#raku-dev on 24 October 2024 - Raku Programming Language Log

00:28 MasterDuke joined

jdv

www.artima.com/forums/flat.jsp?for...read=27898

01:04

dan sugalski, parrot peep

oh, thanks will

ab5tract

Dang, the post mortem is 404

06:40

ugexe: is there a way to ensure that zef uses the raku tree it was installed with instead of the raku tree in the path?

08:50

gist.github.com/ab5tract/77bbce7e7...eb2be4758f

The script works as expected but invoking `.rakubrew/versions/moar-2024.02/install/share/perl6/site/bin/zef locate Time::Crontab` responds with a file installed at `.rakubrew/versions/moar-2024.08/install/share/perl6/site/sources/AF7B5BB0DFEECDDF6640F50E7AC66F120A8776D8`

08:51

this also means that `zef install Foo` is a no-op when it is installed for `moar-2024.08` even though I'm trying to install it for `moar-2024.02`

08:58

I've just checked and unfortunately nothing changes even when I specify the config path relative to the correct version with `--config-path`

09:09

I would prefer not to do anything with the environment

09:14 sena_kun joined 09:24 sena_kun left 10:37 MasterDuke left

timo hurls github.com/MoarVM/MoarVM/issues/1850

10:52

lizmat

feels like a plan... :-)

10:53

timo

now to forget about it and never actually do anything :P

10:55

13:39 [Tux] left 13:42 [Tux] joined

[Tux]

Rakudo v2024.09-140-g2750112c0 (v6.d) on MoarVM 2024.09-10-g939e64dc8

csv-ip5xs	0.258	-	0.262
csv-ip5xs-20	1.097	-	1.118
csv-parser	1.489	-	1.587
csv-test-xs-20	0.141	-	0.143
test	1.842	-	1.911
test-t	0.394	-	0.402
test-t --race	0.268	-	0.274
test-t-20	4.734	-	4.756
test-t-20 --race	1.161	-	1.170

14:44

tux.nl/Talks/CSV6/speed4-20.html / tux.nl/Talks/CSV6/speed4.html tux.nl/Talks/CSV6/speed.log

5 2024-07-18 08:42:10 test-t 0.399

14:45

4 2024-10-21 19:01:30 test-t 0.396

3 2024-08-01 13:22:46 test-t 0.395

2 2024-10-24 16:42:09 test-t 0.394

1 2024-08-19 09:58:29 test-t 0.394

20240718 0.399❙20241021 0.396❙20240801 0.395❙20241024 0.394❙20240819 0.394❙

jdv

changelogs are up. i'll wait an hour or so before starting the releases.

15:12

lizmat: ^

[Coke]

jdv++

15:13

ugexe

ab5tract: I'm not immediately sure how that could happen 🤔 As far as I remember anything (zef or raku internals) invoking raku would be using $*EXECUTABLE. and the paths where modules are installed/listed is determined by the raku ($*EXECUTABLE) used to invoke zef itself

15:20

15:21

i bet the bin shim raku installs does like #!/usr/bin/env raku or something

so zef is getting invoked with whatever raku that would use

15:22

.rakubrew/versions/moar-2024.02/install/raku .rakubrew/versions/moar-2024.02/install/share/perl6/site/bin/zef

15:23

something like that should work though

if you want something shorter you could also do like

15:25

.rakubrew/versions/moar-2024.02/install/raku -MZef::CLI -e '' locate Time::Crontab

15:26

with github.com/Raku/App-Rakubrew/issues/80 it would be even more convient

15:28

jdv

ab5tract: www.sidhe.org/oldblog/archives/000435.html

15:37

just a url change

lizmat

jdv: looking at the changelog now

15:41

ab5tract

ugexe: that makes total sense

15:42

though this isn't installed with the shim raku

the -MZef::CLI works like a charm :)

15:47

though I'll stick to my script for the purposes of locate as I want JSON output

parsing in Java is a yuck fest

15:48

btw, if `zef` is installed in a RO directory like `/usr/...`, will it automagically install into a user-writable location like `/home/...`?

15:50

lizmat

jdv: done with the changelog

15:58

ugexe

ab5tract: the path you listed is certainly a shim because it isnt a sha1 basename

15:59

if you `cat ~/.rakubrew/versions/moar-2024.02/install/share/perl6/site/bin/zef` or whatever you'll see as much

ab5tract

Ah, I see what you mean. I was referring to the fact that I used a non-shimmed raku to install it

16:00

So I was talking about rakubrew shims and I believe you are speaking of a zef provided shim

16:01

ugexe

yeah, although the shim is provided by rakudo itself

github.com/rakudo/rakudo/blob/2750...od#L33-L36

16:02

ab5tract

Hmm.. then maybe we cans make it use it’s absolute path instead of whatever it’s doing now

ugexe

im not sure if that is the right thing to do either tho

16:03

there is relocatable raku, plus im not sure that a CURI should be tied to a specific installation of rakudo or not. remember one can install stuff to a custom location which could be used by any rakudo via e.g RAKULIB

16:04

ab5tract

Good point..

16:05

Maybe a configuration flag would be reasonable

16:06

Or else making the shared CURI be more of a default experience

16:07

jdv

lizmat: thanks

ab5tract

Because I’ve never had the experience when installing a new Raku version of having my existing third party stuff available

16:08

Which, if it works, why not?

ugexe

zef install Foo --install-to='~/foo/bar' and then RAKULIB="~/foo/bar"

16:09

ab5tract

That is absolutely too much work :)

ugexe

the other thing is the default experience will generally be catered to a typical user, not a developer of the core. most users do not have multiple rakudos installed, and would generally be best served by default that avoid any caveats to the aforementioned shared repo

16:10

for example there is a concept of repository version for when e.g. the layout of the files is changed, and which rakudo will "migrate" to a new version upon install if a new version is introduced. any old rakudo that does not know the new format would then break trying to use that repository

16:12

ab5tract

Fair enough. Thanks for making Zef so usable from Raku itself. I can put away my concerns for the shims and move on to pushing the Zef support in Comma

ugexe

actually i lied, there is an easier way to do it (although caveats still apply)

16:14

set github.com/ugexe/zef/blob/45f3ac76...ig.json#L5 to home, or set ZEF_INSTALL_TO=home, or pass zef --install-to=home

16:15

1 bin/ script [zef] installed to:

16:16

/Users/nlogan/.raku/bin

raku -e 'say $*REPO.repo-chain.head'

inst#/Users/nlogan/.raku

the "home" repo is at the head of every raku installation and isn't tied to a specific version

16:17

we actually did use to default to the "home" repo I think, but switched away due to problems such as the caveat I listed earlier

16:18

also since you mentioned comma... you don't *have* to install zef, you could just run it from a shared cloned repo

16:20

i guess a drawback of using environment variables like ZEF_INSTALL_TO= though is that they are consumed in Zef::CLI not Zef::Client or Zef::Config so your example code would not end up using it

16:32

github.com/ugexe/zef/blob/45f3ac76...kumod#L429

i guess ideally Zef::Config should be rewritten to handle the environment variable stuff and provide methods for accessing the final config value of things including env stuff instead of just returning a hash of what is in the config file itself and handling the environment variable stuff at a call site

16:34

in that way the existing Zef::Config might be better thought of as Zef::ConfigFile

16:35

ab5tract

since my experience was that `Zef::Config` worked exactly the way I expected/wanted, I'm a bit ambivalent about all that

16:37

but maybe I will indeed look into using a shared repo eventually

16:38

but probably won't be adopting a shared CURI unless there is a strong demand for it

16:39

it can be useful to switch between Rakudo "SDK" installations specifically because you want to compare how two diverging versions of a module behave

or maybe this is case of me feeling "I just put this bucket of bolts back together, I'm not going to let anything tear it apart" :)

16:42

ugexe: do you find yourself using `-e''` very often? It makes me wonder whether a `-x` or similar might be appreciated by more people than just myself

16:44

lizmat

I have an alias: r='time raku -e'

16:47

and I use that *very* often

ugexe

ab5tract: i think i primarily find myself using it when i dont want to set PATH to whatever/perl6/install/site

although i only ever use `-e ''` in combination with `-MZef::CLI` (or -e 'use Zef::CLI')

16:48

if im remember correctly i found avoiding having to set PATH useful on windows or windows ci systems where I dont want to be bothered figuring out how to set the env vars or to determine the CURI bin directory to set PATH to begin with

16:52

Geth

rakudo/release-2024.10: 19ff0d0fa4 | (Justin DeVuyst)++ | 3 files
Update changelog + announcement

Deliberately not logged:
  [0732084c][d56a34ae][1c96328c][a0626a20][d804906a][c1fd6736]
  [0bd6c486][c719ac22][19420673][021a3bb2][bbd8b65d][25640772]
  [b94ad1d7][2e305032][427f8b7a][a60b4e7d][dde57d94][0e9a0ee3]
  [6528f868][ccf01a59][b7d10895][40cbb194][d38c1140][6c30c590]
  [8b51c4a4][0f5d54d2][4b53990c]

17:08

rakudo/windows_latest_azure_pipeline: 610f2e4e6e | (Timo Paulssen)++ | azure-pipelines.yml
Use moarvm with exception handler generation.

17:36

rakudo/windows_latest_azure_pipeline: e6fc4f89a9 | (Timo Paulssen)++ | azure-pipelines.yml
try getting a minidump from nqp build

17:58

rakudo/windows_latest_azure_pipeline: d104352f0e | (Timo Paulssen)++ | azure-pipelines.yml
up moarvm checkout commit

18:25

rakudo/windows_latest_azure_pipeline: f94fd51d7f | (Timo Paulssen)++ | azure-pipelines.yml
up moarvm checkout commit

18:48

18:49 sena_kun joined

jdv

hmm, looks like a bunch of roast errata fails:(

19:00

lizmat: around?

19:17

Geth

whateverable/coke/max-age: 766e60b6d6 | (Will Coleda)++ | lib/Whateverable/Webhooks.pm6
Add $!max-age

Fixes #390

19:19

whateverable: coke++ created pull request #393:
Add $!max-age

rakudo/windows_latest_azure_pipeline: c6b2c48da6 | (Timo Paulssen)++ | azure-pipelines.yml
try to debug crash ...

19:22

jdv

m: combinations(3,2).raku

19:26

camelia

( no output )

jdv

m: combinations(3,2).raku.say

camelia

((2, 2), (2, 2), (2, 2)).Seq

jdv

m: combinations(3,2).say

19:27

camelia

((0 1) (0 2) (1 2))

jdv

is that first one correct?

ab5tract

that seems quite weird to me

19:28

m: dd combinations(3,2)

camelia

((2, 2), (2, 2), (2, 2)).Seq

ab5tract

m: dd |combinations(3,2)

camelia

(2, 2)
(2, 2)
(2, 2)

ab5tract

m: dd combinations(3,2).gist

19:29

camelia

"((0 1) (0 2) (1 2))"

ab5tract

lizmat: does this seem familiar to you?

jdv

that's the only failing errata test (6.c-errata) that isn't explained easily so far

Geth

roast/6.c-errata: 35c7179af6 | (Elizabeth Mattijsen)++ (committed by Justin DeVuyst) | 2 files
Adapt some .flat tests

in light of github.com/rakudo/rakudo/commit/0fc7d45f66

19:39

ab5tract

jdv: bisectable6 seems to say that this is to be expected (assuming I'm holding it half way correctly). see #whateverable

19:45

jdv

hmm, i'm not in there i don't see that channel on irclogs

19:47

how is that correct though?

ab5tract

it says its the same on all revisions

19:49

bisectable6: exit 1 if combinations(3,2) ~~ ((2,2),(2,2),(2,2)).Seq

bisectable6

ab5tract, Will bisect the whole range automagically because no endpoints were provided, hang tight

jdv

i mean this test passed last release so clearly something changed

19:50

bisectable6

ab5tract, Output on all releases: gist.github.com/e0e90916d1cc7b9e40...91339fe498

ab5tract, Output on all releases and bisected commits: gist.github.com/f85681074fd6dee30e...69e70c674e

ab5tract

then I'm holding bisecavle wrong

jdv

github.com/Raku/roast/blob/6.c-err...ions.t#L17

19:51

ab5tract

m: combinations(3,2).list.perl

19:52

camelia

( no output )

ab5tract

m: combinations(3,2).list.perl.say

camelia

((2, 2), (2, 2), (2, 2))

ab5tract

yeah, I see

jdv

i mean the equivalent in master is missing the .perl so it works so i could just patch that in but why did it change from last month to now...

19:53

Geth

roast/6.d-errata: 5 commits pushed by (Elizabeth Mattijsen)++

19:57

ab5tract

it does feel like something is wrong

20:00

I don't get why bisectable doesn't find it

20:01

jdv

yeah:(

20:02

Geth

rakudo/windows_latest_azure_pipeline: 261f311461 | (Timo Paulssen)++ | azure-pipelines.yml
try to debug crash ...

ab5tract

ah, this probably explains it:

¦«HEAD(2750112)»:

Cannot test this commit (Broken archive)

jdv

what does that mean?:)

20:03

ab5tract

surface level: It means it's not testing against HEAD, where the problem is

20:04

underneath: no idea :)

jdv

so, is this a blocker then?

20:06

hmm

maybe github.com/rakudo/rakudo/commit/693f15ea ?

20:10

we really should bring the *able stuff inhouse. right now alexdaniel manages it.

20:12

ab5tract

jdv: that's got to be it

20:25

jdv

yup. built with that reverted and it passes

20:31

*sigh*

ab5tract

20:35

Geth

whateverable/main: 766e60b6d6 | (Will Coleda)++ | lib/Whateverable/Webhooks.pm6
Add $!max-age

Fixes #390

20:37

whateverable/main: 0b9b6a5a63 | ab5tract++ (committed using GitHub Web editor) | lib/Whateverable/Webhooks.pm6
Merge pull request #393 from Raku/coke/max-age

Add $!max-age

20:41 notable6__ left, bloatable6__ left, coverable6 left, shareable6 left, sourceable6 left, releasable6 left, evalable6 left, unicodable6__ left, greppable6 left, quotable6 left, committable6 left, nativecallable6 left, tellable6 left, bisectable6 left, linkable6 left, benchable6 left

jdv

have to revert and redo parts of the release now...

20:42

[Coke]

ugh, sorry jdv

20:43 bloatable6 joined, releasable6 joined 20:44 benchable6 joined, sourceable6 joined, evalable6 joined, shareable6 joined 20:45 bisectable6 joined, committable6 joined, notable6 joined, linkable6 joined, quotable6 joined 20:46 unicodable6 joined, coverable6 joined, nativecallable6 joined, greppable6 joined 20:47 tellable6 joined

jdv

might need to stop for a bit to do dinner but should get it done in a few hours...

20:48

Geth

rakudo/release-2024.10: e6bf5457bf | (Justin DeVuyst)++ | src/core.c/Rakudo/Iterator.rakumod
Revert "Un-nqpify *.combinations() without slowing it down"

This reverts commit 693f15ead2b2b89fb9e5e6fe976bb364a165d3d5.
This commit is being reverted because it broke this test:
github.com/Raku/roast/blob/6.c-err...ions.t#L17
... (6 more lines)

21:07

rakudo/release-2024.10: 358d27ab26 | (Justin DeVuyst)++ | 3 files
Update changelog + announcement

Deliberately not logged:
  [0732084c][d56a34ae][1c96328c][a0626a20][d804906a][c1fd6736]
  [0bd6c486][c719ac22][19420673][021a3bb2][bbd8b65d][25640772]
  [b94ad1d7][2e305032][427f8b7a][a60b4e7d][dde57d94][0e9a0ee3]
  [6528f868][ccf01a59][b7d10895][40cbb194][d38c1140][6c30c590]
  [8b51c4a4][0f5d54d2][4b53990c][693f15ea][e6bf5457]

rakudo/windows_latest_azure_pipeline: 8f406a8897 | (Timo Paulssen)++ | azure-pipelines.yml
random changes

21:09

ab5tract finally recognizes that patch

21:10

it's a shame it didn't fully cover this scenario, because otherwise I loved the idea of moving back to Raku from NQP in Rakudo

21:11

jdv

i'm sure she can fix it later, just need it fixed now or out.

21:14

there's been a bunch of that work in the past - i'm a fan

21:15

ab5tract

indeed, I said it all in the spirit of its eventual return :)

21:30

Geth

rakudo/windows_latest_azure_pipeline: 7d69fa862b | (Timo Paulssen)++ | azure-pipelines.yml
random changes

21:52

jdv

lizmat: github.com/rakudo/rakudo/commit/e6...f9198038a1

22:01

the release is now progressing past where that caused the redo

22:03

ab5tract

jdv++

22:08

Geth

nqp/main: 416410829e | (Justin DeVuyst)++ | tools/templates/MOAR_REVISION
[release] Bump MoarVM revision to 2024.10

22:11

nqp/main: 2289d6af8b | (Justin DeVuyst)++ | VERSION
[release] Bump VERSION to 2024.10

rakudo/release-2024.10: d48de082b5 | (Justin DeVuyst)++ | tools/templates/NQP_REVISION
[release] Bump NQP revision to 2024.10

rakudo/release-2024.10: 44bc9df693 | (Justin DeVuyst)++ | VERSION
[release] Bump VERSION to 2024.10

22:13 sena_kun left

Geth

rakudo: jdv++ created pull request #5672:
Release 2024.10

22:14

rakudo/main: 5 commits pushed by (Justin DeVuyst)++

22:15

jdv

patrickb: the 2024.10 release happened

22:29

.tell El_Che the 2024.10 release happened

tellable6

jdv, I'll pass your message to El_Che

jdv

release is done. commit away.

nite!

Geth

rakudo/windows_latest_azure_pipeline: e6d9eae9eb | (Timo Paulssen)++ | azure-pipelines.yml
is lldb even there?

22:32

ab5tract

Thank you jdv!!

22:46

lizmat

jdv: thanks for the revert

23:16

Geth

rakudo/windows_latest_azure_pipeline: 3cfd6bc45e | (Timo Paulssen)++ | azure-pipelines.yml
use older nqp now that the release has happened

23:44

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

Thank you!