it’s not required to. I think at the moment it does
guifa . o O ( [Coke]’s mention of aspell reminds me to finish the hunspell port }
[Coke] codesections: ... neither hyper nor race perform things in order, right.
hyper returns the results in order, race does not.
codesections Yep, agreed. 18:15
[Coke] I read that you were disagreeing with the latter point there.
ok. same page.
codesections I did something kind of similar with github.com/Raku/doc/blob/master/ut...methods.p6
[Coke] for stuff not using react blocks in the information gathering, hyper may make sense. Will check.
18:17 molaf left 18:21 ufobat_ left 18:25 jmerelo left
[Coke] yah, as soon as I start using "ok" in the hyper flow, things start exploding, so the "save output in a hash" and then "loop over the hash" method seems to be the way to go. shame. 18:34
timotimo if you can, put the oks in a second stage with a .eager.map 18:35
then it'll be serialized, both in terms of the same order as it comes in, and on a single thread 18:36
[Coke] timotimo: huh. verified, eager helps. Why is it needed? 18:38
(looks like some items are just not received without it. 18:39
timotimo otherwise it'll just become part of the hyper pipeline as well
[Coke] ahhhh 18:40
timotimo eager is along the lazy-hyper spectrum / dimension 18:41
like, lazy, eager, hyper, race ... i guess
18:43 finanalyst_ left 18:45 wamba left 18:46 wamba joined 18:51 domidumont left
[Coke] codesections, timotimo: gist.github.com/coke/61bf77696acb4...b4942acfc5 19:15
if I try to hyper the processing, but keep the test output in the eager at the end, I see the random failures which are not reproducible when run individually.
19:17 aborazmeh left
codesections [Coke]: and, to be clear, you _don't_ see those errors if you do everything with a race + hash behind a lock? 19:18
[Coke] I don't see them in the original version without hyper/race 19:20
I didn't also implement a race version to test
codesections Ok, makes sense. You didn't run into any errors like that with aspell.t's race/lock, though? 19:21
19:21 bdju left
[Coke] nope. 19:22
19:22 bdju joined
tbrowder rypervenche: thnx, looking a zola and github pages at the moment... 19:24
codesections rypervenche: tbrowder: Zola and Hugo are _very_ similar -- Zola exists because the main developer really liked Hugo but couldn't stand its template syntax 19:26
rypervenche Yeah, I've started looking at Zola a bit. I'll see how I like it.
19:34 dotdotdot joined
guifa Ugh, okay, github newbie 19:37
How do I go about ammending a PR?
19:37 molaf joined
codesections If it hasn't been merged, just push to the same branch 19:37
guifa Will both committs end up in the master branch or will git treat them as one? 19:39
codesections two separate ones
[Coke] if it's a branch you control, you can squash them and force push. (In general you don't want to do this on a public branch) 19:40
you can also copy that branch to a new branch, squash locally, push and open a new PR. 19:41
codesections If you want it to be just one commit from the PR, you can do a "squash and merge" commit when you merge the PR via github, you're probably better off closing the PR, using `git rebase` to combine your commits, and then opening a new PR.
s/via github, you're/via github. But if you don't want to do that, then you're/ 19:42
(forgot, like, half a sentence there. Oops)
guifa Heh it’s okay. I basically just totally botched a C&P job when I had already rebased once and that’s hwy I’m in this situation :-) 19:43
19:46 sena_kun joined
MasterDuke no need to close the PR, then you lose comments. just git rebase and force push 19:47
19:47 Altai-man left
MasterDuke (/me has done that many many times to many many PRs) 19:48
guifa Well, for some reason git commandline doesn’t like my password (I normally use gitdesktop)
But it looks like the second one has already been picked up by the PR so … meh 19:49
codesections guifa: use an SSH key instead of a password! It's more convenient and **way** more secure 19:51
codesections gets off soapbox 19:52
MasterDuke i sometimes wish there was a syntax for "there are two arguments, one is of type A, the other is of type B; do X with A and Y with B, but i don't care what order the two are in" when writing multis 19:55
El_Che Altai-man: building locally now 19:58
tellable6 El_Che, I'll pass your message to Altai-man
El_Che Altai-man: profile test failes, but running raku --profile does not give an error 19:59
tellable6 El_Che, I'll pass your message to Altai-man
[Coke] MasterDuke: seems like that could go into a module, something like infix.... is reversible
MasterDuke hm 20:00
El_Che Altai-man: funnily enough the other ubuntu 16.04 pkg builds fine. The only difference I can find is that I build each project separately
tellable6 El_Che, I'll pass your message to Altai-man
codesections MasterDuke: is this the syntax you're trying to avoid? 20:01
m: multi f(Str $s, Int $i) { f($i, $s) }; multi f(Int $i, Str $s) { say "Int $i, Str $s"}; f('answer', 42)
camelia Int 42, Str answer
codesections that's not *that* verbose
MasterDuke m: multi f(Str $s, Int $i) { $s.Int + $i }; multi f(Int $i, Str $s) { $i + $s.Int }; say f("2", 3); say f(4, "23") # this kind of thing 20:04
camelia 5
27
El_Che Altai-man: oh I see now, moarvm has new (and broken) test. Older releases of moarvm did not have test so I didn't run the moarvm test there. Sadly, this mean I have a higher failure rate. I'll check what other OSes are impacted by the new tests
tellable6 El_Che, I'll pass your message to Altai-man
codesections m: multi f(Str $s, Int $i) { f($i, $s) }; multi f(Int $i, Str $s) { $i + $s.Int }; say f("2", 3); say f(4, "23") 20:05
camelia 5
27
codesections MasterDuke: but my point was that you can shorten that to ^^^ 20:06
which isn't _much_ shorter for a 1-line function, but would save duplicating the entire body if the function were longer 20:07
MasterDuke oh sure. and that's not a bad solution. but then you might have (depending on inlining) an extra function call in one 20:08
codesections True. given a sufficiently smart compiler, though... 20:09
(famous last words, I know)
El_Che a sufficiently smart compiler is indistinguishable from magic? 20:10
codesections Let's hope not :D 20:11
guifa Ugh. I give up with git. I’m just gonna delete my branch and start over from scratch. That will require a new PR though, yeah? 20:14
[Coke] yup. 20:15
timotimo you can force-push into the same branch
that will update the PR, too
20:15 rindolf left 20:32 roguelazer left 20:33 roguelazer joined 20:34 wamba left 20:39 approaching236 left 20:53 approaching236 joined
Geth Raku-Steering-Council/main: eb843071c9 | (Elizabeth Mattijsen)++ | nominations/2020/lizmat
Add advent blog posts
21:14
lizmat whee! tyil++ vrurg++
tellable6 2020-09-01T21:11:49Z #raku-dev <vrurg> lizmat tyil has fixed RSC notifications, we have have pushes reported into #raku.
[Coke] lizmat: I put together a voting-form.md in that directory. Please encourage people to follow that when voting. 21:16
21:52 AlexDaniel joined, AlexDaniel left, AlexDaniel joined 22:00 xelxebar left 22:01 xelxebar joined
timotimo who wants to build a cut-down version of CORE.setting that compiles like 5x faster and can still be used for simple tests? :S 22:22
MasterDuke interesting 22:36
timotimo env RAKUDO_MICROCORE=-threads,-bags,-supplies,-hyper make install 22:47
22:58 sena_kun left 23:35 pecastro left
Geth Raku-Steering-Council: codesections++ created pull request #25:
codesections nomination
23:49
Raku-Steering-Council/main: d1c57dee10 | (Daniel Sockwell)++ | 2 files
codesections nomination

Based on issue #20, I believe that my commit-bit to the raku/doc repository allows me to nominate myself to serve on the Raku Steering Council. If I am incorrect about that, please feel free to close this PR.
This PR also updates the voting form to include list my name and github handle.
23:52
Raku-Steering-Council/main: f1d7c4d77c | (Vadim Belman)++ (committed using GitHub Web editor) | 2 files
Merge pull request #25 from codesections-forks/codesections-nomination

  codesections nomination