🦋 Welcome to the MAIN() IRC channel of the Raku Programming Language (raku.org). This channel is logged for the purpose of keeping a history about its development | evalbot usage: 'm: say 3;' or /msg camelia m: ... | Log inspection is getting closer to beta. If you're a beginner, you can also check out the #raku-beginner channel!
Set by lizmat on 25 August 2021.
[Coke] There isn't a named class for \v (see the table on the same page that has, e.g. <alnum>) 03:00
[Coke] any prior art for a module that has a proto MAIN and sets up some data to be available (somehow, don't care how) in script's MAIN? 04:03
Voldenet spacekookie: you're not far from the solution, you're only lacking `method keyword($/) { make ~$/ };` and `method payload($/) { make ~$/ }` in Actions 04:12
and $line<statement> is an array 04:13
this would work ix.io/3JIO
(see that kvp example has `method identifier($/) { make ~$/ }`) 04:15
guifa hmmm 06:39
finishing up Timezones::ZoneInfo, and debating method names 06:42
I've got posix-from-calendar and calendar-from-posix; but thinking about having aliases of `calendar-to-posix` and `posix-to-calendar` 06:44
I had `get-timezone-data($timezone-id)`, but now I'm thinking I should eschew the `get-` prefix there.
lizmat clickbaits rakudoweekly.blog/2021/12/27/2021-...-released/ 11:04
spacekookie Voldenet: Oh interesting! What do the >> do? I saw that in the documentation but wasn't really sure what it meant (especially because it was turned into a ligature by firefox or something) 11:17
moon-child fwiw ligature version (», not really a ligature but) works just the same way 11:18
but anyway see docs.raku.org/language/operators#i..._operators
spacekookie Ah thank you! 11:33
Well this is exciting. I _think_ I understand this now but we'll see how things go once I add more complexity to the grammar 11:38
thanks to everyone who's been helping me <3
Nemokosch this community is really awesome 12:13
El_Che yes, I am... I mean, yes we are :) 12:16
moon-child speak for yourself! 12:17
Anton Antonov 🙂 🙂 13:21
guifa what's the license name to use if we're releasing code into the public domain? 14:36
moon-child cc0, maybe 14:38
guifa I can only find CC-PDDC in the spdx database which I guess works. It just feels weird eneeding to say anything beyond "it's public domain" 14:43
guifa is releasing this module in public domain because the code it was based on was also in the public domain and it feels right to pass it along 14:44
moon-child some jurisdictions don't recognise public domain; cc0 tries to make allowance for that
in principle, if you say 'this is in the public domain', that should indicate to people that they can use your code for w/e they want and you won't sue them 14:45
but some companies refuse to use such code because ... lawyers
guifa github.com/alabamenhu/TimezonesZoneInfo yay fully working. now to teach the old DateTime::Timezones to use it 15:10
lizmat guifa++ 15:17
lizmat El_Che: does Cro::HTTP::Client come with the rakudo package ? 16:26
Od1n hello raku team 16:53
guifa good <morning afternoon evening>.pick ! 17:00
Od1n .[2] with Od1n 17:14
El_Che lizmat: normanlly no, except if zef pulled it in as a dependency 19:06
El_Che is there interest in linux arm64 rakudo packages? It would take considerable work because github actions does not support arm64 (besides a remote selfhosted runner). And there is not JIT for arm64 19:11
japhb I'm more interested in the arm64 JIT support ... but I'm biased. :-) 20:17
[Coke] IANAL, but having worked with them on OSS issues for $DAYJOB, I find that the MIT license is pretty minimal (from my standpoint) and pretty unobjectionable (from theirs). 20:18
jast here's my personal (and not lawyerized) overview for some of the more popular licenses: jan-krueger.net/doc/opensource-licenses.html 20:23
in short, MIT is one of the more minimal licenses, with several others having a similar number of (but slightly different) restrictions 20:26
lizmat El_Che: that explains why I needed to wait so long when loading Cro::HTTP::Client for the first time after the update :-) 20:44
El_Che: I think the advantage of arm64 packages is that one wouldn't need to compile Rakudo on those architectures 20:45
and compiling Rakudo on those *is* a barrier when you have less than let's say 1.2G of RAM available
let alone how long it takes :-) 20:46
japhb It is indeed slow as heck. I'm curious what target audience you're thinking of, though. RPi 3 users? 20:47
(It compiles just fine -- if slowly -- on a 4 GB RPi 4. I don't think you can buy an RPi 4 smaller than 2 GB these days, but maybe it was possible in the past.) 20:48
Wikipedia says there was a 1 GB RPi 4 at one point 20:49
Skarsnik my chromebook is an arm64 platform, could be nice to have a raku on it x) 21:18
El_Che japhb: cloud providers are starting to offer arm64 servers and VMs (aws, oracle) + roi users 21:44
rpi
4 21:45
japhb El_Che: Ah, and the cost (and time) of compiles before actively running a node could get prohibitive. 21:48
El_Che how l9ng does it take on a rpi4? 21:52
japhb El_Che: I don't have a time for just the bare MoarVM/NQP/Rakudo build -- my build script includes a fair number of big modules that I always want installed. 21:54
That said, we're talking hours for the full thing, but way less than a day.
El_Che wow 21:55
lizmat El_Che: yeah, so on those grounds already would a binary build be nice :-) 21:55
El_Che and I would run 20+ in parallel
El_Che it would tequire a different approach than amd64 21:56
El_Che does adding cpus make the buikd faster? 21:58
japhb El_Che: I suspect the answer to that is "yes", but I normally don't widen my builds because on x64 the zef installs are the overall time limiter, and on arm64 I have heat disappation issues 22:03
El_Che last wurstion: do rpi usets mainly run raspian or just regular distros? eg on oracle cloud you'd run arm64 ubuntu or oracle linux 22:04
japhb Raspbian (32-bit) and Raspberry Pi OS (64-bit) are both desktop distros. 22:05
They're not very heavy, obviously, but they're certainly not intended for servers.
(Hmmm, unless Raspberry Pi OS has a server install option now, I haven't looked.)
El_Che thx!
japhb np 22:06
japhb El_Che: BTW, as I was setting up for a 2021.12 build on the RPi 4, I remembered -- RPi 4 build performance is *STRONGLY* related to how fast the SD card is. Early on I had to upgrade to a much faster SD card because the default cards shipped with most RPi kits is S-L-O-W. 22:18
japhb I would *hope* that the data center RPi's are set up to do something like network mount their RW partition, but if you're just testing it locally, be aware. 22:19
:-)
El_Che the idea would be running containers on arm64 servers 22:28
or maybe a ci service
to be investigated, because reaching timeouts is likely 22:29
japhb Oh god yes 22:30
japhb El_Che: Looks like about 35-40 minutes to build up through zef itself. 23:03
lizmat yeah, we really need to look at that, and cut out at least one occurence of precompilation 23:20
japhb zef authors: trying to figure out how to deal with intermittent network connectivity when building a huge mess of Raku modules -- at some point the network flakes out just as zef is trying to pull a tarball, and it blows up. 23:39
Is there a way to separate the phases and download the tarballs separately from trying to extract/test/install them?
And in particular, to be able to tell (via e.g. exit code) whether the tarball was retrieved successfully?
Because then I can wrap that download in a retry loop ... or batch download when I've got good connectivity, and build later when I'm offline. 23:40
lizmat well, there's the --serial option, that tests / installs each module at a time ? 23:47