🦋 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.
01:00 Riviera left 01:39 hulk joined 01:40 kylese left 01:44 Sevalecan left, Sevalecan joined
tbrowder antononcube: thnx for the reminder of lizmat's review of new stuff 02:05
02:15 hulk left, kylese joined
SmokeMachine does anyone here uses nvim + telescope? I'm playing with it to create a plugin for Rak, is anyone interested? usercontent.irccloud-cdn.com/file/....54.41.gif 02:58
stanrifkin_ SmokeMachine: What is telescope? 02:59
SmokeMachine telescope is a fuzyfinder for neovim... it's highly plugabble... 03:00
(on my gif, that's that panel that opens to let me search using rak) 03:01
stanrifkin_: ^^
stanrifkin_: github.com/nvim-telescope/telescope.nvim 03:02
Xliff I have most of the bindings for GLib::Top ready to go. Hope to release in the next week or two: developer-old.gnome.org/libgtop/stable/ 03:19
So obtaining information from /proc in a usable manner is near!
03:34 stanrifkin joined 03:36 stanrifkin_ left 05:12 ponycat left 05:29 kylese left 05:34 kylese joined 06:00 Sgeo left
Geth docker: sorairolake++ created pull request #62:
Add CI status badge
docker: 0dc0527914 | (Shun Sakai)++ (committed using GitHub Web editor) | README.md
Add CI status badge
docker: dbc5eef911 | (Juan Julián Merelo Guervós)++ (committed using GitHub Web editor) | README.md
Merge pull request #62 from sorairolake/add-badge

Add CI status badge
06:25 corwin is now known as catwin 07:04 kst left 07:48 sena_kun joined 08:40 ptc joined 09:32 jpn joined
Geth ecosystem/main: aeb3205dae | (Elizabeth Mattijsen)++ | META.list
MoarVM::Remote now lives in zef ecosystem
patrickb lizmat: Thanks for tackling MoarVM::Remote. Very appreciated. 09:59
lizmat it actually needs more TLC 10:00
I had to disable one test to be able to upload
I also feel that Data::MessagePack could use some updating 10:01
10:28 jpn left 10:32 jpn joined
Tirifto m: role R is Int { ; }; role S is Int { ; }; class C does R does S { ; }; 11:12
camelia ===SORRY!=== Error while compiling <tmp>
Package '<anon|1' already has parent 'Int'
at <tmp>:1
Tirifto ↑ Is it a good thing that this fails? Intuitively, I’d expect that to be okay, since both roles want the new class to inherit from the same class. :o 11:14
At least my reading of this hierarchy isn’t that the author wants to inherit from the same class twice (which is odd and sensibly fails), but that the author wants to mix in two roles, both of which want to add the same class to the inheritance. So there’s a duplicate requirement from two separate roles, which could be met by only inheriting once, thereby avoiding the error. 11:27
ab5tract Tirifto: That's a reasonable take. However the roles are already composed before they are being used for C 11:29
*with C 11:30
which allows them to use / override methods and attributes from the Int class in their definitions 11:32
without this composition, they can't be compiled as such roles (usually) refer to things not within their core definition 11:39
Tirifto ab5tract, how can I imagine a role being ‘composed’ before being used with a class? Since the documentation says the role’s methods are copied into the class, I imagined that mixing in a class was (conceptually) much like writing its body into the body of the class (without overwriting anything) and adding the carried classes to the chain of inheritance. `o` 11:40
ab5tract Yeah, I see what you mean 11:46
11:46 amenonsen left, amenonsen joined
ab5tract Tirifto: I've done a quick search of open issues that would match this and haven't found one. 11:56
Would you mind creating one? 11:57
lizmat I guess the reason for not allowing this, is that that the order of composing roles with already existing parents, could create some unexpected method lookup efffects 12:01
in this simple example it wouldn't
12:08 jpn left
Tirifto ab5tract, I wouldn’t in theory, but I am not in possession of a GitHub account; are there alternative ways of submitting reports? :o 12:09
lizmat Tirifto: do you have reasons for not wanting to have a Github account? 12:10
just OOC
ab5tract Tirifto: No worries. I'll take care of it
tbrowder hi, i'm try file open with other line endings. but there seems to be a newline at the very end when i read a single string from a file 12:17
my file looks like this: 12:18
name, age || go, 32 12:19
i cannot see the newline at the end 12:20
i open the file for reading like this: my $fh = open $f, :nl-in("||"); 12:22
for $fh.lines ...
Tirifto lizmat, mostly various ethical concerns, such as requiring the use of non-libre software, licence violations, being owned by a notoriously user-hostile company, etc. 12:27
lizmat gotcha
Tirifto ab5tract, once again, thank you very much!
12:30 MasterDuke left 12:38 jpn joined
tbrowder i guess i can strip any trailing newlines without interfering with interior newlines 13:08
lizmat without more code and/or the input file in a gist, it's hard to guess what's going on, tbrowder 13:11
tbrowder ok, i think i have it. i'll gist it in a bit 13:21
i'm trying to make my little csv module very powerful as you and [Coke] would like. one issue though is a real test file with embedded newlines. 13:22
lizmat tbrowder: there is already a very powerful Text::CSV: I was merely pointing out potential issues in the "simple" approach 13:23
tbrowder i have one (i'll show in the gist). but trying to visualize how the final product looks.
lizmat as long as everybody is on the same page wrt to known issues with an implementation, we should be ok
tbrowder yes, but the difference in mine is it's a final table (on top of my own parser). the file is read entirely before the user gets it 13:25
obviously not able to handle huge files 13:26
and the input file may have comment lines
which are removed on read and the "raw" csv file can be saves to use in libreoffice or excell 13:27
*excel 13:28
of course my motivation is to use it in another module 😄 13:30
aha, i'll create a test file in excel and output to csv (he said while slapping forehead) 13:32
15:54 haxxelotto_ joined 15:57 haxxelotto_ left
ugexe ptc: in your building and testing in raku blogpost you mention not installing the distribution being tested, but actually you *should* install it as that will precompile everything (catching any compile time errors that might not have been exercised by the test suite) as well as ensuring that the META6.json is listing all the provides/modules it should 16:36
i've seen these type of errors from various developers that would have been caught by installing instead of just running the tests tens of times 16:39
additionally one should just use `zef install .` (i.e. let the install workflow also run the tests) instead of `zef test . && zef install .` (i.e. run the tests and install as seperate workflows) as the former will only precompile things once, whereas the later precompiles twice 16:41
additionally additionally, you should almost always use `-I.` (i.e. -I $path-to-directory-containing-META6) and not `-Ilib`, since the later doesn't use META6.json and thus isn't equivalent to what the installed version might be. For example if you fail to list a module in META6.json it would fail very obviously with -I., whereas -Ilib guesses what files should be included based on the files it sees 16:44
17:05 teatwo left, teatwo joined 17:06 jpn left
ptc ugexe: ok, good to know. I thought one explicitly *doesn't* want to install a dist's modules because then if one changes the code, how does one know that one is testing the right code? I've definitely had the situation in the past where the installed code was being found first and any changes I made weren't having any effect. 17:07
good point about -I though; I didn't know about that effect. Thanks! 17:09
I wonder if there's a workflow whereby one tests such compile time and META6.json errors as a pre-release check. It'd also be necessary to uninstall the dist afterwards too, I think, in case one doesn't want to have it installed locally. 17:12
17:46 vlad joined 19:10 vlad left 19:11 ptc left
ugexe fro CI tests at least you wouldn't worry about having leftover installed code 19:14
but you can also just install it into some temporary directory 19:15
for quickly testing and iterating locallym using `zef test .` or `prove -I. t/` is fine. you would want to do a test install at some point though, likely before pushing any changes 19:17
however, I wouldn't recommend using --install-to= when CI tests, because that prevents the use of the staging repository and thus precompilation will occur twice (once during tests, once during install) 19:18
21:08 antononcube joined 21:22 antononcube left, antononcube joined
librasteve I have started to go zef install . in my workflow ... it was a bit of a gear change, but I agree that it is the best approach (I use --force --/test so that I can run the test manually when I want (since that can take a while) and to avoid having to uninstall (except the last time ;-) 21:39
21:59 Sgeo joined 22:17 antononcube left 22:28 sena_kun left 22:30 MasterDuke joined 22:34 stanrifkin left
tbrowder i do the same as librasteve, works great and i can always fall back to the published module if need be 22:54