00:19 squashable6 left 00:22 squashable6 joined 00:45 ggoebel joined 00:53 lucasb left
Geth roast: vrurg++ created pull request #583:
Roast Path to Raku
03:14 ggoebel left 06:24 ufobat_ is now known as ufobat 06:50 camelia joined
Geth ¦ problem-solving: hankache assigned to AlexDaniel Issue Perl 6 to Raku timeframe github.com/perl6/problem-solving/issues/118 07:04
07:14 discord6 joined 07:18 releasable6 left, releasable6 joined
tyil vrurg: perhaps he has a macro to sends a message and then immediately quits :p 07:48
|Tux| Rakudo version 2019.07.1-398-gf3dda96a8 - MoarVM version 2019.07.1-268-g07b00bbd7
csv-ip5xs0.757 - 0.771
csv-ip5xs-206.407 - 6.634
csv-parser20.429 - 20.497
csv-test-xs-200.420 - 0.423
test6.964 - 7.400
test-t1.737 - 1.766
test-t --race0.782 - 0.783
test-t-2029.641 - 29.702
test-t-20 --race8.860 - 9.166
lizmat is coming down with a cold 10:54
so don't expect too much from me the coming days
jnthn 'tis the season... 10:55
Get well soon
lizmat been out of bed for 1.5 hour, feeling like it's time to go back again :-( 10:57
AlexDaniel lizmat: so what's the situation with the website, remind me please 11:14
currently we have perl6.org which talks about Perl 6
lizmat: now, my understanding is that the easiest solution is to configure raku.org fully and make perl6.org redirect to it
lizmat and we have raku.org that redirects to it 11:15
AlexDaniel then change “Perl 6” on the website to “Raku”
lizmat I would hope that would could make a more visible change when redirecting to raku.org
AlexDaniel I remember you saying something about creating a new website, but that means we'll be stuck with “Perl 6” on the main website for a very long time
currently it has a banner on the main page “Perl 6 'Diwali' 6.d Language Specification Released” why not simply change that to say something about the rename 11:16
lizmat well, yes, creating a new website, and *then* reverse the redirects, so perl6.org redirects to raku.org
I'm not sure about the SEO effects if we would be offering the same content on raku.org 11:17
even though we would be redirecting from perl6.org to raku.org
AlexDaniel lizmat: is it my turn to be impatient now? I don't wanna wait for that, I want: reverse the redirects, change Perl 6 to Raku, change the banner on the main page, done (further changes are welcome later) 11:18
lizmat and what about modules.perl6.org ? and other subdomains? 11:19
really, I think we need to think this through 11:20
AlexDaniel lizmat: all that will redirect to *.raku.org
see github.com/perl6/problem-solving/issues/118
wanna change some text about perl6? Can't fully do it because we don't have urls working 11:21
and JJ is waiting for the domain for docs
lizmat which would be in parallel to docs.perl6.org or not?
AlexDaniel I hope not?
lizmat in any case, JJ prepared the new site *before* the redirect change 11:22
AlexDaniel why does it need to be in parallel? We're not forking
lizmat AlexDaniel: please, let me sleep on this for at least one more night
AlexDaniel I don't see any problem with being redirected to a new domain but being presented with old content, it's fine, we're transitioning
lizmat the cotton in my head is preventing me from thinking straight enough to see all implications 11:23
you don't see an issue with it, but Google might... and then we'd lose all the SEO value that we've built up in the past years
that's my worry
that's not something you build up easily 11:24
AlexDaniel there's also this, btw support.google.com/webmasters/answ...3106?hl=en 11:26
lizmat ok, that looks helpful... 11:28
AlexDaniel it starts with a 301, and the fact that the content is not changing shouldn't hurt 11:30
11:34 kawaii joined
lizmat I think the easiest for me would be to give rba control of the raku.org DNS 11:38
AlexDaniel that's right
kawaii www.irccloud.com/pastebin/FOoWHEXN/
lizmat this will take some time to change and require cooperation from rba 11:39
kawaii AlexDaniel: 🤔
AlexDaniel btw who besides rba has access to dns?
lizmat AlexDaniel: could you coordinate with rba wrt to setting up all necessary DNS for raku.org 11:40
AlexDaniel shareable6: hey what's up
shareable6 AlexDaniel, I cannot recognize this command. See wiki for some examples: github.com/perl6/whateverable/wiki/Shareable
AlexDaniel shareable6: HEAD
shareable6 AlexDaniel, whateverable.6lang.org/HEAD
lizmat and then tell me when that's done and what the IP numbers of those DNS servers are ?
AlexDaniel rba: ↑ :) 11:41
11:41 shareable6 left
AlexDaniel kawaii: yep, an issue on my side… one min 11:41
kawaii 🤭 11:42
AlexDaniel ohhh all my websites are down
kawaii excellent
11:42 shareable6 joined
rba lizmat: at the moment there is just a *raku.org -> * 11:42
lizmat: at the moment there just a raku.org/* -> perl6.org/* isn‘t it? 11:43
AlexDaniel rba: yes
kawaii: fixed 11:44
rba lizmat: So we will use Cloudflare as DNS, ok?
kawaii AlexDaniel: perfect, thanks, will post the results tonight/tomorrow morning 🦜
AlexDaniel u: 🦜 11:46
unicodable6 AlexDaniel, U+1F99C PARROT [So] (🦜)
AlexDaniel kawaii: is that… the right emoji? :D
kawaii AlexDaniel: would you have preferred 🦋?
AlexDaniel just, why? :D 11:47
u: 🦪
unicodable6 AlexDaniel, U+1F9AA OYSTER [So] (🦪)
12:20 ggoebel joined
AlexDaniel ok, Unicode::GCB passes Unicode 12 tests but doesn't pass Unicode 11 tests 12:21
which makes sense
soooo… check for rakudo version and choose the test file depending on that
sounds about right
AlexDaniel will submit a PR
why is it self!arity-error(args) and not self!arity-error($args) 12:52
jnthn: ?
ah, it's probably right 12:53
jnthn AlexDaniel: context?
AlexDaniel jnthn: github.com/rakudo/rakudo/issues/3237
tyil rba: cloudflare for all perl6/raku DNS names?
jnthn AlexDaniel: Yes, args is the arg capture, $args is extracted from it; probably not the best choice of names. 12:55
AlexDaniel jnthn: I have no idea how this could cause any issues, probably a problem somewhere deeper :( 12:56
rba tyil: Yes, we've moved all to Cloudflare.
AlexDaniel assuming it was bisected correctly, which looks about right
tyil :/ 12:57
AlexDaniel tyil: any better way to do it?
tyil anything that's not cloudflare
AlexDaniel like?
tyil anything that's not cloudflare -> not cloudflare, any other DNS provider would be a saner choice already
if privacy is even a minor concern, of course 12:58
and security as well
CF provides nothing, but costs you all your privacy and brings possible (unknown) security issues
idk why people keep moving their stuff to it
jnthn AlexDaniel: Me either, I suspect it's just uncovered an existing issue. 12:59
AlexDaniel I think the logic in Blin is wrong 13:01
it tests a module 4 extra times on endpoint (normally HEAD) to figure out if it's a flapper
which doesn't help if the flapping is a newly introduced regression
the real test is to see if it's not flapping on the start point, but then the bisect result will be bogus… uhhh… 13:02
13:05 tejr joined
tejr Please do not put Perl sites behind CloudFlare 13:05
The looming threat of GitHub is bad enough
lizmat threat in what way ? 13:06
13:06 pmurias joined
AlexDaniel lizmat: well, github thingie sucks really 13:07
tejr It forces people who don't want to use GitHub to use their perverse workflow
AlexDaniel no, not that… 13:08
tyil tejr: or can't, because github refuses to (or can't) operate in certain countries
AlexDaniel the issue is that some projects in their history moved several times for one reason or another
and there's nothing special to github that will suggest that it'll last
tejr It doesn't make sense for a language with such strong free software credentials to lean on shady proprietary tools like GitHub 13:09
Same principle for CloudFlare
AlexDaniel at some point everyone was using sourceforge, but whoops
tejr Please don't do it
AlexDaniel and it's difficult because it's not just git, the biggest problem is the issue tracking
tyil AlexDaniel: that's also a pretty strong point, even giants fall, and you don't want to sit on it's shoulders when it's going down
lizmat tejr: what are the alternatives
tejr To GitHub, or to CloudFlare? 13:10
lizmat well, you can say that we shouldn't do something, but you haven't said what we *should* do
AlexDaniel lizmat: when it comes to github? Not many… you can use notabug or something similar, but not a lot of people use it so we're potentially losing contributors
same when trying to selfhost it
it's an unfortunate tradeoff :( 13:11
rba I choose Cloudflare mainly becaus of the fast DNS.
lizmat we were selfhosting and then off the air for about a week :-(
rba The Cloudflare CND was just a nice add-on.
tyil rba: "fast" in what way? I could set up a DNS server and have it respond within 20ms, which is pretty fast 13:12
CF CDN is just another means of tracking where users go on the 'net
lizmat tyil: but would that also be 20ms for someone in Japan?
rba They provide a anycast network infrastructure around the globe and allow us to switch backend server quickly if needed.
tyil lizmat: that's what we have caching for
CF is oftentimes slower than not using any MITM 13:13
the "lets use CF because its guaranteed faster" is a very harmful fallacy
lizmat fwiw, I run my own DNS and use BuddyDNS for secondaries 13:14
rba tyil: Which server would you recommend?
tyil there's thousands, I don't have a specific one to recommend, I just don't recommend Cloudflare if you have any care for privacy or security for you or your visitors
tejr The meagre benefits of CloudFlare are imo vastly outweighed by the damage its MitM model does to the web 13:15
tyil cryto.net/~joepie91/blog/2016/07/14...a-problem/ starting at "But The Speed" heading
lizmat if you're really that worried, we should think about supporting DNS over https://
AlexDaniel as for github, I'm ready to pack my bags and move my stuff once there is a proper alternative, but simply another website won't do it. Maybe some mix of matrix and bittorrent can be great, or something like that, but not just another website…
rba lizmat: LOL
tyil lizmat: that's not an improvement whatsoever
lizmat it is for privacy, no ?
tyil DoH is a bad joke played on you by people that don't have your best interest at heart
AlexDaniel well, not exactly… 13:16
tyil it's designed such that programs will ignore your machine's global settings
so you can't do any decent filtering
it's mostly pushed by companies that just want to track your behaviour
rba agrees
lizmat ok, that just goes to show how much I still know about these things :-) 13:17
tyil it's incredibly hard to control DoH (which is a design "feature"), and it doesn't really solve any problems DNS is having
DoH is also piggybacking on the entire HTTP protocol, which is bad for performance over a protocol designed for just DNS information
lizmat well, the problem is e.g. that provider monitor DNS requests
tyil lizmat: DoH doesn't solve that
tejr Where "problems" are "we don't have full control of ther web yet" and "DNSSEC is too hard"
lizmat in the case of our provider, if you do a DNS request for certain domains, they will block your internet access
tyil and if your ISP did DoH, then they'd still be doing just that 13:18
DoH doesn't solve shady servers
tejr It has to be plaintext at some point
lizmat in this providers case, because they think you have a virus
but it could be anything, of course
tyil and in a couple years, Cloudflare will blcok access to certain IPs because they don't play nice with their services
that's a very possible scenario, there's already such scenarios in the world 13:19
the last thing you want, is to have all internet traffic going through a single provider
which is what we're moving to by using cloudflare
lizmat well, I have good experience with using my own primary DNS and have BuddyDNS handle the secondaries
is that an idea, rba? 13:20
if not, why? :-)
rba We have discussed this.
I'm open. Yet the following requirements we need:
Access for multiple accounts. 13:21
An audit log.
kawaii There's nothing wrong with Cloudflare, and it gives us a huge amount of control and tooling IMO. 13:22
tyil do we need those things because we've had issues with dns management before, or is this just pre-emptively?
kawaii: both parts of that statement are incorrect
lizmat tyil: because we had access issue before 13:23
kawaii tyil: I worked at their London office
tyil kawaii: sounds like you may be strongly biased, then
lizmat: that's fair, I didn't know that
rba See github.com/perl6/problem-solving/issues/29
tyil then I understand wanting those features
tyil clicks
rba We didn't had DNS control over perl6.org before. 13:24
It was just a wildcard record to single ip.
lizmat and it took a while to find the person in charge of that, thanks to GDPR
kawaii Cloudflare gives us access control, an incredible WAF, arguable performance improvments and caching and near foolproof reverse proxy to mask the origin IPs of our services. 13:25
tyil cloudflare gives you troubles if you use anything to improve privacy, their CDNs don't actually improve speed, and a MITM is still a MITM 13:26
kawaii Their analytics are quite reasonable too, though I wonder if we should have some kind of OSS analytics like Matomo on the site
tyil don't forget cloudbleed was a thing, they're not foolproof over there either 13:27
kawaii I take zero days with a pinch of salt, every software has them, it's not a sensible metric
tyil but it's even less sensible to praise CF as an infallible entity of good
they're a company, working for profit
kawaii I'm running Linix, it takes anyone 5 minutes to get root access on my box if they have a brain
tyil if you're not paying them gratuitous amounts of money, they'll be getting money from you in a different manner 13:28
kawaii Cloudflare mostly just use the free customers to test beta features on
lizmat tyil: so, we need a NotForProfit instance handling our DNS needs ?
afk for a bit&
tyil lizmat: not saying it *must* be, but those are generally more trustworthy when it comes to privacy
the "free" tier is only free until you actually get DDoSed
kawaii tyil: what specific privacy concerns do you have? 13:29
tyil so there's no free tier if you actually want to use their "features"
kawaii tyil: not true, Cloudflare offer unlimited DDoS protection and no longer drop the reverse proxy under an attack
tyil kawaii: constant tracking across the 'net, and them actively blocking people using a privacy vpn
kawaii That policy change was made in the last year
tyil I'm also sceptical about people praising it's mitm features, just because they call it a reverse proxy 13:30
Geth roast/master: 4 commits pushed by (Vadim Belman)++
tyil shows the massive power of marketing
kawaii RE: "tracking", welcome to 2019, I just made an SSH connection to my server in Japan, I passed through about a dozen if not more routers on the way and I guarentee they all logged that activity. I visited the EFF's website last night too, It's more or less impossible to do anything online without someone knowing. 13:32
VPNs are snake-oil too unless you're running your own.
tyil the defeatist mentality isn't going to help this discussion
it's a sad way to defend harmful practices 13:33
kawaii I'm being realistic, we lost the war against privacy decades ago.
tyil we didn't, that's just you thinking that because it's easier/more convenient
rba Regarding DNS: Again I'm open for new ideas. Please put them into github.com/perl6/problem-solving/issues/29 and if agree on another DNS provider this is fine for me to.
kawaii rba: freedns.afraid.org/ might be a second choice 13:34
tyil you're also conflating "some people knowing" with "lets give all our information to a for-profit entity without getting any returns for it"
AlexDaniel vrurg: can you also rebase it, please? 13:35
rba Our current server which hosts most of the websites is really fast and I don't know if the CDN gives much better performance. (At least for region with good internet coverage.)
tyil rba: CDNs generally have worse performance in areas with worse internet coverage, ironically enough 13:36
so it rarely has a big improvement
AlexDaniel I actually think we're winning
it takes time though
tyil its an improvement for massive sites with thousands of visitors every second, since they can offload traffic to multiple servers easily
it's not that the CDN is inherently faster, it's just that you're using more nodes to serve all connections 13:37
AlexDaniel https on most websites, people switching to messengers that do e2e encryption, etc
tyil AlexDaniel: I dont think people have to switch anymore, even FB messenger and whatsapp support e2e (though they have different issues) 13:38
this idea that "privacy is dead" is incredibly short sighted, and incredibly wrong
kawaii Messenger and WhatsApp are probably involved in some kind of private key escrow agreement with the CIA
vrurg AlexDaniel: sorry, don't get you. Rebase where?
rba For me availability would be more important. And I think about setting up redundant instances of our webhosting in the future.
AlexDaniel vrurg: your problem-solving PR, can't merge it because it has conflicts
tyil kawaii: read the part in between brackets of my message 13:39
the world isn't all black and white, please don't stop reading after deciding on a snappy remark
vrurg Ah, right! That's another reason I wanted to resubmit it yesterday, but was overexhausted... Thanks!
tyil rba: that sounds like a good thing, tbh
I'm currently hosting some stuff in my own kubernetes cluster, that might be interesting to solve that problem 13:40
kawaii rba: container services on a kubernetes platform?
either that or floating IP managed by corosync/heartbeat to multiple backends
AlexDaniel vrurg++ # Raku progress 13:41
rba Yes with containers. How to orchestrate them is not decieded yet.
kawaii Swarm is *okay* but pretty sure it's being deprecated eventually
rba I think I will open an problem-solving ticket on this topic soon, so everyone can add his experience too.
AlexDaniel their, yeah 13:43
rba is afk.
13:50 lucasb joined
pmurias kawaii: having read the CIA interogation manuals I guess it's better that they read my messages without keeping me in their basement till I get bored enough to betray all my secrets 14:11
15:07 pmurias left 15:12 |Tux| left 15:30 dogbert17 joined 15:48 |Tux| joined 15:59 ufobat left
jdv79 nine: how hard you guess a Inline::PHP would be as ridiculous as it sounds 16:31
nine jdv79: if what I saw last time I looked is still true, hard. PHP doesn't seem to have a real embedding interface unlike Perl or Python. 16:36
OTOH there's metacpan.org/pod/PHP::Interpreter 16:41
vrurg nine: have a few minutes? I was terrorizing timo yesterday, but he doesn't know how to help. 17:08
nine vrurg: with what? 17:17
vrurg Still the same – chasing lost closure context. There is interesting situation with R#2897: it fails because it can't find &infix:<++>. When I dumped outers of the sub I found that it's closure preserves contexts up to the unit level, but not CORE.setting. 17:20
nine Ah, yeah, I've read that
vrurg I thought that if closure is lost – it'd be lost completely. 17:21
nine So, do you know how CORE.setting gets into the contexts of a sub where it's working?
vrurg The anonymous sub must retain the CORE.setting of Bar.pm6, where it's created. Any other variant is plain wrong. 17:23
nine I concur.
vrurg Basically, the scenario is more-less clear to me: sub-producer returns a clone of sub. The clone is wrapped with p6capturelex. At this point, as I was tracing things down, CORE.setting is an outer of the sub. But when it gets through BEGIN in intermidiate module – the CORE.setting is getting lost. 17:25
Either some code implcitly stops serializing at unit level, or it stops on the first empty frame between !UNIT_MARKER and CORE. That's my wild guess. 17:26
nine Sounds like you almost have it :) 17:27
vrurg But that's exactly where I'm stuck. I have no idea how BEGIN works. 17:29
But at least if you can point me at where exactly the contexts are added to SC, it would spare me a couple of hours of research. 17:30
nine vrurg: it's pretty much all about finish_code_object and compile_in_context 17:31
vrurg Anything in moar? Can p6captureouters be related? 17:32
nine p6captureouters is actually not part of more but an extop installed by rakudo
But it's certainly worth looking at it to see if it does any stopping at borders 17:33
vrurg I know it's not from moar. The point is that it refers to moar's API and I didn't have time to look deeper. 17:35
Ok, anyway, got a little more clues. Thank, nine!
18:13 MasterDuke joined 18:33 cognominal joined
cognominal This name leads my pet subject. A name should just be a convention for an entity to which was given a unique id at creation. With such convention changing name is a no brainer in flat spaces. In hierarchical space like programming languge, raku is well placed to deal with homonymous names because it has conventions to refer to any name accessible in the hierarchy. 18:40
* name change 18:41
18:44 reportable6 left 18:49 reportable6 joined
vrurg BTW, notbody yet bothered to give us a pronouncation of `Raku`. I pronounce it more-less the way google translate does it. 19:10
AlexDaniel vrurg: “Ráku or Rakú”? Last time I asked sena_kun told me it's neither 19:12
vrurg: I guess just do it the way google translate does 19:17
that's about right, right?
vrurg There was also variants like 'ray-kuh' because the first sillable is open. Anyway, google pronounce English variant as `ráku`, Japaneese as 'rákú'
I.e. Japaneese sounds like to split sillables, both on their own stressed out. 19:18
*two split 19:20
19:31 cognomin_ joined 19:33 cognominal left
AlexDaniel lizmat: github.com/perl6/problem-solving/i...-542756891 19:51
Geth Blin: debe6760fd | (Aleks-Daniel Jakimenko-Aleksejev)++ | lib/Blin/Processing.pm6
Don't ignore flapping regressions

It used to test the end point to see if the module is flapping, which is definitely wrong if the flappiness is a new bug in the compiler. Now it will detect flapping regressions, but it won't bisect them correctly yet (which is arguably better than just ignoring them).
19:57 ufobat joined
discord6 <Aearnus> hey, I found something LTA but I couldn't find time to investigate it 19:58
<Aearnus> if you use a module after using Test, and Raku can't find the module, it exits with nothing but the message "Failed" 19:59
<Aearnus> It makes sense as to why (I guess a module not being included would make all the tests fail), but it doesn't seem like a sane default to me at all -- at least there should be some info as to what failed 20:00
nine m: use Test; use FooBar; 20:10
camelia ===SORRY!===
Could not find FooBar at line 1 in:
nine I think the "Failed" has a different reason 20:11
discord6 <Aearnus> huh. Perhaps it was if the module I imported happened to throw an exception 20:18
<Aearnus> I think that was it
<Aearnus> m: module Foo { fail }; use Test; use Foo; 20:19
evalable6 (exit code 1) ===SORRY!===
Could not find Foo at line 1 in:
Aearnus, Full output: gist.github.com/bd018711ff4c0d3dd3...0158c2d623
discord6 <Aearnus> m: module Foo { fail }; use Test; need Foo; 20:22
evalable6 (exit code 1) ===SORRY!===
Could not find Foo at line 1 in:
Aearnus, Full output: gist.github.com/cf45a396dc07ccfd13...6d5a196da8
discord6 <Aearnus> hmm. I'll put together a full proof of concept later
20:37 cognomin_ left 20:54 ggoebel left 21:23 ggoebel joined 21:32 ggoebel left 21:38 patrickb joined
patrickb . 21:43
tellable6 2019-10-16T07:06:12Z #perl6-dev <nine> patrickb since you're already at PERL6_HOME and CompUnit::Repository: what's your plan to bring the Inline::Perl5 using spectests back?
patrickb .tell nine I didn't know there was a problem. Is there a bug report or something I can read up on? 21:45
tellable6 patrickb, I'll pass your message to nine
22:34 patrickb left 23:06 MasterDuke left