guifa meh, just going with .foo with %hash<foo> //= Foo.new; I think it's as concise as I'll get 01:00
01:34 lizmat_ left 01:35 lizmat joined 02:04 hulk joined 02:05 kylese left
timo could make an operator like postfix:<++> that just doesn't increment or anything 02:12
guifa timo: true, but I think my final one is at least recognizable if not quite a pattern 02:15
also, I just realized that this mDNS / Zerconf module / project would make a badass talk on concurrency 02:16
I have supplies leading to supplies leading to supplies
and yet the design is crazy clean
udp port gives a supply, packets are processed by the mDNS server and records are put into a set of supplies based on types. the zeroconf module is subscribed to those and makes additional queries as needed to scope out the landscape and, based on what it finds, emits records onto yet another supply 02:18
and the end user just does my $zc = Zeroconf.new; $zc.service-notification('_printer._tcp.local').tap: -> { ... } 02:19
03:15 hulk left, kylese joined
ab5tract guifa: snazzy! 03:21
guifa I mean, I'm sure if I made it a monolith in C it'd be about a gazillion times more efficient, but so far, it seems more than fast enough for mDNS 03:57
03:59 Sgeo left 04:03 Sgeo joined 04:35 Aedil joined
aruniecrisps @ab5stract tried doing that, doesn't seem to be doing anything it looks like 05:05
ab5tract Boo :( 05:33
Next best thing might be to nuke the whole dependency 05:34
07:22 Xliff joined
Xliff Can someone explain to me the arguments for the following call? - go_path_pie_wedge (GOPath *path, double cx, double cy, double rx, double ry, double th0, double th1) 07:23
I'm fine with the first 2, but the rest of them are confusing.
The first 3 rather. I'm trying to write bindings for this in raku.
I think I lost a brain cell or two within the last 5000 lines of Gnumeric/GOffice 07:24
08:25 Aedil left 08:49 abraxxa-home joined 08:52 abraxxa-home left 08:55 abraxxa-home joined 09:14 Sgeo left, Sgeo_ joined 09:24 melezhik joined
melezhik aruniecrisps: on my memory zef :from<native> thing always been an issue , I still believe there should be a simple way to declare native dependencies , and this should not be in META/zef, in many other language any library depending on native c/c++ lib just through an exception in any form staying it needs the one, and then `apt-get/yum and friends 09:27
install native lib` just do the job ...
@aruniecrisps on my memory zef :from<native> thing always been an issue , I still believe there should be a simple way to declare native dependencies , and this should not be in META/zef, in many other language any library depending on native c/c++ lib just through an exception in any form staying it needs the one, and then `apt-get/yum and friends
install native lib` just do the job ...
09:32 melezhik left 09:50 Xliff left 10:20 sena_kun joined 10:30 itaipu left 10:55 Aedil joined 11:02 Sgeo_ left 11:32 Aedil left
Geth ecosystem/main: 18a85cebd8 | (Elizabeth Mattijsen)++ | META.list
Remove BigRoot

It lives on as a Raku Community module		 11:43
11:52 melezhik joined
melezhik I've put some my ideas on how Dormitory could evolve as wiki pages - github.com/melezhik/Dormitory/wiki - comments are welcome 11:52
El_Che melezhik: what do you mean by Dockerfree? Are you running on bare metal? 11:55
melezhik I mean docker or any container are not used at all 11:56
just build from source code and run thing as is on host
and yeah this could be bare metal 11:57
El_Che so, you stick to straight-forward applications with no os dependencies?
11:59 xinming joined
El_Che the no-docker thing is the simplicity for the dev? 11:59
12:01 melezhik left
Geth ecosystem/main: fdae6d45c7 | (Elizabeth Mattijsen)++ | META.list
Remove Statistics::LinearRegression

I lives on as a Raku Community module		 12:14
ecosystem/main: 6ddc3424d8 | (Elizabeth Mattijsen)++ | META.list
Remove UNIX::Daemonize

It was a WIP and its tests have been failing for many years already		 12:15
12:26 xinming left 12:29 melezhik joined
melezhik El_Che yes, more or less. I think people sometimes overuse docker, where simple `zef install . && cro run` is enough (as an example in Raku world, but the same idea for other stacks).  I mean for production it's valid to use containers/k8s but for dev it could be overkill 12:32
12:35 xinming joined 12:37 melezhik left
El_Che just to be clear, is it a framework for Raku or for all languages? And of so, how do you handling stuff like compiling en dependencies 13:09
it's hard to tackle
Geth ecosystem/main: d47270d150 | (Elizabeth Mattijsen)++ | META.list
Remove overload::constant

It lives on as a Raku Community module		 13:13
aruniecrisps @melezhik what do you think is the solution to this problem? 13:26
13:32 xinming left 13:34 xinming joined 13:45 melezhik joined
melezhik @El_Che - this is for any language framework. Where people need microservises architecture 13:47
I don’t see why installing dependencies could be an issue
Most of the dependencies are handled by language native managers like pip/python go mod tidy /go , zef/Raku etc 13:49
If we talk about system things like databases or any system packages. In case of dormitory all the heavy duty stuff is carried out by Sparky which shines for such a job 13:50
See for example some prebuilt plugins - sparrowhub.io/search?q=sparkyci 13:51
aruniecrisps i generally agree with not having native dependencies as part of META6.json 13:52
melezhik So people don’t rewrite anything and just use it
aruniecrisps, yes that the idea
13:53 melezhik left
aruniecrisps I still don't know why --exclude="pq" doesn't seem to be doing anything 13:54
ugexe why don't you post the full output with --debug because --exclude="pg" specifically has worked for other people and generally works
aruniecrisps $ zef install --debug --exclude="pq" DB::Pg ===> Searching for: DB::Pg ===> Found: DB::Pg:ver<1.1>:auth<cpan:CTILMES>:api<1> [via Zef::Repository::Ecosystems<rea>] ===> Dependencies: NativeCall, NativeLibs, LibUUID, JSON::Fast, pq, Test, Test::When ===> Searching for missing dependencies: LibUUID, pq:ver<5>:from<native> ===> Failed to find dependencies: pq:ver<5>:from<native> Failed to resolve some missing 13:57
dependencies (use e.g. --exclude="pq" to skip)
ugexe fair enough 13:59
yeah exclude indeed seems to have stopped working with native dependencies 14:05
i guess you have to do `--exclude="pq:ver<5>:from<native>"` 14:17
although --exclude="pq" should just work
however i also observe that just --exclude="pq" does not just work
aruniecrisps i do wonder if i basically lose the actual postgres connectivity if i disable it 14:23
ugexe --exclude doesn't disable anything, it just tells zef to ignore looking for that dependency. since it is a native dependency it can't install it anyway 14:29
14:42 xinming left
antononcube How come? > Sadly it has turned out to be impossible to organize a Raku Conference (neither in-person or online) […] 14:42
14:46 xinming joined
lizmat because Shitov didn't want to do it, and nobody else wanted to take it on? 14:51
15:12 Aedil joined
antononcube I see. 15:22
15:28 melezhik joined 15:30 melezhik left 15:35 melezhik joined, melezhik left 15:36 melezhik joined
melezhik . 15:36
15:40 melezhik left 15:43 melezhik joined 15:46 melezhik left 15:57 melezhik joined 15:58 melezhik left 16:27 melezhik joined 16:28 melezhik left
antononcube BTW, there are plenty of unapproved comments of Raku Advent 2024 posts. So, advent authors -- please approve (or reject) the ones of your posts. 16:40
17:04 melezhik joined 17:06 melezhik left
guifa whoa 17:18
just realized
if my $foo = [call] { ... } elsif $foo == [other value] { ... } works 17:19
I don't know why I was only anticipating the $foo to be valid in the first block 17:20
ab5tract guifa: indeed, that's one of the early "hook" features that pulled me into the language. being able to do inline assignments in a sequence of expressions is totally awesome 17:33
if you *don't* want $foo available outside of the first block, you add it as an argument in a pointy block (if [call] -> $foo { ... }) 17:35
guifa ab5tract did you manage to fix the indirect call syntax for Comma? 18:23
because it looks like it's (kind of) working and you don't know the level of joy that this brings me
color is showing up white (not sure if the intent is for it to match sub calls or method calls, or just be different to highlight it's an indirect) but melikes 18:25
19:04 Aedil left
ab5tract hmmm, not sure what's going on there, as I haven't touched the parser guts yet 19:42
19:56 rypervenche joined
antononcube @timo The "BOOTException" I a mentioned two days ago is happening when installing "Jupyter::Chatbook" using an INIT phaser. In that phaser a JSON file is ingested that has new lines or to content of that JSON file is expanded in the phaser to have new lines. Without the new lines no exception is given. I could not make a simpler, smaller code that manifests that failure or exception. 20:13
Turned using a class method and a TWEAK is a better solution. And I do not get the exception when expanding the content. 20:14
guifa ab5tract : ah, well, still... whatever changed, I'm super happy :) 20:25
ab5tract merry christmas! ;) 20:26
guifa $*SCHEDULER.cue: { push: ab5tract, @beers.pick(3) }, :at<next-conference> 20:27
so .raku methods... 20:28
for more ephemeral things like handlers, what are people's thoughts? just put a .new and assume it's bad practice to try to .raku them ? 20:29
timo for that case it's fine to give something that is merely informative, i think 20:57
there is no guarantee that eval-ing the result of a .raku gives a perfect result for everything 20:58
antononcube, I was hoping i could use the error that you are encountering to test if the error message improvement works 21:04
guifa timo: alright, noted. I'm putting a bit too much polish on this module as I go ha 21:08
timo i'm assuming your changes to the json string are causing it to become invalid, the parse error exception is landing in a variable somewhere and getting stored and then serialization reaches it and bails out 21:09
guifa making an IPv6 subset is ... ugly lol 21:10
timo indeed. so many different ways to specify stuff 21:11
guifa I think i'll save that until later 21:15
timo yeah it's probably fine to have the function inside moar that converts the ipv6 string to the correct struct return the error, then throw that as an exception 21:17
we have ways to throw typed exceptions directly from inside moarvm, but it's a little bit fiddly, so it might be good enough to substring match on the X::AdHoc thrown by whatever op does this
if you want to make it structured, that is. which we should probably do at least at some point, but if it stands in the way of implementing the feature, it doesn't have to be immediately, and it doesn't have to be you who does it 21:18
guifa oh this is at a much higher level 21:20
timo if you want to do it really weird, you can expose a syscall that only tries to do the conversion and nothing else, and returns if it was successful or not
then the subset can just be `where { nqp::syscall("is_this_ipv6", $_) }`
guifa so when you advertise a ZeroConf service, you need to have a service name (e.g. 'My HTTP Server._http._tcp.local'), either an IPv4 or IPv6 address and a hostname, and optionally some config text 21:23
I'm trying to be good and valid everything at the higher level with good error messages. by the time it gets to MoarVM level, this will already be tucked into a DNS packet 21:24
timo ooh 21:25
this isn't about UDP broadcast support any more, got it
we don't have a grammar for ipv6 addresses in the ecosystem yet?
guifa maybe 21:26
timo there is an IP::Addr dist 21:27
raku.land/zef:vrurg/IP::Addr 21:28
guifa timo: I haven't been able to figure out broadcasting / receiving on my IPv6 interface. When the IPv6 address goes to MoarVM resolve function, it always comes back the IPv4 equivalent. When I try to force the binding anyways since I know it's a valid hostname, I end up with an address unavailable error
timo say IP::Addr.new( "2001:db8::/123" ).broadcast; 21:29
will fail with "no method" exception because broadcast is not available for IPv6 addresses.
^- i wonder if this has something to do with that?
> Key Concept: IPv6 has unicast and multicast addresses like IPv4. There is, however, no distinct concept of a broadcast address in IPv6. A new type of address, the anycast address, has been added to allow a message to be sent to any one member of a group of devices. Implications of the Changes to Address Types in IPv6 21:30
guifa I'm sending to a multicast and I edited code so that libuv should be adjusting accordingly 21:31
but I'll need to try with some other languages/libraries to see if it's a limitation in libuv or my setup, etc
github.com/vrurg/raku-IP-Addr/blob...kumod#L187 looks like my current approach isn't really any different, except I'm using split then regex heh 21:33
timo A socket can (un)subscribe to a multicast group using: 21:37
where membership is UV_JOIN_GROUP or UV_LEAVE_GROUP.
^- it feels like something went missing here lol
can you send to a multicast without "joining" the correct multicast group first? is that perhaps what's missing? 21:41
i've not written code yet that attempts to multicast or anycast on ipv6 21:42
22:06 Sgeo joined
guifa Yeah, the UV_JOIN_GROUP is what I'm using and it works okay with IPv4 22:13
r = uv_udp_bind(udp_handle, ssi->bind_addr, (ssi->flags & 0b10) ? UV_UDP_REUSEADDR : 0); 22:14
that's actually the line that's causing the error, iirc 22:15
The only other thing I can think of is that the casting from sockaddr_in6 to sockaddr isn't working entirely as intended 22:20
timo did you already try strace or perf trace to see what the syscalls that it actually issues look like? at least strace does some helpful decoding of the arguments to be more human-readable, too 22:21
guifa I haven't yet -- I admit, those are tools I don't really have much experience using 22:23
timo strace is one of the first tools i reach for in situations like this 22:28
22:37 sena_kun left
guifa okay, at least with printf'ing ha, I was able to rtack it down to uv_udp_set_membership 22:41
does okay with a stringy representation of IPv4, but doesn't like IPv6
let me try it fully qualified to see if that changes 22:42
timo can you try stracing the startup of like avahi-daemon if you set it to ipv6-only mode, if that is even a thing it has? 22:46
guifa ugh 22:51
dtrace: system integrity protection is on, some features will not be available
timo oh, macos? 22:53
guifa ya 22:55
I'll see about booting up a VM and trying this in there -- it's a decent set up since I have to recompile with my MoarVM edits and Rakudo edits 22:56
timo sounds like a job for a Containerfile ;) 22:57
guifa ha I'm trying to limit the number of "okay time to learn a new tool" moments on this project :-) 23:02
timo :D 23:03
guifa right now it works just fine with IPv4 and realistically that's probably enough for most people
timo what, you don't measure projects by "number of new tools learned"? ;)
guifa for now, at least
timo right, ipv6 will never actually become a thing
or maybe in ... 20 years :)
guifa I'm sure it will .... eventually lol.
but for home WLANs, it's really probably overkill 23:04
also apparently it's just fine to point the multicast to the same socket from what I was reading
(from IPv4 and IPv6)
oh nm 23:12
I got it!
thank you NodeJS who also uses libuv
stackoverflow.com/questions/384483...th-node-js
timo .o( and you'll get the messages back, too, by default? )
guifa Whereas the IPv4 doesn't need an interface address (and so their examples on libuv's page uses NULL), IPv6 does 23:13
I just passed in "::%en0" (my wifi interface) and voila!
timo ah, makes sense! 23:16
i mean, i'm not entirely sure why it's necessary / forced
and of course the error code and message you got back were ... less than helpful
guifa The catch is there's no way to predict that address (technically called a "zone identifier" i guess) 23:58

Powered by Cro and the Raku® Programming Language.

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

Thank you!