»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg p6eval perl6: ... | irclog: irc.perl6.org/ | UTF-8 is our friend!
Set by sorear on 4 February 2011.
00:30 sudokode joined 00:33 mcglk left 00:34 mcglk joined 00:36 sudokode left 00:40 sudokode joined
[Coke] sorear, masak: yay, I could that on irclog just before boarding. ;) (+%) 00:44
00:44 sudokode left
[Coke] Now the last bug remaining is trying to find the "compreg" 00:45
d Partcl compiler.
00:50 tokuhiro_ joined 01:00 flexibeast left 01:04 FROGGS_ joined 01:06 cjbot left 01:08 FROGGS left
sorear [Coke]: boarding? 01:08
01:15 spider-mario left 01:25 jeffreykegler left, jeffreykegler joined 01:30 broquaint left 01:31 sudokode joined 01:39 am0c_ joined 01:46 mcglk left 01:47 mcglk joined 01:57 mcglk left 01:58 mcglk joined 02:07 dbr left
[Coke] sorear: yes, I've been talking to you from hotels and airports. 02:17
[Coke] is now at his final destination for the day.
02:18 cjbot joined 02:27 tokuhiro_ left 02:31 whiteknight left 02:35 mucker joined 02:48 thelazydeveloper left 02:50 jeffreykegler left 03:02 am0c left 03:04 am0c_ left 03:15 adu joined
[Coke] er, not "I could that", but "I caught that" 03:27
03:30 mucker left 03:31 MikeFair_ joined
MikeFair_ sorear: hey ther 03:31
sorear yo.
MikeFair_ sorear: Actually I was on #perl6 on the Perl Network
sorear: no matter though, all here now :)
sorear: So you were saying (and I was doing my best to keep up) :) 03:32
sorear mm, it that a thing? 03:33
MikeFair_ Guess we're just a bunch of confused souls over there not realizing we belong here? 03:35
sorear I think most of them joined so that they could tell off other people who join
but in doing so they've prevented the channel from expiring
MikeFair_ Well I just left (and then there were 9) 03:36
:)
There are two ways to interact with the GAC: the Global Assembly Cache Tool (gacutil.exe) and the Assembly Cache Viewer (shfusion.dll). 03:39
sorear: Does that help?
sorear 1. it needs to be problematic 03:40
2. the mono gac layout is such that it cannot be quickly scanned on Linux, unless there's a database somewhere I'm missing
s/problematic/programmatic/
MikeFair_ 1. totally agreed, 2. Mono has to load them, how does it do it? 03:42
03:42 bubaflub left
MikeFair_ sorear: I guess I'm just not seeing why you have to scan? Are we building a bunch of *.pm files? 03:42
sorear when mono or mcs wants to load an assembly from the gac, it *knows the name*
oh, I thought you were recommending a scan 03:43
Note
You must have administrator privileges to use Gacutil.exe.
diakopter /o\
MikeFair_ sorear: I wonder what shfusion.dll is like...
sorear it sounds like it's probably related to shell32.dll 03:44
anyways, i don't think a gac scan would buy much at all
MikeFair_ sorear: And one thing I've never known is whether or not the GAC is literally a Cache (like a local copy perhaps that's stored after a JIT has processed it) or is it the actual installation location 03:45
sorear it is an installation location
it does not do line-fill/eviction like most "cache"s
MikeFair_ that location is known at run-time? 03:46
sorear huh? 03:47
MikeFair_ sorear: Perhaps I'm missing something, you were saying that you didn't know the name of the assembly when Perl tried to load it
sorear MikeFair_: I think at this point I don't understand the problem you are trying to solve 03:48
Currently the syntax is CLR::("assembly name")::Namespace::Class
MikeFair_ You were having some crazy ideas on how to redo niecza's name binding mechanism so 'use gtk:from<clr gtk-sharp>' Just Works 03:49
And I thought, likely mistakenly, finding out what assembly to load was a problem 03:50
sorear MikeFair_: No, the problem is with niecza's current insane semantics for scoping of nested names 03:51
MikeFair_ I was also saying that the System.Reflection classes could help with creating an auto-binding
sorear on the CLR side, the problem is solved
sorry 03:52
MikeFair_ ahh ok
So if i'm hearing you correctly, niezca currently needs to be told about an assembly because dynamically finding it based on the namespace is problematic due to scoping for nested names? 03:54
sorear no. 04:04
it is _by design_ that you need the assembly name 04:05
04:05 colomon left
sorear assembly names are the mechanism by which the CLR tames type name conflicts 04:05
04:06 Telgalizer joined
MikeFair_ sorear: ok.... and the GAC for being able to install the same named thing multiple times for different versions 04:07
sorear MikeFair_: the GAC is a weird pariah thing
it's mostly there to hold copies of System.dll et al 04:08
anything third-party should go into your application directory, the CLR looks there for assemblies
MikeFair_ sorear: I guess, I see it as how they can install .Net 2.0 and .Net 3.0 on the same box and have each directory holds its own version of "System"
MikeFair_ nods.
04:09 birdwindupbird joined
MikeFair_ sorear: Why isn't the namespace sufficient for taming type name conflicts? 04:09
sorear well, where it gets interesting is transitivity
say module D comes in two versions, D 1.0 and D 2.0, both of which export a type D.D
module, I mean assembly, B, requires D 1.0 04:10
assembly C requires D 2.0
MikeFair_ sorear: There's a few pages I was reading the other month on when and how you can nest namespaces, mask type names, etc
sorear D versions 1.0 and 2.0 are not compatible
assembly A requires both B and C
this works
it works because of the way the CLR handles class names as subordinate to assembly full names (including version & pubkey) 04:11
Perl 6 is actually supposed to do something very similar
MikeFair_ Right and those get embedded in at compile time in CLR land? 04:12
sorear use Foo:auth<cpan:DCONWAY>:ver<1.0>
yes
MikeFair_ sorear: But in practice I assume everyone just writes use Foo::auth making the construct nearly useless? 04:13
sorear no.
MikeFair_ sorear: Are is it been a adopted and accepted as a best practice?
s/Are/Or
s/is/has/ 04:14
sorear because Foo::auth means nothing unless there is a "use Foo" in scope, and the use statement can define, lexically, an implicit auth/ver
well, it hasn't been adopted mostly because it hasn't been implemented yet :D
MikeFair_ hehe - that could be a problem ;)
Well I guess what you're now saying then is that if you replaced "cpan" with "clr" you could do something similar? 04:15
(Or am I skipping ahead too many steps and ending up in the wrong place)
sorear subtly wrong 04:18
the idea with :auth is that it represents some kind of naming authority
a URI which represents someone or something that issues names 04:19
MikeFair_ nods. "Makes sense"
sorear if cpan:TOM and cpan:DICK decide to use the same module name, there won't be a problem because the authorities are different
presumably cpan:TOM will not use the same name for unrelated modules, if he's right in the head
and cpan: IDs are unique becuase they are centrally allocated 04:20
MikeFair_ right
sorear there is an additional name adverb :from which is more appropriate here 04:21
use Foo means use Foo:from<perl6>:auth(*):ver(*)
MikeFair_ naming conflicts are easy once that gets distributed and there isn't a central naming authority
ahhh
I think I see where you're going now
sorear actually there is one more, :name 04:22
when you say 'use Foo' you are importing a module Foo and you are creating an alias Foo in the lexical scope
usually, you want them to have the same name 04:23
but the lexical alias has to be a valid Perl 6 identifier, and foreign module names don't always oblige
MikeFair_ I can see that
sorear you might do use SWF:from<clr>:name<System.Windows.Forms> 04:25
MikeFair_ Yeah, that's what I was thinking
sorear as for :auth I would ideally do pk:HASHHERE
MikeFair_ Though I'm not exactly clear on the distinction between from and auth
sorear with the usual Perl 6 convention that no auth = any auth is fine, I can't be bothered to check 04:26
moritz \o
sorear o/ moritz
MikeFair_ auth = author? 04:27
which conceptually stands for some unique namespace?
sorear use Cecil:from<clr>:ver<0.9.4.0>:auth<pk:0738eb9f132ed756>:name<Mono.Cecil>
is the most complete syntax 04:28
auth = authority
think like a certificate authority
Larry's standard grammar starts with rammar STD:ver<6.0.0.alpha>:auth<perl.org>; 04:29
s/^/g/
MikeFair_ yeah I was thinking that "auth" might go really well with a domain name or a standards body 04:30
Something that can provide sanity to the names it hosts
Can an assembly have the same version and a different public key? 04:31
04:34 colomon joined
MikeFair_ auth almost seems like it would be the runtime version 04:35
and pk part of the version
well maybe not auth 04:36
use Cecil:from<clr 2.0>:ver<v:0.9.4.0,pk:0738eb9f132ed756>:auth<publisher_name>:name<Mono.Cecil> 04:38
sorear you can't put that into ver 04:41
MikeFair_ bbiab
sorear ver is fixed-format
also, publisher name is not part of the programmatic name of CLR modules
04:46 kaare_ joined 05:07 birdwindupbird left 05:09 birdwindupbird joined
MikeFair_ sorear: Well ok then, that settles that :) 05:10
sorear: auth seems like a great place to put pk :) 05:11
sorear: It's a bummer about version though, kind of prevents coming up with a better way to track versions in the future :) 05:12
sorear: Anyway, regardless of that I can't help but remember how excited I was to read Larry's Apocolypses on what Perl6 was to become, I was thinking so many of the same things at the time I read those 05:14
sorear: a fixed version format seems a trivial price to pay for the rest of all this goodness. :)
sorear It's linguistically defined so it falls under the language update mechanism 05:15
If Perl 6.3 changes the version syntax, you can enable the new syntax with use v6.3;
we do a few things differently from the days of the apocalypses. we're much more focused on the present. 05:16
r: my class A::B {}; { my class A::C {}; say A::B } 05:25
p6eval rakudo 196572: OUTPUT«A::B()␤»
sorear jnthn: low-priority ping 05:27
05:41 crab2313 joined 05:46 adu left 05:48 Telgalizer left 05:50 cjbot left
MikeFair_ sorear: There's a lot here that I'm wanting to learn 05:55
sorear MikeFair_: in my last p6eval line or more generally? 06:01
whee, I have a built Rakudo
3m26s :D
MikeFair_ sorear: that seems fast 06:02
sorear: more generally
sorear MikeFair_: I started Niecza in large part because in Q2 2010, with the computer I was using at the time, Rakudo builds were more of a 10h thing
MikeFair_ seeing the word class in Perl code without some pointer next to it will take some getting used to ;)
sorear "some pointer"? 06:03
MikeFair_ ->
or ::
sorear what's the connection between class and ::/-> in Perl code?
MikeFair_ sorear: i'm just not used to seeing "class" as a keyword, and the various evolutions in Perl5 (and earlier) have used the module stuff 06:04
sorear: It's totally a misrepresentation on my part, just a keyword that's new to me for perl. That's all I'm really saying. :) 06:05
06:09 erkan left, erkan joined
Timbus i think a more visible difference is using . instead of -> 06:09
MikeFair_ Timbus: Yeah that too 06:11
It's going to be cool
Hey question, i've actually been really impressed with what the .Net stuff has done with enumerations
Anything special in Perl6 regarding enumerations? I mean a lot of it probably doesn't apply because of how strict .net is, but I thought I'd ask 06:13
benabik r: enum A <a b c>; say A::a 06:15
p6eval rakudo 196572: OUTPUT«a␤»
sorear n: enum A <a b c>; say A::a # long name 06:16
n: enum A <a b c>; say a # short name
p6eval niecza v19-44-g55e895a: OUTPUT«(timeout)[auto-compiling setting]␤»
sorear ok, I broke niecza on p6eval 06:17
MikeFair_ :)
sorear r: enum A <a b c>; say A::a # long name
benabik perlcabal.org/syn/S12.html#Enumerations
p6eval rakudo 196572: OUTPUT«a␤»
sorear b: enum A <a b c>; say A::a # short name
p6eval b 922500: OUTPUT«0␤»
sorear r: enum A <a b c>; say A::a #
p6eval rakudo 196572: OUTPUT«a␤»
sorear ignore b
r: enum A <a b c>; say +A::c # knows its index
p6eval rakudo 196572: OUTPUT«2␤»
sorear r: enum A <a b c>; say A::c.WHAT() # knows its type 06:18
p6eval rakudo 196572: OUTPUT«A()␤»
sorear r: enum A <a b c>; say A::c.perl # long name
p6eval rakudo 196572: OUTPUT«A::c␤»
MikeFair_ ok that's cool
sorear r: enum A <a b c>; say +A("c"); say +A(1); # niecza supports using enums as coercers, but niecza is broken right now, let's see if rakudo works too 06:19
p6eval rakudo 196572: OUTPUT«use of uninitialized value of type A in numeric context in block at /tmp/9dazyzIKWV:1␤␤0␤1␤»
MikeFair_ ok, so there's this one other thing I've been thinking about lately about replacing the base types of int/float/string with infinite enumerables 06:20
the main reason I've been thinking about it has to do with I/O and internationalization
sorear what do you mean? why?
MikeFair_ So let's take int for a second
1 2 3
sorear int isn't infinite, anyway 06:21
MikeFair_ Well numeric
ok string
it's not as easy to explain with string 06:22
sorear be advised that type names in Perl 6 are case sensitive and 'int' and 'Int' are entirely different things
Int is a reference type which points to an integer of arbitrary size
int is a CPU-level integer
MikeFair_ Ok let's talk Int then 06:23
the idea simply is that the string characters used to represent a number exist inside a character set
and in different langauges even use different symbols
and to some extent, the same number can be represented in different basis 06:24
(well they most definitely can it's just not as common to speak about Int outside of base10) 06:25
sorear I have a bad feeling about this.
MikeFair_ The think I was considering was if there was an enumerable type to line up with the character set or locale, it could mapped back to the actual "Int" 06:26
sorear What would be the point?
MikeFair_ The same number could be expressed by casting it through its enumerable
It creates a presentation layer between the machine and the human 06:27
but it makes it rather transparent to the programmer
sorear dear #perl6, I am having trouble coming up with a tactful reply 06:28
MikeFair_ sounds good
thanks for listening :)
sorear so I'm not going to say anything at all
MikeFair_ withdraws his comments. :)
benabik I'm not following completely, but skimming it looks like "all we need is another layer of indirection". 06:29
MikeFair_ the idea was that my application could have a locale, which would implicitly select enumerables that knew how to express the concept of that number in a locale specific way 06:30
sorear enumerables are for things you can enumerate. 06:31
MikeFair_ benabik: It's actually more like, We have this layer of indirection that we do poorly or not at all, and maybe there's a better way to do it
sorear: You can enumerate Integers
sorear if you want to add a type coercion from Str to Num, say _that_ instead
huf is this for code or for formatting numbers in a locale-specific way in output?
or do you want strange integer literals that change their meaning depending on locale? or ... i'm lost 06:32
MikeFair_ huf: they don't change their meaning, just their representation 06:33
huf ?
sorear MikeFair_: did you come from [email@hidden.address]
MikeFair_ huf: it's primarily for input and output with an end user
sorear: umm, no, I take it that's not a compliment. ;)
huf so this is just serialization
i thought this was about internal representation or literals... 06:34
MikeFair_ it's something that crossed my mind when reading about enumerables in .Net that are programmaticly defined
Like fibonacci
Doing locale specific representation and input 06:35
sorear MikeFair_: I think you are making the mistake of thinking IEnumerable and System.Enum have anything whatsoever to do with each other
MikeFair_ so yeah it's more about serialization
sorear: likely
sorear: plus I'm talking about it in Perl6
sorear: So yeah, I'm fairly easily convincable that I should be locked up :) 06:36
sorear This is, I am sorry to say, not a p6-ready idea. We've been shipping compilers for several years now and there will be no modifications to the core language unless you can 1. explain why they are needed in a way convincing to the people whose production code you've just broken 2. implement it in at least one shipping compiler
MikeFair_ sorear: Totally agreed, I'm not asking for changes, god no, and this _should be_ doable in the existing language 06:38
sorear: Or not do it
sorear: And it shouldn't break any existing code
sorear Then I don't think I understand you
06:38 araujo left
sorear It seems like you want to rip out Str and replace it with a gigantic enum 06:38
MikeFair_ sorear: More like Str->enum->Int 06:39
sorear: If code chooses to do that
sorear for a purpose having ... something ... to do with locales
MikeFair_ Yes, to create a distinction between "how we present and receive" a number as separate from the number itself 06:40
sorear well, in an hour or so the #perl6 core folks from Germany, Poland, Norway, and Sweden will be up and we'll be able to have a proper discussion about locales 06:41
MikeFair_ Another use case is for unit s convestion
err conversions
huf so instead of (pseudocode) int parse_int(str s), you'd like int parse_int(str s, enum map) where the default map is an infinitely large enum that maps "0" => 0 and so on?
but you can replace it to "balin" => 0, "dwalin" => 1, .... ? 06:42
MikeFair_ huf: Yeah, and "infinitely large" to me means "some code does the conversion"
huf: More like "zero"=>0, "one"=>1 06:43
sorear MikeFair_: What advantage would this have over just calling the correct codein the first place?
MikeFair_ And my code could receive "one hundred thousand and sixty" and the number 1,060 could end up in the Int 06:44
sorear FOUR TWO NINER FOUR NINER SIX SEVEN TWO NINER FIVE 06:45
sorear ducks
MikeFair_ sorear: That would be another enumerable type that could be defined 06:46
sorear: hehe :)
sorear I recognize that this is not a valid form of argument, but INTERCAL has had this feature for decades. 06:47
MikeFair_ sorear: That's interesting.... I didn't know that
MikeFair_ feels less insane.
Not that any of us would, but we could use a "Roman Numeral" representation 06:51
Or someone who wanted to have a Roman Numeral calculator
sorear waits for MikeFair_ to find and read a definition of INTERCAL
I would prefer something like Lingua::EN::Number.parse("one")
it's just a function call, no need to make it fancy 06:52
also Lingua::EN::Number is just a module, so someone other than me can write it
MikeFair_ Create the enumeration type to do the conversion, then take any perl calculator program and set the new interface locale which uses that representation for I/O
sorear: that sounds similar to what I'm talking about 06:54
06:54 araujo joined, araujo left, araujo joined
MikeFair_ The idea is that I have no idea what the end user needs to understand the numbers on the screen 06:54
sorear What do you mean by I/O? 06:55
What if the calculator program uses a GUI?
What if the calculator program outputs to HTTP? Content-Length: MLMVII isn't going to fly. 06:56
MikeFair_ sorear: by I/O I mean the act of taking a character set and converting it to an Int and vice versa
huf that sounds like two steps
reading in bytes, taking a charset and getting a string
sorear We don't use character sets here.
huf and _then_ figuring out what int that string might represent 06:57
MikeFair_ huf: yes, this is more about the "figuring out what that string meant" and "outputting the meaning of the Int" 06:59
sorear Anyways we got rid of $" outright so I can't see us _adding_ a globabl variable that controls the stringification of Int
huf but that's not I/O...
MikeFair_ sorear: The environment shell has one
huf: My apologies
huf: What do you call that? 07:00
sorear Str/Int type coercions
MikeFair_ huf: It's such a transparent single thing that rarely gets discussed (at least in the circles I travel in)
sorear anyways, I had a hell of a time getting Niecza to run on tadzik's machine 07:01
because Niecza uses a partially text-based serialization scheme
and Polish versions of .NET have a different and incompatible definition of "number" from the US versions, *even when you don't use locale calls* 07:02
I am not going to make 'print 12345' in Niecza locale-sensitive, and I am not budging on this
MikeFair_ nods. 07:03
sorear I don't beleive in perpetuating cycles of abuse
MikeFair_ Me too, my thought is that a lot of the problem is because there's this unexamined assumption of what "print 12345" means
sorear I could budge on 'say 1234', since say is explicitly intended for human consumption as opposed to protocols 07:04
MikeFair_ sorear: fair enough, what about on the input side
sorear print 12345 means to output the characters "12345
print 12345 means to output the characters "12345"; we define it precisely because it needs to be usable in network protocols 07:05
Perl has no way of telling whether input comes directly from the user or from another program
MikeFair_ sorear: And those are UTF-8 characters? or Unicode
sorear There is no such thing as "UTF-8 characters". 07:06
UTF-8 is the name of a reversable function mapping sequences of Unicode code points to sequences of bytes
If we err on the side of locale sensitivity, code written in the US which does input from other programs will break, often messily, when run in other locales 07:07
MikeFair_ ok, I always just thought of it as an encoding methodology that's different from ASCII and different from Unicode
sorear If we err on the side of consistency code will be less usable but it won't break outright
Unicode is not an encoding methodology.
Some people use Unicode when they mean UTF-16, but they're wrong. 07:08
MikeFair_ Ok, I know that when I'm dealing with character data, if I get the encoding wrong things break 07:09
sorear Perl 6 strings are character data, not byte data, so nothing can break there
Perl 6 also has a Buf type for byte data 07:10
MikeFair_ What is the internal representation of characters? I mean will it take japanese characters?
sorear File handles have an associated encodings which allow them to be dwimmy
Unspecified and undetectable.
It is required to handle all 0x10FFFE Unicode characters.
MikeFair_ sorear: So if Unicode isn't an "encoding methodology" but there are things such as Unicode characters, what do you consider Unicode? 07:11
shachaf If a picture isn't an "encoding methodology" but you can still save it in a file, what do you consider a picture? 07:12
sorear Unicode is a map between 21-bit code points like 0x3080 and abstract characters like 07:13
む
MikeFair_ ok, yeah, I consider "bitmap" an encoding methodology for a picture
you can't put a picture in a file
sorear A character encoding, in general, is an algorithm which transforms sequences of 21-bit code points into (longer) sequences of 8-bit byte 07:14
MikeFair_ you can only put a representation in a file
sorear You're changing definitions in the middle of the conversation. Stop it.
MikeFair_ nods.
Yes, clearly I don't have it as clearly delineated as you do yet
sorear 00:07 < MikeFair_> ok, I always just thought of it as an encoding methodology that's different from ASCII and different from Unicode 07:15
00:07 < sorear> Unicode is not an encoding methodology.
MikeFair_ right
sorear no. I screwed up there.
I should have said first "You mean 'character encoding'"
moritz Unicode is mostly a character repertoire 07:16
sorear yay moritz.
moritz plus some associated semantics
sorear moritz: if I understand the gist correctly, MikeFair_ wants to do something to the effect of { temp $*OUT.number-formatter = &Lingua::EN::Number::format; Some::Other::Module::prints_to_stdout; } 07:17
moritz: as someone who maybe has actual experience with l10n and numerics, what do you make of this? 07:18
moritz I don't really have that 07:19
the way to do it is to subclass $*OUT.WHAT and add custom print/say methods though, I'd guess
07:20 SamuraiJack joined
MikeFair_ moritz: one of the assumptions is that the methods to do it the right way have to be assignable at runtime 07:22
moritz MikeFair_: erm, what? 07:23
methods have to be assignable at runtime?
you can just create the class at compile time, and chose at runtime which one you use
bog-standard OO polymorphism 07:24
sorear MikeFair_: if you haven't already done so, I recommend you look up what INTERCAL is
MikeFair_ sorear: I did. I think I understand that you're saying this sounds like a really "trumped up" idea and is something to be made fun of 07:25
sorear: Obviously I haven't read everything about it yet
moritz: I guess I just considered the routines to convert Str <-> Int weren't "interruptable" 07:27
The main thing I noticed is that programs do "non-US" based thinking and interaction really badly and even the l10n stuff we have is mostly second class and "difficult" 07:29
and yes, I've had to write some, and my conclusion is that "we haven't figured this part out yet" 07:30
The enumerables thing was just a clue that I thought gave me some insight. One of the problems is that the string "12345" and the Int 12,345 are too closely coupled 07:31
moritz MikeFair_: I think the real problem with that l10n is that most code isn't clear about which conversions are user-visible, and which aren't 07:32
and l10n must only change user-visible conversions
MikeFair_ moritz: Well even our code goes into the compiler as strings. It's "user visible" to the developer 07:33
moritz I didn't mean the developer with "user" 07:35
MikeFair_ moritz: I was dawning on me that the concept "12,345" stored in an 'Int' could a bazillion different representations
s/I/It
especially if you start taking into account what the units are as a second part of the 'Int' (like meters, widgets, dollars) 07:37
That's where the enumerated types started showing up.... Maybe they should be full blown classes 07:38
Regardless, you guys have been great at giving me feedback on this here 07:40
It's not a totally baked suggestion yet, and like sorear showed me, I don't have my terminology sorted out yet 07:42
07:42 SamuraiJack left
dalek d: d93fffe | larry++ | / (5 files):
checkpoint STD_P5 mods in case of a bus

currently parsing 831 lines of viv		 07:43
TimToady the basic principle is that we don't like accidentally generic code, but you can have intentionally generic code if you declare it so 07:44
most L18n turns out to be accidentally generic
07:44 SamuraiJack joined
TimToady so we tend to dislike locales intensely 07:45
TimToady goes to see the parade &
MikeFair_ TimToady: That's interesting... My experience was that it was maddeningly abstract because everything had to be done explicitly 07:46
TimToady: I can appreciate not wanting anything to happen by accident. 07:47
TimToady: I haven't quite decided if I like locales yet or not. I clearly appreciate the problem they are attempting to solve and think it's a decent first attempt approach at it. 07:49
sorear: And for what it's worth, I'm not about to ask you to rewrite anything. It's my itch to scratch if I feel so strongly about it right. :) 07:50
Thanks guys, good night! :) 07:51
sorear: So I couldn't resist following up on Intercal before I went to bed. :) 08:00
sorear: It's pretty hilarious, and I clearly described almost every single one of its idiosyncrocies 08:01
sorear: That's funny
gnight 08:04
sorear gnight
masak I can see the use for localization. much of the time even s/use/requirement/. 08:16
(morning, #perl6)
08:17 popl left
masak but I also think I agree about overgeneralizing it being a bad idea. 08:17
i18n libs probably become one of those things where you use <10% of the provided functionality, and the rest is dead weight.
sorear good morning masak. 08:18
08:18 Psyche^ joined
sorear in sweden yet? 08:18
masak sorear: morning. yes, been in Sweden since late wednesday evening. :) 08:19
08:19 Patterner left, Psyche^ is now known as Patterner
masak sorear: how was the flight back? was the jetlag going west easier to take than the one going east? 08:20
sorear I don't think I particularly experienced it either way 08:22
I went to sleep quite early Fri night, but I think that was a function of not sleeping ~at all Thu night 08:23
masak oh, good to hear.
sometimes jet lag can be quite bad, but it seems you didn't have any of that, then.
sorear Probably the best way to put it is that jet lag was the least of my problems 08:26
shachaf always finds GMT+2 -> GMT-8 much easier than GMT-8 -> GMT+2 08:27
sorear also, when I arrived in GMT+2 I landed in an extremely regimented environment 08:28
Liz and Wendy basically planned out all of our waking hours in advance
masak shachaf: it seems to vary from person to person. I find going west slightly bothersome, because I *really* need sleep when I get there. 08:31
moritz just limit your travel speed to what you can do by foot (or horse or ship or bycicle) 08:32
sorear wonders if you can get jet lag on foot deep in the Arctic 08:33
jnthn or anyone else who is interested: I am toying with some ideas for changing lexical package handling in Niecza and I'd like feedback 08:35
masak is interested 08:42
moritz www.pm.org/ has a new design
I am also interested, but I don't think I can provide the feedback level that sorear++ seeks 08:43
masak oh, I don't think so either. still, I'm willing to listen. maybe I can be the teddy bear outside the sysadmin's office ;)
FROGGS_ ya, just listen and nod ;o) 08:44
masak nods
08:44 FROGGS_ is now known as FROGGS
FROGGS I'm interested too btw, but since I started hacking perl6 just a few days ago I won't be of any help 08:45
sorear well we have a couple of obvious use cases that simply don't work 08:46
because lexicality of package names is limited to the first level only 08:47
if you do 'class Foo::Quux {}; { my class Foo::Bar {}; #`( code here ) }' the outer Foo::Quux cannot be seen because Foo is shadowed by the inner lexical package 08:48
so far it seems like a DIHWIDT but it really causes problems with lexical imports
use Test; use Test::More; # this doesn't look quite so contrived now does it? 08:49
niecza & rakudo have no way to create a true lexical Test::More, so they bind the second import to 'More' in the Test:: package
but Test:: isn't private to the compilation unit
you've just polluted the Test module's namespace 08:50
and anything else in your program that does 'use Test' will see Test::More because of the shared package hash
so basically what I want to do is to allow lexical names like Test::More 08:51
possibly in a preparsed format
masak rn: class Foo::Quux { our $x = "OH HAI"; }; { my class Foo::Bar {}; say $Foo::Quux::x }
sorear this would require a bit of smarts in the MY:: name resolver, but it seems doable
p6eval niecza v19-44-g55e895a: OUTPUT«(timeout)[auto-compiling setting]␤» 08:52
..rakudo 196572: OUTPUT«OH HAI␤»
masak sorear: "cannot be seen..."? it seems to work at least in Rakudo.
as I would very much expect it to.
sorear masak: rakudo is breaking the other way, then 08:53
masak could you demonstrate "breaking the other way" on p6eval?
sorear class Foo::Quux { our $x = "OH HAI"; }; { my class Foo::Bar { our $x = "In FooBar"; }; }; say $Foo::Bar::x
rclass Foo::Quux { our $x = "OH HAI"; }; { my class Foo::Bar { our $x = "In FooBar"; }; }; say $Foo::Bar::x
r: class Foo::Quux { our $x = "OH HAI"; }; { my class Foo::Bar { our $x = "In FooBar"; }; }; say $Foo::Bar::x
p6eval rakudo 196572: OUTPUT«In FooBar␤»
masak hm.
yeah, that seems odd. 08:54
sorear I think STD did something similar to what I'm proposing 08:55
I thought it was insane in 2010 but now I'm coming around
moritz doesn't see the breakage 08:56
there's just one $Foo::Bar::x
and you assign twice to it, so of course it prints the second value 08:57
sorear No, I only assign once to it
masak moritz: it's visible outside the lexical scope of Foo::Bar
sorear Foo::Bar is a my package, it ought not to be visible from outside the blcok 08:58
moritz masak: sure, because you said "our"
ok, so the real question is "what does 'our' mean"?
masak moritz: something *inside* a 'my' package is visible outside of that 'my' package's scope.
sorear r: class Foo::Quux {}; { my class Foo::Bar { method hi() { "OH HAI" }; }; }; say Foo::Bar.hi
p6eval rakudo 196572: OUTPUT«OH HAI␤»
sorear # look moritz no our in sight
I used to think this is OK, but MikeFair_ made me take a step back 08:59
masak moritz: it'd be like the hash keys of a lexical hash variable could be declared "our" and then made visible outside of the hash's lexical scope.
moritz masak: that's totally analogous to { our $x = 3; } ; our $x; say $x
masak: there you have a lexical block, and you can still see an our-variable from the outside
sorear r: my class Foo::Quux {}; { my class Foo::Bar { method hi() { "OH HAI" }; }; }; say Foo::Bar.hi
p6eval rakudo 196572: OUTPUT«OH HAI␤»
masak moritz: no, it's not analogous. because the package that $x is declared in there is an 'our' package. 09:00
moritz sorear: so the problem is that namespaces with :: only work as 'our'?
sorear Yes
masak finds that while he may not be very informed on these issues, he certainly is opinionated ;) 09:01
sorear It's clearly working as designed
moritz ok, crazy idea
what if we just stored 'Foo::Bar' in a lexical scope? 09:02
if it's my class Foo::Bar { }
instead of going through the Foo/Bar hierarchy
sorear moritz: that's what I just proposed, only more concisely. moritz++ 09:03
moritz that means I'm not totally off track then :-)
sorear the single biggest flaw I see is that it breaks the Foo::Bar - Foo.WHO<Bar> equivalence 09:04
which ISTR jnthn being fond of
moritz I haven't really thought it through 09:05
what about our-classes?
if we have
our class A::B { our class C::D { } }
and
our class A { our class B::C { } } 09:06
etc.
masak I'm for the above idea too. in fact, I thought it worked like that already.
moritz can we make sure that lookup of A::B::C::D doesn't have to go through ridiculously many possible combinations?
sorear moritz: I was imagining that lexicality only applies to prefixes, and greedily takes the longest 09:07
I'll go look at what STD is doing
moritz tries to process that 09:08
std: A::B
p6eval std d93fffe: OUTPUT«===SORRY!===␤Undeclared name:␤ 'A::B' used at line 1␤Check failed␤FAILED 00:00 40m␤»
moritz std: my class A::B { our class C::D { } }; C::D
p6eval std d93fffe: OUTPUT«===SORRY!===␤Undeclared name:␤ 'C::D' used at line 1␤Check failed␤FAILED 00:00 41m␤»
moritz std: my class A::B { our class C::D { } }; A::B::C::D
p6eval std d93fffe: OUTPUT«ok 00:00 41m␤»
sorear is A::B::C::D in %*CURLEX? No. Is A::B::C? No. Is A::B? Yes! (alias installed by our class). Let's look for C::D under the A::B class alias
masak right. 09:09
moritz and if not, you still have to go looking A, right?
masak so it's not "ridiculously many possible combinations", it's a linear thing.
moritz and then in A, look for B::C::D, recursively
oh no
masak hm, troo.
moritz now I understand
I think what sorear++ meant was that this is only done for the first level 09:10
so no recursion
sorear right
because the package entries point to packages, not lexpads
recursion wouldn't make sense, typologically
masak oh good. 09:11
I don't understand why that is, but I'm releaved that it is :)
moritz masak: a use case for macros \o/ blog.racket-lang.org/2012/08/fully-...-sort.html 09:12
masak moritz: nice. 09:13
moritz that could be kind of a "man-or-boy" test for a macro implementation :-)
masak moritz: I think with what's in nom now, we can do that. let me try during the day :0
:)
moritz++ 09:14
felher wants to see that using perl6 macros :)
sorear o/ felher
felher o/ sorear 09:15
moritz I don't want to discourage you, but I wonder how recursive macros can work when there's a big bug in the hygiene 09:17
masak heh :)
it's big bug, but I expect to find and fix that one quickly.
09:17 birdwindupbird left
masak I'm more concerned about the other, more subtle bug that I know about. 09:17
sorear What big bug?
masak sorear: seems unquotes get the macro's lexical scope, not the mainline's. 09:18
r: macro foo($x) { my $v = "inside"; quasi { {{{$v}}}() } }; my $v = "outside"; say foo sub { $v }
p6eval rakudo 196572: OUTPUT«===SORRY!===␤Can not get non-existent attribute '$!past' on class 'AST'␤»
masak :/ 09:19
ok, that's not the failure mode I wanted to demonstrate....
moritz irclog.perlgeek.de/perl6/2012-08-25#i_5930171 09:20
masak r: macro call($a) { my $x = "how very"; quasi { say $x; {{{$a}}}() } }; my $x = "lexical"; call { say $x }
p6eval rakudo 196572: OUTPUT«how very␤how very␤»
masak r: macro foo($x) { my $v = "inside"; quasi { {{{$v}}}() } }; my $v = "outside"; say foo { say $v }
p6eval rakudo 196572: OUTPUT«===SORRY!===␤Can not get non-existent attribute '$!past' on class 'AST'␤»
masak r: macro foo($x) { my $v = "inside"; quasi { say $v; {{{$v}}}() } }; my $v = "outside"; say foo { say $v } 09:21
p6eval rakudo 196572: OUTPUT«===SORRY!===␤Can not get non-existent attribute '$!past' on class 'AST'␤»
masak hm... what's different between 'call' and 'foo'?
oh!
r: macro foo($x) { my $v = "inside"; quasi { say $v; {{{$x}}}() } }; my $v = "outside"; foo { say $v }
p6eval rakudo 196572: OUTPUT«inside␤inside␤»
masak there we go.
ah, and the LTA error message is because there's not AST typecheck on unquote trees. 09:22
masak makes a note to add that ASAP
ChoHag Well that's a bit shit. Emacs thinks the contents of a heredoc is just more code and tries to indent it. 09:23
(Not perl 6, just randing) 09:24
(ranting)
sorear to be fair multiline contextual constructs like regex blocks and heredocs and pod blocks are pretty hard on editors in general 09:25
masak they're pretty hard on parsers, too :)
sorear I think it would be nice if we had single-line versions of the abovew
ChoHag Single-line multi line heredocs? 09:26
sorear ChoHag: you can make a block comment editor-friendly by prefixing each line with #
what if there was something you could use to prefix each line of a heredoc? 09:27
ChoHag Ah.
sorear masak: a code-aware editor is a parser that has to be able to parse malformed documents starting from the middle of the file :D
masak something like that.
seems to me a savvy editor would save intermediate parse states in order not to have to parse the whole document each time. 09:28
sorear editing STD.pm6 gets very annoying when vim's highlighter decides to start in the middle of a regex and then messes up the entire screen
well, you can do that, but then any edit requires a huge amount of reparsing
ChoHag If when parsing you save the state of the parse every n-hundred bytes, then a change at byte 706 doesn't have to parse the first 700 bytes. 09:30
And the rest can be ignored until it's visible.
Or something like that.
09:32 broquaint joined
sorear wait. 09:33
n: say qq:to[OUTER];␤ hello { qq:to[INNER] } world␤ insidestuff␤ INNER␤ more stuff␤ OUTER␤ 09:34
p6eval niecza v19-44-g55e895a: OUTPUT«(timeout)[auto-compiling setting]␤»
sorear evalbot control restart
09:34 p6eval left 09:36 p6eval joined 09:37 ChanServ sets mode: +v p6eval
sorear moritz, why do I get rsync error: unexplained error (code 255) at io.c(605) [sender=3.0.9] when p6eval restarts? 09:37
n: say qq:to[OUTER];␤ hello { qq:to[INNER] } world␤ insidestuff␤ INNER␤ more stuff␤ OUTER␤
p6eval niecza v19-44-g55e895a: OUTPUT«Unhandled exception: Here document used before body defined␤ at /home/p6eval/niecza/boot/lib/CORE.setting line 1402 (die @ 5) ␤ at /home/p6eval/niecza/src/Op.pm6 line 331 (HereStub.zyg @ 5) ␤ at /home/p6eval/niecza/src/Op.pm6 line 23 (Op.ctxzyg @ 4) ␤ …
sorear perl 6: might be the first language to support NESTED HERE DOCUMENTS 09:39
dalek albot: 3bb2f7e | sorear++ | evalbot.pl:
p6eval only sees the costs of installable mode, so force inplace mode for niecza
 09:40
moritz sorear: I have no clue 09:41
sorear I found the problem
I ran 'git pull' in the evalbot dir and it pulled changes to rsync handling 09:42
which had been in the github repo but not live since Aug 2 or so
09:45 p6eval left, p6eval joined, ChanServ sets mode: +v p6eval
sorear moritz: what is "ulimit -u 80" intended to accomplish? 09:45
looks like on feather3 "ulimit -u" is supported by bash but not by dash 09:47
masak hm, I can't really use dies_ok with macros I know are going to fail during application... 09:48
(because application is at compile time)
I think I will need to use an eval() there.
sorear you need to use ... eval_dies_ok! 09:49
masak ah!
sorear and put the macro inside the eval block, hmm
mst p3rl.org/Test::Fatal
^^ this API replaces all the Test::Exception rubbish nicely
simpler, more orthogonal, more pleasant to work with
masak mst: I've seen that one before. it's nice. 09:50
agree about "more orthogonal".
mst the only reason Test::Exception is useful is that it can do various bits of caller() and etc. hackery and my argument would be that that's more a "working around 5core limitations" thing than a feature
09:51 JimmyZ joined
masak it might be too late to do a full switchover for S24 :/ 09:51
mst what's S24 ?
masak the testing synopsis.
I'm guessing it specifies eval_dies_ok et al.
if people think it's a good idea to make Perl 6's Test.pm API more like Test::Fatal, I'm willing to work on that a bit. 09:52
essentially try it out in a branch and prepare a patch that switches us over.
mst I would suggest that, instead, you add the exception { ... } construct
and then wait and see which you find it easier to explain to people how to write tests using
masak nod.
I like that idea. 09:53
mst since it occurs to me that the converse of my previous statement could be valid as well
moritz there's no reason not provide both for a while
masak right.
mst in that it may be that it's the workarounds ib T::Exception
masak mst++ # suggestion
mst that make Test::Fatal superior
also, one could build the lives_ok etc. -atop- the exception { construct
once they become trivial sugar
I mean
lives_ok = f ($block, $words) { is(!exception($block), $words); }; 09:54
for example
dalek ast: 7e37729 | masak++ | S06-macros/macros-d2.t:
[S06-macros/macros-d2.t] new test file
mst then they're hardly going to be a maint bruden
masak: while I'm here I'd like to help ensure that #perl6 avoids as many perl5 mistakes as possible 09:55
dalek kudo/nom: d21d35d | masak++ | t/spectest.data:
[spectest.data] added macros-d2.t
moritz note that either you'd need to write isa_ok exception({block}), X::Foo # with parens
masak reviews on the above file are very welcome.
mst masak: that way you can make all new mistakes that I can learn something from :D
moritz or we'd have make exception a prefix operator 09:56
masak "Perl 5 and Perl 6: two very good sets of complementary mistakes"
moritz (because we have no function prototypes in p6)
masak: reviewing 09:57
masak: first of all, can haz better tets file name puleze?
d2 makes sense from your perspective, but not from anybody elses
masak moritz: good point. I can do that. 09:58
moritz after reading "macro podolsky" I went like "and where is rosen"?
and then I scrolled down
masak++
masak :) 09:59
moritz masak: the last test is wrong
masak: 'cause eval_dies_ok evals in the context of eval_dies_ok, not in the context of your sub
s/sub/code/ 10:00
so it doesn't see macro bohm
(which is why the test succeeds, even though the feature you are trying to test doesn't work
masak moritz: er. 10:01
10:01 crab2313 left
masak moritz: oh! 10:01
moritz: you're saying eval_dies_ok's context is inside of Test.pm?
moritz masak: correct
masak because it's a string, so it doesn't have closure semantics....
shoulda seen that, yes.
moritz I'd love to change that, it's a very common mistake
but &eval doesn't have an option for evalling in a different context 10:02
sorear masak: what was the bug with einstein? 10:03
02:49 < sorear> and put the macro inside the eval block, hmm
moritz (&eval is probably the wrong abstractioin level to add that though. We should have a &compile function that compiles the code but doesn't run it yet, and that one can have such strings and wires attached like compiling in a different context)
masak sorear: einstein works in rakudo. 10:04
sorear: podolsky doesn't currently, even though it should be equivalent. 10:05
&
moritz [1] 8244
now you can just type 'fg 1' to summon masak++ again :-) 10:06
ChoHag Free ++ with every &? 10:07
sorear OH
10:08 drejk joined, spider-mario joined
sorear r: macro foo() { quasi { 2 }; }; say foo 10:09
p6eval rakudo d21d35: OUTPUT«2␤»
sorear r: macro foo() { my $q = quasi { 0 }; $q = quasi { 1 + {{{$q}}} } for ^5; $q }; say foo
p6eval rakudo d21d35: OUTPUT«1␤»
sorear nm 10:10
&
moritz r: macro upto($a, $b) { return quasi { $a } unless $b; return upto(quasi { 1 + {{{$a}}} }, $b - 1 }; say upto(0, 5) 10:11
p6eval rakudo d21d35: OUTPUT«===SORRY!===␤Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 2, near "}; say upt"␤»
moritz r: macro upto($a, $b) { return quasi { $a } unless $b; return upto(quasi { 1 + {{{$a}}} }, $b - 1) }; say upto(0, 5)
p6eval rakudo d21d35: OUTPUT«===SORRY!===␤Cannot call 'Numeric'; none of these signatures match:␤:(Mu:U \v, Mu *%_)␤␤»
10:15 drejk left
ChoHag Does perl 6 have a DBI-like (specifically, postgres) module? 10:18
arnsholt jnthn: Any particular reason NativeHOW.nativesize expects an argument besides the invocant, since it just returns the size? 10:20
FROGGS ChoHag: github.com/perl6/DBIish/
since its on modules.perl6.org I can guess its in rakudo*?
ChoHag Cheers 10:22
moritz FROGGS: the reasoning isn't valid, but it's still in R* 10:29
and yes, it supports postgresql. With prepared statements and placeholders 10:30
.
arnsholt Mmmm. Debug printf output in my compiler ^_^ 10:47
10:48 MayDaniel joined
moritz are you debugging output from printf? or using printf to debug? 10:53
arnsholt The latter, thankfully =) 10:55
Making sure I get the right numbers in my metamodel hackery (actually storing and using sizes for native types)
FROGGS arnsholt: I can pass my callback sub just like: thing_that_takes_a_cb( &subname ) 11:28
right?
arnsholt Yup
FROGGS passing a anonymous sub and passing a $subref doesnt work right now? 11:29
arnsholt Dunno, TBH 11:30
Did you try?
FROGGS I did, no warnings, but the callback doesnt get called 11:31
arnsholt Huh. That's a bit odd, I guess
I'll try to look into it 11:33
FROGGS cool, great, thanks 11:34
11:35 tokuhiro_ joined
FROGGS passung Mu as the arg will end up as a NULL in C? 11:35
passing
arnsholt I think so 11:36
But most often you'll want to pass the type object
FROGGS I mean, it works as expected (callback get unregistered) but I want to make sure 11:37
type object?
like the type identifier of a subroutine reference? 11:38
is it "Code"?
11:40 whiteknight joined, tokuhiro_ left
FROGGS k, passing Code works too... its like it is some sort of bitch, no complains, just takes everything 11:40
11:46 thelazydeveloper joined 11:49 MayDaniel left
arnsholt FROGGS: The type object is part of the meta-object protocol 11:57
dalek ast: 8e15845 | masak++ | S06-macros/macros-d2.t:
[macros-d2.t] fix mis-context

  moritz++ for pointing this out: strings aren't passed as
closures, and eval_dies_ok evaluates things in Test.pm's lexical context. moving the macro into the string fixes it.		 11:58
arnsholt When you say "my $x = Something.new" or "my Something $x .= new" what happens is that a method is invoked on the Something metaobject
And the type object of a class is used as a typed undef value, which makes sense for NULL values 11:59
dalek ast: 0710d57 | masak++ | S06-macros/ (4 files):
[S06-macros] gave test files more descriptive names

  moritz++ for suggestion. now they are not tied to the
grant milestones, but to what functionality they exercise... just like the rest of the test suite.		 12:00
12:00 kaare_ left
dalek kudo/nom: f7ec305 | masak++ | t/spectest.data:
[spectest.data] chased macro test files rename
 12:02
12:23 tokuhiro_ joined 12:28 tokuhiro_ left 12:29 tokuhiro_ joined 12:34 tokuhiro_ left 12:42 DrEeevil is now known as bonsaikitten
arnsholt is starting to really like tmux 13:02
13:08 cognominal left 13:10 cognominal joined
JimmyZ is using byobu, which is tmux-based 13:13
masak ah; nice name for a screen-like application :) 13:14
arnsholt Hmm. Looks interesting 13:16
Although, I already have much of what it prints out in my xmobar status line 13:17
13:18 birdwindupbird joined
FROGGS arnsholt: is there a way to declare default argument values using NativeCall? like: our sub my_sub( int $a, int $b = -1) is native ... 13:18
when trying it and calling this sub with only one arg I get a coredump
arnsholt Don't think that'll work
What you'll probably want to do if you want that functionality is create a wrapper Perl 6 function that calls the dyncall function 13:19
FROGGS is it on the wishlist already? :P
arnsholt Nope, hadn't really considered that one
FROGGS arnsholt: I did that too (wrapper func), but it crashes as well
arnsholt That's odd...
FROGGS ya
the subs have different names btw, its no multi stuff
arnsholt What happens if you just call the C function directly? 13:20
FROGGS when I call it with all args, it does what it should (plays music actually)
arnsholt That's odd 13:21
Could you gist the relevant bits, perhaps?
FROGGS that crashes: github.com/PerlGameDev/SDL6/blob/f...ls.pm6#L10
its not much though 13:22
arnsholt Huh. Odd 13:23
Not sure what's going on 13:26
Could you try running it through gdb?
See where it segfaults
FROGGS I will, gtg for a walk now, will report to you then 13:29
arnsholt Excellent 13:30
13:33 MayDaniel joined 13:46 josp joined 13:47 josp left, kaare_ joined
masak nwc10++ # nicest "how do I unsubscribe?" reply I've ever read, on p6c 13:49
I also found myself thinking "whatever this guy has a PhD in, I bet it isn't mailing lists" :) people have different backgrounds. 13:50
13:55 daxim joined 14:12 marmay joined
frettled masak: yep. And looking in message headers isn't necessarily trivial in a modern GUI-based email client, they mostly hide it somewhere for experts. 14:13
14:15 simcop2387 left, simcop2387 joined
masak I'm not sure I'd think to look there, actually. 14:17
frettled One of the points of the list-headers is to make it easy for email clients to identify mailinglists _and_ to unsubscribe. Look at how well that's panning out. :( 14:18
masak likes this quote from the distant past: irclog.perlgeek.de/perl6/2005-06-28#i_-584545 14:21
pmichaud good afternoon, #perl6 14:22
ChoHag How do you call up the pod of an installed module? 14:23
geekosaur as for the list headers, they are indeed for clients. don't think I've seen a clientother than gmail that offers unsubscribe etc. to the actual user
14:24 xinming joined
masak good morning, pmichaud. 14:24
ChoHag Reality isn't all it's cracked up to be. 14:25
It can barely be said to exist, for one thing.
pmichaud ...morning?
hmmmm
masak pmichaud: UGT, Universal Greeting Time.
pmichaud wfm
it's morning in #perl6, the dawn of a new era
or something like that.
arnsholt jnthn is at the MOP hacakthon thingy these days, right? 14:38
14:50 kst left
pmichaud arnsholt: yes. 14:53
14:53 Circlepuller joined
spider-mario I now provide arch linux builds of rakudo star. :) 14:56
14:56 Circlepuller_ left
spider-mario a pacman repository. 14:56
[rakudo] 14:57
Server = spidermario.free.fr/archlinux/$repo/$arch
masak \o/ 14:59
ChoHag Are there docs anywhere describing a simple perl6 tcp server? 15:01
Also, has anything like an http daemon been written, or even something more freaky like rest or soap? 15:02
JimmyZ ChoHag: github.com/tadzik/Bailador 15:05
ChoHag That looks even better. 15:06
pmichaud ChoHag: : github.com/mberends/http-server-simple/
I'm afk for a while -- bbl 15:07
JimmyZ ChoHag: and github.com/tadzik/zblog
15:10 GlitchMr joined 15:13 perl5 joined
masak perl5: hi. I saw you on the back of lots of YAPC::Europe t-shirts. :) 15:14
15:16 perl5 left 15:24 ponbiki joined
jnthn ohhai o/ 15:28
arnsholt jnthn: \o/ 15:30
I'm having some trouble extracting the nativesize stuff in the P6{int,num} REPRs 15:32
masak jnthn! \o/ 15:34
jnthn arnsholt: I could tell you how, but I'm gonna change the answer pretty shortly...
arnsholt Aha. Might as well wait for the new answer then =) 15:35
jnthn dammit, libuv builds fine as a .lib but I can't get it to build as a .dll
15:36 JimmyZ left
ingy greetings. seeking 2 things. 1) the current perl6 grammar. 2) A succinct idiom comparison between perl5 and perl6. 15:40
masak ingy: (1) github.com/perl6/std/blob/master/STD.pm6 15:42
diakopter (2) that would be lovely to me as well 15:43
masak I've seen such documents, but it's been a while.
ingy thanks! 15:51
15:51 flussence left, flussence joined
arnsholt jnthn: FROGGS gets what looks like a decontainerize of something that isn't a container in github.com/PerlGameDev/SDL6/blob/f...ls.pm6#L10 (in play_channel) 16:04
Backtrace at nopaste.info/2f9f16692a.html
16:04 marmay_ joined
arnsholt Doesn't look like it's the literal -1 that's it. Any thoughts on what it might be? 16:05
16:05 tokuhiro_ joined
arnsholt FROGGS: Oh. What happens if you remove the type decorations of the args in play_channel? 16:06
FROGGS how should I pass the args then?
is there a $_ or @_? 16:07
or is it like (shift, shift, shift)?
arnsholt Oh, you can leave the argument list, but try making it play_channel( $channel, $sample, $loops ) 16:08
FROGGS ahh, misread
will try
16:08 marmay left
FROGGS its working 16:08
I need to recheck that I did no mistake 16:09
16:10 tokuhiro_ left
arnsholt Excellent! 16:11
16:11 saran joined
arnsholt Looks like native typed args don't end up in containers 16:11
FROGGS ya, it fails again if I add the "int" to the first arg
16:11 saran left
arnsholt Which then explodes when we try to decontainerize 16:11
16:11 spider-mario left 16:16 Circlepuller_ joined 16:17 saran joined 16:19 snearch joined, Circlepuller left 16:24 Circlepuller joined 16:27 Circlepuller_ left 16:31 saran_ joined
arnsholt FROGGS: I think I've figured out what the problem is. I'll prod you when it's fixed 16:31
pmichaud back again 16:32
16:33 saran_ left
pmichaud reading backscroll and email 16:34
16:34 saran left
FROGGS arnsholt++ 16:39
thank you ;o)
arnsholt Well, I didn't fix it yet =) 16:40
FROGGS ya, but guessing the problem is the hardest part most times
16:43 birdwindupbird left 16:47 marmay_ left 17:11 am0c joined 17:15 benabik left 17:16 benabik joined 17:21 spider-mario joined
arnsholt jnthn: What's the best way to get access to smo_id (in sixmodelobject.c) from nqp_dyncall.ops? 17:38
dalek c: a5c90a5 | moritz++ | / (2 files):
X::Mixin::NotComposable
 17:40
tadzik 0901 sorear | anyways, I had a hell of a time getting Niecza to run on 17:43
| tadzik's machine
dalek ecs: c80acf9 | moritz++ | S32-setting-library/Exception.pod:
[S32::Exception] rename an exception class, for consistency
tadzik ...when were you... oh, I remember the case now 17:44
good evening #perl6
moritz \o tadzik
arnsholt Nm. Found it 17:45
dalek c: 5f9c9f4 | moritz++ | lib/X/NoDispatcher.pod:
X::NoDispatcher
 17:46
c: 278a4cd | moritz++ | lib/X/Declaration/Scope/Multi.pod:
X::Declaration::Scope::Multi
 17:48
c: 0edd594 | moritz++ | lib/X/Syntax/Variable/Twigil.pod:
X::Syntax::Variable::Twigil
 17:52
c: 4bb78ba | moritz++ | lib/X/Syntax/Self/WithoutObject.pod:
X::Syntax::Self::WithoutObject
 17:54
masak good morning, tadzik. 17:56
rn: constant o = 42; \o
p6eval rakudo f7ec30, niecza v19-44-g55e895a: ( no output )
masak rn: constant o = 42; say \o
p6eval niecza v19-44-g55e895a: OUTPUT«\(42)␤»
..rakudo f7ec30: OUTPUT«42␤»
colomon rn: constant o = 42; say o 18:04
p6eval rakudo f7ec30, niecza v19-44-g55e895a: OUTPUT«42␤»
benabik rn: constant o = 42; say \o/ 1 18:07
p6eval rakudo f7ec30: OUTPUT«1␤»
..niecza v19-44-g55e895a: OUTPUT«Unhandled exception: Cannot use value like Capture as a number␤ at <unknown> line 0 (ExitRunloop @ 0) ␤ at /home/p6eval/niecza/lib/CORE.setting line 293 (Any.Numeric @ 6) ␤ at <unknown> line 0 (ExitRunloop @ 0) ␤ at /tmp/veIlHDuYsG line 1 (mainline @ 4…
benabik Amusing.
18:07 snearch left
moritz nr: say Capture ~~ Cool 18:09
p6eval rakudo f7ec30, niecza v19-44-g55e895a: OUTPUT«False␤»
dalek kudo/nom: bd23ad4 | moritz++ | src/core/ (2 files):
rename an exception class
 18:10
ast: d4fa5e8 | moritz++ | S32-exceptions/misc.t:
follow exception class rename
moritz nr: sub f(::T) { class A is T { } } 18:24
p6eval niecza v19-44-g55e895a: OUTPUT«===SORRY!===␤␤GLOBAL::T does not name any package at /tmp/EYl_nY1i4o line 1:␤------> sub f(::T⏏) { class A is T { } }␤␤A type must be provided at /tmp/EYl_nY1i4o line 1:␤------> sub f(::T⏏) { class A is T…
..rakudo f7ec30: OUTPUT«===SORRY!===␤Cannot type check against type variable T␤at /tmp/aUJgGyqScX:1␤»
18:28 odoacre left
masak how is '\o/' parsed above in benabik's evaluation? 18:30
benabik masak: It seems to be parsed as \o / 1, given that neicza complained about capture to number.
dalek blets: 4eff6da | (Herbert Breunung)++ | docs/ (3 files):
removing noneexisting book
 18:34
masak r: constant o = 42; say o/
p6eval rakudo f7ec30: OUTPUT«===SORRY!===␤Confused␤at /tmp/Q8b9OJmTkQ:1␤»
masak r: constant o = 42; say \o/
p6eval rakudo f7ec30: OUTPUT«===SORRY!===␤Confused␤at /tmp/ATzNEqDZUn:1␤»
benabik masak: I used \o/ 1
masak oh! :)
benabik :-D
masak mystery solved :)
sneaky ;)
18:35 xinming left 18:36 xinming joined 18:37 daxim left 18:47 larks left
dalek c: 19f2c5a | moritz++ | lib/X/Syntax/P5.pod:
X::Syntax::P5
 18:47
18:49 larks joined 18:51 popl joined
masak .oO( Perl 6 is so different, Perl 5 is represented in it as an exception type ) :P 18:51
moritz the Pod:: classes need documentation 18:52
any takers? 18:53
masak: oh, and the Macro class needs documentation
r: macro a { }; say &a.WHAT
p6eval rakudo f7ec30: OUTPUT«Macro()␤»
masak noted.
moritz \o/
moritz imagines the memo percolating to the different masak clones 18:54
masak "A macro is a Routine whose invocation typically happens during parsing."
or s/m/M/ perhaps.
moritz r: say Macro.^mro 18:56
p6eval rakudo f7ec30: OUTPUT«Macro() Routine() Block() Code() Any() Mu()␤»
moritz why doesn't it inherit from Sub?
masak why should it? 18:57
why doesn't Method inherit from Sub?
moritz because it's very sub-like
masak so is Method.
moritz nonono
masak sure it is.
moritz methods are scoped differently by default
and are usually called in a different way
masak so are macros.
moritz called, yes. Scoped? 18:58
I hope not :-)
masak methods are *scoped* the same too. they are registered differently.
methods adhere to lexical scoping just as much as macros and subs do.
but please, tell me what would be different in practice if I made Macro a subclass of Sub. that should be the determiner of this. 18:59
not some fuzzy philosophical similarity.
see, src/core/Sub.pm is empty :) 19:00
sorear good * #perl6 19:01
moritz you're right, it's probably not important
just different from what I expected
good morning sorear
masak moritz: I will remember your surprise and take it in consideration in case any empirical difference manifests itself. 19:02
good morning, sorear.
dalek c: 9795da9 | moritz++ | lib/Macro.pod:
stub documentation for Macro
 19:03
19:06 larks left, larks joined
masak moritz++ 19:06
19:08 ObseLeTe joined 19:10 gongyiliao joined
masak gongyiliao: hi. 19:10
gongyiliao hi
masak rn: say "Welcome to #perl{ 1 + 2 + 3 }" 19:11
p6eval rakudo bd23ad, niecza v19-44-g55e895a: OUTPUT«Welcome to #perl6␤»
19:12 birdwindupbird joined
moritz rn: say "Welcome to #perl{ [+] ^4 }" 19:13
p6eval rakudo bd23ad, niecza v19-44-g55e895a: OUTPUT«Welcome to #perl6␤»
sorear haven't I seen that name before?
ah, only in the sense that gongyiliao has joined #perl6 53 times since my records begin 19:15
masak oh :) 19:16
sorear was momentarily confused by "nwc10" 19:18
silly internetwork name differences
FROGGS is there a pod2html or something that can handle several =for pod blocks? 19:21
19:22 GlitchMr left
moritz there's Pod::To::HTML 19:23
FROGGS ohh, cool
moritz that's what we use to generate the HTML on doc.perl6.org 19:24
sirrobert is there a special var that knows the name of the file currently being worked on?
(and: hi p6 =) 19:25
sorear rn: say $*FILE 19:28
p6eval rakudo bd23ad: OUTPUT«Dynamic variable name not found␤ in method gist at src/gen/CORE.setting:9757␤ in sub say at src/gen/CORE.setting:7125␤ in block at /tmp/28zi11Lu1v:1␤␤»
..niecza v19-44-g55e895a: OUTPUT«Any()␤»
sorear rn: say $?FILE
p6eval rakudo bd23ad: OUTPUT«/tmp/vPw47tI1pw␤»
..niecza v19-44-g55e895a: OUTPUT«/tmp/t0BITImhAC␤»
sirrobert huh
I haven't seen $? before
sorear rn: say callframe(0).file
p6eval niecza v19-44-g55e895a: OUTPUT«/tmp/snl5S79OLb␤»
..rakudo bd23ad: OUTPUT«/tmp/0Se6AQrLK4␤»
sirrobert if $* means global, what does $? mean? 19:29
sorear $? is for constants
lexically scoped compile time values
sirrobert ahh... cool
sorear in a way it's closely related to $*
felher sirrobert: you may want to take a look at: perlcabal.org/syn/S02.html#Twigils
FROGGS so you are talking about __FILE__?
moritz nr: say $?FILE
p6eval niecza v19-44-g55e895a: OUTPUT«/tmp/FWp6A5Lqde␤»
..rakudo bd23ad: OUTPUT«/tmp/vC7WDrAdxp␤»
sorear because with recursive descent parsing, lexical scopes in the program correspond to dynamic scopes in the compiler
$? support is spotty 19:30
sirrobert sorear|felher: thanks--reading
sorear: hm ok
sorear niecza has &?BLOCK, &?ROUTINE, $?FILE, $?LINE, $?POSITION, $?ORIG, and I think that's it
moritz doc.perl6.org/language/variables#Twigils # needs much more work
sorear rakudo I think has about the same number, it adds $?CLASS and friends but lacks $?LINE? 19:31
nr: say $?LINE
p6eval rakudo bd23ad, niecza v19-44-g55e895a: OUTPUT«1␤»
sorear nr: 1␤1␤say $?LINE
p6eval niecza v19-44-g55e895a: OUTPUT«===SORRY!===␤␤Two terms in a row (previous line missing its semicolon?) at /tmp/Ug7FX83imF line 2:␤------> <BOL>⏏1␤␤Parse failed␤␤»
..rakudo bd23ad: OUTPUT«===SORRY!===␤Confused␤at /tmp/HoGtJg5BT4:1␤»
sorear nr: 1;␤1;␤say $?LINE
p6eval rakudo bd23ad, niecza v19-44-g55e895a: OUTPUT«3␤»
sorear mmh
nr: say &?BLOCK
p6eval rakudo bd23ad: OUTPUT«Nil␤»
..niecza v19-44-g55e895a: OUTPUT«routine mainline() { ... }␤»
sorear nr: say &?MOO
p6eval rakudo bd23ad: OUTPUT«Nil␤»
..niecza v19-44-g55e895a: OUTPUT«Any()␤»
sorear ok, there's one niecza has 19:32
MikeFair_ howdy #perl6 19:33
moritz nr: say Nil ~~ Cool
p6eval rakudo bd23ad, niecza v19-44-g55e895a: OUTPUT«True␤»
sirrobert hi
MikeFair_ gmorning
masak MikeFair_: morning!
felher moritz: maybe i can do a bit of work on twigils in p6doc on wednesday. Today/Mon/Tue are filled up with work ($job). 19:34
o/ MikeFair_
moritz felher: that would be great 19:35
dalek kudo/uncool-iterable: c7e1713 | moritz++ | src/ (3 files):
Iterable should not inherit from Cool

while that might be cool for core classes, it is not in the general case
kudo/uncool-iterable: c6fac5b | moritz++ | src/core/Nil.pm:
Nil is Cool
sirrobert r: dir 19:37
p6eval rakudo bd23ad: ( no output )
sirrobert r: dir.WHAT
p6eval rakudo bd23ad: ( no output )
felher r: dir.perl.say
p6eval rakudo bd23ad: OUTPUT«Array.new(IO::Path.new(basename => "star", directory => "."), IO::Path.new(basename => "src", directory => "."), IO::Path.new(basename => ".bashrc", directory => "."), IO::Path.new(basename => "nom-inst1", directory => "."), IO::Path.new(basename => "toqast", direc…
sirrobert r: for dir -> $f { say $f}
p6eval rakudo bd23ad: OUTPUT«===SORRY!===␤Missing block␤at /tmp/Evirw6da_2:1␤»
sirrobert why is that a missing block? 19:38
moritz r: for dir() -> $f { say $f }
p6eval rakudo bd23ad: OUTPUT«IO::Path<star>␤IO::Path<src>␤IO::Path<.bashrc>␤IO::Path<nom-inst1>␤IO::Path<toqast>␤IO::Path<test3.pl>␤IO::Path<.profile>␤IO::Path<t>␤IO::Path<nom-inst2>␤IO::Path<nom-inst>␤IO::Path<toqast-inst>␤IO::Path<toqast-inst2>␤IO::Path<examples>␤IO::Path<Makefile>␤IO::Path<…
sirrobert hm
moritz sirrobert: because the block is taken as an argument to the dir function
sirrobert ohh
moritz n: for dir -> $f { say $f }
p6eval niecza v19-44-g55e895a: OUTPUT«===SORRY!===␤␤Missing block at /tmp/GET_0Y3r4G line 1 (EOF):␤------> for dir -> $f { say $f }⏏<EOL>␤␤Parse failed␤␤»
moritz std: for dir -> $f { say $f }
p6eval std d93fffe: OUTPUT«===SORRY!===␤Function 'dir' needs parens to avoid gobbling block at /tmp/eX_uOAhYR_ line 1:␤------> for dir⏏ -> $f { say $f }␤Missing block (apparently gobbled by 'dir') at /tmp/eX_uOAhYR_ line 1:␤------> for dir -> $f { say …
moritz std has a better message
sirrobert yeah
hey... you know what would be cool 19:39
a separate project housed only error messages. When an error got thrown out, it has something like a code
a separate github repo houses messages to make them helpful?
then is fed in during compile?
I would happily contribute better error messages when I came across weird stuff, but don't have time to dig into the guts 19:40
moritz sirrobert: the problem is not the message itself
sirrobert: the hard part is detecting the error condition in the first place
sirrobert well, in this case
but having a lexicon of errors could be helpful in general =) 19:41
moritz github.com/rakudo/rakudo/blob/nom/...ception.pm isn't too bad to hack on
sirrobert: well, doc.perl6.org has a nearly complete list
sirrobert moritz: fair enough =)
19:42 tokuhiro_ joined
jnthn back from nom 19:43
moritz wb jnthn 19:44
std: rxs// 19:45
p6eval std d93fffe: OUTPUT«===SORRY!===␤Bogus term at /tmp/5MniyHlIb6 line 1 (EOF):␤------> rxs//⏏<EOL>␤Undeclared routine:␤ 'rxs' used at line 1␤Parse failed␤FAILED 00:00 42m␤»
jnthn Method and Sub inherit from Routine.
That's what factors out their commonalities.
It's fine. :)
moritz jnthn: yes, that much isn't debated at all
dalek c: 3afc269 | moritz++ | lib/X/Syntax/Regex/Adverb.pod:
X::Syntax::Regex::Adverb
 19:46
moritz jnthn: the question was really whether Macro should inherit from Sub, or directly from Routine
jnthn Routine. :)
imho ;)
19:46 kst joined, tokuhiro_ left
jnthn I'd not really want a macro to ~~ Sub 19:46
That feels excessively surprising.
dalek c: d1d8a9d | moritz++ | lib/X/Syntax/Reserved.pod:
X::Syntax::Reserved
 19:49
p: a886a77 | (Arne Skjærholt)++ | src/ops/nqp_dyncall.ops:
Properly handle non-Sixmodel objects in dyncall ops.

Two issues:
  - decontainerize in nqp_dyncall.ops was out of sync with decontainerize in
   sixmodelobject.c (didn't consider possibility of non-SMO arguments)
  - unmarshalling code (to int/float types) need to dispatch to VTABLE_get_* in
   case of non-SMO arguments.
  FROGGS++ for uncovering the problem.		 19:51
jnthn nemesys: Hmm...I'm curious why the second bit was needed. 19:52
grr
arnsholt: ^^
arnsholt The VTABLE stuff? 19:53
If it isn't SMO, REPR(value)->box_funcs->get_foo(...) kinda isn't there, no?
dalek c: 87275f7 | moritz++ | lib/X/Constructor/Positional.pod:
X::Constructor::Positional
 19:54
jnthn arnsholt: Right. I meant...where on earth is the non-6model object coming from?
arnsholt Native types it seems 19:55
moritz erm, what? 19:56
you mean, sub f(int) is native(...) ?
dalek volaj: d47cd3f | (Arne Skjærholt)++ | t/02-simple-args. (2 files):
Add test for wrapper sub with native parameter types.

Closes issue #12 in conjunction with NQP commit a886a77		 19:58
volaj: 3d6ee45 | (Arne Skjærholt)++ | lib/NativeCall.pm6:
Merge branch 'master' of github.com:jnthn/zavolaj
sorear JNTHN!
jnthn SOREAR!
moritz if those box into parrot PMCs, then they need more care on the callee side, I guess
arnsholt jnthn, moritz: See the new test in zavolaj d47cd3f up there
dalek c: b84221e | moritz++ | lib/X/Syntax/Obsolete.pod:
X::Syntax::Obsolete
 20:00
arnsholt A wrapper around a native function, where the wrapper has native types on its args and passes them through to the function apparently passes the natives through
sirrobert this is cool, but syntactically, why does it work? 20:01
r: my $a = {a => 1}; my $b = {b => 2}; my %h = $a, $b; say %h;
p6eval rakudo bd23ad: OUTPUT«("a" => 1, "b" => 2).hash␤»
sorear jnthn: I hear you're in Norway now. Anyways, I'd like you to read irclog.perlgeek.de/perl6/2012-08-26#i_5931019
sirrobert why is ($h1,$h2) a hash?
s/$/%/
20:01 sftp left
sorear jnthn: 8:46 to 9:11 20:02
moritz sirrobert: it seems that hash assignment flattens out hashes regardless of the sigil
sorear sirrobert: it's not, it's a parcel
moritz sirrobert: which looks like a bug to me
sirrobert hm
20:02 sftp joined
moritz n: my $a = {a => 1}; my $b = {b => 2}; my %h = $a, $b; say %h.perl 20:02
p6eval niecza v19-44-g55e895a: OUTPUT«Unhandled exception: Unmatched key in Hash.LISTSTORE␤ at /tmp/jAmTUPUiEk line 1 (mainline @ 16) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4138 (ANON @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4139 (module-CORE @ 571) ␤ at /home/p6eval/nie…
sirrobert r: say ({a => 1}, {b=>2}).WHAT
p6eval rakudo bd23ad: OUTPUT«Parcel()␤»
sirrobert r: %h = ({a => 1}, {b=>2}); say %h.WHAT; say %h; 20:03
p6eval rakudo bd23ad: OUTPUT«===SORRY!===␤Variable %h is not declared␤at /tmp/1r2mmum6ga:1␤»
sirrobert r: my %h = ({a => 1}, {b=>2}); say %h.WHAT; say %h;
p6eval rakudo bd23ad: OUTPUT«Hash()␤("a" => 1, "b" => 2).hash␤»
moritz sorear: is that error the same as "odd numer of elements in hash assignment"?
arnsholt moritz: I dunno. If Rakudo possibly passes around bare Parrot PMCs I think the dyncall stuff is the right place to handle the possibility
sorear moritz: yes
moritz: except you can get it with even numbers too 20:04
n: my %h = (a => 2), 'b';
p6eval niecza v19-44-g55e895a: OUTPUT«Potential difficulties:␤ %h is declared but not used at /tmp/tZi5QKRtaz line 1:␤------> my ⏏%h = (a => 2), 'b';␤␤Unhandled exception: Unmatched key in Hash.LISTSTORE␤ at /tmp/tZi5QKRtaz line 1 (mainline @ 2) ␤ at /home/p6eval/niecza/l…
sirrobert huh...
moritz sorear: so it specifically handles hashes iin hash assignment?
sirrobert r: my %h = {a=>1, b => 3}; %h ,= {c=>3}; say %h;
p6eval rakudo bd23ad: OUTPUT«("a" => 1, "b" => 3, "c" => 3).hash␤»
sirrobert that's pretty cool.
sorear moritz: yes 20:05
moritz sorear: is that specced? (I'm not saying it's wrong, I'm just curious :-)
sorear I know I've seen real-world code that does 'my %h = $h'
niecza might be wrong in allowing Hashes to be freely admixed into lists of pairs, keys, and values 20:06
sorear looks for relevant spec
The relevant spec is S02:4711, which does NOT allow passing Hash objects to Hash.LISTSTORE 20:07
20:08 SamuraiJack left 20:09 cognominal left, cognominal joined, cognominal left, cognominal joined
moritz let's wait for a word from TimToady though. It might be an omission, since in p5 a hash would simply flatten out 20:11
felher should teach his address bar to recognize things like S02:4711 20:14
20:16 gongyiliao left
sirrobert I love it when you get to that magic moment when your subs and modules and such turn into a perfect DSL. 20:19
masak sirrobert: that's why I tend to start most designs by pretending I have the perfect DSL already :) kinda top-down. 20:21
sirrobert masak: me too... super smooth =)
it tends to be multi-pass for me
usually 3 or 4 20:22
sorear sirrobert: The Forth community has a wonderful additude to this. Every sub you define is defining an incremental new language
20:22 bruges left
sirrobert huh 20:22
never heard of "The Forth"
sorear The (Forth community)
sirrobert ohhh 20:23
heh
jnthn sorear: Yeah, there are some issues in Rakudo around that area.
sorear Forth is a languagey thing from many years ago
sirrobert I think I've heard of but, but know nothing about it
20:24 bruges joined
sorear Forth's niche is a local optimum very low on the expressiveness/implementation complexity curve 20:26
you can use it for fair-sized apps, it has a REPL, and it can be implemented in about 1-4k of machine code 20:27
dalek Heuristic branch merge: pushed 30 commits to nqp/toqast by jnthn
benabik Forth is used in RedPower.
(Mod that adds a 6502 processor to minecraft)
sorear these days it gets most use in embedded systems, for the last reason
masak they totally should've called the language "The Forth", though. :D 20:28
colomon Forth++
\
sorear it may be a language, but it's the kind of language that has builtins for accessing a block storage device
benabik May the Forth be with you?
sorear 42 LOAD 20:29
masak when I last studied the Forth culture, I got the same kind of vibes from it of "we hold the ultimate truth to how computing should be done" that I sometimes get from Lisp or Smalltalk people. 20:33
not putting any value judgement in that. that's how it felt.
sorear to put it bluntly, when a language loses its core niche, most of the people who remains are True Beleivers. 20:34
pmichaud good evening, #perl6
sorear good evening pmichaud 20:35
20:36 REPLeffect left
colomon masak: actually, I think there are really interesting parallels between Forth and Lisp. 20:36
20:37 fgomez left
sirrobert masak: Forth is Force with a lisp? 20:37
sorear Forth is a fourth-generation language on a limited OS that had 5-character command names 20:38
same reason we have SCHEME instead of the originally intended SCHEMER.
.oO( if Scheme were created today, it would obviously have been named Schemr)
masak :P 20:40
and it would've had rounded corners and would only hire rock stars and ninjas.
benabik roundrects everywhere 20:44
20:45 birdwindupbird left 20:50 REPLeffect joined
jnthn Do we *need* multi-jointed lexical packages? 20:54
What happens if we kill them?
ChoHag If you were working on a project for fun, and you weren't really sure what it was, what would you call it? 20:59
20:59 kaare_ left
pmichaud ChoHag: pick a fun name 20:59
ChoHag I ask here because it will feature perl 6 for a significant part of it.
But what's fun about a name? 21:00
sirrobert ChoHag: pick a commonplace thing in the same conceptual broad swath as the project. 21:01
ChoHag Hmm.
It's a swathe of quite significant proportions.
sirrobert ChoHag: e.g. if you're working on something to do brownian motion, dfo "
do "riverbed"
or something
ChoHag hmm 21:02
sirrobert what's the project (in like... 10 words or less)?
ChoHag Code that writes itself. 21:03
Not in the quine sense.
9.
sirrobert back 21:04
how about something like "meta"
oh, here you go 21:05
ChoHag I dunno.
Isn't that a bit ... meta?
sirrobert Ouroboris
FROGGS or "cloud", everything is a cloud nowadays
masak jnthn: I wouldn't mind, I think.
ChoHag This has emphatically nothing to do with clouds.
sirrobert Ouroboros, I guess 21:06
(s/i/o)
ChoHag That might be a good choice.
Maybe Youroboros.
sirrobert or if you want something with more popular recognition,
Escher
ChoHag Because it's not really mine.
sirrobert anyway, you get the idea
ChoHag Yeah that'll do. 21:07
I'm not a big one for names.
sirrobert it's a different skill, for sure.
ChoHag My little girl's only 14 months and I've called her the cat's name on numerous occasions.
sirrobert heh
FROGGS I always swap the names of my two sons :/ 21:08
sirrobert When my son was 5 last year, he said to his mom: Hey ... <short pause> ... you, with the glasses.
FROGGS hehe
ChoHag That wouldn't work in our house.
sirrobert it didn't work in ours either
(we both wear them)
ChoHag My dad and brother have the same name too, and sound the same on the phone. 21:09
That was always fun.
FROGGS your dad, your brother and you have the same name?
thats basically not the idea of names 21:10
ChoHag My dad and brother have the same name in addition to most of my (previous) household wearing glasses.
Sorry I thought this was #perl.
Weird language being the order of the day, month and indeed year.
FROGGS gnight 21:11
ChoHag What's @INC changed to and how do I make it include . (or lib)? 21:12
(In rakudo)
sirrobert use lib?
ChoHag Are you asking me? 21:13
I'm asking you.
sirrobert I'm asking you if that didn't work
FROGGS use lib 'lib';
I use it
ChoHag Apparently it does. 21:14
Too often questions have the answer 'the same as perl 5'. 21:15
felher Yeah, and if you want it for a quick test/check/whatever, you may use: 21:18
export PERL6LIB=yourPathHere
And then start your program :)
21:21 fgomez joined
sorear ChoHag: pass -I. to the compiler 21:22
"use lib" is deeply problematic
jnthn: What do you mean by multi-joined lexical packages?
ChoHag deeply? 21:23
21:24 grondilu joined
sorear yes, deeply. 21:25
it's kind of like very. 21:26
grondilu There has been no news for a while on rakudo.org. What's up?
sorear I just went to Rakudo.org and it has a post from last month concerning the release 21:27
sirrobert where do I find the Test module?
sorear All Perl 6 compilers come with a Test module preinstalled
jnthn sorear: my class Foo::Bar { }
sirrobert sorear: yeah, but where do I find it in the/a github repo? 21:28
sorear: I want to look at the code
sorear sirrobert: each compiler has a different (but mostly compatible) Test module
pmichaud sirrobert: github.com/rakudo/rakudo/blob/nom/lib/Test.pm 21:29
sirrobert pmichaud: ahh, thanks
sorear sirrobert: github.com/sorear/niecza/blob/mast...b/Test.pm6 21:30
pmichaud time for sleep here -- be back tomorrow for some real hacking :)
sirrobert wave
sorear: thanks
sorear sirrobert: github.com/perl6/Pugs.hs/blob/mast...ib/Test.pm 21:31
github.com/fglock/Perlito/blob/mas...o6/Test.pm 21:32
github.com/fglock/Perlito/blob/mas...o5/Test.pm
are there any productiony implementations I'm missing here? 21:33
sirrobert dunno, but pmichaud had a dwimmier parser =) 21:37
ChoHag Is there any automagic done on the import list of a module? 21:38
I can see _that_ you write an import sub, but can't find examples of any that take arguments.
And just parsing @_ myself seems so perl 5 .. ago. 21:39
sirrobert ChoHag: sorry, dunno =) 21:40
21:42 Circlepuller left, Circlepuller joined
jnthn ChoHag: If you mean what I think you do, see lib.pm6 in Rakudo 21:42
sirrobert ok, clean tests 21:43
ChoHag I don't think that was.
sirrobert heading home; later all
21:43 sirrobert left
sorear wait, it that defining GLOBAL::<&EXPORT> ? 21:43
ChoHag I want to know what to do about the foo bar in 'use Module <foo bar>;'
sorear our sub EXPORT(*@a) 21:44
ChoHag: sub foo is export () { }
ChoHag: that's how you make it work.
ChoHag So that would call foo and bar as subs when Module is loaded?
dalek p/toqast: 9b48c83 | jonathan++ | src/QAST/Operations.nqp:
Don't re-use the result register of the protected region in handle op for being the overall result.
p/toqast: 442c735 | jonathan++ | src/NQPQ/ (2 files):
Update try, CATCH and CONTROL for QAST.
p/toqast: c2b37e4 | jonathan++ | t/nqp/44-try-catch.t:
Eliminate a bogus test, plus some other updating to the exception tests.
p/toqast: 62dddd0 | jonathan++ | src/QAST/ (2 files):
Fix register allocation bug involving handlers and resumable exceptions.
21:45 ObseLeTe left 21:46 ObseLeTe joined
sorear ChoHag: no, it would copy them to the caller's namespace when Module is loaded 21:46
A.pm: 'module A; sub x is export { }' B.pm: 'use A;' 21:47
erm
A.pm: 'module A; sub x is export { }' B.pm: 'use A "x";'
ChoHag So does nothing replace the Module->import(LIST) from perl 5?
sorear Did you read lib.pm6 like jnthn said?
ChoHag It was 4 lines.
One of them was } 21:48
Didn't take long.
sorear Can you understand line 2?
ChoHag Unless the semantics of our are radically different, yes. 21:49
sorear line 2 is the answer to your question
ChoHag Right.
sorear line 2 replaces the Module->import(LIST) from perl 5
if you have the answer, why are you asking? *confused* 21:50
ChoHag Because I only had the answer just now when you showed me.
22:01 worr joined, worr left
sorear jnthn: I am going to make fairly significant changes to the way niecza handles 'my class Foo::Bar' and 'use Foo::Bar'. I think that these changes are necessary for proper handling of the multiload provisions of S11. I do not expect much short-term support. 22:10
22:12 Circlepuller_ joined, preflex_ joined 22:13 preflex left, preflex_ is now known as preflex 22:14 Circlepuller left 22:17 Circlepuller_ left
dalek p/toqast: 8d4da1b | jonathan++ | src/NQPQ/Actions.pm:
Fix closure interpolation.
 22:18
jnthn sorear: multiload as in, multiple versions? 22:20
sorear jnthn: yes 22:26
jnthn sorear: Happy hacking. If you can write up a description of what you do, I'd be interested to see it. 22:27
sleep & 22:35
masak 'night, #perl6 22:56
23:05 s1n left 23:13 MayDaniel left 23:18 tokuhiro_ joined 23:23 tokuhiro_ left, grondilu left 23:29 adu joined, gongyiliao joined 23:40 colomon_ joined 23:42 colomon left, colomon_ is now known as colomon 23:54 spider-mario 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!