#perl6-dev on 21 May 2018 - Raku Programming Language Log

Perl 6 language and compiler development 2.0 | Logs at irclog.perlgeek.de/perl6-dev/today | For toolchain/installation stuff see #perl6-toolchain | For MoarVM see #moarvm
Set by moderator on 20 May 2018.

00:00 lizmat joined

jnthn

MasterDuke: No, missed that. I think it's still a sensible feature. About .* and .+, actually they are only about walking the MRO, not multi-candidates, so it can probably apply for methods too

00:18

MasterDuke

interesting

00:22

and you think the static optimizer is the best place to check?

00:23

jnthn

Maybe. An alternative is to do it right after CHECK time

00:30

MasterDuke

when/where is that?

jnthn

m: class A { ... }

00:31

camelia

The following packages were stubbed but not defined:
A

jnthn

It's the place that we handle that

MasterDuke

thanks, i'll look there

jnthn

Yeah, at the moment there's

00:32

# Checks.

$*W.assert_stubs_defined($/);

$*W.sort_protos();

I'd add it around there

Actually

00:33

sort_protos is doing candidate sorting, I think

You could potentially do the check right then since in theory that already has the list of things that'd want checking

MasterDuke

nice

00:34

jnthn

Sleep time for me, happy hacking :)

MasterDuke

later...

01:56 shareable6 joined 01:58 ilbot3 joined

moderator

Perl 6 language and compiler development 2.0 | Logs at irclog.perlgeek.de/perl6-dev/today | For toolchain/installation stuff see #perl6-toolchain | For MoarVM see #moarvm

Kaiepi

i'm adding a test to perl docs to ensure nbsp is used whenever perl 6 appears

02:46

AlexDaniel

Kaiepi: cool but… didn't we have a test for that already?

02:47

Kaiepi

i didn't think we did, but i'll check in a second

02:48

doesn't look like it

02:50

02:51 Zoffix joined

Zoffix

Kaiepi: are you looking in t/ ? There's another set in xt/ github.com/perl6/doc/tree/master/xt

02:51

xt/perl-nbsp.t should be the one you're talking about

Kaiepi

what tests typically go in xt?

also, i guess what i could do is write a util that automatically replaces Perl 6 without the nbsp with the one with it

02:53

Zoffix

Author tests. Basically stuff that end users aren't meant to be running. Or phrased differently, covering problems that aren't important enough to stop installation of the dist for end users if those problems exist, like typos and nbsps

Kaiepi

MasterDuke

i'm getting a `Cannot find method 'gist' on object of type Signature`. however, this is the code `note('a = ' ~ nqp::can(%a<signature>, 'gist') ?? %a<signature>.gist !! "can't .gist");`

02:55

Zoffix

MasterDuke: precedence

02:56

m: use nqp; dd "a" ~ nqp::can(42, "meows") ?? "yes" !! "no"

camelia

"yes"

Zoffix

m: use nqp; dd "a" ~ (nqp::can(42, "meows") ?? "yes" !! "no")

camelia

"ano"

MasterDuke

02:57

i didn't realize ~ bound that tightly

Kaiepi

how do i make an array like [1,2,3,4] into one like [[1,2],[3,4]]?

03:06

AlexDaniel

m: say [1,2,3,4].rotor(2)

03:07

camelia

((1 2) (3 4))

Kaiepi

thanks

Geth

¦ rakudo: zoffixznet self-assigned X::TypeCheck chops off .perl too soon github.com/rakudo/rakudo/issues/1848

03:25

Zoffix

Kaiepi: note that that will toss the last element if there's an odd number of them. Use :partial to keep it

03:33

dd [[1,2,3,4].map: {[$^a, $^b]}]

03:34

evalable6

[[1, 2], [3, 4]]

Kaiepi

aight, thanks

Zoffix

m: dd [1,2,3,4].map: *RR,*

03:35

camelia

((1, 2), (3, 4)).Seq

Zoffix

Neat. You a way to close over a comma :D

Geth

rakudo/post-release-2018.05: a79de42904 | (Zoffix Znet)++ | 2 files
Fix premature .perl chopping in X::TypeCheck

Fixes github.com/rakudo/rakudo/issues/1848 R#1848
Before chopping .perl, check if object's .^name is part of it, and if it is, bump the chop length by the length of the .^name.

03:39

synopsebot

R#1848 [open]: github.com/rakudo/rakudo/issues/1848 X::TypeCheck chops off .perl too soon

Zoffix

m: class Foo is Int {}; class Bar does Rational[Foo, Foo] {}; dd (Bar.new: Foo.new(1), Foo.new(2)).numerator.^name

03:52

camelia

"Int"

Zoffix

turns out we were never creating our Rationals correctly :P

03:59 Zoffix left 04:54 Zoffix joined

Zoffix

Turns out there's a 6.c-errata spectest that blocks zero-denominator rationals normalization: github.com/perl6/roast/commit/60dd...7c992b5929

04:55

BUT, the commit is a change from original tests that expected normalization. The commit message doesn't say the reasons for the change and the IRC chat around that commit doesn't say the reason for the change.

04:56

.ask lizmat do you recall the reason for this commit, but any chance? github.com/perl6/roast/commit/60dd...7c992b5929 The first two tests, that expected numerator to be normalized to 1 or -1. By CaR Grant's logic, the normalization is desired, but that commit is part of 6.c-errata

04:58

yoleaux

Zoffix: I'll pass your message to lizmat.

Zoffix

What sucks is normalization alone doesn't fix the ZDR math bugs -_-

m: say Num(<1/0> + <0/0>)

camelia

Inf

Zoffix

m: say Num(<1/0>.Num + <0/0>.Num)

04:59

camelia

NaN

Zoffix

ffffff...

05:12

note to self. git reset --hard doesn't undo last commit >_<

Geth

roast/car-grant-unreduce: 798b571f5f | (Zoffix Znet)++ | S32-num/rat.t
[CaR Grant] Change spec of zero-denominator rationals normalization

I deem lack of normalization a bug, and thus the tests wrong.
Since they're numerically equivalent, the .numerator/.denominator should be the same and we should normalize them right away. This lets us also use simpler logic in things like == and === and to have .WHICH give the same thing. ... (8 more lines)