lizmat jnthn: the problem is that $lhs_ast is not a QAST::Var in the case of "my $foo is Bar" 00:02
I guess we need to probe deeper into $lhs_ast to find the var and *then* get the sigil 00:03
jnthn Ah, hmm 00:05
Interesting
Sleep time for me, alas
But it should be possible to get it to work. Actually, if we do, we might generate better code for a bunch of cases
lizmat good night!
jnthn In that if it doesn't know the sigil it generates more general code 00:06
'night
lizmat also calls it a night 00:20
Zoffix First victim of .subst $/ fix: HTTP::Parser dies with "[4] Cannot assign to a readonly variable ($/)" 01:40
Nobody upgrade geth :P
ohhh. and it's actually using $/ in substr, so its code was busted in the past 01:41
ohhh. and it's actually using $/ in substr, so its code was busted in the past 01:43
oops
Ah nah, not busted 'cause they used -> $/ as param 01:44
Fix: github.com/tokuhirom/p6-HTTP-Parse...ll/7/files 01:46
Feeling early signs of a burn out, so I'm gonna try to disappear for a week and chill 02:30
Zoffix &
[Tux] Rakudo version 2017.10-4-g4fca94743 - MoarVM version 2017.10-1-g213fc774
csv-ip5xs1.189 - 1.223
csv-ip5xs-2014.069 - 14.223
csv-parser11.968 - 12.479
csv-test-xs-200.433 - 0.480
test11.733 - 11.752
test-t3.154 - 3.329
test-t-2058.327 - 59.796
test-t-20 --race19.943 - 21.598
07:44
Geth nqp: 626ea02068 | (Stefan Seifert)++ | tools/build/MOAR_REVISION
Bump MoarVM for NativeCall fix
08:11
¦ nqp: version bump brought these changes: github.com/MoarVM/MoarVM/compare/2...g116c7a133
rakudo: 948af00b59 | (Stefan Seifert)++ | 2 files
Pass type objects to the VM for arguments of native functions

The VM may need this information for making decisions about how to build a native call site, e.g. if it's possible to create JIT compiled code or whether it needs to refresh a CArray's elements after a call. Fixes GH #1220
08:14
synopsebot RAKUDO#1220 [open]: github.com/rakudo/rakudo/issues/1220 [regression] Regression since 2017.10 breaks IO::Socket::Async::SSL tests
nine [Tux]: that rakudo version is quite a bit behind. We're now at 2017.10-40-g948af00b5 08:18
[Tux] <headdesk />'s (forgot to move from nom to master 08:19
module FAIL: Data::Dump:ver<0.0.5> 08:40
nine -win 16 08:45
[Tux] Rakudo version 2017.10-40-g948af00b5 - MoarVM version 2017.10-29-g116c7a13
csv-ip5xs1.116 - 1.119
csv-ip5xs-2012.749 - 13.069
csv-parser12.219 - 12.333
csv-test-xs-200.436 - 0.443
test12.031 - 12.163
test-t3.230 - 3.243
test-t-2058.999 - 59.149
test-t-20 --race20.508 - 20.530
08:53
lizmat Files=1229, Tests=75779, 318 wallclock secs (14.68 usr 5.33 sys + 2181.52 cusr 215.15 csys = 2416.68 CPU) 09:13
good *, #perl6-dev! 09:25
.tell jnthn re .STORE, rather than sending a named :initialize to .STORE, why not invoke it with a type object on initialization, and bind the result to the var? 09:26
yoleaux lizmat: I'll pass your message to jnthn.
lizmat .tell jnthn that way, we could use MMD to handle the different cases
yoleaux lizmat: I'll pass your message to jnthn.
jnthn m: class C { multi method m() { say "later" }; multi method m(:$initial!) { say "first" } }; given C.new { .m(:initial); .m } 09:33
camelia first
later
yoleaux 09:26Z <lizmat> jnthn: re .STORE, rather than sending a named :initialize to .STORE, why not invoke it with a type object on initialization, and bind the result to the var?
09:26Z <lizmat> jnthn: that way, we could use MMD to handle the different cases
jnthn Already *can* use MMD to do that :) 09:34
lizmat true
jnthn The multi cache has been able to handle things with named args for a while now
lizmat sorry, not awake enough yet :-) 09:35
the problem is really that .STORE is being called from the result of Type.new 09:36
in the case of Set, that would be Set.new
but that is the empty Set, that is immutable
so by the time Set.STORE is called, it is handed an object it's not allowed to change 09:37
now, you could argue that Set.new could create a new instantiated Set object all of the time
but that felt wasteful to me :-)
jnthn It's not an optimization if it makes things wrong. 09:38
lizmat so I was thinking, why not combine the useful with the efficient thing to do
jnthn 'cus STORE magically making an instance too sounds like awful design
lizmat well, then maybe we should call the method CREATE-AND-STORE when we call it for initialization 09:40
with a Mu.CREATE-AND-STORE() { nqp::create(self).STORE } for compatibility
jnthn And all this just to make Set.new return (perhaps surprisingly) a singleton, when we already have an empty set term that does precisely that? 09:41
lizmat Set.new returns the empty set term
that's the point
jnthn Yes, I'm saying why do that when we already have a term for the interned thing 09:43
lizmat the codegen of "my %h is Set" does a Set.new underneath, to call .STORE on 09:44
and Set.new returns the empty set()
jnthn Yes, and I'm saying it should not! 09:45
lizmat I'm not convinced it should not :-) 09:46
not just to make my %h is Set work
is there a deeper reason for you to have Set.new not return the set() singleton ?
jnthn It's a bit surprising for .new to not make something new. 09:48
lizmat well, if the new is just like the old, why have a new ?
I mean, it makes checking for the empty set so much faster (nqp::eqaddr) 09:49
jnthn OK, then let's make this simpler: no STORE in Set. Only SetHash.
Done.
Time for $other-job.
Geth rakudo: 6ac2b15cfa | (Elizabeth Mattijsen)++ | 2 files
Revert "Make 'my %h is Set(|Hash) = ...;' DWIM"

This reverts commit 1949a2bc13085f05c21f4a6b627d2f435f76182a.
09:52
lizmat jnthn: re Set.new() returning a new empty Set rather than set(), we have a spectest since May 2017 testing for it returning set(): I assume you also want that changed then 10:20
jnthn lizmat: Yes, that's certainly since 6.c, so easy to loosen up 10:29
lizmat m: dd Mix.new =:= mix(); dd Bag.new =:= bag() # rince and repeat for Mix and Bag I assume ? 10:32
camelia Bool::True
Bool::True
lizmat m: use nqp; my $a = 42; my $b = 42; dd nqp::eqaddr(nqp::decont($a),nqp::decont($b)) # but this is not unexpected
camelia 1
lizmat anyways, /me revs up the chainsaw 10:33
AlexDaniel squashable6: next
squashable6 AlexDaniel, ⚠🍕 Next SQUASHathon in 3 days and ≈23 hours (2017-11-04 UTC-12⌁UTC+14). See github.com/rakudo/rakudo/wiki/Mont...Squash-Day
jnthn m: use nqp; say nqp::eqaddr(Int.new(5), Int.new(5)) # :) 10:38
camelia 0
jnthn lizmat: ^^ is the equivalent thing though, and we get it right there
lizmat m: dd bag().Set =:= set() # jnthn, is that ok, or should that also be a Set.new ? 10:49
camelia Bool::True
jnthn lizmat: That's probably OK 10:56
lizmat so you're only concerned by Set.new(), Bag.new() and Mix.new() ? 10:57
jnthn Yes; note that those are also precisely the ones that we'd need to fix for the my %h is Set = ...; to work, which - despite my grumpiness - I'd kinda like to, in that it makes working with immutable Sets a bit nicer. 10:58
lizmat ok 10:59
lizmat revs down the chainsaw a bit
Geth rakudo: aab2b98305 | (Elizabeth Mattijsen)++ | 8 files
Set|Bag|Mix.new() no longer return set()|bag()|mix()

  - to pave the way for making "my %h is Set|Bag|Mix" work
  - Set|Bag|Mix.new() returning sentinels not considered worthwhile having
11:47
roast: cbfc93971f | (Elizabeth Mattijsen)++ | 3 files
Adapt tests to new Set|Bag|Mix.new() behaviour

As implemented in aab2b98305f04ec7
11:48
tbrowder lizmat: any pointers to debugging nqp other than jnthn’s course? 11:59
lizmat MIUAYGA ? 12:00
Make It Up As You Go Along :-)
tbrowder and zoffix’s blog...
funny! but seriously, other than the zoffix info, is there an “easy” way to find the source of a “===SORRY!===“? 12:03
i’m getting close with print statements but... 12:04
lizmat --ll-exception
?
tbrowder ok, i haven’t tried that in awhile and forget about it, thnx 12:05
dogbert2 Zoffix: RaR suggestion for you, www.youtube.com/watch?v=iDT7Jbg_4zU 12:57
MasterDuke re my question from yesterday about changing proto bodies from `{*}` to `{ * }`, here's a diff of what i've done gist.github.com/MasterDuke17/1412a...df9ca7e428 12:58
hopefully a fresh pair of eyes can spot a trivial error 12:59
perlpilot MasterDuke: why are proto bodies changing to "{ * }" ?
MasterDuke just to normalize with how most of them a written. there are about 630 written as `{ * }` now, i just thought it might be nice to have them all consistent 13:00
perlpilot I thought that {*} was a token unto itself? 13:01
MasterDuke yeah, `proto token foo {*}` is different, changing them does something different 13:02
dogbert2 lizmat: does 1202 still SEGV for you? 13:03
lizmat not segv, but MoarVM panic: Heap corruption detected: pointer 0x11225cea0 to past fromspace 13:04
dogbert2 have you tried running it in gdb with a breakpoint. I can't get it to SEGV or panic, it only hangs form me 13:06
lizmat haz no gdb on MacOS :-( 13:07
jnthn Ain't it lldb or something there? 13:09
lizmat jnthn++
jnthn Don't ask me how to use it though ;) 13:10
Though I think in the next month or so I'll see if I've budget to get my office a biscuit tin
Uh, mac mini :)
dogbert2 wonders what can cause the code to hang after writing 'Created initial affinity worker thread' 13:11
dogbert2 tries asan again 13:16
AlexDaniel this looks about right? gist.github.com/AlexDaniel/5c40302...ddd6860323 13:20
.tell lizmat gist.github.com/AlexDaniel/5c40302...ddd6860323 13:22
yoleaux AlexDaniel: I'll pass your message to lizmat.
Geth rakudo: b6a4d5b555 | (Elizabeth Mattijsen)++ | 7 files
Make 'my %h is Set = ...' DWIM

  - also for SetHash, Bag, BagHash, Mix, MixHash
  - by implementing .STORE for all QuantHash types
  - .STORE is now also sent a :$initialize flag which is True for initialization
   - True for: my %h is Set = ...;
   - not set for: %h = ...;
  - throws X::Assignment::RO when trying to change Set|Bag|Mix
   - as in: %h = ...;
13:23
lizmat .
yoleaux 13:22Z <AlexDaniel> lizmat: gist.github.com/AlexDaniel/5c40302...ddd6860323
lizmat AlexDaniel++ 13:25
AlexDaniel eventually I'll put it into a bot so that you can generate it any time you want
lizmat AlexDaniel++ (again!)
m: say 42 13:27
camelia 42
lizmat m: say 42 13:34
camelia 42
lizmat m: my %d is SetHash = Date.today .. Date.new("2017-11-30"); %d .= grep: *.key.day-of-week > 5; dd %d # this now works :-) 13:35
camelia SetHash.new(Date.new(2017,11,5),Date.new(2017,11,12),Date.new(2017,11,26),Date.new(2017,11,4),Date.new(2017,11,19),Date.new(2017,11,11),Date.new(2017,11,18),Date.new(2017,11,25))
Geth rakudo: 497e0582e6 | (Elizabeth Mattijsen)++ | src/core/Exception.pm
Make X::Immutable::RO not show all of its gist

If it gets too long. It's just there for you to get a hint :-). This fixes RT #132353
13:56
synopsebot RT#132353 [new]: rt.perl.org/Ticket/Display.html?id=132353 [LTA] error with using meta assign ops with bound SetHash
lizmat jnthn: before I take on auto-generating BUILD_LEAST_DERIVED, do you have any idea whether we should include autogenerated methods in .^methods or not: 14:02
m: class A { has $.a = 42 }; dd A.^methods.map: *.name
camelia ("a", "BUILDALL").Seq
lizmat options I see:
- a and BUILDALL should *not* be returned, unless a ":autogenerated" is given as a parameter 14:03
- only a should be returned, as people expect that to happen, being autogenerated or not
- BUILDALL / BUILD_LEAST_DERIVED should never be returned if they are autogenerated 14:04
jnthn I think the "a" has to appear 14:09
It's too surprising for it not to
lizmat ok
jnthn I guess one could argue that the user *did* declare the method
lizmat ok
jnthn If we take `has $.a` as being both an attribute and a method declaration 14:10
lizmat yeah, sure
so, should I just check the name, or should I tack something onto the Method object: hidden-from-methods ?
if the latter, how do I do that in QAST :-) 14:11
jnthn Maybe methods can get a .generated or .synthetic property
And we filter on that
lizmat just add an attribute to the Method object ?
jnthn Well maybe, though we might find there's a flags bitfield we can re-use
To keep the size down
The usual way to then write these things up is to write a trait_mod candidate for is generated or so 14:12
That tweaks the flag
And then just find/invoke it
iirc, there's code in routine_def's action method doing exactly that for some other property related to wrapping 14:13
travis-ci Rakudo build failed. Elizabeth Mattijsen 'Make 'my %h is Set = ...' DWIM 14:17
travis-ci.org/rakudo/rakudo/builds/294853229 github.com/rakudo/rakudo/compare/a...a4d5b55552
buggable [travis build above] ✓ All failures are due to: GitHub connectivity (1 failure).
MasterDuke i really don't understand why this is happening: github.com/rakudo/rakudo/blob/mast...on.pm#L42: `my str $s = nqp::atpos($strings,$i);` 14:56
after my patch nqp::atpos($strings,$i) is a Whatever
lizmat that feels... weird ? 14:59
MasterDuke `elsif $s.comb(/:r '*' || \d+ || <.alpha>+/).eager -> @s { my $strings := nqp::getattr(@s,List,'$!reified'); <...>; my str $s = nqp::atpos($strings,$i);` # how is this at all affected!?!? 15:00
Geth roast: bd6c8f5df4 | (Elizabeth Mattijsen)++ | 3 files
Add some tests for 'my %h is Set|Bag|Mix = ...;'
15:30
MasterDuke narrowed it down to one of the changes in Str.pm... 15:37
Geth roast: d615e019d3 | (Elizabeth Mattijsen)++ | 3 files
Add some tests for 'my %h is SetHash|BagHash|MixHash = ...;'
15:43
MasterDuke it's this line github.com/rakudo/rakudo/blob/mast...r.pm#L950: `proto method match(|) { $/ := nqp::getlexcaller('$/'); {*} }` # changing that to `{ * }` breaks things 15:49
which seems wrong...
jnthn {*} is a special form 15:51
github.com/rakudo/rakudo/blob/mast....nqp#L3518
Parsed as a single token
A { * } is just a normal bare block that evaluates to a Whatever 15:52
MasterDuke i've tried to trigger than .panic, but haven't been able to
jnthn m: sub foo { {*} }
camelia 5===SORRY!5=== Error while compiling <tmp>
{*} may only appear in proto
at <tmp>:1
------> 3sub foo { {*}7⏏5 }
expecting any of:
horizontal whitespace
term
jnthn m: proto sub foo { {*} }
camelia ( no output )
MasterDuke doh
i wasn't putting the other set of braces, no wonder i didn't get it 15:53
so `proto sub foo {*}` is the same as `proto sub foo { * }`, but `proto sub foo { {*} }` is *not* the same as `proto sub foo { { * } }` 15:54
ugexe m: proto sub foo(|) { { * } }; multi sub foo { }; say foo(1); # kinda odd
camelia *
jnthn Yeah. If you're standardizing on anything, probably better to go without the whitespace 15:55
ugexe but i guess the {*} redispatches, which is why i dont get the error message I thought
MasterDuke well, there are about ~150 `{*}` now and ~630 `{ * }`
jnthn {*} is a "magic" token that means "do the next bit of the multi-dispatch here"
ugexe m: proto sub foo(|) { {*} }; multi sub foo { }; say foo(1); # was confused initially that this errored but the other did not 15:56
camelia Cannot resolve caller foo(Int); none of these signatures match:
()
in sub foo at <tmp> line 1
in block <unit> at <tmp> line 1
MasterDuke should i go the other way, turn the `{ * }`s into `{*}`s? 15:57
lizmat m: use Telemetry; snap for ^5; snap; .say for periods 15:58
camelia 177 / 175
16 / 14
12 / 12
11 / 10
10 / 11
lizmat documentation: gist.github.com/lizmat/9dfe921efe7...4dfdbcd68c
jnthn MasterDuke: I dunno. I'm the wrong person to ask. Such tiny discrepancies don't bother me in the slightest. :P
lizmat and gist.github.com/lizmat/e601e91c90e...6b965ec770 </plug>
lizmat hopes for feedback from jnthn masak timotimo moritz Zoffix AlexDaniel` 15:59
ugexe if you name it something.pod I think it'll show up as rendered 16:01
lizmat it didn't render well :-( 16:02
ugexe yeah i notice pod always seems to render bigger than I want when github renders it (so I always write readmes in .md now)
lizmat well, I want to add this to the doc repo 16:03
MasterDuke anyone else care about `{*}` vs `{ * }`? should i bother?
lizmat but I guess that would only be appropriate if it can actually live in lib
MasterDuke: consistency is nice... :-) 16:04
MasterDuke lizmat: care which direction to make the change? i.e., add spaces where possible, or remove all them? 16:05
lizmat well, as jnthn made clear, you can't remove the spaces always 16:06
so I would go for removing spaces, making the {*} more visible
you can't *add* spaces always :-)
MasterDuke heh, was just going to ask which you meant... 16:07
ok, i'll give that a try
lizmat so how old was the "nom" branch anyway? 16:09
timotimo oh, how does the "cpus" value work 16:14
lizmat cpu / wallclock
timotimo isn't that the usage percentage? 16:15
lizmat if cpu is twice as large as wallclock, how many CPU's were in use on average in that period ?
[Coke] lizmat: rakudo.org/2011/09/09/nom-branch-default/ 16:16
timotimo the docs don't say "on average" :)
jnthn 6 years of damn muscle memory
lizmat timotimo: noted
timotimo so i thought there's some magic trick that tells us how many different CPUs got threads from our program scheduled on it or something
[Coke] (that's not quite the right answer, since that's just when it was default)
lizmat April 2011 according to rakudo.org/2011/09/09/nom-branch-default/ 16:17
[Coke] lizmat: ... I apparently didn't read the whole thing. :)
seems older.
lizmat a lot of things have happened in the past 6.5 years :-) 16:18
timotimo lizmat: i'm still confused, what's the difference between cpus and utilization? 16:19
lizmat cpus gives a factor
timotimo ah
so no need to * 100 to get percentage
lizmat utilization gives a percentage
up to 100% if all cpus were used in full for the given period 16:20
timotimo OK
Geth rakudo: ae1f0fdafc | (Elizabeth Mattijsen)++ | lib/Telemetry.pm6
Reset default @snaps as soon as we start periods
16:29
lizmat m: 42 16:31
camelia WARNINGS for <tmp>:
Useless use of constant integer 42 in sink context (line 1)
Geth rakudo: MasterDuke17++ created pull request #1221:
Normalize proto bodies
16:42
synopsebot RAKUDO#1221 [open]: github.com/rakudo/rakudo/pull/1221 Normalize proto bodies
lizmat m: 42 17:07
camelia WARNINGS for <tmp>:
Useless use of constant integer 42 in sink context (line 1)
lizmat m: snap; Nil for ^10000000; snap; say periods[0].cpus 17:08
camelia 5===SORRY!5=== Error while compiling <tmp>
Undeclared routines:
periods used at line 1
snap used at line 1
lizmat m: use Telemetry; snap; Nil for ^10000000; snap; say periods[0].cpus
camelia 1.0050972
lizmat m: use Telemetry; snap; await start { Nil for ^10000000 } xx 4; snap; say periods[0].cpus 17:09
camelia 2.67922708
lizmat interesting
dinner&
Zoffix: iBooks on my MBP doesn't seem to like temp.perl6.party/pub/2017/PSix-Doc...-2017.epub : "This book is corrupt" 18:59
just an FYI 19:00
Zoffix vOv
m: await ^60 .map: { start sleep ⅓ }
camelia MoarVM panic: Could not spawn thread: errorcode -11
stmuk lizmat: I've some "blurb" about R* which could go in to the weekly if you want 19:12
yoleaux 28 Oct 2017 07:32Z <piojo_> stmuk: When I was talking about archlinux support for rakudo star, I was really talking about DESTDIR support. That's all that's needed for us to be able to make packages, both third party and official.
lizmat stmuk: yes please!
stmuk OK I can gist it in a few mins
lizmat cool, thanks1
! 19:13
MasterDuke anybody have any thoughts/comments/etc about github.com/MoarVM/MoarVM/pull/737 and github.com/rakudo/rakudo/pull/1217? 19:18
lizmat it has conflicts ? 19:20
MasterDuke not according to github? 19:22
the rakudo PR requires the moarvm and nqp PRs though
[Coke] "this branch has conflicts" (737)
MasterDuke oops, hadn't reloaded the moarvm PR tab 19:23
that'll be because nine++ added some new ops, i'll resolve that 19:24
updated 19:28
stmuk lizmat: gist.github.com/stmuk/65d15b8b4877...c42a1a0472
lizmat stmuk++ 19:29
stmuk the TLDR is "no panda anymore" and experimental linux AppImage 19:30
lizmat stmuk: you're sure you want your email address in there? 19:37
stmuk err maybe not I did wonder
lizmat :-)
stmuk maybe "feedback on IRC" better
lizmat will do that
stmuk lizmat++ 19:38
or github.com/rakudo/star/issues 19:39
jnareb Hello. I have just read rakudo.org/2017/10/27/main-developm...to-master/ (which unfortunately is on a 'blog' with no user comments) 20:38
So I thought to post comment(s) here 20:39
Is there a reason why 'nom' branch cannot be a symref to 'master'?
timotimo oh, does github allow for that? 20:51
AlexDaniel it's not github, it's git that has to allow that
and I don't think it does
maybe something has changed over the years 20:52
jnareb: you sure it is possible? 20:53
and if yes, how? 20:54
Zoffix Looks like `git symbolic-ref refs/heads/nom refs/heads/master` per stackoverflow.com/questions/549920...949#549949 21:00
jnareb `git symbolic-ref -m "create symbolic ref to 'master'" refs/heads/nom refs/heads/master` 21:02
it's low-level command, so there are no user-friendly shortcuts
`git branch` would show it as "nom -> master"
the "-m <reason for reflog>" is optional 21:03
though the problem is/might be that it doesn't survive fetching (yet) 21:04
I think that refs/remotes/origin/HEAD is a special case, created on client side 21:05
so it would be a command to use by the person cloning the repository, after cloning it
AlexDaniel that sounds more painful than continuing to use “nom” :) 21:07
currently the only “issue” is that people who were using nom will have whatever there is in nom currently, which is 2017.10 release (almost) 21:08
lizmat feels a bit like mustard after a meal to me 21:09
as in: we've bitten the bullet of the change already, why make it more complicated now? 21:10
[Coke] googles that phrase.
huh. lizmat++
AlexDaniel lizmat: well the question still remains: was there any better way to do it? 21:11
jnthn AlexDaniel: Leave it alone?
But nooo, nobody here can leave names of things alone :P :P
timotimo heh heh
hey jnthn how are you? :)
AlexDaniel leave what alone? It's already “master”
jnthn timotimo: Seems I'm slowly getting rid of my obligatory autumn cold... 21:12
lizmat and it's already part of the title of this P6W :-)
timotimo my roommate just got hers and it's really bad in her case :(
she got it the day i left to visit my parents
jnthn AlexDaniel: My point was we didn't have to do this at all :)
Zoffix dogbert2: thanks, but that Wolfenstein release looks to have terrible reviews on Steam. Also, I didn't quite enjoy the mood of The New Order; I like games where you take a nice breather from battles (e.g. Far Cry 3/4) 21:13
jnthn AlexDaniel: But now we have...so.
AlexDaniel facepalms
jnthn timotimo: Mine has been thankfully mild...so far. 21:14
timotimo glad to hear it
i might have been able to sidestep it :)
jnareb actually it is not fetched as a symref, but it doesn't matter, because when fetching client would advance both 'master' and 'nom', and they would point to the same commit
which I have just checked
AlexDaniel six years of pain and one hour to fix it. Goddamit why do we have to go through this naysaying every time.
jnareb just I am not sure if you can actually create symbolic ref on GitHub...
well, the true source needs to have symref, that is the repository from which is pushed to GitHub 21:16
timotimo haha, "true source"? :)
Zoffix AlexDaniel: change is scary and people don't like things that are different. Mr. Plinket gets it right
jnareb everyone else would magically get 'nom' and 'master' pointing to the same commit
timotimo so everybody who pushes to github regularly has to set it up locally and that'll cause nom and master to get the same commits pushed at the same time whenever you push to either of them?
Zoffix AlexDaniel: www.youtube.com/watch?v=FxKtZmQgxr...DE6F720A2D
jnareb well, 'refs/remotes/origin/master' and 'refs/remotes/origin/nom' 21:17
AlexDaniel Zoffix: and also “it works for me, I don't care about anybody else”, no video though :)
jnareb timotimo: Yup!
well, if you do 'push all'
Zoffix I rather we go through the grace period and then pop that blocking message back in on nom rather than aliasing nom and master and having things that shouldn't work work forever. 21:18
jnareb if you push only single branch, it wouldn't
timotimo that's not terribly much win all in all
jnareb but when you push 'nom', it would be the same as 'master'
timotimo that's useful if you accidentally push to nom
AlexDaniel we already have pushes to nom blocked 21:19
a little bit
jnareb Well, either 'push all', or have configured `push.default` to `matching` (`git config push.default matching`) 21:20
AlexDaniel jnareb: that's very interesting, thanks :) 21:21
jnareb Which is configuration that maintainer (the person pushing to GitHub repo) should use anyway, in my opinion
[Coke] I am happy with pushes to nom blocked and no magic. 21:22
jnthn If our goal is to ween people off using nom, then it's probably best to leave it frozen.
jnareb AlexDaniel: blocking pushes to 'nom' is also a good solution
jnthn And eventually eliminate it
jnareb though 'nom' as symref could be a good interim solution, until everyone updates their clients / scripts / buildenv / behavior / ... 21:23
well, could have been
[Coke] has anyone else come and complained? 21:24
Zoffix Did anyone complain at all? :) 21:25
AlexDaniel Well, maybe a little bit. Tux was benching nom for some time before finally realizing :)
Zoffix Pushes to nom are currently blocked already. Need to have a PR that another reviewer approves first or something.
AlexDaniel there are things like that, but nobody really complained
Zoffix Tux would realize instantly if the blocking message was in place :P 21:26
AlexDaniel :)
jnareb just my €0.03
bye
Zoffix jnareb: thanks.
Zoffix looks up the sub-currency of euro 21:27
oh, they're called cents too
AlexDaniel that's about 0.00001 BTC :) 21:28
Zoffix :)
AlexDaniel jnareb had a good point that there was a *different* (not necessarily better) way to do it. That's great and I loved that piece of information. Why did we have to make them feel unwelcome is beyond me. 21:36
maybe I misunderstood something 21:37
lizmat and another Perl 6 Weekly hits the Net: p6weekly.wordpress.com/2017/10/30/...-mastered/ 21:47
jnthn lizmat++ 21:51
Zoffix AlexDaniel: which part made them feel unwelcome? 21:52
AlexDaniel lizmat: on code-golf.io you can't see the solutions and that seems to be an intended part of the game :) 21:56
Zoffix: not sure
Zoffix lizmat++ good weekly 21:57
AlexDaniel lizmat++ yeah!
Zoffix AlexDaniel: to my eye, looked like that just used the web irc interface to deliver the 2 cents and then just went on their merry ways. No unwelcomness occured. 21:58
Zoffix &
AlexDaniel okay then
AH! 22:00
hm no… nvm 22:01