🦋 Welcome to the MAIN() IRC channel of the Raku Programming Language (raku.org). Log available at irclogs.raku.org/raku/live.html . If you're a beginner, you can also check out the #raku-beginner channel!
Set by lizmat on 6 September 2022.
00:00 reportable6 left 00:03 reportable6 joined 00:35 kjp joined 00:43 Manifest0 left 01:04 xinming left 01:06 xinming joined
rf How's everyone doing today? 01:21
tea3po I am well, ty for asking. You?
rf Good, just had an excellent dinner and some ice-cream :^) 01:22
Anton Antonov @rf So, no more cinnamon rolls ?! 01:33
shmup @ 01:34
rf @Anton, no I ate them all D:
01:37 frost joined 01:40 frost left
Anton Antonov @rf Of course you cook some new ones, no ? Also, do you know how to make ice-cream? 01:42
rf I know how to make ice-cream but it is too much work, I am a lazy programmer after all 01:43
I think I am going to make cinnamon rolls a yearly thing, too many calories haha 01:44
Anton Antonov @rf I see. At this point I completely delegate cooking. When was doing my PhD, I considered spreading butter on bread cooking... 01:47
rf Lol, what field did you study? I sure hope it wasn't culinary arts ;D 01:48
01:51 frost joined
Anton Antonov @rf Good point. My PhD is/was in Large Scale Air-Pollution Simulations. 01:51
rf Nice that is a very interesting topic to cover 01:53
Anton Antonov @rf Complex systems can be very "entertaining" : github.com/antononcube/SystemModeling . 01:59
02:00 frost left
rf Mathematica is interesting haha, I have never used it for anything useful though 02:01
Anton Antonov Until 2014 I used Mathematica for type of scientific and mathematical modeling and/or data analysis. The corresponding "productized" versions in C++. Then I started doing the data analysis with R. I still do mathematical optimization and modeling with Mathematica; the corresponding productized versions in R or Python. 02:06
rf Cool, maybe productized versions in Raku one day 02:07
Anton Antonov @rf I doubt it. 🙂 02:09
rf Nemokosch, Template6 is kind of borked too with for loops not having the variable, for example: [% for foo in something %] [% foo.value %] [% end %] If something isn't set it simply prints foo.value 02:11
Anton ;(
Anton Antonov @rf I am writing a blog post about it. 02:23
rf We can always dream
Anton Antonov @rf In color or black-&-white ? 02:25
rf I think Raku is a fairly colorful language, so let's do color 02:26
Nemokosch hm, shouldn't that throw one level up, if something is not set in the first place?
rf Not sure, but I have to do: [% if something %] [% for foo in something %] [% foo.value %] [% end %] [% end %] 02:28
to get around it
Nemokosch tbh I don't quite get why you would want to loop over something that doesn't even exist 02:30
rf Well sometimes it will exist sometimes it wont I don't want to write a new template for both cirumstances lol 02:37
jdv is Template6 worth it? i havent looked. but doesnt cro have some template thing or raku's interpolation is pretty awesome in the first place. 02:39
tellable6 2023-02-27T15:13:10Z #raku-dev <AlexDaniel> jdv that you can fix yourself :) github.com/Raku/whateverable/blob/...le.p6#L268
2023-02-27T15:21:39Z #raku-dev <AlexDaniel> jdv I patched releasable with the fix right on the server, but please correct it in the repo as well
jdv .tell AlexDaniel ok 02:40
tellable6 jdv, I'll pass your message to AlexDaniel
rf jdv, I think it's OK, I'm using it for a fairly large project but there are some rough edges 02:43
Nemokosch idk it seems a bit weird to not know what exists on top level lol 02:44
anyway
m: for Nil -> $big-brain { say 'didgeridoo' }
Raku eval didgeridoo
Nemokosch KHM KHM
the usual "strict workaround" won't do because even Nil can convert into a perfectly valid one-element array... 02:45
m: for Empty -> $big-brain { say 'didgeridoo' } 02:46
Raku eval
Nemokosch thank heavens 02:47
I wonder if the "strict lookup" could just always return Empty 02:48
rf That would work like a charm I think 02:51
Gotta head off the for the night, cya folks 02:58
02:58 rf left 03:08 razetime joined 04:08 evalable6 left, linkable6 left, notable6 left, quotable6 left, statisfiable6 left, reportable6 left, nativecallable6 left, benchable6 left, squashable6 left, coverable6 left, shareable6 left, bisectable6 left, sourceable6 left, committable6 left, bloatable6 left, releasable6 left, greppable6 left, unicodable6 left, tellable6 left, benchable6 joined, greppable6 joined, squashable6 joined, quotable6 joined, linkable6 joined, committable6 joined 04:09 shareable6 joined, coverable6 joined, notable6 joined, unicodable6 joined, tellable6 joined, bloatable6 joined, sourceable6 joined 04:10 releasable6 joined, bisectable6 joined, reportable6 joined, nativecallable6 joined 04:11 evalable6 joined, statisfiable6 joined 05:11 statisfiable6 left, tellable6 left, notable6 left, evalable6 left, linkable6 left, bloatable6 left, squashable6 left, unicodable6 left, reportable6 left, bisectable6 left, releasable6 left, quotable6 left, benchable6 left, notable6 joined 05:12 tellable6 joined, linkable6 joined, reportable6 joined, bloatable6 joined, statisfiable6 joined 05:13 releasable6 joined, bisectable6 joined, benchable6 joined 05:14 quotable6 joined, evalable6 joined, unicodable6 joined, squashable6 joined 05:18 razetime left 05:31 razetime joined 06:00 reportable6 left 06:03 reportable6 joined 06:16 razetime left 06:47 teatwo joined 06:50 tea3po left 07:02 teatwo left 07:03 teatwo joined 07:06 xinming left 07:31 razetime joined 08:13 Sgeo left 08:27 ProperN[out] left 08:28 ProperNoun joined 09:09 Manifest0 joined 09:12 dakkar joined 09:14 razetime left
Nemokosch rf: I did the patch 😛 09:24
09:35 nort left 10:35 unicodable6 left, evalable6 left, committable6 left, linkable6 left, sourceable6 left, shareable6 left, coverable6 left, statisfiable6 left, quotable6 left, tellable6 left, squashable6 left, bisectable6 left, benchable6 left, releasable6 left, notable6 left 10:36 unicodable6 joined, tellable6 joined, releasable6 joined, notable6 joined, quotable6 joined, shareable6 joined 10:37 sourceable6 joined, bisectable6 joined, linkable6 joined, squashable6 joined, committable6 joined, benchable6 joined, coverable6 joined 10:38 evalable6 joined, statisfiable6 joined 11:03 Johanna100 joined
Johanna100 I was trying to find a scripting language appropriate for my projects. I considered Guile but I find Scheme difficult to read and Guile doesn't build on MacOS without a hack (for some bizarre reason the build script depends on non-standard sed; the bug has been discussed in their mailing list but nobody thought MacOS was important enough to fix it, 11:11
and the barrier to entry for me to fix it myself is too high). Raku builds and tests without any problem whatsoever. Keep kicking butt guys
11:20 thundergnat joined, frost joined
lizmat Johanna100: we try :-) and the girls also :-) 11:21
11:21 frost left
thundergnat Glad it worked out for you. Raku still has a few rough edges here and there and the performance isn't where we would like it to be yet, but it is pretty usable for a large cross section of tasks. 11:23
11:25 thundergnat left 11:26 frost joined 11:28 frost left
Nemokosch seems like Guile is GNU software so maybe it's not surprising it comes with "non-standard sed" (I suppose gsed) 11:29
11:54 grondilu joined 12:00 reportable6 left
Geth Raku-Steering-Council/main: 79cdc3a435 | (Elizabeth Mattijsen)++ | minutes/20230218.md
Add minutes of 18 Feb 2023 meeting
 12:01
lizmat seems this one fell through the cracks :-(
12:03 reportable6 joined
Nemokosch will another weekly hit the net today? 12:05
lizmat it's about to drop :-) 12:07
final proofreading as we speak :-) 12:10
Nemokosch 🥁 🥳 12:12
lizmat and yet another Rakudo Weekly News hits the Net: rakudoweekly.blog/2023/03/07/2023-10-toronto/ 12:17
12:25 exp joined
exp I'm trying to write a small script using explicitly sized integers, ie `my uint8 $a;`, all operations on $a return `Cannot invoke object of type 'NQPMu'`. Am I doing something idiotic or did this previously work? 12:27
2023.02 apparently, freshly built 12:28
Nemokosch you could give a snippet to bisectable6 12:30
lizmat exp: could you gist an example? 12:31
exp lizmat: literally `my uint8 $a; say $a` 12:33
ooh actually, no
lizmat % raku -e 'my uint8 $a; say $a'
0
exp on one line it works
on two lines it fails
(the linebreak coming after the first ;)
apologies for not noticing that
lizmat still cannot reproduce. can you make a gist ? 12:34
exp is this sufficient? pastebin.mozilla.org/SXud5UhN 12:35
lizmat ah, you're using the REPL
exp yes i failed to mention that sorry
Nemokosch then maybe it can be reproduced by some EVAL? 12:36
lizmat the REPL has some known issues... I guess this is one more :-(
exp wow ok, that's a real shame 12:37
lizmat yeah... the problem is that each line in the REPL is a separate EVAL
with their own scope, and there are some issues specifically related to natives that don't pass on from one EVAL to the next 12:38
exp ah well, thank you for the fast answer lizmat 12:39
lizmat you could try running with the RakuAST grammar by prefixing: RAKUDO_RAKUAST=1
so: $ RAKUDO_RAKUAST=1 raku 12:40
at least the errors are a bit more understandable
exp yeah, at little improvement at least, i just implemented it in C so no big deal, just very surprising
lizmat I guess once we have RakuAST at a stage where the setting is also compiled with it, we will have an opportunity to fix the REPL in that respect 12:41
so no solution before the next release of Rakudo, I'm afraid 12:42
12:49 derpydoo left
Geth doc/main: 1d11030921 | (Elizabeth Mattijsen)++ | doc/Type/DateTime.rakudoc
Document DateTime.posix(:real)
 13:01
13:02 derpydoo joined
Nemokosch is it possible to turn a string into a character group for regex? 13:06
if you know what I mean 👉 👈
Geth doc/main: b3e0c1d99f | (Elizabeth Mattijsen)++ | 2 files
Document Exception|Cool.Failure coercer
 13:07
lizmat you can with RakuAST 13:09
m: say Q|$_ ~~ / <[abcd]> /|.AST 13:10
camelia RakuAST::StatementList.new(
RakuAST::Statement::Expression.new(
expression => RakuAST::ApplyInfix.new(
left => RakuAST::Var::Lexical.new("\$_"),
infix => RakuAST::Infix.new("~~"),
right => RakuAST::QuotedRegex.new(…
lizmat replace the "elements" arguments with $string.comb.List
Nemokosch what do I need to set to access it? 13:12
lizmat you should copy the .AST output (well, as much as you need) 13:13
do the change to the elements argument
call .EVAL on the ast, and you'll a regex that you can use? well, I think :-) 13:14
also: assuming you're on HEAD
specify either "use v6.e.PREVIEW" or "use experimental :rakuast"
Nemokosch I only get RakuAST::CompUnit.new 😦 13:15
lizmat gist ?
Nemokosch > raku -e 'use experimental :rakuast; say Q|$_ ~~ / <[abcd]> /|.AST'
lizmat then you're not on HEAD 13:16
m: say Q|$_ ~~ / <[abcd]> /|.AST
camelia RakuAST::StatementList.new(
RakuAST::Statement::Expression.new(
expression => RakuAST::ApplyInfix.new(
left => RakuAST::Var::Lexical.new("\$_"),
infix => RakuAST::Infix.new("~~"),
right => RakuAST::QuotedRegex.new(…
lizmat you should get something like that ^^
Nemokosch pfff 13:17
will the generated code run on 2023.02 at least? 13:18
lizmat there's a good chance it would
13:21 rf joined
Nemokosch hm, okay, building HEAD 13:21
rf Good morning folks
Nemokosch hello 13:26
lizmat Nemokosch proof of concept: gist.github.com/lizmat/d1c090e8dcf...b1263f38d7 13:34
Nemokosch actually, I wanted to ask - what can you call EVAL on, and what can you expect to get back? 13:36
13:37 seekr joined
rf lizmat++ for the Weekly News :^) 13:40
lizmat on HEAD, you can call EVAL on Cool and RakuAST::Node objects
as a method
as a sub you could already, but then you also need a "use MONKEY-SEE-NO-EVAL" 13:41
grondilu so, what is RakuAST status? Can we generate AST from arbitrary raku code now? 13:56
tellable6 2023-03-04T12:14:32Z #raku <guifa_> grondilu: you can use the NFC, NFD, NFKC, NFKD ops
lizmat grondilu: you can if you don't use features that haven't been implemented
grondilu is it documented somewhere? 13:57
lizmat m: the Str.AST method will do that for you
camelia ===SORRY!=== Error while compiling <tmp>
Two terms in a row
at <tmp>:1
------> the Str.AST⏏ method will do that for you
expecting any of:
infix
infix stopper
postfix
statement end
…
lizmat no, that's being worked on as we speak
grondilu 🤔 13:58
from a Str?
lizmat there's a *lot* to document
m: say Q|say "yes"|.AST
camelia RakuAST::StatementList.new(
RakuAST::Statement::Expression.new(
expression => RakuAST::Call::Name.new(
name => RakuAST::Name.from-identifier("say"),
args => RakuAST::ArgList.new(
RakuAST::QuotedString.new(
…
grondilu nice
lizmat see also the gist I just posted
it creates a Regex object for you, that you can use in your code 13:59
Anton Antonov @grondilu Some of the current efforts of guifa use RakuAST — that code can be a learning point. 14:00
grondilu what about the other way around? Everything is done through the RakuAST package (I'm assuming it's a package)?
lizmat there's a RakuAST package, but most classes inherit from RakuAST::Node 14:01
grondilu m: print RakuAST
camelia ===SORRY!=== Error while compiling <tmp>
Use of RakuAST is experimental; please 'use experimental :rakuast'
at <tmp>:1
------> print ⏏RakuAST
expecting any of:
argument list
term
grondilu m: use experimental :rakuast; print RakuAST
camelia Use of uninitialized value of type RakuAST in string context.
Methods .^name, .raku, .gist, or .say can be used to stringify it to something meaningful.
in block <unit> at <tmp> line 1
grondilu m: use experimental :rakuast; print RakuAST.raku
camelia RakuAST
grondilu I see
lizmat you will HEAD to get the full .raku support 14:02
that didn't make it to the 2023.02 release
grondilu well, I guess soon I will not have any more excuse not to try to implement google's protobuf. Some time ago you guys told me it would be easier once rakuast is out.
lizmat I'm not sure right now, lack of context, but probably yes 14:03
and if you're on HEAD, there's a good chance you can already now
grondilu once we get an AST, how do we turn it into executable code? 14:04
lizmat but of course, this is really bleeding edge stuff :-)
EVAL it
grondilu oh, ok
lizmat either with $ast.EVAL or EVAL $ast
see the gist I just posted
re documentation: there are about 250 classes to document 14:05
and probably more will still come
grondilu I see. Very cool.
lizmat getting the .raku right for those classes was already.... interesting and a lot of work :-)
grondilu well .raku is not critical for anything is it? I mean it's not like $ast.EVAL actually means $ast.raku.EVAL, is it? That would defeat the purpose of an ast, right? 14:07
(lol hopefully what I just wrote does not too dumb/obvious)
*sound
lizmat not dumb at all
there's two things: .raku on an AST will give you the RakuAST::Node calls needed to build that AST 14:08
Anton Antonov @grondilu It doesn’t to me. (So, you are safish…)
lizmat secondly: the .DEPARSE method will give you back the Raku code for the given AST
m: say Q|say "hello world"|.AST.DEPARSE 14:09
camelia say("hello world")
grondilu wow that seems ambitious.
is it needed anyway? I mean I don't quite see the use of turning an AST back into raku code. It's a bit like decompiling, isn't it? 14:10
lizmat well, the class responsible for deparsing is subclassable
so in a way, it's like a "tidy" functionality 14:11
m: say Q|if 42 { say "hello world" }|.AST.DEPARSE
camelia if 42 { say("hello world") }
lizmat hmmm I sorta expected a newline after the opening { there
anyways :-)
rf m: say Q|if 42 { say "hello world" }|.AST.EVAL 14:12
camelia hello world
True
lizmat rf: that's a bit roundabout way, but yeah, that works 14:13
rf Yeah haha, just wondering if it was possible
lizmat m: say Q|if 42 { say "hello world" }|.EVAL
camelia hello world
True
grondilu in any case I think it's very cool that rakudo can compile into a syntax tree. It reminds me of the Wolfram Language. It's very elegant and it may in the future help interoperability with other programming languages. 14:14
IMHO
lizmat possibly, yes
rf So what happens, Raku Code -> Raku AST -> MoarVM Byte Code?
grondilu that would be pretty cool.
lizmat yeah, that's what normally happens 14:15
grondilu but maybe Raku AST -> Lisp
lizmat except that we now have a different way to create Raku AST
rf New backends would be simpler to build for sure
lizmat not just from Raku source
rf: no, those would still just need to support nqp (mostly)
it's all still built on nqp 14:16
rf Ah I see
Nemokosch still, NQP is easier to implement than the whole Raku language
lizmat indeed it is :-) 14:17
which was the whole point of NQP to begin with
Nemokosch so does this mean there won't be more QAST hackery in the frontend? 14:21
grondilu 👀 14:22
Nemokosch and surely not in the core, right? 🥺 14:23
lizmat all of the QAST hackery should happen inside RakuAST::Node classes 14:25
now, I don't see a reason why one wouldn't be able to add a RakuAST::Node class in the module ecosystem
in that case, QAST hacker there *would* be needed
but the idea is that the core RakuAST::Node classes should allow you to do everything you want 14:26
grondilu including macros, right?
lizmat that'd be the idea, yes 14:30
although the exact syntax and way to work that, is still undecided
first we need to get spectest to run clean, *then* we need to be able to build the setting with RakuAST 14:31
and then we can think about macros :-)
Nemokosch it would feel more relieving if QAST was completely abstracted out of RakuAST. After all, that's purely runtime-relevant code.
lizmat we basically replace src/Perl6/Grammar|Actions|World by src/Raku/Grammar|actions
in the old code, a lot of the QAST generation was done in World 14:32
which was basically an organically grown mess that works
the RakuAST classes should provide a clean interface for the logic
and that includes generating QAST, at least for the foreseeable future 14:33
the RakuAST classes *are* the way of abstracting out QAST out of the normal compilation process 14:34
now, one could maybe think about skipping the QAST step in AST -> QAST -> MAST 14:35
Nemokosch yes but at the same time it was promised that RakuAST would become standard Raku, and I still can't see any plans to merge "standard Raku" into Rakudo basically
hence it would be good to have actually pure interfaces to it
lizmat que? the t/12rakuast/* tests are intended to become part of spectest
once they have completely stabilized 14:36
so the RakuAST classes *will* be standard classes in 6.e and higher
Nemokosch then it's not fortunate that they come with a QAST baggage 14:37
that was my point
14:37 evalable6 left, linkable6 left 14:38 linkable6 joined
lizmat Nemokosch as a *user* of RakuAST classes,. you don't have to deal with QAST at all, that's the point 14:38
Nemokosch and as far as I can remember Jonathan Worthington's presentations, the plan appeared a bit different
lizmat then I think you're misremembering... but prove me wrong :-) 14:39
Nemokosch that is, the runtimes (compiler backends, in the broader sense) would consume RakuAST
not that RakuAST itself would be the compiler backend
lizmat well, eventually that *may* become possible
14:39 evalable6 joined
lizmat but that'd be akin to implementing Raku in Raku 14:39
and yes, that *can* be done (with NQP as the example) 14:40
Nemokosch a related thought: nobody seems to be that interested in keeping Raku abstract from Rakudo overall; I don't think it's coincidental that there haven't been attempts of basically any sort. I can elaborate on that in case but for now I'd just say: what if it's really unnecessary to (try to) maintain a separate standard from Rakudo? 14:44
Or would that cause too many Perl flashbacks? 😅
moritz the most reliable way to keep the two separated would be to have another implementation 14:45
Geth doc/main: 92686de1bd | (Elizabeth Mattijsen)++ | 2 files
Document ThreadPoolScheduler.new(:max_threads) a bit better

And the associated RAKUDO_MAX_THREADS environment variable.
lizmat what is in roast, determines what is Raku 14:46
RakuAST classes *will* become part of roast
ergo RakuAST is Raku
Nemokosch Yes but my point is exactly that the standard is not sufficient to actually reason about something that isn't Rakudo 14:47
neither comprehensive enough (huge lack of pragmatism wrt metamodel stuff), nor clear enough on how it should be applied regarding versions
I have heard this (informal?) term "full Raku implementation", as in supporting all possible versions of Raku. it's hard to figure out whether this is meant to be normative; one thing is sure, it's sort of obvious for Rakudo (by design) that it supports all versions of Raku - as long as Raku doesn't innovate too much 14:51
this sort of "full implementation" also more or less defeats the purpose of versions 14:52
of course it's astonishing for backwards compatibility if you just ship a runtime that covers all language versions but 1. it really holds back a lot, especially if your language was huge at the beginning 2. I'm afraid backwards compatibility will never be the major selling point of a language that notoriously broke backwards compatibility by its own existence and we know the rest... 14:57
lizmat as far as I'm concerned, Raku will at least support 3 language versions at a time 14:58
but at some point 6.c will *not* be supported anymore
Geth doc/main: 438708cd09 | cfa++ | doc/Type/DateTime.rakudoc
Fix signature typo
 15:02
15:09 [Coke] joined
[Coke] github.com/perlconference/tprc-202...i/Raku-BOF - please let me know if anyone is thinking about going. 15:09
I can drive and will go hang out even if I don't attend the conference proper.
Geth doc/main: bd8282ece8 | cfa++ | doc/Type/ThreadPoolScheduler.rakudoc
Add code preamble
 15:11
doc/main: 8fd5f52b1a | (Elizabeth Mattijsen)++ | doc/Type/ThreadPoolScheduler.rakudoc
Change pseudocode into actual code

Now not needing the preamble		 15:14
doc: cfa++ created pull request #4259:
Amend documented signatures for `Failure` methods		 15:18
doc/main: ec0bc09b51 | cfa++ (committed using GitHub Web editor) | 2 files
Amend documented signatures for Failure methods (#4259)
 15:23
Anton Antonov Is there a video recording of this presentation, "Grammatical (dis)agreement: Mixing grammars in Raku" ? tprc2022.sched.com/event/11neo 15:24
I could not find one within 5 min search 15:25
lizmat guifa might know :-) I seem to recall that presentation never actually was given, but I hope to be wrong :-) 15:26
Anton Antonov @lizmat Agh, thanks!
I am basically trying to decide should I go to Toronto's conference or not. 15:27
So, I am perusing schedules from previous years.
lucs Is having my own local zef repo as simple (so to speak) as copying/editing/using its config file? 15:30
I'd like to try out stuff, pushing and pulling (with zef and fez) from that repo, and that repo only, without polluting the community shared ones.
Geth doc/main: 57206323e1 | (Elizabeth Mattijsen)++ | doc/Type/IO/Path.rakudoc
Document IO::Path.dir-with-entries
 15:32
lizmat lucs: if you can use one of the standard storage methods (such as used for fez), it *should* be just a matter of configuration 15:35
15:37 grondilu left 15:39 razetime joined, jgaz joined 15:40 razetime left 15:45 Sgeo joined
Geth doc/main: ef66f20a7b | (Elizabeth Mattijsen)++ | doc/Programs/03-environment-variables.rakudoc
Document support for INSIDE_EMACS environment variable
 15:47
lucs @lizmat: Where can I learn about those "standard storage methods"? 15:49
(about to look at fez documentation...) 15:50
lizmat argh... it's been a while... 15:55
the default zef config file should be tell you which classes to look at
the default zef config file should be tell you which classes to look at 15:56
now, where does that live again ... 15:57
tonyo zef -h 15:59
shows it at the bottom
lucs: are you trying to host your own version of fez?
lucs @tonyo Yes, and have zef read from there only (just for trying out stuff). 16:01
I mean, have zef and fez working hand in hand in that (those?) repo only.
tonyo ah - if it's laid out similarly to fez then it should just be config on the zef side (i've already written the bits that fetch/build/install from that type of repo)..conversely on the fez side it's just a matter of setting up the right end points for it to hit and modifying the default url 16:03
lucs @tonyo If I understand what you're saying, I need to have a properly tweaked zef config, and fez will use that? (I think I can tell zef which config file to use, but not sure how to make fez aware of it.) 16:07
tonyo fez has it's own but the urls are all in the source, so on the fez side it requires code changes (they're configurable in the functions, just no way to set it globally or in the commands themselves) 16:08
zef just needs a config file update
lizmat see also: raku.land/zef:lizmat/Zef::Configuration 16:10
Nemokosch I'm back. So anyway, to wrap it up, I don't think it's among the most important things for now to clarify the situation with Rakudo vs Raku, however eventually it needs to be addressed because the current situation is kind of a false sense of security
tonyo in what way? 16:11
16:13 derpydoo left
Nemokosch It comes with the illusion that there could be other implementations when in reality it doesn't account for them 16:14
lucs @tonyo: Okay, gotcha, thanks.
Nemokosch roast, that is
lucs @lizmat: Thanks for the link. 16:15
16:16 Johanna100 left 16:23 grondilu joined
grondilu In the 5to6 perlfunc doc page, it is written that ioctl (a perlfunc) is NYI in raku. docs.raku.org/language/5to6-perlfunc.html#ioctl Is it still true and if so, is there a userspace solution to do it? 16:24
16:28 nommef joined 16:29 derpydoo joined
lizmat grondilu: perhaps any of the Terminal:: modules? raku.land/?q=Terminal 16:29
16:30 swagg_boi left, nommef left
Geth doc/main: 52c2ea05e7 | (Elizabeth Mattijsen)++ | doc/Type/List.rakudoc
Document roundrobin( ..., :slip)
 16:31
doc/main: 08b4ee46c5 | (Elizabeth Mattijsen)++ | doc/Type/Cool.rakudoc
DOcument Cool.Order
 16:35
16:36 falsifian joined
Geth doc/main: 33a6b17f1f | (Elizabeth Mattijsen)++ | doc/Language/variables.rakudoc
Bring version statement more in line with others
 16:41
16:50 swaggboi joined
grondilu well, I thought I could make a NativeCall binding real quick, but then I learn that apparently '...' in a parameter list is a thing in C. 16:55
I had never seen that before. 16:56
and I bet NativeCall can't deal with that. 16:57
dakkar ah, the terrible varargs
s.thenautilus.net/notes/999qqjv162 I wrote some vaguely-working code to deal with varargs 16:58
grondilu oh apparently that's how printf works. I should have guessed.
lizmat yeah, varargs are a pain 16:59
grondilu: what are you trying to achieve specifically ?
dakkar btw, ioctl has a terrible interface for historical reasons (essentially, it became a dumping ground for random features that should have been separate syscalls)
grondilu I want to get window size 17:01
Geth doc/main: 9d53594e5f | (Elizabeth Mattijsen)++ | doc/Programs/03-environment-variables.rakudoc
Document RAKUDO_OPT for now
 17:02
grondilu with the TIOCGWINSZ command
for a broader context, I'm interested in experimenting with the kitty graphics terminal protocol.
In raku that is 17:03
dakkar I suggest you do something like s.thenautilus.net/notes/999oh9hlyu
a bunch of wrapper functions, each declaring their "own" `sub ioctl` with the right parameter types
lizmat also: raku.land/zef:terminal-printers/Terminal::Print
?
grondilu install Terminal::Print 17:07
(
(s)
tonyo grondilu: printf works that way 17:08
oh, you found that already
dakkar printf is the *simple* case of varargs ☹
ioctly is the hard one
ooh, Terminal::Print "cheats", it runs `tput lines` + `tput cols` 17:09
github.com/ab5tract/Terminal-Print...#L104-L105
grondilu oh yeah there is tput
well I guess I can use that
Geth doc/main: 3db8263a4e | cfa++ | doc/Type/List.rakudoc
Normalise link (remove ".html")
 17:10
grondilu the number of lines and columns might be enough, but the ioctl call also gives sizes in pixels. Any way to get those too with tput? 17:12
ugexe no tput on windows fwiw 17:20
tonyo docs.raku.org/syntax/Coercion%20type
grondilu looks up tput man page, then the terminfo man page 17:21
no occurence of "pixel" in the terminfo man page. They talk about dots, though.
dakkar gist.github.com/dakkar/38703016cb0...a8c5c5a525
works on my machine
correction: rows & cols are in the other order 17:23
17:23 Johanna88 joined
dakkar grondilu: try that? ☝ 17:23
ugexe nifty, does it work after resizing the window? 17:24
dakkar yes
in a real program, you'd hook on the WINCH signal and call the ioctl again, to keep your cached values up to date
grondilu dakkar: weird, I get a 'no_fallback' unexpected argument error 17:26
dakkar !
ugexe mmhmm, thats what i used to do with github.com/ugexe/zef/blob/48ed6a2b...emInfo.pm6 but that was with tput and mode... using that nativecall code would have been much nicer
grondilu dakkar: apparently it was because I included your code in a module. I don't get the error if I copy your code as is. 17:27
dakkar grondilu: that may be some weird precomp issue…
ugexe: keep in mind that the actual value of TIOCGWINSZ may well change between different machines! 17:28
so the portability is pretty limited
ugexe ah
grondilu try it if you want to reproduci it : use module ioctl; # your code ...
anyway very nice. I love it when people write code for me :-) 17:29
also thanks :)
dakkar grondilu: if you call the module `ioctl`, there's some ambiguity with the `sub ioctl`
change one of the names, and it works again 17:30
coleman Nice newsletter this week <3
dakkar (I have no clue why it gets confused / how it should work)
coleman love the published minutes
grondilu dakkar: I don't think there should be any ambiguity, should it? 17:31
m: package foo { sub foo {} }; foo()
camelia WARNINGS for <tmp>:
Useless use of constant value foo() in sink context (line 1)
grondilu m: package foo { sub foo {} }; dd foo()
camelia foo(Any)
17:31 vrurg left
dakkar m: module foo { sub foo {}; dd foo() } 17:31
camelia foo(Any)
dakkar yep, typenames are preferred over sub names 17:32
grondilu yeah that is more accurate
dakkar m: module Foo { sub foo {}; dd foo() }
camelia Nil
dakkar m: module foo { sub foo {}; dd &foo() }
camelia Nil
17:32 vrurg joined
dakkar there's always that option, of course 17:32
grondilu normally you don't name modules in all low caps, but Ioctl looked very wrong. 17:33
dakkar loctI
locntI
(dammit, typo in a joke…)
grondilu this code ended up being much simpler than expected 17:35
dakkar nativecall is pretty magical
until you need a constant value from a C header file… 17:36
grondilu wait you sure about uint16?
not uint32?
dakkar manpage says `unsigned short` 17:37
(which, again, may be 16 bits, may be 32, may be 24… good luck)
grondilu hum
dakkar the C ABI is terrible
(mainly because it doesn't actually exist)
grondilu it is unfortunate that we can't just say uint and let NativeCall figure out the best size.
s/best/correct/ 17:38
dakkar because only the C compiler knows ☹ 17:39
(memories of "perl XS modules must be compiled with the same C compiler you build perl itself with")
Nemokosch XS looks worse than vanilla C by far tbh 17:42
dakkar eh, it's a decent compromise for all the complexity it hides
Nemokosch well, one can just use C I guess, lol
17:42 jpn joined
it seems so horrible that it's really hard to make it worthy 17:43
dakkar not really, you need a lot of knowledge about perl internals to write a library that can work as a perl extension
ugexe if you just use C you are quite limited in what you can do
Nemokosch I mean, if I had to do that, I would definitely say, who cares about Perl, after all, the whole thing is simpler to keep in C 17:44
dakkar disconnects
ugexe because often you don't want to write/rewrite your entire system in C and its easier/more efficient to just write the hot section in C
if you're working on a perl code base you can't just say who cares about Perl afterall 17:45
Nemokosch for that, NativeCall is actually pretty good, I'm absolutely bought that devs didn't want to repeat the XS precedent
17:48 dakkar left 17:51 derpydoo left 17:57 derpydoo joined 18:00 reportable6 left 18:02 reportable6 joined
grondilu lol just after I go back to reading about kitty's protocol, the next paragraph tells me about an ANSI escape code to get the number of cols and lines of the terminal. 18:07
grondilu wonders what that does on the irc bot 18:08
m: print "\e[14t";
camelia
grondilu lol, nothing
18:13 jgaz left
rf Because the bot isn't in a TTY I think 18:13
grondilu yeah that would make sense 18:19
Nemokosch the RakuAST regex generation worked 🥳 lizmat thank you 18:23
now it only needs a little finetuning to case insensitive
lizmat m: say Q|/:i /|.AST 18:28
camelia RakuAST::StatementList.new(
RakuAST::Statement::Expression.new(
expression => RakuAST::QuotedRegex.new(
match-immediately => False,
body => RakuAST::Regex::InternalModifier::IgnoreCase.new(
negated => F…
lizmat RakuAST::Regex::InternalModifier::IgnoreCase.new() 18:29
Nemokosch RakuAST::Regex::InternalModifier::IgnoreCase.new 18:31
oops right, lol
lizmat that's what I like about the new RakuAST::Node.raku functionality 18:32
Nemokosch very understandable 😋 18:34
lizmat also a great help when writing RakuAST::Node tests 18:37
18:38 jpn left
Nemokosch okay this needs to be inside the top level regex node, not wrapping it 18:41
it works now, cool
18:42 Johanna88 left 18:46 teatwo left, teatwo joined 18:50 tea3po joined 18:53 teatwo left
Geth doc/main: 1e8ea8a48a | (Elizabeth Mattijsen)++ | doc/Programs/03-environment-variables.rakudoc
Document RAKUDO_PRECOMPILATION_PROGRESS
 18:55
doc/main: a327de2e7d | (Elizabeth Mattijsen)++ | doc/Type/DateTime.rakudoc
Mark availability of DateTime.Numeric
 18:57
tonyo m: $*SPEC = IO::Spec::Win32; '\\some-server/share'.IO.say; 18:58
camelia "\some-server/share".IO
tonyo m: $*SPEC = IO::Spec::Win32; '\\\\some-server/share'.IO.say;
camelia "\\some-server/share".IO
Geth doc/main: 03a7c1dd7a | (Elizabeth Mattijsen)++ | doc/Type/IO/Path.rakudoc
Document IO::Path.add(<bar baz>)
 19:02
19:07 grondilu left
Geth ¦ doc: coke self-assigned xt test for missing C in C<> (and let's use a allow-list for doc text) github.com/Raku/doc/issues/2306 19:10
doc/main: e2eb0f28c5 | (Will Coleda)++ (committed using GitHub Web editor) | 20 files
Add test for <> (missing formatcode) (#4257)

  * Add test for <> (missing formatcode)
  * Fixup some missing C<> in pod.		 19:12
19:12 jgaz joined
tonyo m: $*SPEC = IO::Spec::Win32; my $x = '\\\\some-server/share'.IO; $x.add('walt').absolute.say; 19:13
camelia \\some-server\share\walt
tonyo m: $*SPEC = IO::Spec::Win32; my $x = '\\\\some-server/share'.IO; $x.add('walt').relative.say; 19:16
camelia \\some-server\share\walt
tonyo m: $*SPEC = IO::Spec::Win32; my $x = '\\\\some-server/share'.IO; $x.add('walt').say;
camelia "\\some-server/share\walt".IO
Geth doc/main: 9f10710a53 | (Elizabeth Mattijsen)++ | doc/Language/control.rakudoc
Document next/last with a value in 6.e
 19:20
19:22 jpn joined
Geth doc/main: 63ca40c6b0 | (Elizabeth Mattijsen)++ | doc/Type/List.rakudoc
Document .pick(**)
 19:28
19:29 jpn left 19:35 dgrigoro joined
lizmat afk& 19:40
dgrigoro Hello everyone! I am new to Raku. I'm reading docs.raku.org/language/containers.html and I'm trying to execute the following code snippet: 19:41
my $a = 0;
my $b = 0;
$a := $b;
The snippet fails when I type it interactively in the REPL, but succeeds when when I type it in a file and execute it.
Why is that?
19:43 jpn joined
gfldex dgrigoro: Because Raku is not as dynamic as it might appear at first. The REPL got quite a few limitations. 19:43
dgrigoro okay, so would you suggest that I avoid the REPL while I'm a beginner to avoid confusion? 19:45
gfldex I use a text file and some simple binds in Vim and that works fine for me. The result (over the years) looks like this: gist.github.com/3bb53da68917dee089...143f6a1245 19:46
When I'm done playing, I comment out a section. That way I keep a simple record of my play sessions. 19:47
Nemokosch I'd say one develops a feel for when the REPL won't cut it
hopefully by the time you learn it, it won't be actually needed because the rework arrives 19:48
with more clever logic than literal EVAL
19:49 jpn left
binding, operators and low-level stuff, these are probably the three notorious weaknesses 19:49
by operators I mean custom-defined operators
dgrigoro I see, thanks Nemokosch and gfldex. I just realized there's also a #raku-beginner channel, I'll post there in the future :) 19:50
20:03 jpn joined 20:08 jpn left
rf I use REPL for simple things like testing equalities 20:20
m: say [1,2,3,4] =~= [4,3,2,1]
camelia True
Voldenet dgrigoro: I never use repl and instead do `raku -e 'whatever code'` 20:26
it's nice because I can simply copy the history line into the file if it gets too long 20:27
(as long as it's one line, I don't bother)
Nemokosch I never even knew about =~= and that it works like that for an array 🤯
rf m: say [1,2,3,4] == [4,3,2,1]
camelia True
rf Also works :^)
But for the wrong reasons I think? 20:28
m: say [1,2,3,4] == [5,5]
camelia False
rf nvm
Voldenet it's correct
== is set equality
20:29 [Coke] left
Voldenet m: say [1,2,3,4] eqv [4,3,2,1] 20:29
camelia False
gfldex m: say +[1,2,3,4] == +[4,3,2,1] 20:31
camelia True
gfldex == will always coerce
Voldenet nevermind, == is not set equality 20:33
in this context
m: say [1,2,3,4] == [1, 1, 1, 1] 20:34
camelia True
gfldex m: say [1,2,3,4] ~~ [1, 1, 1, 1]
camelia False
gfldex m: say [1,2,3,4] ~~ [1,2,3,4]; 20:35
camelia True
Geth doc/main: 094c08a618 | (Will Coleda)++ | doc/Programs/03-environment-variables.rakudoc
typo,specify lang
 20:38
doc/main: 7904706d86 | (Will Coleda)++ | doc/Type/List.rakudoc
fix signature.
doc/main: dfb3dd27d2 | (Will Coleda)++ | xt/pws/words.pws
new word
Nemokosch == is numeric equivality 20:43
20:43 jpn joined 20:49 NemokoschKiwi joined 20:50 jpn left 20:51 jpn joined 20:52 NemokoschKiwi left
Geth doc/main: 7258d52c0c | (Will Coleda)++ | 3 files
Remove old util

Use xt/check-signatures.rakutest instead.
Closes #4223		 20:54
¦ doc: coke unassigned from codesections Issue util/list-missing-methods failure github.com/Raku/doc/issues/4223
¦ doc: coke self-assigned Check for indexing marks that try to index the page they're in github.com/Raku/doc/issues/4057
21:03 jpn left 21:13 TieUpYourCamel joined
Voldenet m: say [1,2,3,4].Set == [1,2,3,4].Set;say [1,2,3,4].Set == [1,2,3,5].Set 21:29
camelia True
True
Voldenet ah
m: say [1,2,3,4].Set (==) [1,2,3,4].Set;say [1,2,3,4].Set (==) [1,2,3,5].Set
camelia True
False
Voldenet that was the one I was looking for
tonyo rf: github.com/tony-o/raku-protone/tree/rf 21:30
Nemokosch it might work without the explicit conversions as well
m: say [1,2,3,4] (==) [1,2,4,3];say [1,2,3,4] (==) [1,2,3,5] 21:31
Raku eval True False
Voldenet of course
Nemokosch what I like about this is that it is explicit - about the trust towards (==) that it serves a dedicated purpose (Setty comparison) 21:32
tonyo rf: that is the version of protone that allows direct raku code in the template (it's unsafe for user input) 21:33
Voldenet I don't usually use it, but maybe I should 21:34
Nemokosch I'm kinda fixated on this, I mean that the dedicated operators should not be overloaded - at least not in a way that breaks the basic "it just does a coercion" semantics 21:35
Voldenet m: say [1,2,3,4] ~~ [1,2,3,4].Set;
camelia True
Voldenet smartmatch should be the only "cool equality" op 21:36
Nemokosch that's why I wanted coercions for Date and DateTime
instead of overloading ==
Voldenet: don't forget that smartmatch has a bit of a baggage 21:37
Match ~~ Match is sacrificed for the "regex quote" operations (m//, s/// kind of stuff)
that one I'd think can be fixed if the decision that takes there is made 21:38
Voldenet Yes, it's sometimes not obvious what smartmatch might do, imo the Java Way of having .equals method is a nice convention
Nemokosch however, Bool ~~ Bool, that one I don't think could be made equality-like
that's because when tries to be useful in if-like scenarios as well 21:39
so it must specially handle the "boolean context", i.e the smartmatch to Bool 21:40
Voldenet in fact, having .EQUALS for precise coercions and type requirements in the future is not impossible
rf tonyo: Sweet! I assume it doesn't do any escaping or anything right, just eval?
tonyo yea
there's only one shortcut
Nemokosch I think the intended .EQUALS is the eqv operator
it has some problems but actually one that I know comes straight from the reliance on smartmatching 21:42
Voldenet Yes, but how do you define eqv for a class? 21:43
Nemokosch with a multi?
tonyo it coerces {{[a-Z]+}} -> $out =~ %ctx<...>
Nemokosch what could stop you? 🙂
Voldenet Ah, right, I got stuck in java mode 21:44
Nemokosch what might be a problem, though, is to access private fields 21:46
I kind of have a feeling that this whole path hasn't been walked down on, yet 21:47
21:47 derpydoo left
Voldenet I like how COERCE works, that's why I assumed EQUALS would fit the picture similarily 21:49
Nemokosch actually I'm thinking maybe there isn't even a contradiction 21:50
there could be an EQUALS method that backs up eqv
Voldenet that sounds sane 21:51
Nemokosch hmm, actually... 21:53
in Raku, private is literally instance private
so a method might not even help tremendously much with making a comparison based on private data
but this is the point where I'd really start thinking if what one wants is rather hashability for the class 21:56
which is governed by WHICH, and grants you use of hashy types and === 21:57
21:59 jpn joined 22:05 dgrigoro left 22:09 xinming joined 22:14 jgaz left
rf Nemokosch: How do I clear the stash in Template6 22:16
tellable6 rf, I'll pass your message to Nemokosch
rf Ah localise looks like what I'm after. 22:21
@Nemokosch: Basically I want to clear the stash between renders, is this possible? 22:27
Nemokosch I probably won't know 22:37
but so far I don't know what you mean 22:38
rf Right now if I do .process('template', foo => 'bar'); when I can .process('other-template'); foo is still bar, I want it to be cleared between calls 22:39
22:39 jpn left
rf Or it might be that if I do .process('template', foo => bar), then process the same template with just .process('template') foo is still bar 22:43
guifa grondilu: in fact, I'm working on a code tidier that's based on RakuAST nodes. It's still very basic now, but while core should only provide rudimentary node to string capabilities, moduels are a great place to do even more 22:47
tellable6 guifa, I'll pass your message to grondilu
guifa lizmat & AntonAntonov I gave the talk indeed, not sure if they ended up posting the video of it — we had some recording issues IIRC. 22:48
lizmat maybe there are some slide still ?
guifa I still have the slides somewhere, and happy to send them out
I seem to recall I had one or two small errors on the slides. Just opened them up and I'll try to find and fix 22:51
lizmat Anton Antonov will be happy :-) 22:52
Nemokosch rf: the latter sounded saner at least - but is the former also true?
because that clearly sounds like a bug
rf I will test, I think it's just the former 22:53
lizmat m: say Q|constant Int $foo = 42|.AST # ah, at least it's consistent :-)
camelia ===SORRY!=== Error while compiling
Term definition requires an initializer
------> constant Int⏏ $foo = 42
lizmat I'll try to fix that in RakuAST tomorrow
rf @Nemokosch: It's actually the latter. It happens on both of my templates 22:55
Nemokosch okay but both of them only with their processed stash, right? 22:59
not cross-contaminating
rf No, they are cross contaminating 23:02
I do .process('form', :%errors); and .process('show'); and show gets errors too
Nemokosch argh
seems like there is one context per service 23:19
I mean easy to work around it (multiple instances for requests) but like I can't see why this would be a good idea in the first place...
apparently it's intended... there is a reset option you can pass to the instance 23:20
if I'm not wrong, it propagates from the main instance towards the right place 23:21
setting :reset should cause all .process calls start from an empty state
something is weird here, though 23:26
supposedly it calls reset on the context
I can't see such method
haha yeah, this is missing! 23:28
rf Hmm, that is fairly inconvenient 23:35
Do you think you could add a flag such that each call to process will cause a new context to be born 23:38
23:39 linkable6 left, evalable6 left, evalable6 joined
Nemokosch I mean, the flag is there, it just calls a non-existent method 😄 23:42
23:42 linkable6 joined
what about, for now, reset will simply clear the stash? 23:43
23:48 deoac joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!