»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'p6: say 3;' or rakudo:, std:, or /msg camelia p6: ... | irclog: irc.perl6.org | UTF-8 is our friend!
Set by moritz on 25 December 2014.
andreoss raydiak: reported 00:04
raydiak andreoss++ 00:07
00:08 raiph left
timotimo Cannot find method 'value' 00:13
at src/Perl6/World.nqp:651 (blib/Perl6/World.moarvm:install_lexical_container:376)
i wonder how i did that
seems like my $_ = ... in a block did that 00:14
raydiak m: for ^3 { my $_ = "foo"; .say } 00:15
camelia rakudo-moar 8f447f: OUTPUT«Potential difficulties:␤ Redeclaration of symbol $_␤ at /tmp/qjZQlordFN:1␤ ------> for ^3 { my $_ ⏏= "foo"; .say }␤Cannot assign to an immutable value␤ in block <unit> at /tmp/qjZQlordFN:1␤␤»
raydiak m: do { my $_ = "foo"; .say } 00:16
camelia rakudo-moar 8f447f: OUTPUT«===SORRY!===␤Cannot find method 'value'␤»
raydiak m: do { my $_ }
camelia rakudo-moar 8f447f: OUTPUT«===SORRY!===␤Cannot find method 'value'␤»
raydiak m: { my $_ }
camelia rakudo-moar 8f447f: OUTPUT«Potential difficulties:␤ Redeclaration of symbol $_␤ at /tmp/zPA58pBXMj:1␤ ------> { my $_ ⏏}␤»
raydiak m: -> { my $_ }
camelia rakudo-moar 8f447f: OUTPUT«===SORRY!===␤Cannot find method 'value'␤» 00:17
raydiak m: { my $_ },
camelia rakudo-moar 8f447f: OUTPUT«Potential difficulties:␤ Redeclaration of symbol $_␤ at /tmp/kAJiU0iva0:1␤ ------> { my $_ ⏏},␤»
raydiak m: -> { my $/ }
camelia ( no output )
00:17 kjs_ joined
raydiak m: .say 00:19
camelia rakudo-moar 8f447f: OUTPUT«Nil␤»
00:22 adu left 00:23 broski left
psch m: { temp $_ = "foo" } 00:25
camelia ( no output )
psch m: { my $_ = "foo" } 00:26
camelia rakudo-moar 8f447f: OUTPUT«Potential difficulties:␤ Redeclaration of symbol $_␤ at /tmp/BFXaizliST:1␤ ------> { my $_ ⏏= "foo" }␤»
00:29 leont joined
timotimo Failed to open pipe: 12 00:33
in QX
psch timotimo: i hope you're also golfing and ticketing ;)
timotimo no energy right now 00:34
this is from a locally modified spesh_diff.p6
but the changes shouldn't give me trouble ...
i shall add some debug output
00:43 colomon left 00:44 ssutch left 00:50 virtualsue left 00:51 colomon joined, Rounin joined
timotimo cannot figure it out 00:54
geekosaur ENOMEM... 00:56
00:57 kjs_ left
timotimo oh? 00:57
geekosaur if that's an OS errno
but the only place I can think of that could generate that other than synthetically via malloc() is fork() in which case it means you're out of virtual memory 00:58
which seems moderately unlikely since it would be rather obvious from more than just p6 00:59
01:02 andreoss left 01:09 [Sno] left 01:11 cognominal left 01:14 BenGoldberg left 01:23 adu joined, adu left 01:24 raiph joined 01:28 BenGoldberg joined 01:31 BenGoldberg left 01:32 BenGoldberg joined 01:40 Mso150 left 01:41 Mso150 joined, adu joined 01:56 adu left 02:20 leont left 02:23 rmgk_ joined, rmgk left, rmgk_ is now known as rmgk 02:27 kaare__ joined 02:33 adu joined
adu what's the difference between associative_bind and bindkey? 02:33
and positional_get and atpos? 02:36
02:46 colomon left
Timbus associative_bind and positional_get appear to be NQP ops 02:48
uh, or more like VM ops 02:49
02:52 gfldex left 02:54 colomon joined
vendethiel was confused for a moment 02:58
colomon is usually confused longer than that 03:02
03:08 Mouq joined, Mso150_i joined 03:09 Mso150 left, dj_goku_ joined 03:10 dj_goku left 03:18 jakesyl_insomnia left 03:20 jakesyl_insomnia joined 03:30 noganex joined, amkrankruleuen left 03:31 djruffkutz joined 03:32 djruffkutz left 03:33 noganex_ left 03:39 dj_goku_ left
adu what does SCWB mean? 03:41
03:43 dj_goku joined, dj_goku left, dj_goku joined 04:20 raiph left 04:25 Mso150_i_q joined 04:26 Mso150_i left, skids left 04:29 ggoebel111111112 left 04:30 ggoebel111111112 joined 04:40 skids joined 04:46 dj_goku left 04:49 dj_goku joined 05:04 vendethiel left 05:10 dj_goku left 05:21 sirdancealot joined 05:23 davido_ left 05:24 dj_goku joined, dj_goku left, dj_goku joined, davido_ joined 05:44 KCL_ joined 05:52 telex left 05:54 Mouq left, telex joined 05:58 dj_goku_ joined, dj_goku left 05:59 xinming left 06:00 xinming joined 06:07 IllvilJa left 06:09 BenGoldberg left, IllvilJa joined, skids left 06:13 Mso150_i_q left 06:18 skids joined 07:46 xfix joined 07:48 berekuk joined 07:54 skids left 08:01 berekuk left 08:04 berekuk joined, berekuk left 08:10 darutoko joined 08:13 novice666 joined
novice666 what is the mechanism of incrementing immutable variable $ in oneliner perl6 -ne ' say ++$ ' file.txt ? 08:18
08:19 mr-foobar left
moritz novice666: it's not immutable, it's just anonymous 08:25
but, uhm, does that even work? 08:26
seems it does 08:27
08:27 lumimies left
moritz it's weird, I'd have though you have to declare it first 08:27
--target=ast says it generates lexical $ANON_VAR__1
for the $
novice666: does that answer your question? 08:28
08:28 mr-foobar joined 08:30 ugexe left 08:31 ugexe joined
JimmyZ m: $; my $b; 08:36
camelia rakudo-moar 8f447f: OUTPUT«===SORRY!===␤Cannot find method 'orig'␤»
JimmyZ m: my $; my $b;
camelia ( no output )
08:39 virtualsue joined
adu do you see anything wrong with this? pastie.org/9825136 08:49
moritz adu: that's Perl 6 code (not NQP), right? 08:51
well, the indentation of the 'make' line is off
JimmyZ @<pointer>?
you mean @$<pointer>? 08:52
moritz m: 'ab' ~~ /$<a>=(.)+/; .say for @<a>
camelia rakudo-moar 8f447f: OUTPUT«｢a｣␤｢b｣␤»
moritz JimmyZ: @<pointer> is a valid shortcut 08:53
adu moritz: yes
JimmyZ ah, I didn't know it ..
adu JimmyZ: I don't know if that's what I mean
I've never seen @$ before
JimmyZ m: my $@a = [ 3, 5]; say %@a; 08:55
camelia rakudo-moar 8f447f: OUTPUT«===SORRY!=== Error while compiling /tmp/sQkFxNVcBf␤Cannot declare a variable by indirect name (use a hash instead?)␤at /tmp/sQkFxNVcBf:1␤------> my $⏏@a = [ 3, 5]; say %@a;␤»
JimmyZ m: my @a = 3, 5; say %@a;
camelia rakudo-moar 8f447f: OUTPUT«3 => 5␤»
moritz adu: ah, one thing that's wrong is that the recursion into $ast doesn't seem to terminate anywhere 08:56
adu moritz: so the behaviour I'm seeing is that if I have a pointer ("char * s"), then it runs forever, but if I dont ("char c"), then it parses pretty quick
moritz adu: but it would be much easier to help if you'd tell us why you suspect something might be wrong 08:57
yes, the recursion into $ast would explain that 08:58
adu but it's not recursion
it's a completely different ast
moritz $ast = -> $type, $init { $ast($pointer($type), $init) }
there are two occurences of $ast here. They refer to the same object. 08:59
adu maybe if I do 2 of them $ast2 = -> ... {}; $ast = $ast2?
moritz Thus, recursion
yes, you need a separate lexical for it to work as a closure 09:00
adu I thought {} did that?
moritz {} creates a block and thus a scope
but not a new lexical 09:01
'my' does that
adu how do I accomplish this? 09:02
maybe I could call make instead? 09:03
moritz my $wrapped = $ast; $ast = -> $type, $init { $wrapped($pointer($type), $init) } 09:05
adu: but are you sure all of those need to be callables? they aren't as nicely debuggable as other data structures 09:06
adu if you can find a better at to pass info from one end of the grammar to the other end, I'm all ears 09:07
moritz well, make() already gives you that capability 09:08
adu ok, but can I call make multiple times?
moritz well, the last one will take effect
fwiw I'm not talking just about this one action sub 09:09
erm, method
I wonder why you seem to passing callables as code objects everywhere
want multiple values? make() an array
adu s/at/way/
ok, but an array of [a, b, c] will still have to be turned into a(b(c(x))) at some point 09:10
moritz why? why must a, b and c be callables?
adu moritz: the alternative is to find a way to insert the current declaration specifiers in the parameterized types encountered later on 09:11
moritz why not simply a class Pointer { has $.pointee } or something like that? 09:12
adu because the fact that it's a pointer, and the type of the pointee are in two separate locations of the grammar
moritz I don't see the problem 09:13
Pointer.new(pointee => $other.ast)
adu $.pointee comes from $<declaration-specifiers>, which doesn't know about pointers, and $<pointer> doesn't know about the other
moritz adu: I think I'd have to see the whole source code to be able to discuss this properly 09:14
adu moritz: I would love to do that, but both pieces of information are not available until $<declaration>
github.com/andydude/p6-c-parser/bl...Parser.pm6
so that's the grammar
09:15 jack_rabbit left
adu I have 2 action classes, CASTActions, and DROXActions, and both action classes have a similar issue, but I kind of gave up on CAST for right now, so I am currently trying out the function/block/lambda idea with DROXActions 09:16
moritz can't find any DROXActions 09:19
adu github.com/andydude/droxtools/blob...ctions.pm6 09:20
it's in a different repo because it has nothing to do with general C parsing
09:22 molaf__ joined
adu moritz: the CAST is very OOP, iirc it does have a Pointer class as you described 09:22
09:26 molaf_ left
adu but if the function/block/lambda idea works, I'd like to do the same thing with CASTActions 09:28
09:35 denis_boyun_ joined
adu I was actually discussing this with raydiak a couple days ago 09:36
I was saying that I wanted 2 ASTs, one that represents syntax, and one that represents symantics 09:37
and raydiak said that I should have one AST and do the transformations in the actions
moritz or simply add another transformation stage for syntax tree -> sematnic tree 09:39
adu is that easy? 09:40
masak bzzt. guten antenoon, #perl6.
adu hi masak
masak what does it mean for an AST to "represent semantics"? 09:41
moritz it becomes an Abstract Semantic Tree :-) 09:42
masak well, that clears that up...
09:42 mr-foobar left
adu masak: well, the issue is with C declarations. they are generally of the form: <declaration-specifier>+ [[<declarator> ['=' <initializer>]?]* % ','] 09:43
the declaration-specifiers form a complete type. 09:44
but the declarators can be of the form myArray[2][3][4] or myFunc(double, double) or myArrayFunc(double)[3] or myFuncArray[3](double), etc. 09:45
masak with you so far.
sounds like a kind of expression parsing, if you ask me. 09:46
adu the "complete type" for the entire declaration must combine information from the declaration-specifiers with information about whether or not it's an array, function, pointer etc from later in the grammar
moritz wait
what piece of information exactly comes from later in the grammer? 09:47
adu moritz: -> $type { Pointer.new(pointee => $type) }
the $type comes from the declaration-specifiers
moritz adu: in terms of syntax, please 09:48
myArrayFunc(double)[3]
which piece of information about that comes from later in the grammer?
*grammar
adu moritz: myArrayFunc comes from direct-declarator-first:sym<identifier> 09:49
moritz: (double) comes from direct-declarator-rest:sym<p-parameter-type-list> 09:50
moritz: and [3] comes from direct-declarator-rest:sym<b-assignment-expression>
moritz adu: but when you are at the point where the whole string 'myArrayFunc(double)[3]' has been parsed, you have all the ASTs for all of those tokens already
adu moritz: yes, but you don't have the return type of the function! 09:51
moritz adu: because the function can be post-declared? 09:52
adu my point is that <declarator> has all the information it needs to build any AST, except for what is in <declaration-specifiers>
all of the information should be available to <declaration>
but I don't know how to combine it
moritz not for a syntax tree
a syntax tree shouldn't encompass information that comes later on 09:53
adu so declarator has (Name, PartialType), and declaration has (Type)
moritz so you seem to need a separate pass that first builds a symbol table, and then resolves forward-references
adu and I want the AST to have name and type as fields
moritz that's not an AST anymore 09:54
AST = Abstract *Syntax* Tree
adu so I should have a single AST that represents syntax
not semantics
moritz now we can start to argue whether type information that's not obviosu from syntax is still syntactic 09:55
but from a practical POV, you just need another pass
whethere you build a separate tree, or simply add some more information later on is pretty much up to you 09:56
adu so, how do I do that? pass an array of actions to parse() ?
moritz no. Simply traverse the tree you build
if that tree has a somewhat regular structure, that's trivial
adu that's a lot of if statements
moritz do that after you called .parse 09:57
not necessarily
first of all, mulltis can help a lot
or methods on the varios types that make up your AST
adu oh, right, I can match on the types
moritz finally, you should make the AST somewhat uniform 09:58
your users (including yourself) will thank you for it
09:59 gfldex joined
adu so by uniform, do you mean simple? 09:59
like QAST has Stmt 10:00
but CAST has several classes of statements
moritz the QAST tree conists only of QAST::Node subtypes (except for leaves)
which makes it very easy to check if something is a branch of a leaf
also there's a uniform API for accessing QAST children 10:01
so if you want to do a complete tree traversal, you basically just have to do (paraphrased) multi visit(QAST::Node $n) { vist($_) for $n.list }; multi visit(Any) { # do nothing, leaf node } 10:02
adu wow
moritz so if our AST is OO-y, it helps to give those classes a common API, via a superclass or role 10:03
adu is it ok if I subclass QAST to make CAST?
moritz don't do that
first, QAST is NQP-defined
which can lead to oddities if you subclass it in Perl 6
adu ok
moritz second, you don't want to support all the operations that QAST::Node does, so there's not much point 10:04
role CAST { has @.children handles <list> }
not so complicated, really :-) 10:05
adu I've never seen "handles" before
moritz it's for delegation
it just forwards the method 'list' to attribute @.children 10:06
role CAST { has @.children; method list { @.children.list } } # equivalent
adu ok 10:07
masak ab5tract: you didn't offend me. I agree with you that Perl 6 should not be considered immature. 10:08
adu can you pass extra parameters to methods in the actions even if the grammar doesn't have those actions? 10:09
masak ab5tract: Perl 6 faces a rather high bar of success, because it is a second system, and the first system is widely deployed.
ab5tract: thus, it needs to be not just better, but (a) *significantly* better, and (b) not significantly worse.
as to (a), I feel Perl 6 has some points of pride now. I've felt that since sometime one-two years ago. 10:10
as to (b), the keywords are "performance" and "stability". we can still improve there, and we should. 10:11
if I sounded offended, it was probably more that I typed while scowling at seeing yet another projection of intricate facts down to a one-dimensional ("mature"/"immature") metric.
moritz adu: not really. But you can have attributes in action objects that carry additional information 10:12
adu oh like $<x> = y
moritz oh, captures
those come in via $/
adu hmm, is that not what you meant?
moritz no, but maybe it's what you meant :-)
you were asking the question 10:13
adu I was also wondering if I could subclass an action class
moritz sure
actions are in no way special 10:14
adu how do I call a superclass's method with the same name?
moritz the typical way is the callwith/callsame/nextwith/nextsame dispatchers 10:15
or you can explicitly call self.YourSuperClass::methodname(args) 10:16
masak did anyone have any opinion about whether irclog.perlgeek.de/perl6/2015-01-10#i_9914995 exposes a bug about enum values exported from many places? 10:17
adu interesting
moritz masak: exportation is currently scoped to compunits
masak: you can consider that a bug or a feature 10:18
masak moritz: that seems orthogonal to what I was asking.
moritz masak: no
masak: let me explain :-)
... or maybe not 10:19
masak moritz: my question, more exactly, is this: if two modules export two different enum values B, should that value be considered "poisoned" in the importing scope?
(rather than fail due to a double import, as it does now)
moritz m: module M1 { enum A is export <B C> }; module M2 { enum X is export <B Z> };
camelia rakudo-moar 8f447f: OUTPUT«===SORRY!===␤A symbol 'B' has already been exported␤»
moritz masak: it doesn't fail as a double import
masak innerestin'
moritz masak: it fails as a double *export*
masak then I see what you mean. I think. 10:20
moritz masak: and that's because exportation is scoped to the compunit
masak yeah, that's not the behavior I wanted to evoke.
hold on.
masak splits the stuff into files
10:20 denis_boyun_ left
moritz and we have no posining for exports 10:20
masak only enum values, IIUC.
10:21 spider-mario joined, denis_boyun joined
masak here: gist.github.com/masak/406d687c3e4e0684c013 10:23
10:23 denis_boyun___ joined
masak by my reading of S12, this should work, but then you're not allowed to refer to just `B`; you have to say `X::B` or `Y::B`. 10:23
(and if you also import a `sub B { ... }` from somewhere, it takes precedence and "un-poisons" the slot for itself) 10:24
moritz that takes slot &B
10:25 denis_boyun left
masak no doubt -- but I'm talking about what happens when you write `B` in code after that. 10:26
if the sub has been imported, that's a call to `B`.
moritz that should resolve to &B, yes
masak to `&B`, sorry.
moritz skims S12 10:27
masak but if it hasn't, then it's still a compile-time error, because it's not clear if you're referring to `X::B` or `Y::B`.
10:29 virtualsue_ joined
moritz Since there is an enum C<OK>, the function C<OK> may only be 10:29
called using parentheses, never in list operator form.
10:29 virtualsue left, virtualsue_ is now known as virtualsue
moritz so you'd be able to call B() when the B slot is poisoned, but not just 'B args', iiuc 10:29
ah yes 10:30
timotimo .tell japhb now the log is filled with lines of Run command exited with exit status 1: install/bin/perl6 --optimize=3 /home/timo/oldhome/timo/perl6/bench/perl6/rc-forest-fire 16 16 inf
yoleaux timotimo: I'll pass your message to japhb.
moritz "If there is
a collision on two enum values that cancels them both, the function
still may only be called with parentheses, since the enum key
is "poisoned"."
masak right.
that's the only reference to the concept of poisoning I've found. 10:31
it mentions the effect in passing, but it's pretty clear about it.
masak submits rakudobug
moritz masak: S12 doesn't talk about in which circumstances poisoning, so I'd assume it happens everywhere
so, also in importation/exportation
masak moritz: ah, you mean between any two imported symbols?
moritz: that's very possible.
moritz masak: no
masak moritz: but enum values are a bit special in that they're essentially short-form aliases for their real unambiguous names. 10:32
moritz masak: I mean between two imported enum items vs. two declared-here enum items
masak moritz: like, `True` is short for `Bool::True`.
moritz yes
masak moritz: oh, that's entirely possible, yes.
moritz: though I could equally well argue that the declared-here enum value takes precedence...
moritz with "everywhere" I meant "at declaration and at {ex,im}portation"
masak declaring something feels like much more of a "conscious act" than importation. 10:37
like, you might not know you're importing a `B`, but you definitely know you're declaring one!
10:40 lumimies joined, IllvilJa left 10:42 IllvilJa joined, denis_boyun___ left 10:43 denis_boyun_ joined 10:45 denis_boyun_ left 10:46 denis_boyun_ joined 10:47 mvuets joined 10:49 virtualsue left
masak finally watches TimToady++'s APW keynote 10:53
El_Che TimToady: The Fosdem Language track schedule is online: fosdem.org/2015/schedule/track/languages/ 10:59
You'll be closing the track. Sunday 13-13:50 11:00
A good time and position
masak \o/ 11:01
El_Che ah nice, there is an other perl talk just before
masak .oO( Perl 6: a last language )
El_Che (which I didn't know noting about)
fosdem.org/2015/schedule/event/des...ry_module/
so the dev room and larry's talk don't collide 11:02
great
11:05 rindolf joined 11:31 adu left 11:33 rurban joined 11:35 mr-foobar joined 12:04 Alina-malina left 12:05 Alina-malina joined
lizmat good *, #perl6 12:07
m: $;1 # trying to do a "useless use of variable" warning but not finding the line number
camelia rakudo-moar 8f447f: OUTPUT«===SORRY!===␤Cannot find method 'orig'␤»
lizmat it would seem that .orig is not being set in the case of an unnamed variable 12:08
does anybody have any clue where that would need to be set ?
dalek kudo/nom: 7ad8639 | lizmat++ | src/Perl6/Optimizer.nqp:
Improve 'Useless use' warning for unnamed vars
 12:09
lizmat ^^^ does not fix the problem, will just produce a better warning when the problem is fixed
FROGGS_ I'd leave the word 'variable' in the warnings 12:12
masak lizmat+ # there's another '+' waiting for you ;)
lizmat :-)
FROGGS_ *g*
lizmat m: class A { has $!a; method a { $!a; 42 } } # why no warning here, FROGGS_ masak ? 12:13
camelia ( no output ) 12:14
12:14 konsolebox joined
lizmat I think showing the sigil with the name, implies variable, no? 12:14
are there contexts where parameters should give that warning? 12:15
FROGGS_ std: sub foo($a) { }; foo(42)
camelia std f9b7f55: OUTPUT«Potential difficulties:␤ $a is declared but not used at /tmp/pOnBFaMNSA line 1:␤------> sub foo(⏏$a) { }; foo(42)␤ok 00:01 139m␤»
FROGGS_ sort of
lizmat std: $;1 12:16
camelia std f9b7f55: OUTPUT«ok 00:00 135m␤»
lizmat std: $b;1
camelia std f9b7f55: OUTPUT«===SORRY!===␤Variable $b is not predeclared at /tmp/UOGOkK6nT5 line 1:␤------> <BOL>⏏$b;1␤Check failed␤FAILED 00:00 135m␤»
lizmat std: my $b;$b;1
camelia std f9b7f55: OUTPUT«ok 00:00 135m␤»
lizmat hmm... we seem to be more strict than std :-)
dalek kudo/nom: 09c5420 | lizmat++ | src/Perl6/Optimizer.nqp:
Add variable
 12:17
kudo/newio: 76afb6c | peschwa++ | src/vm/jvm/runtime/org/perl6/rakudo/RakudoJavaInterop.java:
Static classes won't have a constructor; honor that.

This commit fixes generation of constructor dispatch to not generate any dispatch code if we don't have constructors in the first place.		 12:18
kudo/newio: 28429ec | peschwa++ | / (2 files):
Correct marshalling of listy types to Java.

  raiph++ for pointing me at this.
kudo/newio: 8f447f8 | FROGGS++ | / (2 files):
Merge pull request #350 from peschwa/jvminterop

More JVM-Interop improvements
kudo/newio: 7ad8639 | lizmat++ | src/Perl6/Optimizer.nqp:
Improve 'Useless use' warning for unnamed vars
kudo/newio: 09c5420 | lizmat++ | src/Perl6/Optimizer.nqp:
Add variable
kudo/newio: 5c5a060 | lizmat++ | / (3 files):
Merge branch 'nom' into newio
ecs/newio: 48af140 | TimToady++ | S02-bits.pod:
Mention unspace disallowed in regex (except \#)

The current engine disallows unspace in regexex as too ambiguous. This is fine for actual whitespace, but is counterintuitive when people try to backslash the # character. So \# is now specifically allowed to match a literal # character.		 12:19
ecs/newio: 518438a | (Stéphane Payrard)++ | S99-glossary.pod:
laziness and other entries
ecs/newio: 1eee389 | lizmat++ | S (2 files):
Merge branch 'master' into newio
12:23 rurban left 12:25 ggoebel111111112 left 12:30 leont joined, kjs_ joined 12:35 Alina-malina left 12:37 ggoebel111111112 joined
dalek kudo/nom: 58ae622 | lizmat++ | / (4 files):
Remove unneeded stub
 12:41
FROGGS_ lizmat: all warnings/exceptions talk about "variable $foo"... and I'd not imply anything, a newby that does not know the idea of sigils well might be more confused than needed
masak lizmat: re parameters: I used to gnash my teeth at Niecza warning about all kinds of unused paramters. 12:45
lizmat: it's my perogative as a programmer to name a parameter in my routine, and then not use it.
lizmat: in some way, that's a kind of documentation too (and may happen easily, especially with multis), and it happens *often* as one goes through quick refactor cycles. 12:46
it feels silly to warn about it.
FROGGS_ well, I could imaging to only warn about that in a pedantic mode
masak IntelliJ IDEA strikes the perfect balance here, if you ask me.
FROGGS_ because, that's what it is
masak it colors unused declarations grey. 12:47
that way, people who don't care just blissfully tune it out, easily.
and people who do care pick up on it, and handle it when they feel like it.
...I don't quite know how to express this yet, but a lot of things that we think crudely of as "warnings" and "errors" right now will distribute themselves over the next 20 years or so into slightly different buckets all having to do with different levels of notifications inside an IDE like IntelliJ. 12:48
arnsholt And sometimes an API requires you to include a parameter you don't actually need
masak the notion of an "abstract interpreter" constantly checking your code while you write it is just too good not to win. 12:49
12:49 rurban joined
arnsholt CL lets you say "nevermind that this variable is unused" 12:49
masak arnsholt: yes, but that argument is weaker in Perl 6, since you can always make a positional anonymous.
arnsholt Troo
12:50 vendethiel joined
dalek kudo/nom: 574192c | ab5tract++ | src/core/set_operators.pm:
on our way to first class hashes
 12:54
kudo/nom: a5a1832 | ab5tract++ | src/core/set_operators.pm:
baggy handling of ops now takes mixes into account
kudo/nom: 671bbcc | ab5tract++ | src/core/set_operators.pm:
the mix/bag ops should convert arguments to Bag in cases where no Mix is present
kudo/nom: 3f625fb | ab5tract++ | t/spectest.data:
Add the mix test to the spectest list
kudo/nom: f7bf2fb | ab5tract++ | src/core/set_operators.pm:
Implement a few performance enhancements as suggested by lizmat
kudo/nom: bcea51a | ab5tract++ | src/core/set_operators.pm:
While in the vein of performance, remove some junctions
kudo/nom: 13d9afe | lizmat++ | / (2 files):
Merge pull request #351 from ab5tract/555/quanthash_ops

Get Mixes working properly with the QuantHash ops
FROGGS_ masak: would a trait be an option? 12:58
'is unused'
12:58 Alina-malina joined
FROGGS_ is good for documentation too 12:58
masak FROGGS_: my point is, to me it doesn't feel like the user should have to defend her declaration of something unused *at all*. 12:59
I've never felt helped by such a warning. only annoyed.
more broadly, I am against, and will keep being against, any error or warning that assumes the programmer is a beginner rather than a power user. 13:03
that's ultimately why I have a beef with `say;` being erroneous, too.
it would be darn useful for it not to be!
lizmat .tell ab5tract seems t/spec/S03-operators/mix.t is missing from roast ? 13:04
yoleaux lizmat: I'll pass your message to ab5tract.
lizmat m: say() # 2 extra characters to show that you really mean it 13:05
camelia rakudo-moar 8f447f: OUTPUT«␤»
lizmat possibly, the error should mention the use of a heredoc 13:06
.tell ab5tract also, some Set/Bag interaction with (-) seems wrong: say (<a b c>.Set (-) <a b>.Bag).perl -> ("c"=>1).Bag 13:12
yoleaux lizmat: I'll pass your message to ab5tract.
13:15 brrt joined, sirdancealot left 13:22 sirdancealot joined 13:25 fwilson left 13:26 yeahnoob joined, yeahnoob left 13:27 yeahnoob joined, yeahnoob left 13:28 yeahnoob joined, yeahnoob left, n0den1te joined 13:29 yeahnoob joined, yeahnoob left, yeahnoob joined 13:30 yeahnoob left, yeahnoob joined, yeahnoob left 13:31 yeahnoob joined, yeahnoob left 13:32 yeahnoob joined, [Sno] joined
dalek kudo/nom: 2aef1f7 | lizmat++ | src/core/set_operators.pm:
Make (<+) and (>+) 2x faster in worst case
 13:35
13:39 yeahnoob left
FROGGS_ masak: I agree with you about 'say;' 13:40
moritz \o 13:41
masak well, the pattern is the same in both cases: the warning or error is "this *could* be wrong, so I won't allow you to do that"
when there are perfectly legitimate cases for an experienced user to do that thing.
FROGGS_ hi moritz
n0den1te m: say; 3; 13:42
camelia rakudo-moar 13d9af: OUTPUT«===SORRY!=== Error while compiling /tmp/sL_TNzIL0W␤Unsupported use of bare 'say'; in Perl 6 please use .say if you meant $_, or use an explicit invocant or argument␤at /tmp/sL_TNzIL0W:1␤------> say⏏; 3;␤ ex…»
FROGGS_ masak: but '$foo is unused' also makes perfectly sense... imagine you read a foreign code you have to maintain
unused params always raise the question if this was intentional or just forgotten to take into account 13:43
13:44 kjs_ left
masak FROGGS_: the compiler shouldn't second-guess the programmer! 13:44
FROGGS_ no, but maintainable code is important 13:45
13:45 denis_boyun_ left
masak FROGGS_: I don't see the structural difference between "this `say;` *might* be you being confused and meaning `.say;`, so I'm not going to allow you to write that" and "this parameter is undeclared and I don't know if you are aware, so I'm going to warn about that by default" 13:46
n0den1te masak: nor should the programmer second guess the compiler *shudder*
FROGGS_ so when you stub a signature, implement the routine partly and accidently pass your tests is bad... hints are nice in that regard
13:46 denis_boyun joined
masak no, they are not nice. 13:46
FROGGS_ I do like hints if I can satisfy them in either way... see, that's why I *love* clang 13:47
masak if I wanted suggestions for where my code declares unused things which might point to structural deficiencies, I'd use a linter.
13:47 kjs_ joined
masak I don't like warnings that punish me for writing the easiest possible code to accomplish something. 13:48
if I have to sprinkle the code with "#OK" comments to quiet the compiler, I feel like the compiler's errand-boy.
FROGGS_ these #OK comments are weird anyway 13:50
13:51 fwilson joined
FROGGS_ anyway, my motivations for an 'is unused' traits are 1) self documenting code, especially if others read it, 2) tell the compiler what you are up to, which leads to 3) silence that bloody warnings and possible optimizations 13:53
masak I feel you are misguided, since those warnings shouldn't happen in the first place.
FROGGS_ not 'silence optimization' of course :o)
masak: that's your opinion :o) 13:54
masak yes, it is!
masak starts a campaign: "Don't Bother Me While I Program" (DBMWIP)
FROGGS_ I welcome this warning, since I rarely (never) add params I don't use
masak if the compiler emits a warning *just because it can*, then the compiler should shut up.
FROGGS_: that's a narrow-minded view of what happens during programming.
when I code, I copy-paste stuff. I change it on the fly. I compiler half-finished things to see if they work. 13:55
13:55 yeahnoob joined
n0den1te FROGGS_: what masak is saying is, "Don't baby-glove programmers" 13:55
FROGGS_ no, it helps me to do my job
13:55 yeahnoob left
masak in a lot of transition states, there will be unused parameters and variables. 13:55
13:55 brrt left
masak the compiler flagging them up as something I should care about is a bloody annoyance. 13:55
13:55 yeahnoob joined
FROGGS_ I welcome every warning that results in a: "right, I still need to do that part" 13:55
masak if I wanted that information, I'd ask for it. 13:56
FROGGS_ how do you ask for it?
13:56 yeahnoob left
n0den1te turn on warnings? 13:56
masak FROGGS_: those are low-level tasks, not something that should show up during every compile.
FROGGS_: that's why it's more like linting.
13:56 denis_boyun___ joined, yeahnoob joined, [Sno] left
masak I can have a *valid* program with some unused parameters. 13:56
13:57 yeahnoob left, raiph joined
masak I want to know about things that are wrong, not TODO-list things. 13:57
13:57 yeahnoob joined
FROGGS_ I like todo-list things, because I usually work on several places at once, and that helps me keeping track 13:57
13:57 yeahnoob left, denis_boyun left
masak I cannot help that you like this useless warning. there are plenty of other ways for you to remember that you have unused parameters, though. ways that wouldn't bother me while I code. 13:58
13:58 n0den1te left
FROGGS_ masak: you can have valid programs in many different ways, shall we silence everything that could be caused just because the program is not done yet? 13:58
masak FROGGS_: you're twisting my argument. 13:59
13:59 yeahnoob joined, yeahnoob left
FROGGS_ no, I am just using your argument for other bits that match it as well :o) 13:59
14:00 yeahnoob joined
masak just so we don't have to strawman each other, here are some warnings I like: 14:00
FROGGS_ so, I don't mind if I get extra warnings while I am about to add things and recompile... the compiler should take the program as if it was done
masak "redeclaration of variable in scope" -- fine, that was probably a thinko on my part.
vendethiel FROGGS_: in C/C++, what I see most of the time is `void foo(Type /*paramname*/)` 14:01
masak "use of undefined value" -- this (runtime) warning I always pay attention to, immediately.
various deprecation warnings -- nice. and, usually, lizmat++ 14:03
FROGGS_: for every time I commit, I might have compiled my program ~100 times. if I get warnings about unused parameters even 10 of those times, that's a whole lot of noise in my compile output for something I don't really care to know. 14:05
vendethiel
.oO( let's not follow go there... )
(go errors for every unused import, and even variables I think) 14:06
14:07 jluis joined
masak vendethiel: I think it was Go's "error on unused import" that made me start to think about these things more clearly. 14:09
more precisely, there are some possible problematic things that can happen in transitions and during refactors, that I *just don't care* to be errored/warned about.
[Tux] WTF 14:13
6 my int $i = 0; $i--;
m: my int $i = 0; $i--; 14:14
camelia rakudo-moar 13d9af: OUTPUT«Cannot assign to an immutable value␤ in sub postfix:<--> at src/gen/m-CORE.setting:4980␤ in block <unit> at /tmp/dmsr5BPDZQ:1␤␤»
masak today's autopun spotting: twitter.com/davidcoverdale/status/...3574686722
[Tux] m: my int $i = 0; $i = $i - 1; 14:16
camelia ( no output )
14:16 brrt joined
masak [Tux]: there's wide agreement that that should work, but it's difficult to implement. here's (I think) why: usually the $value gets replaced by $value.pred and then put back into the variable's container. but natives have no containers, only the value itself. 14:17
[Tux] I'll go with $i - 1 for now
[Tux] tries to implement all the CSV options, so we at least will be able to SHOW that it works 14:18
moritz smoke.perl6.org/report is now up and running 14:20
thanks to colomon++ and a wee bit of infrastructure by me
masak \o/ 14:21
the only kind of smoke I like!
14:22 brrt left
masak why is the author of 007 listed as "Unknown"? 14:22
moritz masak: did you specify it in META.info? 14:23
14:23 BenGoldberg joined
masak good question. I didn't write the META.info :) 14:24
seems to be in there, though: github.com/masak/007/blob/master/META.info#L4
moritz huh 14:26
smoke.perl6.org/project/007 reports it
(which takes it from ecosystem-api.p6c.org)
github.com/colomon/SmokeResults/bl...ts.pm#L130 this is where the 'Unknown' comes from 14:29
github.com/colomon/SmokeResults/bl...ts.pm#L201 ummm
14:34 ab5tract joined
ab5tract hi lizmat 14:35
yoleaux 13:04Z <lizmat> ab5tract: seems t/spec/S03-operators/mix.t is missing from roast ?
13:12Z <lizmat> ab5tract: also, some Set/Bag interaction with (-) seems wrong: say (<a b c>.Set (-) <a b>.Bag).perl -> ("c"=>1).Bag
ab5tract github.com/perl6/roast/pull/52/files
both addressed in this PR 14:36
for the record, i do not see any sane way to allow LoL semantics for (-) if it is expected to return a Set in the case that a Set is provided first
another set related PR (which TimToady has expressed support for) would explicitly make the set semantics to always promote its arguments to the most complex present 14:41
lizmat: for reference -- github.com/rakudo/rakudo/pull/335 14:50
14:50 brrt joined
ab5tract masak: glad to hear i didn't offend :) 14:52
14:55 _4d47 joined
ab5tract moritz: FWIW, it is strongly recommended by the Dancer devs to only use Dancer2 14:56
(i noticed that SmokeResults is a Dancer 1 app) 14:57
moritz ab5tract: patches welcome :-) 14:59
ab5tract :)
maybe later
moritz (I didn't develop that app, just patched it a bit here and there)
ab5tract: and ask colomon++ first, he'll be the one accepting or rejecting patches
ab5tract sounds like a plan 15:00
moritz: not sure if you want to merge that roast PR or wait for lizmat
15:00 Alina-malina left
moritz ab5tract: I'm not firm at all in the mix/bag department, so I'd like to leave the decisions to those who know it better 15:01
ab5tract fair enough :) 15:02
15:02 berekuk joined
ab5tract feels good to have a patch in core :D 15:03
El_Che as long your patch doesn't results in core files :)
15:03 skids joined 15:05 zakharyas joined 15:06 frew left, sirdancealot left, sirdancealot joined
ab5tract okay, now onto the airplane back to Amsterdam 15:07
15:07 sirdancealot left
ab5tract .tell lizmat check out github.com/perl6/roast/pull/52/files for the mix.t .. and checkout out github.com/rakudo/rakudo/pull/335 for what i think about `Set (-) Bag ~~ Set` ;) 15:08
yoleaux ab5tract: I'll pass your message to lizmat.
masak a new blog post about Second Systems: www.anton-pirker.at/the-big-rewrite-war-story/ -- HN discussion: news.ycombinator.com/item?id=8869608 15:09
15:10 ab5tract left 15:11 FROGGS__ joined, FROGGS_ left, yeahnoob left 15:14 kjs_ left 15:15 sirdancealot joined
moritz github.com/rakudo/rakudo/pull/324 # some more opinions (preferably from rakudo developers) would be very welcome 15:17
15:17 denis_boyun___ left 15:19 broski joined
broski brah 15:19
m: "Bruh".say
camelia rakudo-moar 2aef1f: OUTPUT«Bruh␤»
psch moritz: irclog.perlgeek.de/perl6/2014-10-05#i_9458502 # the NULL was originally jnthn++'s idea, but i might still have implemented it less-than-sensible 15:21
moritz: that's a data point, not really a justification... :)
15:24 KCL_ left 15:25 mvuets left 15:30 Rounin left
moritz m: class broski { method brah { say 'Bruh' } }; broski.brah # could not resist :-) 15:35
camelia rakudo-moar 2aef1f: OUTPUT«Bruh␤»
moritz psch: if jnthn++ endorses the whole idea (and the implementation), I'd like to see some tests for the command line parser 15:36
psch: but I won't ask you to write any before we're confident that it'll be merged at all
psch moritz: S19-command-line-optins has a few tests, but we could probably use more 15:38
moritz psch: those are only very basic, and only for rakudo's interpretation of built-in options, not for the general mechanism
psch 04-negation.t seems rather mechanism-oriented to me 15:39
although the message for the first one seems to be wrong, it's help and not stagestats 15:40
moritz psch: maybe I should rephrase
psch moritz: the parser should probably be tested in isolation though, i think that's what you mean 15:41
moritz psch: the tests in roast are integration tests. But the command line parser is isolated enough to warrant unit tests, IMHO
yes
psch: one more piece of small-scale feedback: why does the new class inherit from HLL::CommandLine::Parser? what functionality does it reuse? 15:42
dalek ast: 926a7fa | ab5tract++ | S0 (2 files):
something like mixes working
 15:43
ast: e9b66c3 | ab5tract++ | S03-operators/mix.t:
getting closer:
ast: 8646069 | ab5tract++ | S03-operators/ (2 files):
a bit closer every day
ast: 1090421 | ab5tract++ | S03-operators/set.t:
I do not see a way around returning Bag, nor do I see a reason why the order of the subtraction should determine it's type (really this needs to be consistent, as already discussed in my PR for sane promotion of set types)
ast: 6df4ae1 | ab5tract++ | S03-operators/mix.t:
Mixes now have first class support in the operators
ast: 50c7fb6 | lizmat++ | S0 (4 files):
Merge pull request #52 from ab5tract/555/quanthashy_ops

Add a mix.t test
15:44 virtualsue joined
psch moritz: iirc the inheritance was for polymorphism reason - less reuse of features but fitting into the same spot 15:44
moritz: i'd have to dig to see if i misthought that though
moritz psch: since the code that uses the CLI parser doesn't have any type constraints, there's no need for inheritance to achieve polymorphism 15:45
(and if it were using type constraints, it would be better to constrain to a common interface role)
brrt spectest has set/bag confusion 15:48
psch moritz: true. apparently the inheritance isn't neccessary. it might be an artifact from trying to extend the current mechanism in HLL::CommandLine::Parser to handle delimited options from before i decided to implement it via a grammar
moritz: removing the inheritance fails with missing method 'stop-after-first-arg' 15:55
moritz: i think that means the interface role should be implemented... :) 15:56
moritz: i'll have a closer look again during next week, don't really have time right now. i'm fairly sure i can clean it up at least a bit
brrt SO3-operators/set.t 15:57
16:00 scsix joined 16:05 jluis left, berekuk left
novice666 m: ++$++ 16:06
camelia rakudo-moar 2aef1f: OUTPUT«Cannot assign to an immutable value␤ in sub postfix:<++> at src/gen/m-CORE.setting:4975␤ in block <unit> at /tmp/AXtqwobfez:1␤␤»
dalek ecs/newio: c1656c4 | lizmat++ | S16-io.pod:
Some elaboration about classes
lizmat brrt: set/bag confusion noted, but I'm not sure what the right answer is there (anymore) 16:07
yoleaux 15:08Z <ab5tract> lizmat: check out github.com/perl6/roast/pull/52/files for the mix.t .. and checkout out github.com/rakudo/rakudo/pull/335 for what i think about `Set (-) Bag ~~ Set` ;)
novice666 moritz: ++$++ OUTPUT«Cannot assign to an immutable value␤
lizmat brrt: my feeling is that we should keep as much information as possible
16:09 Sir_Ragnarok left 16:12 berekuk joined 16:18 Sir_Ragnarok joined, raiph left, broski left 16:20 denis_boyun_ joined
moritz novice666: $++ returns a value, not a variable 16:22
novice666: so you can't prefix-++ it anymore
16:23 berekuk left
moritz psch: well, as I said before, wait with cleanup before you've got a general approval; otherwise it's sunk effort 16:24
psch moritz: right, i'll do it like that :) 16:25
dalek kudo/newio: f3a420f | lizmat++ | src/core/io_operators.pm:
dir(...., :absolute) is no more
 16:31
colomon moritz: just moved SmokeResults to perl6-community-modules 16:35
moritz colomon++ 16:37
colomon: did you see my email about where to sync the results to?
lumimies moritz: So why does that error mention postfix:<++>? 16:41
moritz m: (++$)++
camelia rakudo-moar 2aef1f: OUTPUT«Cannot assign to an immutable value␤ in sub postfix:<++> at src/gen/m-CORE.setting:4975␤ in block <unit> at /tmp/PcNrC5qUma:1␤␤»
moritz m: ++($++) 16:42
camelia rakudo-moar 2aef1f: OUTPUT«Cannot assign to an immutable value␤ in sub prefix:<++> at src/gen/m-CORE.setting:4967␤ in block <unit> at /tmp/epVjkDbOBB:1␤␤»
moritz lumimies: seems neither returns a container
lumimies moritz: I thought ++$ would
lizmat $ perl -e '++$a++' # syntax error at -e line 1, near "$a++" in P5, maybe we need that in P6 as well ? 16:45
moritz lumimies: I thought so too, but it doesn't
16:45 Alina-malina joined 16:48 berekuk joined 16:51 kaare__ left 16:53 telex left 16:54 telex joined 16:59 mayuresh joined
mayuresh p6: say 3; 17:00
yoleaux 6 Dec 2014 18:28Z <psch> mayuresh: github.com/BizarreCake/Arane is the previous project i mentioned. it's written in C++11 though, not C89
camelia rakudo-{parrot,moar} 2aef1f: OUTPUT«3␤»
dalek kudo/newio: a20bee0 | lizmat++ | src/core/ (9 files):
Match end of FILETEST to case of -x file tests

To easier allow extension to uppercase -x file tests
mayuresh cool :) 17:01
17:01 zakharyas left
mayuresh psch: it's worth it, don't stop, do complete it :) 17:01
i am going to be working on p6 interpreter because; 17:02
1. it's a rite of passage for my cs education
masak lizmat: the table at the top of S03 has the "Autoincrement" level as being non-associative, which would mean that `++$a++` is indeed a syntax error.
mayuresh 2. i really admire mr. wall's work
okay, ubuntu just installed updates, needs to restart my computer! 17:03
dang, also time for me to hit bed. 17:04
masak mayuresh: if you want inspiration for an interpreter of a language a little bit like Perl 6 -- but much smaller -- do check out masak.github.io/007/
mayuresh thanks psch :)
masak: i have been told to write a scheme r4rs interpreter first 17:05
then go for any language of my choice
chose perl6. :)
17:05 frew joined
psch mayuresh: that's not my project 17:05
mayuresh psch: oh, sorry
you just pointed it out to me! thanks. :)
alrighty, see you all later. 17:06
17:06 mayuresh left
dalek kudo/newio: 5cce0e9 | lizmat++ | src/core/IOU.pm:
Make error reporting more sturdy, jnthn++
 17:07
colomon moritz: saw it but hadn't yet fully grasped that I needed to do something to make it work. moritz++ 17:09
17:10 slavik1 left, araujo left 17:11 araujo joined, molaf__ left
dalek kudo/newio: 94bd202 | lizmat++ | src/core/io_operators.pm:
Make single mkdir($path,$mode?) return Bool again
 17:19
lizmat .tell jnthn looks like nqp::mkdir on moar doesn't fail if the actual mkdir fails, it does on parrot and jvm 17:25
yoleaux lizmat: I'll pass your message to jnthn.
17:25 slavik joined
b2gills `int` is defined as being the largest integer that runs at full speed. Which means it can be 64-bit on a 32-bit machine ( the Pentium processor may be an example ) or even 32-bit on a 64-bit machine ( I doubt there were any in general use, if indeed they exist at all ). 17:34
17:42 denis_boyun_ left
lizmat .tell jnthn or perhaps I'm not understanding fail() correctly :-( 17:46
yoleaux lizmat: I'll pass your message to jnthn.
geekosaur I think there were some transitional 32/64 RISC architectures, all of course pretty much dead 17:49
colomon is getting completely lost in a maze of p6c.org machines and accounts 17:59
moritz++
18:03 zakharyas joined 18:13 FROGGS_ joined 18:16 Sqirrel left 18:17 FROGGS__ left
b2gills I think there should be a native type that is defined as being the same size as a far pointer. ( most processors now have the same size for far pointers and near pointers ) 18:18
18:24 mvuets joined, Ugator joined
b2gills There are near-pointers, far pointers, maximum full speed integers, bus-width, and architecture. There is not necessarily any correlation among them. The Maximum full speed integers is most likely the same as the largest of the rest of the set, but it can also be smaller. ( It is always the same as at least one of the others I think ) 18:30
18:31 zakharyas left
dalek kudo/nom: 319c9c3 | lizmat++ | src/core/Str.pm:
Make Str.Bool between 2.5x and 4x as fast
 18:32
colomon moritz: I updated the authors file and now it's completely borked. I'm more than a little bit puzzled by this. 18:37
moritz: I'm trying out running a smoke test on hack.p6c.org at this very moment.
b2gills x64 processors start out in 16-bit mode which has 16-bit near pointers, and 32-bit far pointers. To get to 32-bit mode you have to do a 32-bit jump ( typically to the next instruction ). At that point both pointer types are 32-bits, but allows 64-bit integer math. I'm not sure how you get to 64-bit mode but I think it's a similar operation. At that point the pointers and max integer math are all 64-bit. ( I may have been confused about somet 18:39
At any rate 64-bit mode is the only mode in x86 processors where near pointers, far pointers, and the largest full-speed integer size are all the same. 18:41
timotimo cool, lizmat
18:44 brrt left 18:45 scsix left
b2gills ( except for some 32-bit processors which didn't have native 64-bit integers. I think the 80286 was an example ) 18:45
18:53 dj_goku_ left
flussence
.oO( why didn't 64-bit use 32-bit near pointers? We could've avoided that x32 thing entirely... )
 18:54
19:02 dj_goku joined 19:04 rindolf left 19:14 raiph joined, Alina-malina left 19:17 xfix left 19:18 berekuk left 19:21 kjs_ joined
b2gills Because 32-bit and 64-bit x86 processors use virtual addressing, all programs have access to what they "think" is all of the memory of the computer ( possibly to even more memory than is actually installed with page swapping ). There is little point to have different near pointers and far pointers when two pages that seem to be far apart, may in fact be right next to each other in RAM. It also probably simplified the logic in the processor f 19:24
Also since 32-bit had the same size for near and far pointers, it makes little sense for AMD to have added that to 64-bit mode. 19:26
19:28 Sqirrel joined 19:30 konsolebox left 19:37 kjs_ left 19:40 jakesyl_insomnia left 19:42 jakesyl_insomnia joined
moritz 19:42
19:43 [Sno] joined 19:47 sirdancealot left 19:52 kaare__ joined 20:02 FROGGS_ left
dalek ast: c8e2a0c | usev6++ | S12-enums/misc.t:
Add test for RT #123114
 20:02
synopsebot Link: rt.perl.org/rt3//Public/Bug/Displa...?id=123114
20:02 FROGGS_ joined 20:07 Mouq joined 20:15 darutoko left
masak b2gills: in what situations would it be useful to have a native Perl 6 type the size of a far pointer? 20:21
20:39 denis_boyun_ joined 20:48 spider-mario left 20:49 raiph left 20:50 spider-mario joined 20:54 kurahaupo1 joined 20:56 brrt joined
avuserow_ Hi all. Is there any way to get logs for a module's test failures on smoke.perl6.org? 20:58
20:59 mvuets left
avuserow_ I'm trying to get Compress::Snappy to passing, and I'm not able to reproduce the failure locally, so I wonder if its C library isn't installed 20:59
21:03 kurahaupo1 is now known as kurahaupo, [Sno] left
brrt lizmat - seems mkdir fails for me (linux) 21:03
FROGGS_ avuserow_: you'd have to ask colomon
21:03 [Sno] joined 21:05 virtualsue left 21:06 raiph joined 21:09 kjs_ joined
b2gills Masak: for NativeCall on 32-bit x86 processors where `int` should be a 64-bit type 21:11
masak oh, nativecall. 21:12
that makes sense.
21:13 kjs_ left
arnsholt ATM int is int64_t on all platforms, IIRC 21:15
21:15 broski joined
brrt b2gills - i'd argue if possible against a lot of effort for 32 bit x86 21:18
21:21 kjs_ joined 21:22 raiph left 21:31 raiph joined 21:32 kjs_ left 21:33 raiph left
ugexe is 'or' supposed to be required to be on the same line as a closing bracket? try { say 1 } or die; says undeclared routine 'or' if there is a line break before the 'or' 21:38
lizmat ugexe: I guess that's a consequence of the fact that a closing curly brace at the end of the line signifies end of statement 21:40
vendethiel no, that's why we have semicolons
mmh
21:40 kurahaupo left
FROGGS_ lizmat is right 21:41
m: my $h = { a => 42 }␤say $h
camelia rakudo-moar 319c9c: OUTPUT«a => 42␤»
lizmat m: try { say 1 }; or die # same error
camelia rakudo-moar 319c9c: OUTPUT«===SORRY!=== Error while compiling /tmp/TAQhWubnjA␤Undeclared routine:␤ or used at line 1. Did you mean 'on'?␤␤»
FROGGS_ m: try { say 1 }\␤or die 21:42
camelia rakudo-moar 319c9c: OUTPUT«1␤»
dalek kudo/nom: 98b23d6 | lizmat++ | src/core/Str.pm:
Make smartmatch on Str about 40% faster
FROGGS_ you can unspace it though if you want
ugexe aha cool 21:43
FROGGS_ m: try { say 1 } \␤or die # I'd probably put another space there... I do like whitespace :o) 21:44
camelia rakudo-moar 319c9c: OUTPUT«1␤»
b2gills brrt: You're assuming that's the only architecture that we may want to support where the pointer size is something other than the largest full speed integer (`int`)
broski i have some questions, may sound total noobe but I'm curious 21:45
brrt no, i'm assuming that if we want to get 6.0 out the door, supporting umpteen different architectures while everybody's running x64 may be a distraction :-)
although the fact that we're talking about it is a pretty good sign 21:46
b2gills ^
broski: fire away :-)
broski :D
timotimo heyo brrt
brrt \o timotimo
broski brrt: wondering, why are there choices for different backends
brrt you on #moarvm? could you possibly comment on the repr op devirt issues
that's a good question broski :-) 21:47
basically we distinguish between perl6-the-language and perl6-the-compiler
FROGGS_ broski: rakudo started with only one backend (Parrot), but the goal was to support several, like the JVM to get a foot in that market
brrt and there have been many compilers 21:48
timotimo today i've been wondering, idly, about how to build stuff like "first we look for this constant string that we know follows after a known-length regex, and then we skip back a bit and match the other stuff in the regex"
21:48 Alina-malina joined
FROGGS_ broski: and when Parrot still was too slow and now sign of major improvements were seen, MoarVM was started to please rakudo's needs 21:48
brrt basically, each backend has a different stability / feature profile 21:49
broski i see
Mouq timotimo: I was also wondering last week if there was an opt we could do, where if we see /…foo$/ we could just skip to the end of the string
FROGGS_ broski: you could watch this: www.youtube.com/watch?v=XgPh5Li3k4g 21:50
timotimo ah, yes indeed
Mouq: if you want, go ahead ;)
FROGGS_ broski: also, here is a nice collection: www.youtube.com/watch?v=YW75CXY-P3...FwobZHdXdK 21:51
21:51 donaldh joined
Mouq is still trying to figure out QRegex and repr AST :P 21:52
timotimo "repr AST"?
Mouq s/AST/NFA
timotimo oh, hehe.
Mouq Too many acronyms man
timotimo TMAM
Mouq :9
timotimo well, as long as you don't invent new types or subtypes for regexes, you don't need to do anything with the NFA generator/runner 21:53
broski those are great resources, thanks, they are bookmarked
^^ FROGGS_:
FROGGS_ :o)
donaldh In NQP, is it possible for a role's method to call the method it replaces on the class the role gets applied to? 21:54
Mouq timotimo: I do think it would be much less misleading if we called QRegex::NFA QRegex::NFAGenerator or the like
org.perl6.nqp.qregex.nfa.factory 21:55
donaldh Specifically, QAST::Children.shallow_clone replaces QAST::Node.shallow_clone and I'd like to call the original.
timotimo that's in nqp? 21:56
donaldh yep 21:57
timotimo i know in perl6 you can specify precisely by putting the name of the class you want into the method name part
not 100% sure what it looks like, haven't ever used it so far
Mouq foo.Class::bar, isn't it? 21:59
donaldh so maybe self.QAST::Node::shallow_clone() ? 22:00
timotimo that's what i thought
22:03 raiph joined 22:04 lue left 22:05 kurahaupo joined
dalek kudo/nom: 04ce883 | lizmat++ | src/core/Str.pm:
Make Int ~~ "a" silent and faster
 22:08
22:08 berekuk joined 22:11 jack_rabbit joined
dalek kudo/newio: 574192c | ab5tract++ | src/core/set_operators.pm:
on our way to first class hashes
 22:12
rakudo/newio: a5a1832 | ab5tract++ | src/core/set_operators.pm:
rakudo/newio: baggy handling of ops now takes mixes into account
22:12 dalek left
lizmat sorry dalel 22:13
*dalek
22:13 dalek joined, ChanServ sets mode: +v dalek
dalek ecs/newio: 0e734c4 | lizmat++ | S32-setting-library/IO.pod:
Got as far as mkdir()
 22:14
22:18 lue joined
FROGGS_ donaldh: in nqp you'd do: nqp::findmethod(...)(args) 22:18
donaldh: grep for that
timotimo good point 22:19
FROGGS_ the foo.class::routine doesn't work in nqp AFAIK
donaldh no it doesn't :) 22:20
FROGGS_ and using findmethod you can provide an object and a type
22:22 virtualsue joined 22:24 kurahaupo left
donaldh Hmm. findmethod only takes an object and a method name 22:24
timotimo does findmethod return the invocable object or just the "slot number" or something? 22:25
22:31 colomon left, jbotz joined
donaldh I suspect I am discovering why the code is the way it is. 22:32
jbotz hello...
newbie here...
masak welcome.
jbotz i'm studying, and I'm curious about something... 22:33
what should be the value of a for statement?
masak depends a little.
jbotz i.e. should "say for 1..5" output anything?
masak that doesn't mean what you think it means 22:34
jbotz I don't think anything...
lizmat good night, #perl6!
jbotz but I saw this in the perl docs...
timotimo correct, this is a case of an "after statement" for loop
jbotz sub divisors (Int $n) { $_ if $n %% $_ for 1..$n } 22:35
brrt good night lizmat
timotimo so it'll just do "say" the 5 times
masak except it's disallowed :/
jbotz oh, got it
timotimo usually the result of a for loop is the list of what the last statement inside the for loop's block gives you
22:36 kjs_ joined
jbotz ok, but the sub divisors doesn't work in rakudo 22:36
masak no, it doesn't
need parens around that statement
used to work
spec changed
timotimo good to know
jbotz ok, got it... with parens it works! 22:37
timotimo please provide a link to where you found that so we can fix it
raiph m: for Array.new(1,2), Buf.new(1,2) { say .WHAT } # rakudobug? specbug?
camelia rakudo-moar 04ce88: OUTPUT«(Int)␤(Int)␤(Buf)␤»
timotimo m: say Array ~~ Iterable; say Buf ~~ Iterable
camelia rakudo-moar 04ce88: OUTPUT«True␤False␤»
timotimo m: say Array ~~ Positional; say Buf ~~ Positional 22:38
camelia rakudo-moar 04ce88: OUTPUT«True␤True␤»
raiph a Buf is a role, whereas an Array is a class
timotimo so ... ?
masak raiph: for all I know it's correct
jbotz timotimo: doc.perl6.org/type/Signature
the part about type constraints 22:39
masak jbotz++
timotimo heh. in a totally unrelated chapter of the docs it b0rks :\
it's hard to keep all the docs up to date to every change :( 22:40
anyway, it's bedtime for me now
good night :)
raiph timotimo: I was asking why Array does Positional as I'd expect but Buf doesn't, even though it `does Blob` which `does Positional` 22:41
22:41 kjs_ left 22:44 BenGoldberg left 22:54 kurahaupo joined
brrt TimToady's talk has been moved to sunday? 22:54
22:56 BenGoldberg joined
masak 'night, #perl6 22:56
b2gills jbotz: `.say for 1..5` is another way of writing `for 1..5 { .say }` ( copied from Perl5 ) 23:00
The `.say` is short for `$_.say`
or `say $_` 23:01
23:02 brrt left
b2gills jbotz: That sub could have been written: `sub divisors (Int $n){ (1..$n).grep( {$n %% $_} ) }` 23:05
broski just read about junctions 23:19
m: sqrt(4|9|16).say 23:20
camelia rakudo-moar 04ce88: OUTPUT«any(2, 3, 4)␤»
broski nice
23:21 telex left 23:22 telex joined
broski I have a question! 23:28
23:33 spider-mario left 23:34 broski left
FROGGS_ broski: just ask... someone will reply, even when not immediately 23:43
23:50 jbotz left 23:53 Ugator left 23:58 denis_boyun_ left

Powered by Cro and the Raku® Programming Language.

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

Thank you!