»ö« | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, alpha:, pugs:, std:, or /msg p6eval perl6: ... | irclog: irc.pugscode.org/ | UTF-8 is our friend!
Set by moritz_ on 25 June 2010.
00:11 supernovus joined 00:24 xinzp joined, ciphertext_ joined 00:25 xinzp left
supernovus The one disadvantage of working on my netbook: the core setting take 10+ minutes to compile... :( 00:31
00:34 rgrau_ left
TimToady std: /[ ]/ 00:38
p6eval std 31520: OUTPUT«===SORRY!===␤Null pattern not allowed at /tmp/zsKkBtPwao line 1:␤------> /[ ⏏]/␤ expecting any of:␤ quantifier␤ regex atom␤Parse failed␤FAILED 00:01 111m␤»
supernovus Bah, what is the best way to share subs between different classes in the Rakudo core setting? I want to make assert-valid-date, days-in-month and is-leap from the Date class available to Temporal as well.
00:42 lkk- joined
TimToady you could try exporting from a class; dunno if rakudo supports that yet though (a class is supposed to "do" module...) 00:43
frew moritz_: hey, I saw that you ported (or are porting) CGI::Application
moritz_: you might want to port at least the basic parts of Plack 00:44
sorear yay it's TimToady
diakopter: reping
supernovus TimToady: Thanks, I will try that and see if the current rakudo supports it. 00:50
I'm guessing subs in roles don't get merged in classes that do those roles? 00:51
00:56 cls_bsd joined 00:57 masonkramer joined 00:58 ash_ joined 01:03 songmaster joined
supernovus Well, using our sub in a role worked. I'm out for dinner, but soon Temporal will have working validation on the parameters of new() 01:08
01:09 supernovus left 01:14 whiteknight left
dukeleto New blog post about PL/Parrot and PL/Perl6: leto.net/dukeleto.pl/2010/06/rakudo...abase.html 01:30
01:40 supernovus joined
supernovus Hmm, I'm not too familiar with how the core setting works, but if you have an 'our sub' in a class, role or module should it leak through to the global namespace? 01:57
ashleydev Is any of the viv / gimmy5 / STD bootstrapping on Perl 5 using Regexp::Grammars? 01:59
ash_ ashleydev: no, i don't believe they do 02:00
ashleydev hmm
i just discovered that and have almost wet myself... I'm shocked to have found it by searching CPAN rather than through the usual news channels... 02:01
(I wasn't looking for it, but for something else actually and have spend the last hour reading the doc)
[Coke] I only heard about it because Damian just showcased it at YAPC. 02:06
so it's not just you. =-)
ashleydev bummed I missed YAPC, and I didn't see any blogs talking about it yet... I googled it and the only ref I saw was for OSCON (future) 02:07
what was the response at YAPC?
02:09 agentzh joined
[Coke] something akin to "OOOOH IT'S DAMIAAAAAN" 02:16
ashleydev i still can't believe it... WOW
pmichaud exit
ww
sorear ashleydev: viv implements a -real- Perl 6 parsing engine; it' 02:17
s far more complete than even the one in rakudo
we haven't been using the Perl 5 regex engine for a couple weeks now
02:19 supernovus left
ashleydev so it's perl 5 implementation of a perl 6 grammar engine? 02:19
(viv that is)
sorear yes and no 02:22
ashleydev where can I get viv?
sorear viv is a compiler; it takes in Perl 6 and generates Perl 5 with a similar structure
the grammar engine is in Cursor and CursorBaser
src/perl6 in the pugssvn
ashleydev k
sorear note that viv relies on STD.pmc in order to parse STD.pm6; it's a bit bootstrapped 02:23
ashleydev so is it aiming to be an implementation of perl 6? 02:24
or just a parsing engine? -- I keep seeing people test syntax with the std: bot here.. but is viv aiming to be more than that? 02:25
sorear viv/Cursor is a sandbox where we play with parsing technology on a mature VM 02:29
TimToady
.oO(mature == "for mature audiences only")
 02:39
ashleydev so is that sandbox for proving via real world use the Perl6 grammar syntax? 02:40
02:42 TiMBuS joined
tylercurtis sorear: is there a way to use viv outside of pugs/src/perl6/? When I try, I get "cannot open unicode maps from ./lib . : No such file or directory" as an error. 02:43
02:44 justatheory joined
sorear tylercurtis: viv needs to find syml/CORE.lex.store in $STD5PREFIX, uniprops in $PERL6LIB, and RE_ast.pm et al in $PERL5LIB 02:45
tylercurtis: try running 'make dist', the files will be automatically sorted
tylercurtis sorear: thanks. I'll try that once I finish what I'm doing. 02:48
sorear TimToady: After calling .Bool, how does Perl 6 know if the return value of that is true or false? 02:55
03:02 jaldhar joined 03:09 jaldhar left, jaldhar joined, agentzh left
ash_ doesn't .Bool --> Bool? 03:13
sorear Given a Bool, can I just *assume* it's either Bool::True or Bool::False? 03:15
03:15 agentzh joined
lue how can I test if a variable is rw? 03:16
pmichaud latest thinking has been that Bool is defined as enum bool Bool <True False>
(although jnthn++ and I aren't sure that's going to hold up) 03:17
sorear How is bool defined?
lue sorta likes the sound of enum
pmichaud bool is a native type, like int
sorear is just now trying to figure out how to make enum work
pmichaud in general, bool has been described as a bit1
sorear: yeah, there's some unclear aspects of enums at the moment, at least for jnthn++ and I 03:18
lue is it possible to use filetest operators on a variable to test for rw/ro?
03:19 felliott joined
ash_ lue: yes 03:19
rakudo: my $a = '/etc/passwd'; say $a ~~ :r 03:20
p6eval rakudo 7df2c2: OUTPUT«Method 'r' not found for invocant of class 'Str'␤ in 'Enum::ACCEPTS' at line 4488:CORE.setting␤ in 'infix:<~~>' at line 322:CORE.setting␤ in main program body at line 11:/tmp/btUNrXWpTJ␤»
tylercurtis pmichaud: are all NQP regexes isa Regex::Method or a subclass?
pmichaud tylercurtis: checking 03:21
lue no, to test if the variable itself is rw/ro.
ash_ rakudo: my $a = '/etc/passwd'; say $a ~~ :e
p6eval rakudo 7df2c2: OUTPUT«1␤»
ash_ oh
hmm
pmichaud tylercurtis: only those that are being used in rvalue (non-void) context 03:22
sorear pmichaud: I need to pick your brain on just how Perl 6 lists are supposed to work
pmichaud lue: there's been speculation that testing for ro-ness would be handled by a role, or some other aspects of a trait 03:23
sorear: I'll answer as best I can :-)
lue because I noticed that := already has a definition in some P6 (well, it's just a die statement, but still), and thought of coding there. 03:24
sorear lue: in niecza you can test if a variable is read-only using VAR($x).readonly, as described in S06
pmichaud tylercurtis: NQP doesn't wrap all regexes in Regex::Method objects. I suppose it could do that.
sorear (actually you can't because the parser isn't written yet, but that's a SMOP)
tylercurtis pmichaud: if I'm understanding you correctly, any regex that got passed as a function parameter would be in non-void context and thus would be a Regex::Method? 03:25
pmichaud tylercurtis: correct.
tylercurtis: that's how we can get .match and other methods on the regex object itself. 03:26
tylercurtis pmichaud: okay. Thanks.
sorear pmichaud: so, uh. What are the important types for lists? 03:28
lue Well, I think something like $a ~~ :rw is appropriate. 03:29
pmichaud sorear: at the moment, Rakudo has a base List type
well, let me back up a bit further 03:30
here's a list of classes involved in lists and iterators
Iterable - things that support .iterator . Also the base class for things that can flatten in list context
sorear I'm trying to implement lists in niecza from first principles
pmichaud s/flatten/interpolate/
sorear I have the mechanics of scalar containers fully implemented and I have native CLR arrays/dictionaries/etc 03:31
pmichaud List - the basic lazy list type. A list is simply a list of objects, which can be containers and/or values
Array - a List where each element is its own scalar container.
sorear What, mechanistically, is "list context"?
pmichaud in general, list context is created by &infix:<,> 03:32
wait, that's not right.
sorear is currently implementing ClassHOW, but wants a head start on understanding lists 03:33
pmichaud &infix:<,> is used to create a Parcel. flattening list context on a parcel creates a List where list elements of the Parcel are considered to be "interpolated" into the overall list
so, with (1, 2, @a, 4).flat the elements of @a are considered to be part of the list 03:34
sorear I have this vague understanding that contexts in Perl 6 are mediated by methods
ie .Bool
pmichaud I'm not convinced that's the case (more) 03:35
sorear so... the compiler automatically calls .flat on values in some place?
ok...
(more) ack
03:35 agentzh left
pmichaud about a year ago there was a push to try to get all of the contexts to correspond to types, but that hasn't worked out 03:35
but yes, the compiler automatically calls .flat on values in some places
for example, assignment to an array variable imposes .flat 03:36
also, when a Parcel is converted to a Seq (e.g., in item context), it's flattened.
sorear What constitutes item context? When is .item called?
pmichaud it can be called explicitly; it's also called implicitly when assigning a value to a scalar container 03:37
my $x = (1, 2, @a); imposes item context on the parcel
rakudo: say (1,2,3).WHAT; my $x = (1,2,3); say $x.WHAT;
p6eval rakudo 7df2c2: OUTPUT«Parcel()␤Seq()␤»
pmichaud for Parcels, .item returns a Seq containing the values of the Parcel (flattened) 03:39
for Iterables, .item returns a non-flattening reference to the Iterable
for most everything else (that I've encountered thus far), .item returns self 03:40
sorear What does a non-flattening reference look like? 03:42
What about binding?
pmichaud in Rakudo, it's an ObjectRef with the 'scalar' property set. (I'm expecting to change the property name to 'item' at some point)
but basically, flattening behavior does not appear to be purely a property of the type of a value
i.e., if we have
my @a = 1,2,3; my $b := @a; say (4, 5, @a).elems; say (4, 5, $b).elems; 03:43
sorear I can do (in Mu) the equivalent of method item ($self is rw:) is rw { $self }, in niecza
pmichaud even though @a and $b are the same object, the first say will report 5 while the second say will report 3
sorear rw invocants are possible as a quirk of my model
I'm thinking this will come in handy here 03:44
03:44 dimid joined
pmichaud sure, Rakudo defines Mu.item as { self } 03:44
sorear does that work for rw?
03:45 sftp joined
pmichaud rakudo isn't yet distinguishing is rw, but yes, it would work for that. 03:45
(i.e., rakudo tends to treat all methods as 'is rw' at the moment)
(if the method returns a container)
sorear Why is self a container? 03:48
pmichaud it isn't always a container
sorear I thought all methods in Rakudo were delegated to the value
pmichaud oh, that's a good point.
you're correct, they are, so in Mu.item it's currently returning the value 03:49
anyway, I suspect that in general using .item should remove any containers from the invocant
sorear How does binding to $x is rw work, if .item is called?
pmichaud ah, that.
03:50 skids joined
pmichaud I suspect that binding isn't currently calling .item. checking 03:50
sorear in niecza, there is no such thing as container(s)
every variable has one and only one container
it feels like a much nicer model
pmichaud rakudo: sub xyz($a) { say $a.WHAT; }; xyz((3,4,5));
sorear I'm trying to keep it
p6eval rakudo 7df2c2: OUTPUT«Parcel()␤»
pmichaud looks like Rakudo isn't imposing .item on binding yet 03:51
sorear ok.
pmichaud in rakudo, every variable also has only one container, I think.
sorear Would it be possible to do away with 'scalar' by instead saying that the compiler doesn't .flat $-variables?
When exactly is .flat called?
frew so Mu is what p5 would call UNIVERSAL right? 03:52
pmichaud when something is put into flattening context
...which somewhat answers your question -- the compiler cannot statically determine flattening context 03:53
(it might be able to do so if it has complete knowledge of all possible signature binds... but in the general case that's not always available) 03:54
sorear What creates flattening context? 03:55
pmichaud Arrays, Seq, and '&flat'
(there might be more -- that's what I know of at the moment)
sorear What is '&flat' and where is it used?
pmichaud it's a contextualizer 03:56
along with &list, &slice, and &item
(I'm not sure that &list is useful yet.) 03:57
oh, and &hash
sorear What stops the return value of &flat from being contextualized again?
pmichaud nothing.
that's even normal (more) 03:58
for example, with my $x = (1,2,@a);
what happens is that a Seq is created that imposes flat context on the Parcel, and then item context is imposed on the Seq
so if I had 03:59
sorear What is the difference between foo(|@arr) and foo(|$arr)?
pmichaud my @a = (1, 2, 3, (item flat @c, @d)); then @a would end up with 4 elements, the last of which is a Seq of the flattened elements of @c and @d 04:00
it's generally more interesting to think of flat in terms of the return value of various parcel-generating operators 04:01
as in
my @a = (1, 2, 3, (item flat @c Z @d))
lue afk
pmichaud here, flat imposes flattening context on the results of @c Z @d, and item context converts that to a Seq
I'm not sure about the difference between foo(|@arr) and foo(|$arr) 04:04
jnthn++ might know. 04:05
I mean, I know what foo(|@arr) does, but foo(|$arr) seems a bit weird.
04:06 skids left
pmichaud oh, wait. I know. 04:06
in any case, foo(|$arr) basically does $arr.Capture
so it's up to the .Capture method to return a Capture object that then gets used in the binding 04:07
sorear What happens after my $x := @y; my @z := $x; for (@z) { .say } 04:09
?
What's the difference between &flat($a) and $a.flat? 04:10
04:10 chronniff joined, felliott left 04:11 chronniff left
ingy is there a text that describes the difference between token rule and regex in grammars? 04:12
sorear S05
ingy where
pmichaud sorear: @z gets flattened.
ingy I just read it
pmichaud ingy: regex is a normal regex -- it automatically backtracks on various quantifiers 04:13
ingy: token is a regex where backtracking is disabled by default
04:13 kaare joined
pmichaud ingy: rule is a token (backtracking disabled) where whitespace in the pattern corresponds to intertoken whitespace in the match 04:13
04:14 kaare is now known as Guest36062
pmichaud rakudo: say 'aaaaa' ~~ regex { a* a } # matches 04:14
p6eval rakudo 7df2c2: OUTPUT«aaaaa␤»
pmichaud rakudo: say 'aaaaa' ~~ token { a* a } # doesn't match
p6eval rakudo 7df2c2: OUTPUT«␤»
pmichaud sorear: the difference between &flat($a) and $a.flat is that the former flattens its parcel while the latter returns the flattened version of $a 04:15
ingy pmichaud: thank you 04:16
pmichaud rakudo: say "you're welcome" ~~ rule { \S* \S* }
p6eval rakudo 7df2c2: OUTPUT«you're welcome␤»
pmichaud note that in the rule I didn't have to specify anything to match the whitespace
ash_ does rule chomp front and back? 04:17
ingy rakudo: say "you're welcome" ~~ rule { \S* \S* }
p6eval rakudo 7df2c2: OUTPUT«you're welcome␤»
pmichaud rakudo: say " you're \n welcome " ~~ rule { \S* \S* }
p6eval rakudo 7df2c2: OUTPUT« you're ␤ welcome ␤»
ingy rakudo: say "you'rewelcome" ~~ rule { \S* \S* } 04:18
p6eval rakudo 7df2c2: OUTPUT«you'rewelcome␤»
sorear pmichaud: so, if $a is the Seq <a b c>, my @b := $a results in @b.elems == 3, but my @b = $a results in @b.elems == 1?
pmichaud + is more interesting here
rakudo: say "you'rewelcome" ~~ rule { \S+ \S+ } # fails
p6eval rakudo 7df2c2: OUTPUT«␤»
ingy ah 04:19
pmichaud sorear: correct.
ingy: rule { \S+ \S+ } is the same as token { <.ws> \S+ <.ws> \S+ <.ws> }
ingy ahrakudo: say "you're welcome" ~~ rule { \S+ \S+ }
rakudo: say "you're welcome" ~~ rule { \S+ \S+ } 04:20
p6eval rakudo 7df2c2: OUTPUT«you're welcome␤»
ingy pmichaud: maybe not :)
pmichaud ?
ingy isn't that <.ws>+
pmichaud <.ws> already allows for multiple whitespace characters
ingy I see 04:21
pmichaud it's basically defined as { <!ww> \s* }
ingy backs down
pmichaud where <ww> is true when in the middle of a word
ingy ;)
pmichaud (i.e., between two word characters)
so, <.ws> requires at least one space between two word characters, and zero or more spaces anywhere else 04:22
ingy ETOOMUCHPOWER!!
;)
ingy gets back to ascending the learning curve
pmichaud I should write a canonical blog post about regex, token, and rule -- it's a FAQ 04:23
afk for 15 -- walking furry canine creature that seems to inhabit my house
04:31 ash_ left 04:35 davidfetter left 04:45 tedv left 04:57 Guest36062 is now known as kaare_
kaare_ pmichaud: Doesn't the book contain a canonical chapter? 05:06
kaare_ downloaded book and will bring on vacation
05:25 LionMadeOfLions joined 05:26 justatheory left
sorear pmichaud: What I still don't quite get is binding to @ variables 05:41
moritz_ it explains the difference 05:42
sorear with $ variables, it is a universal truth that my $a := X1; $a = X2 means (X1).item = X2
but the identity does not seem to hold with @a and .flat 05:43
05:57 cono joined 05:59 ram__ joined, ram__ left 06:03 uniejo joined 06:05 mberends joined
mberends for MiniDBD::Pg, it's hard to choose between Parrot/Pg.pir and Zavolaj to reach the underlying libpq. There may be a way to have the cake and eat it, by including code for both options and then switching the module one way or another. That looks like a job for a macro. hmm 06:14
sorear After my $x; my @y := $x; what do fetches and stores to @y do? 06:20
mberends just a guess, but would @y[0] be bound to $x, and @y[1] and up be bound to an anonymous array? 06:23
it looks like a terribly confusing mess :/ 06:25
kaare_ mberends: Too me it looks like Zavolaj has a more fine grained access to libpg. 06:27
But that's without having looked very closely at it.
mberends kaare_: working with Zavolaj is nice, because it points directly into native libraries. Its main drawback is that it's an external dependency that needs to be installed. Rakudo * will bundle it though. 06:30
moritz_ mberends: Pg.pir doesn't (yet) give access to the error messages yet 06:31
and I think the right approach would be to have separate DBD modules for each approach 06:33
mberends moritz_: ah, didn't see that yet. The libpq API is so friendly though, getting error messages should be very easy.
moritz_ mberends: should I push my not-yet-quite-workiing Pg.pir based DBD module? 06:34
kaare_ I would be happy to work on a DBD::PgPir version, except my vacation makes me unstable :-)
moritz_ that's fine 06:35
mberends moritz_: yes, go for it!
kaare_ - and the fact that Pg.pir needs some updating.
mberends kaare_: the name PgPir is useful for having cake and eating it :) 06:36
moritz_ one thing I'd like to improve over the DBI interface is passing around paramaters in DBI.connect as a hash instead of a string 06:37
at least optionally
letting the user do the quoting seems wrong
mberends moritz_: as long as it's optional. I'd like to keep the P5->P6 migration path as easy as possible. 06:38
moritz_ understood. 06:39
kaare_ It is _so_ wrong. But a lot of these things are there because of age. DBI 2.0 should be designed from new. MiniDBI is OK as a 1:1 from P5 06:41
_someone_ with Parrot knowledge and tuits should look at Pg.pir (or postgres.pir?) and www.postgresql.org/docs/9.0/static/libpq.html 06:42
moritz_ pushed MiniDBD::PgPir
mberends moritz_++ 06:43
06:44 hugme joined
kaare_ pulled 06:45
mberends kaare_: yes, that's the Postgresql manual I used to make a Zavolaj client. I'm currently low on Parrot tuits, getting perldancer into P6 matters more to me ;) 06:46
kaare_ mberends: I was hoping maybe dukeleto *hint hint* ;-) 06:49
moritz_ I'm testing a last_error method in Pg.pir right now 06:51
mberends kaare_: yes :) I like his blog posting (of course;)
the Postgresql C API was so much easier to work with than the MySQL one, because the parameters passed and returned were simpler. Conversely, MySQL probably performs slightly faster because it passes more information per call. 06:56
sorear pmichaud: Are the current list-context tests all correct? 06:57
kaare_ moritz_: I get Typename MiniDBD::StatementHandle must be pre-declared to use it with does at line 3, near "{\n has " 06:58
I guess it's because MiniDBI isn't installed here. I hoped that a export PERL6LIB=lib would help, but no. 07:01
moritz_ I guess I have to 'use MiniDBD'
yep
07:01 mj41_ joined
moritz_ fix pushed 07:01
07:01 [particle]1 joined
kaare_ sorry. Forgot you mentioned it's untestet :-) 07:01
07:02 p6eval joined, songmaster left, PZt left, stef_ left, opx^away left 07:03 mj41_ is now known as mj41
kaare_ I changed the Makefile to include PgPir. Is it OK to push? 07:03
moritz_ sure
07:04 songmaster joined, PZt joined, stef_ joined, opx^away joined, felipe joined, thepler joined, tomaw joined, Gothmog_ joined 07:05 felipe left 07:16 rgrau joined 07:20 cuppe joined 07:21 cuppe_ joined
songmaster \o/ My p6 sudoku solver works! 07:29
mathw \o/ 07:32
07:35 songmaster left 07:40 agentzh joined 07:43 Ross joined 07:58 wallberg joined 07:59 felipe joined 08:00 timbunce joined 08:14 cygx joined 08:26 tadzik joined
moritz_ rakudo: say @*ARGS.perl 08:26
p6eval rakudo 7df2c2: OUTPUT«[]␤»
08:31 dakkar joined 08:32 Ross left
pugssvn r31521 | moritz++ | [t/spec] test @*ARGS 08:32
mathw moritz_: I must thank you again for introducing me to Wise Guys. Wenn Sie Tantzt is currently the only thing keeping me attached to my last tenuous thread of sanity. 08:33
moritz_ wise guys and sanity. /me chuckles :-) 08:34
IllvilJa mathw: isn't sanity overrated? GROSSLY overrated...?
dalek kudo: 3d2cb82 | moritz++ | t/spectest.data:
run test file for @*ARGS
 08:35
mathw IllvilJa: It depends which bit of sanity you lose...
Right now I'm about ready to start throwing people out of windows and through floorboards
IllvilJa mathw: please don't. It messes up the windows and floorboards very badly... 08:36
mathw I know
I doubt my employer would be very happy with the damage to the facilities 08:37
08:43 molaf joined 08:46 thebird joined 08:49 molaf left 08:52 masak joined
masak oh hai, #perl6! 08:52
phenny masak: 30 Jun 18:39Z <moritz_> tell masak github.com/moritz/modules.perl6.org - my approach at phasing out the web/ folder from proto
masak moritz_++ 08:53
I like 'modules'. it's not ideal, but it's far better than 'proto'.
08:54 Trashlord left 08:55 azert0x joined 08:56 azert0x left 08:57 azert0x joined, azert0x left
mathw oh hai masak 08:57
sorear Hello! Masak!
mathw enjoys a brief moment of seeing the SVN log as rather like 1080p CCTV footage of somebody breaking into a bank vault 08:59
09:01 azert0x joined, azert0x left, lkk- left 09:02 azert0x joined, azert0x left
masak rakudo: say +Failure 09:03
p6eval rakudo 7df2c2: OUTPUT«Use of uninitialized value in numeric context␤0␤»
09:05 azert0x joined 09:10 tylercurtis left 09:11 silug joined
sorear yay, niecza has a working-seeming rudimentary ClassHOW now 09:25
09:28 cygx left
mathw sorear++ 09:28
09:30 mikehh joined
sorear niecza has gotten to the point where typing in ASTs to test is getting tiring, so tomorrow I'm going to try and hook it up to STD.pm6 09:37
STD.pm6 will probably need a few changes for this to work; niecza is AFAIK the first compiler to be directly connected to STD 09:38
(viv, mildew, and elf work by processing match trees, not providing action methods, which is what I want to try and do)
mberends sorear: very nice. Mildew also ran STD.pmc in-process afair, but yes, sharing data only at the tree level 09:40
09:41 sorear sets mode: +o mberends
sorear mberends: did mildew or elf define any extensions to Perl 6? 09:41
niecza is going to have to, since I'm trying to define my setting in extended-Perl-6 09:42
(you can't well define a class that is its own metaclass, to name an egregious example)
mberends sorear: no, nothing beyond STD.pm6 language
masak I'm glad Perl 6 is among the answers here: stackoverflow.com/questions/23930/f...-languages 09:45
sorear out
mberends hi masak, I like their lolcode answer too :) 09:53
masak :)
jnthn o/ 09:56
09:56 sorear sets mode: +o masak
masak \o 09:56
09:57 sorear sets mode: +vv hugme p6eval, sorear sets mode: +oo [particle]1 ingy 09:58 sorear sets mode: +v IllvilJa, sorear sets mode: +v ilogger2, sorear sets mode: -v IllvilJa
bbkr Is there any direct replacement of $. from P5? or do I have to use ZIP operator like "for .lines Z 1 .. * -> $line_content, $line_number { ... }" ? 10:07
mathw masak: Awesome, they have it in Inform 7 as well :) 10:09
"To decide what number is the factorial of (n - a number):"
masak the more I learn about Inform 7, the more it looks like a complex incarnation of HyperTalk. 10:13
mathw It's bizarre 10:14
rather fun from time to time
not to mention that it's the only language in which I've ever successfully written a game :)
10:16 pmurias joined
pugssvn r31522 | pmurias++ | [mildew] [smop] say 7 works with an installed SMOP 10:18
jnthn But what about "say 42"?
pmurias checks ;) 10:19
phenny pmurias: 30 Jun 18:18Z <sorear> tell pmurias pong.
pmurias jnthn: works too
jnthn \o/ 10:20
;)
masak lunch & 10:25
pragma_ checks ;) 10:27
oh, it was some on-next-visibility-send-memo thing. 10:28
10:29 Su-Shee joined
Su-Shee good morning everyone. 10:29
pragma_ what's good about it?
Su-Shee everything you make good of it. ;) 10:30
10:35 agentzh left 10:40 araujo joined 10:42 clintongormley joined
pmurias pragma_: on-next-visibility-send-memo thing? 10:45
10:46 tadzik left
pragma_ pmurias: yes. 10:46
pmurias which thing did you refer to? 10:49
pragma_ phenny's thing. 10:51
pmurias understands 10:52
pragma_: i checked if "say 42" worked in mildew
10:59 tadzik joined 11:06 amkrankruleuen joined
sorear pmurias: you're very lucky I can't sleep tonight. 11:09
is there any chance you could be online at a time other than 4am US west coast time? Or should we take this to email?
bbkr: try for .lines.kv -> $n, $text { ... 11:10
pmurias checks what does 4am translate to in his timezone
arnsholt is occasionally convinced that everyone should just move to Swatch beats 11:11
bbkr sorear++ :) I always forget there is kv for arrays available now :)
pmurias sorear: i'm in UTC+2 11:13
sorear: what's the current time in your place?
s/time/hour/ 11:14
sorear 4:15 11:15
arnsholt California time, isn't it? 11:18
pmurias sorear: i could be on the channel any time while i'm awake (uni ended and i don't have anything planned for this week) 11:19
sorear arnsholt: I don't expect random Internet people with other-side-of-the-world timezones to know where California is 11:22
arnsholt Yeah, that's reasonable. I ask primarily so that I can place you on my mental people-time-map
For some reason I find it easier to remember California or New York than -9 or -6 hours 11:23
pmurias sorear: are you awake enough to answer STD questions? 11:30
sorear maybe
if not try again in ~14 hours
pmurias what would it take to make STD installable? 11:33
sorear if by "installable" you mean "like any other perl module" - that would require the CPAN folks to figure out what data files are 11:34
if you're willing to assume more of the issues yourself, you can use 'make dist', install the result of that somewhere, then set PERL5LIB, PERL6LIB, and STD5PREFIX to the right paths 11:35
pmurias you can install data file with File::ShareDir
sorear note: I don't recommend installing the PERL5LIB files into the system PERL5LIB
STD and Cursor are reserved module names in P6, not in P5 11:36
I'll have to check out File::ShareDir. When I'm awake. 11:38
pmurias ok 11:39
amkrankruleuen ls
pmurias sorear: if you need something done to help with STD being installable tell me 11:40
sorear: mildew needs an installable STD
masak std: STD, Cursor; 11:41
p6eval std 31522: OUTPUT«===SORRY!===␤Undeclared names:␤ 'Cursor' used at line 1␤ 'STD' used at line 1␤Check failed␤FAILED 00:01 112m␤»
masak TimToady: ^
sorear: it'd be very nice to be able to simply and reliably install STD as a module. 11:42
bbkr are there any differences between Perl6::Grammar available in Rakudo and STD.pm from Pugs? I'm thinking of nasty mutation testing: generate a lot of random trash like "{.{}}#@" and run it through STD.pm and if passes then on Rakudo eval_lives_ok. so we may find rare differences between STD.pm and Rakudo 11:43
masak bbkr: I had a similar idea (lots of random trash) for Rakudo once. I never realized it. Rakudo has gotten a lot more stable since. 11:44
11:44 hugme left
masak bbkr: I say go for it. if nothing else, you'll get a cool blog post out of it. 11:44
should be *fairly* simple to set up with a shell script, methinks. 11:45
11:45 pugssvn joined 11:46 pmichaud joined, Juerd joined
bbkr masak: yes, it's simple. I plan to do it when I finish RT queue checking for Kiev build, probably weekend. can I match against Perl6::Grammar to achieve the same result as on "std" bot here? 11:46
rakudo: say Perl6::Grammar.parse('say 1;') 11:47
11:47 clintongormley left
p6eval rakudo 3d2cb8: OUTPUT«say 1;␤» 11:47
bbkr rakudo: say Perl6::Grammar.parse('blablablabla')
colomon rakudo: say Perl6::Grammar.parse('say 1;').perl
p6eval rakudo 3d2cb8: OUTPUT«blablablabla␤»
rakudo 3d2cb8: OUTPUT«Method 'perl' not found for invocant of class 'Regex;Match'␤ in main program body at line 11:/tmp/eCBml2ykRT␤» 11:48
masak bbkr: no, Rakudo's Perl6::Grammar is simply the parser in Rakudo.
11:48 clintongormley joined
masak bbkr: to use STD, you need to use STD.pm6. 11:48
arnsholt rakudo: Perl6::Grammar.parse('oogabooga') && say $/ ?? "pass" !! "fail"
p6eval rakudo 3d2cb8: OUTPUT«fail␤»
arnsholt rakudo: Perl6::Grammar.parse('say 1;') && say $/ ?? "pass" !! "fail"
p6eval rakudo 3d2cb8: OUTPUT«fail␤»
arnsholt Aw. Dang =)
bbkr colomon: lack of .perl on Match() already seen it in RT 11:49
colomon rakudo: Perl6::Grammar.parse('say 1;') ?? "pass" !! "fail"
p6eval rakudo 3d2cb8: ( no output )
colomon rakudo: say Perl6::Grammar.parse('say 1;') ?? "pass" !! "fail"
p6eval rakudo 3d2cb8: OUTPUT«pass␤»
colomon I don't think .parse sets $/ 11:50
arnsholt Oh, right
11:51 Util joined
bbkr masak: do you know where can I get the code for std bot? I cannot just use STD.pm6 in Rakudo afaik, so I need to check how it works. 11:54
masak bbkr: STD.pm6 isn't in Rakudo. it's in the repo formerly known as the Pugs repo.
bbkr: in src/perl6
sorear No Perl 6 implementation can run STD.pm6
bbkr svn.pugscode.org/pugs/src/perl6/std - foud it :) 11:55
sorear So if you're looking to "use STD" - give up now
bbkr found*
11:57 ruoso joined
pmurias ruoso: hi 11:57
11:58 cono left 12:00 bluescreen joined, bluescreen left, bluescreen joined 12:01 bluescreen is now known as Guest93363 12:04 cono joined 12:06 kaare_ left
ruoso hi pmurias 12:11
pmurias ruoso: say 42 works with the installed SMOP, know fixing the bitrot on some tests 12:13
ruoso pmurias, that's awesome...
it should then be close to get "use v6-mildew" then ;) 12:14
pmurias i need to make STD installable and fix the bit rot so that Mildew builds core before that can happen 12:17
* CORE
ruoso Maybe we can convince TimToady to get STD released to CPAN, instead of embedding it in mildew 12:30
pmurias, ^
12:31 felliott joined
pmurias ruoso: working with sorear/TimToady to get STD released would be better then releasing Mildew::STD 12:32
12:35 Trashlord joined 12:37 cygx joined
ruoso pmurias, agreed... 12:38
12:39 rgrau left, rgrau_ joined
masak std: my $h = {}; say (%$h).keys 12:39
p6eval std 31522: OUTPUT«ok 00:01 114m␤»
masak rakudo: my $h = {}; say (%$h).keys
p6eval rakudo 3d2cb8: OUTPUT«===SORRY!===␤Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 11␤»
pugssvn r31523 | pmurias++ | [mildew] enable option bundling
r31524 | pmurias++ | [mildew] fix knowhow handling, ++FRONTEND --debug ++/FRONTEND
masak submits rakudobug
rakudo: my $h = {}; (%$h) 12:40
p6eval rakudo 3d2cb8: OUTPUT«===SORRY!===␤Confused at line 11, near "(%$h)"␤»
masak rakudo: my $h = {}; %$h
p6eval rakudo 3d2cb8: OUTPUT«===SORRY!===␤Confused at line 11, near "%$h"␤»
masak std: my $h = {}; %$h
p6eval std 31522: OUTPUT«ok 00:01 110m␤» 12:41
masak alpha: my $h = {}; %$h; say "alive"
p6eval alpha 30e0ed: OUTPUT«Symbol '%$h' not predeclared in <anonymous> (/tmp/sADQXLXV_d:10)␤in Main (file <unknown>, line <unknown>)␤»
masak huh.
sorear pmurias: next time I see TimToady, I'll pester him to find out it renaming STD to Perl6::STD is plausable
masak alpha: my $h = {}; %($h); say "alive"
p6eval alpha 30e0ed: ( no output )
masak :(
pmurias sorear: you removed $::ORIG? 12:42
sorear: VAST::Base->Str seems to complain that $ORIG is undefined 12:43
masak Match.caps rocks! moritz_++ 12:47
12:48 Guest23195 joined
colomon Match.caps? 12:50
colomon has been distracted by $work the last few days...
arnsholt captures, I think
sorear pmurias: yes... I was trying to make STD closer to reentrant 12:51
see viv:198 12:52
pmurias sorear: i need a way to set it from mildew 12:54
12:54 cuppe left 12:55 cuppe_ left
pmurias sorear: i shouldn't be keeping you awake, i'll change it back to an our and will work out a proper way tommorrow 12:55
12:55 cuppe joined, cygx left 12:56 cuppe_ joined
pugssvn r31525 | pmurias++ | [viv] make $ORIG a global 12:56
r31525 | [mildew] a lot of tests pass again
12:57 masonkramer left, masonkramer joined
sorear pmurias: um, what are you doing in mildew that you can't set it? 12:58
13:00 skids joined
diakopter pmurias: how goes gsoc project 13:01
13:01 foodoo joined
pmichaud good morning, #perl6 13:04
jnthn morning, pmichaud
masak morning, pmichaud 13:06
sorear good morning pmichaud 13:08
my $a = \( (1,2) ); my @b = 3, $a[0], 4; @b.elems ?
colomon o/
13:09 [particle]1 is now known as [particle]
masak ECMAScript 5 has "use strict"; -- ejohn.org/blog/ecmascript-5-strict-...-and-more/ 13:09
13:10 sorear sets mode: +voo pugssvn pmichaud ruoso, pmurias left
masak perhaps not news to people, I don't know... 13:10
13:10 dalek joined
sorear not quite news to me, but I had occasion to reread the ecmascript spec a few weeks ago 13:10
13:11 sorear sets mode: +v dalek 13:12 pmurias joined
diakopter strict mode in es5 is entirely to enable compiler optimizations by disallowing "highly-pessimizing" certain JS language features and enforcing several swaths of code contracts 13:13
pmurias: how goes gsoc project 13:14
13:18 isBEKaml joined
frettled diakopter: and I suspect that is a Good Idea, considering how people maul their JS code :) 13:18
pmurias diakopter: i'm behing schedule, an installed smop works, working on mildew now
cycling&
* behind
diakopter phenny: ask pmurias since mildew depends on pugscode's STD.pm6, will you rely on a p5 snapshot? and which one? 13:22
phenny diakopter: I'll pass that on when pmurias is around.
sorear pmurias asked me earlier about installable STD
diakopter oh 13:25
pmurias diakopter: if STD is installable and on CPAN, the version of mildew on CPAN should depend on that 13:26
phenny pmurias: 13:22Z <diakopter> ask pmurias since mildew depends on pugscode's STD.pm6, will you rely on a p5 snapshot? and which one?
diakopter I think perhaps sorear would be the most suited to prepare such a distribution 13:27
sorear yeah 13:28
I've been working with azawawi on that
to prepare the next version of Syntax::Highlight::Perl6 13:29
Juerd Oh wow, ecmascript "use strict" is not just literally "use strict", it is that *including* the quotes. A string in void context that *does something*. Eww! 13:30
That's almost as scary as meaningful comments (#line)
sorear Juerd: python docstrings have this beat by half a decade at least
13:30 felliott left
sorear Juerd: I recommend you never look up Smart::Comments on CPAN 13:30
Juerd Too late for that 13:31
sorear phenny: ask TimToady In the interests of pmurias & azawawi wanting to build installs, how would you feel about me moving src/perl6 into the Perl6::STD:: p5 namespace? 13:32
phenny sorear: I'll pass that on when TimToady is around.
13:33 foodoo left 13:34 felliott joined
arnsholt sorear: And the Python docstrings are probably nicked from Common Lisp =) 13:38
13:39 felliott left 13:40 envi^home joined 13:41 ash_ joined, wallberg left
masak in what ways do Python docstrings have semantics? 13:46
sorear they can be introspected 13:48
masak cf Perl 6. :)
mberends #!/usr/bin/env perl # this is not a comment ;)
masak mberends: on Windows, it is :P 13:49
sorear it's a comment to Perl
masak but yes, it's a form of out-of-band deus ex machina thing.
sorear it may or may not be a comment to fs/binfmt_script / execve, depending on where in the file it is 13:50
also that should be spelled #! /
masak why?
Trashlord yeah, I'm curious too as to why
sorear according to the autoconf portability guide, there are a fair few unices that parse a 4 byte magic number for scripts
ash_ masak: comments and python can do other things too, like run tests (see doctest) 13:53
masak ash_: sounds a lot like annotations, then. 13:54
sorear: huh. learn something every day.
ash_ masak: yeah, but like sorear said, they can be introspected, so you can do that dynamically at run time with anything, even C extensions to python have a place to specify the docstring 13:55
almost every object has a .__doc__ in python
which is kinda cool, in the interactive terminal if you say help(somefunction) it pulls up the docstring in a man-page like interface 13:56
masak nice.
ash_ its also nice because when you leave the manpage like interface you go right back to where you were in the interactive terminal, so no need to leave the terminal to check docs 13:57
wolverian now python only needs howdoIusethisthing(object) 13:58
ash_ well, in python there is only one way to do anything
13:58 uniejo left
ash_ if you try anything else, well people say your doing it wrong... 13:58
ash_ has some biased on python obviously...
masak is considering learning a bit more of Python 13:59
ash_ its useful to know, especially considering how much stuff is being done in/with python these days... 14:02
masak any reason not to make Match.keys return the names of the captured submatches?
14:04 pnate2 joined
ash_ masak: do you mean like / <ws> / ~~ ' '; $/.keys == <ws>? 14:06
masak yes, if by the last <ws> you mean 'list of one string "ws"'.
ash_ yea, like <a b c> (not a token) 14:07
or any other named regex
14:07 Siddy joined
ash_ what happens if you have 2 with the same name? 14:07
would it just list it twice?
14:08 pnate2 left
masak no, just like a hash wouldn't. 14:08
14:08 pnate left
masak Match already does Associative... 14:08
14:10 Trashlord left
ash_ so, if you captured more than 1 of a named something you would only have access to like the last capture? 14:11
masak no, .keys only gives the names of the subrules captured.
arnsholt ash_: No, it'll be a list, I think
ash_ ah, i see the difference now, thanks 14:12
masak according to S05, it'd be a list, yes.
but you'd access that as usual, through $/{$subrule}.
14:14 gbacon joined
colomon ooo, that reminds me. 14:21
I tried building the ABC module.
regex assertion not terminated by angle bracket at line 83, near "::basenote"
thats from the regex <ABC::basenote>.
nyi in master? 14:22
14:22 Guest93363 left
masak nyi in nqp-rx. 14:23
hm, if I knew how to contact LastOfTheCarelessMen, I'd suggest that he remove Makefile.in, Configure and lib/Configure.pm from his ABC and Vector repositories... 14:25
so I can remove those projects from the bottom of github.com/masak/ufo/blob/master/README :)
colomon Neither of them actually compiles under master anyway, it seems. :( 14:26
is there an effective workaround for the regex thing? 14:29
arnsholt masak: Doesn't he hang around here, but under some other handle?
colomon "last", but I don't think he's logged on in a long time.
masak huh. I had forgotten that. 14:31
colomon: re effective workaround -- not that I know of. if you find one, please let me know.
colomon bother.
14:34 rgrau_ left 14:36 Guest93363 joined
masak rakudo: my ($a, $) = 4, 5; say $a # this is the way to explicitly assign to nothing, right? 14:37
p6eval rakudo 3d2cb8: OUTPUT«4␤»
sorear right 14:38
masak rakudo: my ($a, *) = 4, 5; say $a # what about this?
p6eval rakudo 3d2cb8: OUTPUT«===SORRY!===␤Malformed parameter at line 11, near "*) = 4, 5;"␤»
masak could it be made to work?
is it a good idea?
the two forms would be synonymous, as far as I can see. 14:39
arnsholt Sounds reasonable to me
14:39 Mowah joined
arnsholt But I've come to understand that my understanding of Whatever is a bit limited 14:39
isBEKaml Hi, what does this do here? What exactly is "times" ? >>> return (time, times) »-« @s; # Benchmark.pm 14:40
masak I suppose * would be more of a special case than $ in the implementation.
rakudo: say times 14:41
isBEKaml I worked around this by initalising two variables with time around the loop that runs the code and returning back the difference.
p6eval rakudo 3d2cb8: OUTPUT«Could not find sub &times␤ in main program body at line 11:/tmp/jltqpeaHsP␤»
isBEKaml masak: yes. times gave me nothing. time did. system.pm I guess..
PerlJam isBEKaml: Are you converting the Perl 5 Benchmark?
ash_ masak: so would you loose the 5 in your example?
isBEKaml PerlJam: No, Pugs has Benchmark module. I'm trying to get it to work. :) 14:42
masak isBEKaml: perldoc -f times
isBEKaml PerlJam: made it work with the workaround I mentioned above.
masak ash_: yes.
PerlJam isBEKaml: what masak said. :)
ash_ masak: in perl5 wasn't that my ($a, undef) = 4, 5; ? 14:43
14:43 patspam joined
PerlJam ash_: still is :) 14:43
masak ash_: yes, and the corresponding (s/undef/Mu/) would work here too, I think.
isBEKaml: also, github.com/perlpilot/benchmark/
PerlJam ash_: well, with () around the RHS 14:44
ash_ rakudo: my ($a, Mu) = 4, 5;
p6eval rakudo 3d2cb8: ( no output )
isBEKaml Ah, ok. I guess I could get this to work by calling something to akin to system("time").. Until we have it in Rakudo.. :)
ash_ cool
isBEKaml rakudo: say time 14:45
p6eval rakudo 3d2cb8: OUTPUT«1277994371.64924␤»
masak isBEKaml: &times is only mentioned twoce in very unlit corners of S29 right now. 14:46
14:46 timbunce left 14:47 felliott joined
isBEKaml masak: times is just mentioned under Other Functions. :) No definition. 14:47
masak s/twoce/twice/ 14:48
isBEKaml the second one being part of "set priority times"
masak: I just looked at perlpilot++'s Benchmark module. The code looks similar to what I did in Pugs' Benchmark module. 14:52
:)
14:52 jaldhar left, Siddy left
isBEKaml I mean, it doesn't use times - just time for the difference. Btw, how do I run the system('time') ? exec::run? 14:53
14:54 timbunce joined
isBEKaml D'oh. system.pm -> run("time") 14:55
PerlJam isBEKaml: run("time $proggy");
isBEKaml PerlJam: thanks. :) 14:56
14:56 felliott left 14:57 mberends left
pugssvn r31526 | sorear++ | [STD] add a hook for subclasses to override initial slangs 14:58
PerlJam wow, pugs' Benchmark.pm does look remarkably like mine. I guess even in a world of TIMTOWTDI, there's only so many ways to express a given thing. 14:59
15:00 proller joined
PerlJam though pugs' version was continuing the trend of tightly coupling the output and the timing info. 15:02
That's something that bothers me for some reason
pmichaud re-good morning #perl6
PerlJam greets pm
isBEKaml greetings, pm!
PerlJam: yeah, I was kinda confused when I saw that module. particularly, how it went about timing the loops. 15:03
colomon \o
isBEKaml wow, I really should start thinking in terms of unicode ops... :| 15:08
pmichaud interesting. It's my ($a, $) = 4, 5; if declaring $a but ($a, *) = 4, 5; if $a is already declared.
jnthn Aye 15:09
$ is not a valid term
std: $
p6eval std 31525: OUTPUT«===SORRY!===␤Non-declarative sigil is missing its name at /tmp/kKTvV_drkM line 1:␤------> <BOL>⏏$␤Bogus statement at /tmp/kKTvV_drkM line 1:␤------> $⏏<EOL>␤ expecting twigil␤Parse failed␤FAILED 00:01 112m␤»
pmichaud I'm more curious that * isn't a valid param. :-)
jnthn Guess we maybe could make it one
pmichaud jnthn: I have a question about roles ... :-)
jnthn Though is *** then a slice-slurpy *? ;-) 15:10
pmichaud: If you're really lucky, I may have an answer. ;-)
15:10 alester joined
pmichaud I'm trying to re-implement WHENCE 15:10
masak pmichaud: could 'my ($a, *) = 4, 5;' be made legal?
pmichaud the standard mechanism is Foo but WHENCE({...})
how should WHENCE be defined here?
a sub returning a role? 15:11
15:11 mikehh left
pmichaud a role itself? 15:11
masak pmichaud: ah, you already said it could.
pmichaud masak: I don't know if that could be made legal or not... seems like an odd disjointness if it isn't, though.
jnthn thinks
pmichaud: Give me a moment to check something in spec...
pmichaud well, is there a syntax for a role that would accept parens? 15:12
masak pmichaud: yes, I agree.
jnthn pmichaud: OK, checked what I wanted to
pmichaud: does Role($param) is syntactic sugar, afaict from S14
Here is the key langauge from S14: 15:13
A role applied with C<does> may be parameterized with an initializer
in parentheses, but only if the role supplies exactly one attribute
to the mixin class:
$fido does Wag($tail);
$line does taint($istainted);
Note that the parenthesized form is I<not> a subroutine or method call.
It's just special initializing syntax for roles that contain a single
property.
pmichaud jnthn: looking
jnthn We had that in alpha but it ain't in master again yet.
But this would suggest that WHENCE here is the role
pmichaud agreed, and it mixes in a property
jnthn And it has one "has $.foo" (maybe even it's $.WHENCE) 15:14
pmichaud right
but nyi in master, yes?
jnthn That syntax isn't, no
It does seem to be clearly syntactic.
pmichaud yes.
jnthn But it should be easy
pmichaud I'd be a little concerned about a potential conflict with the coercion syntax, however.
jnthn It's just an extra arg to does and but
Yeah, but you'd not do a coercion on the RHS of does and but, probably? 15:15
I suspect it's a reducecheck thingy.
pmichaud oh, you're saying it's a coercion tied to the actual &infix:<does>
jnthn Others have remarked that they'd a less constrained syntax here
pmichaud hmmmmm
jnthn For setting up more than one attr.
pmichaud: Yes, I've seen it as another arg given to does and but. Just another multi candidate 15:16
pmichaud okay.
jnthn It'd delegate the first args to the two arg form of does and but
And then apply the value for the third.
sorear I'd like to argue for enums NOT being subtypes of their base type
jnthn All of does and but are implemetned in the setting btw.
pmichaud yes, I saw that. 15:17
jnthn Just in terms of the meta-model.
sorear enum Foo <a b c>; Foo ~~ Int; # this blocks a lot of optimizations since Int tends to be final otherwise
jnthn I'm not sure if Foo ~~ Int so much as all(a, b, c) ~~ Int
ash_ rakudo: sub foo { try { return 42; }; }; say foo; # is that right? 15:18
p6eval rakudo 3d2cb8: OUTPUT«Null PMC access in type()␤ in main program body at line 11:/tmp/4oJG5TUxUl␤»
sorear either way
jnthn No, a Null PMC access is never right.
sorear a ~~ Int is bad type safety juju and breaks auto-finalize
ash_ err, i meant, it should return right? try shouldn't catch the return should it?
jnthn ash_: No, it should not
I guess that's the real bug there.
pmichaud jnthn: so, is it worth trying to get the does Wag($property) syntax working, or should I cheat a bit on WHENCE for now? 15:19
I'm fine either way. 15:20
jnthn rakudo: role Foo { method lol { "omg" } }; my $x = Int but Foo; say $x.lol
p6eval rakudo 3d2cb8: OUTPUT«Attempt to use rebless_subclass where the new class was not a subclass␤ in 'infix:<does>' at line 38␤ in 'infix:<does>' at line 456:CORE.setting␤ in 'infix:<but>' at line 483:CORE.setting␤ in main program body at line 11:/tmp/V5_6xznVkB␤»
jnthn oh grrr
pmichaud: It's worth getting that working
pmichaud rakudo: role Foo { method lol { "omg" } }; my $x = 3 but Foo; say $x.lol'
p6eval rakudo 3d2cb8: OUTPUT«===SORRY!===␤Confused at line 11, near "say $x.lol"␤»
jnthn pmichaud: Though I think we have a slight problem with mixing into type objects at the moment too
pmichaud rakudo: role Foo { method lol { "omg" } }; my $x = 3 but Foo; say $x.lol;
p6eval rakudo 3d2cb8: OUTPUT«omg␤»
pmichaud yes, it appears we do.
rakudo: role Foo { method lol { "omg" } }; my $x = Int.clone but Foo; say $x.lol; 15:21
p6eval rakudo 3d2cb8: OUTPUT«Attempt to use rebless_subclass where the new class was not a subclass␤ in 'infix:<does>' at line 38␤ in 'infix:<does>' at line 456:CORE.setting␤ in 'infix:<but>' at line 483:CORE.setting␤ in main program body at line 11:/tmp/VC_NbZTcob␤»
jnthn Let me see if I can spot why.
pmichaud okay.
jnthn (src/metamodel/RoleToInstanceApplier.nqp if you want to look too, btw) 15:22
pmichaud I'm heading to lunch
if it ends up being an easy fix, the I'm good with that.
If not, I'll cheat around it, but keep the model in mind so it's an easy switch later.
jnthn oooh
It's annoying.
arnsholt pmichaud: Is implementing languages on Parrot with NQP fit under your plan for the NQP book? 15:23
jnthn $subclass.HOW.add_parent($subclass, $target.WHAT);
pmichaud arnsholt: yes.
frew is there a clever way I can do an ordered reduce in p6?
arnsholt Cool. Then I'll try to write up my recent experiences for that
pmichaud arnsholt: excellent.
colomon frew: they are always ordered.
jnthn pmichaud: The thing is that type objects today are actually a Parrot subclass.
frew the idea being [&&] @fn.map: { $_() } 15:24
colomon frew: you mean you want to change the order?
jnthn pmichaud: But add_parent does some "unwrapping" to get to the Parrot class from the type object.
frew colomon: no, I want it to stay in order, and since it's a bool I want it to short circuit
jnthn pmichaud: Thus it ends up making the Parrot class that represents non-type object instances the parent of the generated subclass. 15:25
colomon frew: I don't think reduce short-circuits at the moment.
frew does map do that stream thing yet? because for the short-circuiting to matter the entire map shouldn't get calculated
jnthn pmichaud: I think it's probably a p6object behavior.
colomon frew: map is lazy, yes. 15:26
frew ok, cool
pmichaud frew: reductions don't short circuit
frew so reduce should be short-circuiting but NYI?
pmichaud let me rephrase that
frew so what's the right way to do it? junctions?
pmichaud a reduction is allowed to short-circuit, but it still evaluates all of its arguments
so, for example, [&&] 0, 1, 2, foo() 15:27
can still call foo() prior to invoking [&&]
15:27 meppl joined
pmichaud however [&&] fn-returning-lazy-list() 15:27
isn't required to eagerly evaluate the entire list before [&&] gets to it 15:28
colomon frew: does that mean you do not care about the return value of the reduced?
*reduce
pmichaud I don't know if [&&] is required to evaluate its list lazily. We could probably make it do that. 15:29
jnthn rakudo: sub foo() { say "oops" }; say [&&] gather { take 0; take 1; take foo(); }; # curious
p6eval rakudo 3d2cb8: OUTPUT«===SORRY!===␤Could not find sub &infix:<&&>␤»
frew hm
colomon pmichaud: I believe it is lazy but not short-circuiting at the moment.
pmichaud checks. 15:30
jnthn huh, I thought that was added the other day
pmichaud it was added, yes.
p6eval not up-to-date, likely.
jnthn And is tested.
Oh
:S
frew so gather is lazy?
pmichaud frew: yes, almost by definition :-) 15:31
frew I can see that
15:31 Ross joined
jnthn > sub foo() { say "oops" }; say [&&] gather { take 0; take 1; take foo(); }; 15:31
oops
0
aww
colomon I think for a little while reducewith did short-circuit, but we moved away from that for some reason. 15:32
masak IIRC [&&] shouldn't short-circuit...
pmichaud colomon: I think in order to have short-circuiting reduce we'd have to define it as multi sub infix:<[&&]>(*@values) { ... }
frew why shouldn't it?
15:32 Su-Shee left
frew I can understand that reduce may not always want to be lazy, but sometimes it would be nice 15:32
colomon pmichaud: I mean short-circuiting in the lazy sense, not in the not evaluating its arguments sense. 15:33
arnsholt It'd be hard to implement general-purpose short-circuiting for reduce I think
pmichaud colomon: yes, that's what I mean also.
frew arnsholt: right, it mostly just applies to booleans
colomon I don't think there's any issue implementing the lazy sense at all.
masak S03:4431: "Similarly, list-associative operators that have the thunk-izing characteristics of macros (such as short-circuit operators) lose those macro-like characteristics." 15:34
frew maybe something clever could be done with reduce in boolean context?
colomon it's the $chaining block in the non-triangle portion.
pmichaud colomon: && isn't chained.
colomon Oh, right!
never mind, then.
15:35 synth left
pmichaud masak: I think that S03:4431 says that all of the arguments are evaluated 15:35
15:35 Ambiguity joined
pmichaud masak: but that doesn't say anything about whether the evaluation of [&&] could short-circuit on the first false argument it finds 15:35
sorear what is the difference between ?&, &&, and, and andthen?
pmichaud ?& is boolean and
&& is short-circuit and
'and' is short-circuit and, low precedence 15:36
ash_ & is also useable, it makes a junction
15:36 Su-Shee joined
masak pmichaud: er. I almost don't see that distinction. 15:36
pmichaud masak: for example, [&&] 0..* could return false without having to evaluate the infinite list.
masak ah, right. infinite lists.
pmichaud in that sense, [&&] is lazy but not thunking 15:37
masak I was going to say "I see only a difference in speed".
pmichaud or, to make it more concrete
frew pmichaud: so how is that different than my use case?
pmichaud frew: it's not.
frew ok, so it's just NYI then?
frew can't understand which is spec and which is impl :-)
sorear if (node && node->valid) { ... // the #1 reason short circuiting exists in any language
pmichaud frew: we're trying to decide what the spec is.
frew heh
pmichaud frew: currently there's nothing that says that [&&] has to be lazy. 15:38
I don't know if there's something that says that [op] is eager.
frew ok, well, I hope the spec supports my use case, because I know that at the very least it will be useful for some of the stuff we do at work and in DBIC often
pmichaud I'm guessing no -- that in general reduce tries to be lazy.
15:38 synth joined
ash_ sorear: if ($node.?valid) { # woo perl6 15:38
pmichaud we're also exploring the distinction between laziness of reduce and thunking behavior
frew ok, so I'm not the first person to discuss this then 15:39
pmichaud for example, 0 && foo() would mean that foo() never gets invoked. It's "thunked"
frew right
and that's what I care about
(I think)
pmichaud but [&&] 0, foo() would still result in foo() being invoked, even though [&&] might short-circuit and stop evaluating its argument list as soon as it sees the 0.
frew because the list isn't lazy?
pmichaud because [&&] doesn't thunk its arguments. 15:40
while && does.
jnthn With [&&] at the moment, you have to go thunk yourself.
frew wow, can you tell p6 a function arg to be thunked?
for anything?
pmichaud frew: I don't think that's specced yet.
frew ok
well
that would be really powerful
jnthn frew: No, at the moemnt it's just handled by the compiler as a special case.
pmichaud (and we've been curious about ways to spec it :)
frew ok, that's less surprising 15:41
generic thunking would be crazy (in a good way)
but then so is p6 so w/e
15:41 mmcleric joined 15:43 rgrau joined
colomon btw, the triangle form of reduce *is* lazy: 15:44
pmichaud yes, triangle form is explicitly lazy.
colomon rakudo: say ([\+] 1...*).munch(20)
frew ok, well, I was just doing it in p6 and was curious how to do it, since of course I have to use a for loop and a return in the middle
p6eval rakudo 3d2cb8: OUTPUT«13610152128364555667891105120136153171190210␤»
colomon rakudo: say ([\+] 1...*).munch(20).perl
p6eval rakudo 3d2cb8: OUTPUT«(1, 3, 6, 10, 15, 21, 28, 36, 45, 55, 66, 78, 91, 105, 120, 136, 153, 171, 190, 210)␤» 15:45
colomon frew: Can you tell us more about what you are trying to do?
frew colomon: ure!
sure*
pmichaud afk, lunch
frew basically, we have an alarm system (p5) that checks to see if any of the alarms conditions are met 15:46
15:46 macdaddy joined
frew if any of them are met then a bunch of notifications are sent out 15:46
15:46 macdaddy is now known as Guest42363
frew checking conditions is expensive 15:47
so I need to NOT do that if possible
colomon so the things in your map statement there are expensive
frew very yes :-) 15:48
colomon and you still need to know if the result is true or false.
frew otherwise I wouldn't care if they got ran
right
so normally I just do return unless $_->check($observation, $dms);
colomon are their values significant other than true or false?
frew nope 15:49
not in the lease
least*
sorear Bleh. 15:50
When TimToady fixed Test::Util to generate Test/Util.syml, he broke STD:auth<perl.org>
colomon frew: so you need something like for @fn { return False unless $_() }; return True; 15:52
colomon is trying to think if there is a shorter way to do that.... 15:57
oooo, here's an evil one:
[==] 1, @fn.map({ $_() })
of course, that only works if chaining functions can stop early in reduce, which I think is still ambigious. 15:58
*ambiguous
15:58 isBEKaml left
colomon can't decide if he loves that one or hates it. 15:59
16:02 charsbar joined, Grrrr joined 16:03 cdarroch joined, cdarroch left, cdarroch joined 16:16 gbacon left 16:17 TiMBuS left
dalek psi: cc725ab | masak++ | (4 files):
[Yapsi] implemented if, if/else statements

  (jump unconditionally).		 16:18
moritz_ back
masak moritz_! \o/ 16:19
moritz_: so, what's the next step for github.com/moritz/modules.perl6.org ? 16:20
16:20 justatheory joined
moritz_ masak: getting a nod from somebody; then installling it 16:20
[Coke] ~~ at frew.
masak moritz_: sir, I nod at you. 16:21
moritz_ feels nodded
PerlJam what is modules.perl6.org exactly? 16:22
moritz_ PerlJam: it's basically the same as proto.perl6.org, but without proto
16:22 christine joined
PerlJam gotcha 16:22
16:23 Sarten-X joined
masak just like pls is basically the same as proto, but without proto. :) 16:27
jnthn And a more pleasing name. 16:28
masak *rimshot*
16:29 xinming_ joined
moritz_ modules.perl6.org/ there you go 16:31
PerlJam masak: funny. I always insert a mental *rimshot* rather than type it because there are so many times in #perl6 that it would be appropriate that most of our conversations would be *rimshot* punctuated by pun
masak PerlJam: the conversations would be rimshot to death. 16:32
ash_ how is pls coming along?
masak ash_: it works as a proof-of-concept.
ash_: still need to test it for all projects, and likely generalize it in places.
might get a round tuit this weekend.
jnthn eyes a small wooden item on his desk 16:33
masak today is Yapsi release day :)
16:35 hugme joined
moritz_ hugme: add masak to modules.perl6.org 16:35
hugme hugs masak. Welcome to modules.perl6.org!
masak \o/
[Coke] ponders bringing a drumkit noise maker to the next hackathon. 16:36
masak moritz_: is it OK if I add all the proto contributors to modules.perl6.org?
moritz_ masak: proto.perl6.org now redirects to modules/
masak: sure
masak gotta run now, but I'll do it laterz.
moritz_ it's a project we could migrate to the perl6 account over time
masak o/
16:36 masak left
moritz_ \o 16:37
[Coke] "has tests" (but do they pass? ;)
bbkr rakudo: $_ = {.()}; .() # bomb! hide! 16:38
p6eval rakudo 3d2cb8: OUTPUT«maximum recursion depth exceeded␤ in <anon> at line 1:/tmp/feZlmxWl0P␤ in <anon> at line 11:/tmp/feZlmxWl0P␤ in <anon> at line 11:/tmp/feZlmxWl0P␤ in <anon> at line 11:/tmp/feZlmxWl0P␤ in <anon> at line 11:/tmp/feZlmxWl0P␤ in <anon> at line 11:/tmp/feZlmxWl0P␤ in <anon>
..at l…
16:38 amkrankruleuen left
moritz_ that's a nice one 16:39
[Coke]: determining if tests pass is far more tricky than detecting their presence
PerlJam anyone who used proto "has tests" 16:40
jnthn rakudo: .() given {.()}
p6eval rakudo 3d2cb8: OUTPUT«invoke() not implemented in class 'Any'␤ in <anon> at line 11:/tmp/bN9jE6hdr4␤ in main program body at line 11:/tmp/bN9jE6hdr4␤»
jnthn Aww
16:48 rv2733 joined
ash_ jnthn: you use windows, right? 16:51
jnthn right
ash_ is there any way to make an installer for a .dll and header files?
i don't know windows so...
pmichaud back from lunch 16:52
jnthn ash_: Should be possible.
It's putting the header files in a sensible place that's maybe the slightly trickier bit.
But don't see why it wouldn't be doable. 16:53
ash_ is there a standard format for that sort of stuff if you use like Visual Studio? 16:56
jnthn "standard format"?
For installers?
bbkr rakudo: [{.()}, *.() .. *] # /me wonders why it runs infinitely instead of reaching recursion depth allowed
ash_ yeah, for installers
p6eval rakudo 3d2cb8: ( no output )
jnthn The MS one is MSI (MicroSoft Installer)
ash_ for comparision, i know on OS X you can make a 'framework' thats a bundle of the .dylib, and .a files, as well as the headers, and they are versioned so you can include multiple versions, etc. 16:57
jnthn There's various tools that produce them
It's not specific to installing just developer stuff though
ash_ hmm, alright
the only other projects I have seen that use libffi on windows include the source with their project and have modified build instructions, so they aren't building it as a separate .dll 16:58
wolverian look at what Gtk+ does 16:59
wolverian shrugs
16:59 dakkar left 17:01 tylercurtis joined 17:02 Ross left, Ross joined
pmichaud jnthn: I'm going to fix closures first before continuing with WHENCE, since WHENCE will need working closures for autoviv :-) 17:10
(so I expect to fix closures today.)
colomon \o/ 17:14
17:15 isBEKaml joined
bbkr std: for sub{1}, {2} {.WHAT.say} 17:16
p6eval std 31526: OUTPUT«ok 00:01 113m␤»
bbkr rakudo: for sub{1}, {2} {.WHAT.say} # works... impressive :) 17:17
p6eval rakudo 3d2cb8: OUTPUT«Sub()␤Block()␤»
pmichaud yes, sometimes rakudo actually works. :-)
jnthn :P
pmichaud: +1
17:18 thebird left
jnthn pmichaud: In a sense, that's even more important. :-) 17:18
pmichaud well, it's also likely to involve a fair bit of refactoring :-|
but I've been doing a lot of that lately :)
jnthn pmichaud: I may be able to work something out for this mixing in to proto-objects bug too.
isBEKaml beyond all sanity of the devs themselves.. Now that's impressive Rakudo power! :)
pmichaud jnthn: okay, that will be great if so. 17:19
but me working on closures today will likely mean I won't be on WHENCE again until late tonight or sometimes tomorrow (thus you have a bit more time :-)
jnthn Aye
I'll be doing a lot of Rakudo hacking at the weekend 17:20
17:20 pyrimidine joined
jnthn (Hackathon!) 17:20
isBEKaml 3 weeks out and I badly want to get back into Rakudo.. :(
jnthn But hope to have a bit of time tonight too.
bbkr std: 1 X 1 X 1 17:23
p6eval std 31526: OUTPUT«ok 00:01 109m␤»
bbkr rakudo: 1 X 1 X 1
p6eval rakudo 3d2cb8: OUTPUT«No applicable candidates found to dispatch to for 'infix:<X>'. Available candidates are:␤:(Any $lhs, Any $rhs)␤␤ in main program body at line 11:/tmp/0guV9ORhE8␤»
bbkr known bug?
pmichaud rakudo needs slice context parameters
(yes, known bug) 17:24
I've been thinking I might be able to cheat with a Parcel and .slice, though.
17:24 clintongormley left
lue ohai 17:24
17:25 hercynium_ left 17:27 eternaleye joined
bbkr pmichaud: thanks 17:28
17:29 Siddy joined
bbkr rakudo: nextsame # hmm, error message is LTA 17:31
p6eval rakudo 3d2cb8: OUTPUT«Null PMC access in clone()␤ in main program body at line 1␤»
bbkr reports
17:33 Siddy left
pmichaud jnthn: ping 17:38
jnthn pmichaud: pong 17:39
pmichaud perhaps you could help me walk through block construction? 17:40
we have PAST::Block nodes, which translate into the parrot subs
jnthn Right.
pmichaud we have Code objects that wrap the Parrot subs
17:41 clintongormley joined
pmichaud signatures get attached to... ? 17:41
jnthn Unless they are immediate blocks, we then call create_code_object to wrap them.
The sub add_signature is always used to attach a signature to a block.
pmichaud to PAST::Block ?
jnthn It expects to be passed the PAST::Block.
It adds the call to the signature binder 17:42
What then happens is either:
pmichaud what is $!signature and $!lazy_sig_init in Code, then?
jnthn If the $lazy parameter is set to true, then it returns a reference to a block that will create the signature 17:43
(A PAST::Val node with the block as the value)
We provide this to create_code_object.
If lazy is not set, it just emits a :load :init block that attaches the signature at loadinit time.
Well
It uses .loadinit actually
Rather than setting the PIR flags on the block directly.
See line 2677 in Actions.pm 17:44
pmichaud okay, I see that it's still working with the PAST::Block 17:45
jnthn Yes
The wrap is the last thing we do.
$!lazy_sig_init is one of the things we pass along to the Code constructor.
That is, it's provided as an argument to it.
Note that $!signature is the high level, Perl 6 Signature object 17:46
It's not the sig PMC
That still gets attached to the block as a property for now.
pmichaud (just a sec, background events) 17:47
jnthn
.oO( can we upgrade pmichaud++ to multi-core? )
 17:48
tylercurtis jnthn: not without having to worry about him ending up in a mental hospital. 17:49
pmichaud okay, so the $!signature property on a Parrot sub is a P6LowLevelSig 17:50
?
jnthn Correct.
That's what the binder cares about.
pmichaud and the $!signature attribute on a Code object is a Signature
jnthn Right
pmichaud (Perl 6)
jnthn Which the binder doesn't care about - it's mostly just for the sake of enabling introspection.
We build it and cache it
Since it costs a little to build. 17:51
Plus when we get custom parameter traits we'll need to have done so anyway.
pmichaud so, there are two places that can produce a lazy sig 17:52
one is in Code.signature
another is in the binder, when invoking the sub?
wait, I said that wrong
there are two places that evaluate the lazy sig 17:53
one is in Code.signature
the other is in the binder
?
bbkr rakudo: use MONKEY_TYPING; augment class Any { method invoke {1} }; .() # monkey_typing bug?
p6eval rakudo 3d2cb8: OUTPUT«invoke() not implemented in class 'Any'␤ in main program body at line 11:/tmp/e7TjCZoxtm␤»
jnthn pmichaud: Actually the code that does it lives in the dynop bind_signature 17:54
pmichaud hmmmm
jnthn pmichaud: At the moment it only actually invokes .signature
pmichaud oh
so it's all in .signature
jnthn Right
Maybe a tad wasteful in a sense
arnsholt hates the blank page effect
jnthn But at least means it's only done in one place. :-)
pmichaud well, not only that, but I think I'd prefer it to be in the parrot sub somehow
jnthn prefer...what to be in the Parrot sub somehow? 17:55
pmichaud I'm worried about the possibility of ending up with a PAST::Block with no Code wrapper
the lazy sig evaluation
17:55 rgrau left
jnthn The lazy sig thing and the code wrapper only ever happen together 17:55
pmichaud i.e., I don't think the Code object should be keeping track of the lazy sig generator -- that seems like it should be entirely a property of the Parrot sub
jnthn Then we'd have to have a load init to attach the lazy signature builder 17:56
Which we get to avoid now.
Saving all those invocations at startup was kinda the point.
pmichaud thinking.
17:58 synth left
pmichaud does every perl 6 block end up with exactly one Code.new invocation? 17:58
17:58 sorear sets mode: +v hugme
pmichaud or do we somehow end up with wrapper objects (e.g., in a loop)? 17:58
*multiple wrapper objects
jnthn In a loop I think it's an immediate block so we don't get that 17:59
pmichaud I mean a closure within a loop
jnthn Otherwise I think there are situations where we may end up with multiple
pmichaud for @list { my $sub = { ... } }
not the loop block itself.
jnthn Ah
pugssvn r31527 | sorear++ | [STD] Unbreak 'use STD' (the longname changes got more than they bargained for)
jnthn I'm not sure what happens there today.
That's probably the source of our problems though, iiuc. 18:00
pmichaud it's a part of it, yes.
jnthn I think they end up sharing
pmichaud I'm expecting to do some refactoring of Code generation.
jnthn When we shoulda cloned or some such.
pmichaud yes, but we also have to be careful about cloning
we don't want to clone the lazy_sig_init, or the sig
I mean, we *can*, but that gets expensive too :-)
jnthn Right 18:01
I'm pretty sure .clone today is smart about what it clones.
pmichaud when I last checked, it was smart but not smart enough. :)
jnthn Feel free to educate it. ;-) 18:02
pmichaud something about the current factoring also bugs/confuses me a bit. I agree we want $lazy_sig_init to not have to be attached with loadinit, though.
jnthn OK, that's the motivation for lazy sig things being the way they are today. 18:03
pmichaud agreed.
jnthn If you've a better/cleaner way that still gives us that win, I'm all for it.
pmichaud I'll certainly preserve that feature.
jnthn I just didn't think of one at the time I was doing it. :-)
pmichaud would you see an issue with us having all blocks be wrapped in Code objects? 18:04
(not sure I'll do that, just curious if you have any bad reactions to it :-)
18:04 mberends joined
jnthn It just feels (a) wasteful for immediate blocks and (b) I'm not sure how PAST::Compiler will cope if the thing it expects to be a PAST::Block isn't one, but maybe it won't expect that really. 18:05
Other than that, in theory I can't think of any immediate problems you'd hit.
18:05 PZt left
pmichaud PAST::Compiler generally doesn't care. 18:06
and yes, I think I can optimize out immediate blocks
but I'm thinking I'd prefer to always have the actions build a Code object, then optimize them out in the (very few) places that immediate blocks occur
rather than what we have now, which tries to be overly smart about immediate versus non-immediate blocks 18:07
in other words, I'm thinking of pushing the create_code_block and make_lazy_sig functions a little farther down the generation tree 18:08
right now they occur a bit high in the tree
jnthn Could work.
See in the statementlist action method btw 18:09
(on immediate handling)
pmichaud that's the main one that will change, yes :-)
the only time that statement list ends up with an immediate block is when it comes from circumfix:<{ }>
jnthn Troo. 18:10
pmichaud so it's much easier to have circumfix:<{ }> simply tag its past with "I'm immediate in statementlist" than it is to have statementlist try to figure out when a block might be immediate. 18:11
I did this in NQP and it made things much cleaner.
18:11 supernovus joined 18:12 synth joined
jnthn I can easily believe that. 18:12
pmichaud okay, I think I understand the code much better now. Thanks. 18:13
jnthn OK, great. :-) 18:14
pmichaud oh, one last question 18:15
what happens with routine declarations? Do we depend on the outer block being executed before they get invoked?
or do we suppress lazy signature initialization? 18:16
ingy greetings
jnthn rakudo: sub foo() { sub bar($a) { say $a } }; bar(42); # like this? 18:18
p6eval rakudo 3d2cb8: OUTPUT«Could not find sub &bar␤ in main program body at line 11:/tmp/GDHjY41yYQ␤»
jnthn pmichaud: Won't happen here because bar isn't visible :-)
pmichaud rakudo: our &bar; bar(42); sub foo() { our sub bar($a) { say $a } }; # like this.
p6eval rakudo 3d2cb8: OUTPUT«42␤»
18:19 SmokeMachine joined
pmichaud rakudo: our &bar; say &bar.WHAT; sub foo() { our sub bar($a) { say $a } }; # like this. 18:19
p6eval rakudo 3d2cb8: OUTPUT«Sub()␤»
jnthn pmichaud: iirc, all the our ones are set up together no matter how they are nested in the package.
pmichaud rakudo: say &bar.WHAT; sub foo() { our sub bar($a) { say $a } }; # like this.
p6eval rakudo 3d2cb8: OUTPUT«Sub()␤»
pmichaud rakudo: say bar(42); sub foo() { our sub bar($a) { say $a } }; # like this.
p6eval rakudo 3d2cb8: OUTPUT«42␤1␤»
jnthn yes
@PACKAGE[0].block.loadinit.push(PAST::Op.new( 18:20
:pasttype('bind'),
PAST::Var.new( :name($name), :scope('package') ),
$past
));
pmichaud so in that case, the creation of the Code object is .loadinit
jnthn Yes.
pmichaud but the signature may still be lazy. Okay.
jnthn (was going to type laodinit before but wanted to make absolutely sure first :-))
pmichaud but for my-scoped subs, we create a new Code object each time we enter the scope? 18:21
jnthn Yes 18:22
I was hoping one day we'd have .once() on PAST nodes, as once speculated a while back, to make that more efficient.
pmichaud actually, it's almost correct now.
jnthn Or that we'd have proto-lexpads and could set it up once in there.
pmichaud it doesn't want to be set up just once.
jnthn ah
pmichaud we want a new Code object at each scope entry. 18:23
that's what a closure does :-)
but we would like to keep from re-building the signature each time.
i.e., we'd want the signature kept somewhere.
(once we've computed it) 18:24
jnthn Ah, yes...that's what I want "once"
But yes, you're right.
pmichaud that's why in some sense I'd like the properties to be held in the actual Parrot sub and not each individual code object 18:25
but I think I can make this work quite nicely. 18:26
jnthn looks forward to seeing it :-) 18:28
pmichaud well, I have to depart in ~15 for a while, but will be able to pencil-and-paper some thoughts while I'm out.
(I've found that pencil-and-paper seems to work really well for figuring out these design issues, like lists and iterators. Don't know why.) 18:29
18:29 envi^home left
pmichaud okay, thanks. will work more on it later. 18:29
supernovus So, is it correct behavior that any 'our sub' declared in a core setting library will automatically be available globally even if you didn't specify 'is export' ? 18:37
pmichaud no. 18:39
18:39 proller left
pmichaud many of the subs should probably go inside the class declarations and be declared 'is export' 18:39
rakudo has them outside as a cheat for now.
supernovus pmichaud: I have some 'our sub' statements in a role (yes, a role) and they seem to leak through so that you can call them directly from any user level code. 18:41
pmichaud those should not be exporting if you have them in a role. 18:42
that's definitely a bug.
supernovus I didn't think so
pmichaud rakudo: role XYZ { our sub foo() { say "oops"; } }; foo() # like this? 18:43
p6eval rakudo 3d2cb8: OUTPUT«Could not find sub &say␤ in 'foo' at line 11:/tmp/LlANweUR4_␤ in main program body at line 11:/tmp/LlANweUR4_␤»
pmichaud yes, bug.
supernovus pmichaud: Yeah, like that.
pmichaud please to file a ticket. :-)
18:43 Mowah left
pmichaud afk 18:43
18:43 rv2733 left
lue so, as of now, there's no P6 way of testing if a variable is rw, right? 18:44
ash_ lue: yes, i think you can do it in PIR though 18:45
lue I know you can. I just wanted to avoid defining := in two places, and just edit operators.pm. 18:46
And then I got the crazy idea to avoid inline PIR until the actual binding :)
18:47 mmcleric left 18:48 patspam left
PerlJam speaking completely from ignorance here, but it doesn't seem like it would be too difficult to add a way to introspect on traits just like you can on methods. 18:53
lue (unless of course there's a way to avoid PIR at that part as well; but that would require the Hand of God)
I agree, something like $a ~~ :rw
(although I'm speaking from ignorance as well :P) 18:54
ash_ rakudo: my $a; Q:PIR { .local pmc a, b␤ a = find_lex '$a'␤ getprop b, 'rw', a␤ say b };
p6eval rakudo 3d2cb8: OUTPUT«1␤»
ash_ $a is rw, :P
lue methinks that ($a ~~ :rw) is not specced though. .oO(Ah well. That can always be fixed.) 18:56
18:57 constant joined, constant left, constant joined
PerlJam lue: that could get confusing: $file ~~ :r 18:59
is the file readable or does $file have the :r trait? 19:00
19:02 rgrau joined
ash_ $a ~~ rw (since there is an rw role...) 19:02
rakudo: use MONKEY_TYPING; augment class Mu { method rw { Q:PIR { .local pmc self, is_rw␤ self = find_lex 'self'␤ getprop is_rw, 'rw', self␤ %r = box is_rw } } }; my $a; say $a ~~ :rw # almost, i think its not boxing right 19:03
p6eval rakudo 3d2cb8: OUTPUT«===SORRY!===␤The opcode 'box_p_p' (box<2>) was not found. Check the type and number of the arguments␤»
ash_ or i am doing boxing wrong
either way
slavik ash_: if the other guy beats you, you are doing boxing wrong :P 19:05
ash_ slavik: thats a good point too
slavik and then you need to watch more Rocky movies 19:06
ash_ aww man, do i have to?
slavik fine, you can watch rambo
lue rakudo: say $a ~~ rw 19:07
p6eval rakudo 3d2cb8: OUTPUT«===SORRY!===␤Symbol '$a' not predeclared in <anonymous> (/tmp/IsbSpSp8f3:11)␤»
ash_ lol, "He eats things that would make a billy got sick"
lue rakudo: my $a = 3; say $a ~~ rw
p6eval rakudo 3d2cb8: OUTPUT«Could not find sub &rw␤ in main program body at line 11:/tmp/hfrzX2vqAP␤»
ash_ rakudo: my $a is rw = 3; # surprised this works, but i have no idea what it means.... 19:09
p6eval rakudo 3d2cb8: ( no output )
lue rakudo: my $a = 3; my ($b is rw) = $a; $b = 99; say $a 19:10
p6eval rakudo 3d2cb8: OUTPUT«3␤»
lue rakudo: my $a = 3; my ($b is rw) := $a; $b = 99; say $a
p6eval rakudo 3d2cb8: OUTPUT«99␤»
[particle] rakudo: my $b is ro = 3; say $b; $b = 5; say $b;
lue I get the feeling the ($b is rw) causes some hidden side-effects though...
p6eval rakudo 3d2cb8: OUTPUT«No applicable candidates found to dispatch to for 'trait_mod:<is>'. Available candidates are:␤:(Mu $child, Role $r)␤:(Code $block, Any $arg?, Any :export($export)!)␤:(Mu $child, Mu $parent)␤␤ in main program body at line 11:/tmp/iA1cPNx8HH␤»
[particle] rakudo: my ($b is ro) = 3; say $b; $b = 5; say $b; 19:11
p6eval rakudo 3d2cb8: OUTPUT«No applicable candidates found to dispatch to for 'trait_mod:<is>'. Available candidates are:␤:(Mu $child, Role $r)␤:(Code $block, Any $arg?, Any :export($export)!)␤:(Mu $child, Mu $parent)␤␤ in main program body at line 11:/tmp/IGT7C1nhFZ␤»
ash_ rakudo: my $a ::= 3; $a := 2;
p6eval rakudo 3d2cb8: OUTPUT«===SORRY!===␤::= binding not yet implemented at line 11, near " 3; $a := "␤»
lue apparently, ro doesn't exist.
[particle] hrmm, so it seems
jnthn I think $a.VAR.rw could perhaps work 19:13
(if not now, it's the way that probably should)
No, it's readonly
19:13 Guest42363 left
jnthn not ro 19:13
ash_ jnthn: any idea what "my $a is rw" would mean? 19:14
$a's container is read/write? i guess? 19:15
because with class attributes it makes an accessor in reality, so...
jnthn yes
it's kinda the default though
ash_ well, my $a is readonly might be a better examply 19:16
example*
rakudo: my $a is readonly
p6eval rakudo 3d2cb8: ( no output ) 19:17
ash_ rakudo: my $a is readonly = 3; $a = 5;
p6eval rakudo 3d2cb8: OUTPUT«Cannot assign to readonly value␤ in '&infix:<=>' at line 1␤ in main program body at line 11:/tmp/6oT0kt8ODx␤»
ash_ oh, that does work, neat...
lue I once tried $a.rw 19:18
19:18 hercynium joined
jnthn That ain't going to work :-) 19:18
($a.rw)
rakudo: my $a = 42; say $a.VAR.rw; 19:19
p6eval rakudo 3d2cb8: OUTPUT«Method 'VAR' not found for invocant of class 'Int'␤ in main program body at line 11:/tmp/uvUklino82␤»
19:19 macdaddy joined
jnthn ENOTYeT 19:19
lue ESHOUTINGMETHODS
19:19 macdaddy is now known as Guest14798
lue I get the idea that .VAR is more than justs adding a bunch of methods. 19:24
moritz_ ~. 19:25
pmurias sorear: if i want to make STD handle a symbol that is not in the CORE ($LexicalPrelude) how should i do that?
lue (or else it would work right now)
jnthn lue: It's meant to be a macro, not meth. 19:26
lue ah, well, that explains everything! :)
19:30 SmokeMachine left 19:34 supernovus left 19:35 SmokeMachine joined 19:38 PZt joined 19:46 HarryS joined 19:48 SmokeMachine left 19:59 tylercurtis left, brill joined, tylercurtis joined 20:01 Guest23195 left
brill What is the status of perl 6? 20:03
sbp currently: awesome 20:04
getting: awesomer
20:04 Guest23195 joined
sbp perl6 is different from previous big number versions in that there is not one single reference implementation 20:04
20:04 SmokeMachine joined
brill Leaving Python in the dust? 20:04
sbp so there are several efforts in that domain, at different stages of "completion"
and of course a big part of a programming language is its libraries and communities. those things are still developing, but they've already been developing for years. the community is doing better than the libraries so far 20:05
slavik blizkost will fix that ;)
lue (it adopts the C/C++ model)
slavik or pvipvi
Perl5 in Perl6
lue of implementations, that is. 20:06
slavik we need that
brill So if I need sting manipulation perl 6 is worth a look?
20:06 Intensity joined
brill ie as awesome as Perl 5? 20:06
20:07 Transformer joined
jnthn brill: Perl 6 as a language is awesome at that - more so than Perl 5. The implementation on the other hand is still very much a work in progress. A lot of stuff works but (a) slowly and (b) we're still finding and fixing bugs so it's not always a smooth ride. 20:08
brill: There's some really nice Perl 6 code written and running today though that does parsing and other bits. 20:09
tylercurtis brill: Are you willing to deal with no implementation nearly completely implementing the spec, the spec being unfinished and still changing, and not-very-good performance of at least some of the implementations? That said, as jnthn said, Perl 6 is a very awesome language
20:09 Transformer left
brill Thanks all. This pretty much answers my question. 20:11
tylercurtis brill: It's even awesome to work with already, as long as the performance problems and the mutability of the spec(and thus the risk you might have to update your code as new things come along) aren't too big of an issue for you(I wouldn't build a company on it, certainly, but for side projects and throwaway scripts, I think it's a pretty good choice). 20:15
brill So you're still using Perl5 for the more important scripts? 20:18
ash_ jnthn: is it possible to have a custom meta class in rakudo right now? 20:19
20:20 patspam joined
lue jnthn: are there any unwanted side effects of ($b is rw) := $a as opposed to $b := $a ? 20:20
jnthn ash_: short answer: no 20:23
ash_: long answer: the main missing piece to start playing with that is being able to declare your own package declarator to associated with a metaclass. 20:24
*associate
lue: Do you mean my ($b is rw) := $a; ?
lue yes.
jnthn brill: Can't speak for everyone, but that's the case for me. 20:25
lue (I forgot the my's)
ash_ is it possible to give a role a custom meta class?
jnthn lue: is rw is actually the default
ash_ rakudo: my $a = 4; my ($b is readonly) := $a; $b = 21;
p6eval rakudo 3d2cb8: OUTPUT«Cannot assign to readonly value␤ in '&infix:<=>' at line 1␤ in main program body at line 11:/tmp/7hg1YL77xr␤»
jnthn ash_: Actually that's how roles are done - their meta-object is a RoleHOW, not a ClassHOW.
lue so, in a perfect rakudo, my ($b is rw) := $a is a perfect workaround for my $b := a ? 20:26
jnthn Yes
Ah, I see what you're getting at now
That does a signature bind...
ash_ jnthn: really, my eventual goal is to make a phaser for roles and classes, INHERITED (for classes) and COMPOSED (for roles)
i just was hoping i could toy with it without hacking core bits of rakudo for now
lue rakudo: my @a = (1,2,3); (@a[3..5] is rw) := @a[0..2]; say @a; 20:28
p6eval rakudo 3d2cb8: OUTPUT«===SORRY!===␤Confused at line 11, near "(@a[3..5] "␤»
ash_ lue: slices don't work right now
jnthn std: my @a = (1,2,3); (@a[3..5] is rw) := @a[0..2]; say @a;
p6eval std 31527: OUTPUT«===SORRY!===␤Two terms in a row at /tmp/lLZzqoDIRg line 1:␤------> my @a = (1,2,3); (@a[3..5] ⏏is rw) := @a[0..2]; say @a;␤ expecting any of:␤ bracketed infix␤ infix or meta-infix␤ statement modifier loop␤Parse failed␤FAILED 00:01
..115m␤»
jnthn Also it's a parse error 20:29
20:29 pmurias left
ash_ std: my @a = 1, 2, 3; @a[3..5] := @a[0..2]; 20:29
p6eval std 31527: OUTPUT«ok 00:01 112m␤»
lue that's what I'd like to do someday :) . (stupid NES and it's "shadow memory" *grumble*) 20:30
so, apparently, I'll have to wait for/go code the real thing.
tylercurtis brill: I don't really have anything I work on that is in the "important enough that Perl 6 is too risky" category, so I can't really answer that. Most of my time is spent on my GSoC project, where I can't use Perl 6 because Perl 6 isn't available in the Parrot repository. There are a few side projects I'm planning to work on that I won't use Perl 6 for, but that's because I plan to eventually develop into a commercial project. My latest side project is 20:31
written in Perl 6, but given that it's a tool for writing articles including Perl 6 example code, that's unsurprising. I did use Perl 5 instead of Perl 6 for a very trivial script yesterday, but that's because I remembered how @ARGV in Perl 5 interpolated but didn't remember how @*ARGS interpolates into strings in Perl 6 and didn't feel like checking.
20:32 brill left
jnthn @*ARGS[] 20:32
20:33 clintongormley left 20:34 Su-Shee left
ash_ jnthn: do you think it would be difficult to add a phaser? or do you know of anything that would be a problem, if I added one to rakudo for INHERITED and COMPOSED 20:34
20:34 patspam left
jnthn ash_: Well, it's only a problem if it's not in the spec. 20:35
;-)
tylercurtis jnthn: for some reason I was thinking that the string created by "@*ARGS[]" would be different from Perl 5's "@ARGV". 20:36
jnthn ash_: By all means go ahead and try it out though
ash_ jnthn: i was going to modify the spec too, i just wanted to prototype it also to see how it would work, or where it might be difficult to use
lue afk
jnthn ash_: I think from previous discussions it's generally felt that it's OK if something along those lines ends up spec'd.
ash_: Oh, prototype first is fine. :-) 20:37
Even encouraged. :-)
Just that if I apply it to Rakudo I'd like it to be spec'd at that point. ;-
ash_ i can modify the spec too, i don't mind doing that, although i am a bit more hesitant, because its really easy to fork rakudo and throw it away if i fail miserably
jnthn Right, so fork and try it first, and then we can tie up the spec and merging later on. :-) 20:38
tylercurtis Does anyone know a way to get Emacs cperl-mode to syntax highlight Perl 6 POD properly? Whenever I view a file with it, it doesn't realize that the POD ends. 20:48
ash_ is that the one from the pugs repo? 20:51
or is that the modern cperl-mode from perl's main repo?
tylercurtis ash_: I don't really remember. 20:53
ash_: does that mean that one of those does do it properly?
20:54 masak joined
masak ahoy, #perl6! 20:54
masak goes into Yapsi release mode 20:55
20:55 patspam joined
dalek psi: 4037362 | masak++ | (3 files):
[Yapsi] implemented 'while' loops

SIC instructions turned out to be just the same ones. Also added prefix:<--> so that it's easier to write 'while' loops that halt.		 20:56
masak if statements and while loops, all in one day. :) 20:57
colomon \o/
masak oh right, I promised myself to write a docs/LOLHALP file, with low-hanging fruit for interested people.
masak does that now
ash_ tylercurtis: I am not sure, i haven't checked them recently, but svn.pugscode.org/pugs/util/cperl-mode.el is the one in pugs (according to svn last updated on 2010-05-12) 20:59
21:00 skids left, g3oM1 joined
tylercurtis ash_, thanks. the one from the pugs repo works! 21:04
ash_ cool
masak rakudo: my $a is readonly = 3; # should "real" Perl 6 allow the assignment here? 21:06
p6eval rakudo 3d2cb8: OUTPUT«Cannot assign to readonly value␤ in '&infix:<=>' at line 1␤ in main program body at line 11:/tmp/M6tRHTv2Cs␤»
21:07 patspam1 joined, patspam left, patspam1 is now known as patspam
ash_ masak: i'd think so... how else will it get a value? 21:07
masak ash_: that's what I'm thinking, too. 21:08
unless the 'readonly' trait is to be taken *very* literally :)
ash_ my $var_might_as_well_be_Mu is readonly; 21:09
masak well, Any, but point taken. :)
cxreg does 'prompt' have a noecho mode? :) 21:11
masak cxreg: not by default. 21:12
21:12 mand joined 21:15 Guest93363 left
masak I'd be delighted if some people would like to proof-read github.com/masak/yapsi/blob/master/...ce/2010.07 21:25
21:26 mand left
dalek psi: 758338d | masak++ | README:
[README] updated
 21:26
psi: 89a8f27 | masak++ | doc/ (2 files):
[docs] added announcement and LOLHALP
sbp masak: docs/LOLHALP - heh! 21:27
masak :)
21:27 tadzik left
masak think of Yapsi as Rakudo's "spunky little sister"... 21:28
sbp at first I thought you really wanted this proofread
but now I suspect you want praise 21:29
and you deserve it. that is a very entertaining release announcement
masak++
masak sbp: thanks. I certainly didn't have praise in mind, but I'm glad to get it.
sbp if people don't download this release, they must be a few pecans short of a byte 21:30
masak :D
tylercurtis masak: Yapsi's runtime will require a working Perl 6 implementation to run any code compiled with it, right? Any plans for someday branching off into a full-fledged VM implementation(with the core implemented in a subset like Pypi, Rubinius, Squeak, etc.)? (I'm going to be afk for a while but you don't have to wait until I return to answer; 'tis just a little throw-away question of curiosity) 21:31
masak creates the .tar.gz download
tylercurtis: who knows? the more crazy stuff we manage to pull into the Yapsi project, the more people will likely take an interest and help keep it alive. 21:32
I'm increasingly realizing that some things are really easy to implement really soon; things that Rakudo doesn't have yet. we should probably endeavour to put some such things in.
some of the more fun phasers, for example. 21:33
colomon masak: just got ABC to build with a simple (but ugly) rewrite of the one troubled regex.
masak colomon: cool! you rock!
colomon bad news is I think all the existing tests will fail without the ability to use things like <ABC::basenote> in regexes.
not having a way to test individual grammar rules by themselves makes life hard... 21:34
masak colomon: yeah. feels like a deplorable regression in Rakudo.
colomon +1
masak colomon: hm, or maybe that syntax is gone entirely?
pmichaud back again
colomon oh? 21:35
masak pmichaud: will <Foo::bar> ever work again in Rakudo?
or is that syntax ancient?
I mean, methods aren't our-scoped anymore.
colomon (in a grammar, he means) 21:36
pmichaud "It's official
because there's no official definition of "official"." LOL
masak well, in a regex outside of a grammar, too.
:) 21:37
ash_ rakudo: class bar { our sub foo { 'foo' }}; say bar::foo; role foo { our sub bar { 'bar' } }; say foo::bar; 21:38
p6eval rakudo 3d2cb8: OUTPUT«foo␤Can not find sub foo::bar␤ in main program body at line 1␤»
ash_ is there a reason i can' call the sub in the role?
21:38 whiteknight joined
masak ash_: heh. 21:38
ash_: that looks like a bug to me. 21:39
but jnthn would know for certain.
ash_ i thought role's still made a namespace
jnthn reaches for his dictionary and prepares to read masak's Yapsi release announcement
ash_: It's, erm, awkward. :-) 21:40
21:40 ruoso left
jnthn ash_: our scoped stuff in a role is horrible. 21:40
pmichaud masak: I'm looking up the rule-in-another-grammar syntax now
masak jnthn: no miens this time, I promise. :)
pmichaud I don't remember if it's <Foo::bar> or <Foo.bar> now (or something else)
jnthn ash_: Consider if the role was parametric and the our scoped sub referred to a type parameter.
er, method
oh, sub
well either way
21:41 patspam left
colomon masak: I don't see any sign of <Foo::bar> in S05. :\ 21:42
masak colomon: let's hope pmichaud comes back with something.
ash_ jnthn: so..... role foo[::T] { sub bar { say T }; foo[Int]::bar # wouldn't work?
:P
colomon <ABC.basenote> doesn't work in practice.
pmichaud colomon: why not? 21:43
jnthn ash_: Well no way there because the sub is lexically scoped. :P
ash_ oops
colomon pmichaud: I mean, I tried it and it doesn't work.
pmichaud std: / <XYZ::foo> /
p6eval std 31527: OUTPUT«ok 00:01 112m␤»
ash_ role foo[::T] { our sub bar { say T }; foo[Int]::bar
jnthn ash_: I'm not sure if that'll parse though
pmichaud std: / <XYZ.foo> /
p6eval std 31527: OUTPUT«ok 00:01 109m␤»
jnthn std: role foo[::T] { our sub bar { say T }; foo[Int]::bar
p6eval std 31527: OUTPUT«===SORRY!===␤Confused at /tmp/nNLqZkjII4 line 1:␤------> [::T] { our sub bar { say T }; foo[Int]:⏏:bar␤ expecting any of:␤ coloncircumfix␤ signature␤Parse failed␤FAILED 00:01 114m␤»
ash_ std: role foo[::T] { our sub bar { say T }; foo[Int]::bar # just checking
p6eval std 31527: OUTPUT«===SORRY!===␤Confused at /tmp/8vfzcz8YD7 line 1:␤------> [::T] { our sub bar { say T }; foo[Int]:⏏:bar # just checking␤ expecting any of:␤ coloncircumfix␤ signature␤Parse failed␤FAILED 00:01 114m␤»
ash_ oh well
pmichaud std doesn't seem to provide much insight here.
jnthn ash_: If anything is going to work though, that will be it.
ash_: But it's probably going to have to actually make the role exist and then .WHO it or some such. 21:44
pmichaud aha
answer is Pm-2 in pm.txt
ash_ alright, so, in theory a our sub in a role should work, one day
jnthn ash_: Somehow yes 21:45
ash_ kk, just curious
jnthn ash_: I'm not quite sure exactly how it should work off hand
pmichaud svn.pugscode.org/pugs/misc/pm.txt
jnthn There'll be a way though.
colomon pmichaud: so it's just a matter of getting it implemented, eh? 21:46
21:46 Guest14798 left
pmichaud colomon: smop, yes. 21:46
21:46 pyrimidine left, patspam joined
pmichaud std: / <.XYZ::foo> / # also curious 21:46
p6eval std 31527: OUTPUT«ok 00:01 109m␤»
ash_ maybe i'll pester TimToady about what it means, or if he has any thoughts on role_name[Type]::foo for calling a parametric role with an our scoped sub 21:47
colomon pmichaud: is that ironic smop or actual smop? ;)
pmichaud colomon: don't know which, yet.
colomon ah
jnthn masak: Wow I understood every word.
21:49 masak left
dalek psi: f32fadd | masak++ | doc/announce/2010.07:
[announce/2010.07] fixed tiny typo

in time. Oh well.		 21:49
21:49 masak joined
masak stands down Yapsi release mode 21:51
pmichaud masak++ # regular yapsi releases 21:52
why is yapsi still tied to alpha, ooc?
masak pmichaud: at one point, it was waiting for GGE.
pmichaud: now, there's no reason anymore, really.
pmichaud masak: wfm
masak pmichaud: but it's... strangely nice... not to have to Red Queen all the time. not used to that with my Perl 6 software. 21:53
arnsholt "Red Queen"?
masak arnsholt: en.wikipedia.org/wiki/Red_Queen's_race
arnsholt Ah, right. I need to read those books 21:55
masak you do.
and with that, I bid you folks a Good Night(TM).
dalek psi: f2020b2 | masak++ | lib/Yapsi.pm:
[Yapsi] bumped SIC version

upcoming release.
masak tomorrow, I'll go to jnthn++'s place, and we'll have a hackathon. \o/ 21:56
21:56 g3oM1 left
tylercurtis masak: if you're looking for a really crazy idea to get pulled into Yapsi, native code compilation works for the Lisp folks. ;) 21:57
21:58 christine left
masak tylercurtis: I have been toying with the idea to make a C runtime... 21:58
tylercurtis: it would blow all the other implementations out of the water. unfortunately, it'd also be increasingly difficult to keep up-to-date with the regular Yapsi runtime, the one written in Perl 6. 21:59
jnthn omg masak is coming!
:-) 22:00
masak tylercurtis: if you really mean "compilation", then I'm afraid we're already talking about a subset of Perl 6. cf eval.
jnthn: see you tomorrow evening. 'night.
o/
22:00 masak left 22:04 christine joined 22:09 Guest23195 left 22:14 alester left
pmichaud jnthn: I think I figured out that it's just as easy to set a lazysig property on the Parrot sub as it is to include it as an attribute to a Code wrapper. 22:17
22:20 SmokeMachine left
pmichaud (and still not have it be done at .loadinit) 22:20
then by using setprophash and getprophash, we can make sure that all clones of a given Parrot sub end up with the same properties. 22:21
so that once the signature is set for one of the clones, it's set for all.
jnthn: would it be reasonable to have .arity and .count stored as part of a Signature object instead of a Code object? 22:24
afk for a moment, walk.
jnthn pmichaud: Well, a Signature holds all the stuff needed to compute arity and count 22:29
pmichaud: What do you mean by "stored"?
pugssvn r31528 | sorear++ | [STD] A hook, to allow %*LANG to be overriden, must come AFTER %*LANG IS INITIALIZED. Who would have thought before does no good?
pmichaud Signature has $!arity and has $!count 22:30
and Signature.arity / Signature.count methods that cache the results
right now we handle .arity / .count in Code
but since we'll have clones of Code objects floating about, it seems better to cache that information in the thing that doesn't get copied
(and arity/count is really a property of a signature more than a code object) 22:31
of course, we'll keep Code.arity and Code.count, but they can delegate to the Signature
related question -- currently the lazy_sig_init block creates a lowlevsig, yes? 22:32
jnthn pmichaud: +1 to delegating .count and .arity to Signature and caching it. 22:33
pmichaud: (lowlevel) yes
pmichaud perfect
jnthn pmichaud: Those are very cheap to create
pmichaud: And normally all we need unless folks go introspecting.
pmichaud right.
this will preserve that. 22:34
oh, wait
(re-reading backscroll) 22:35
17:54 <jnthn> pmichaud: Actually the code that does it lives in the dynop bind_signature
jnthn pmichaud: "the code" = what detects we need to build a signature 22:36
And triggers it
Not the actual building.
pmichaud right
it triggers it by calling Code.signature ?
perl6.ops:612 22:38
jnthn Correct.
Though that probably builds more than it needs to...
pmichaud so, that builds a Signature object even though it's not needed for introspection
right
jnthn Yeah, mentioned that earlier
pmichaud I'm going to refactor that to have it just build the llsig
jnthn We've room for optimization there.
OK, good.
Not sure why I didn't get around to that before now, it's probably an easy win. :-) 22:39
pmichaud so, the refactoring I'm thinking of keeps all of the llsig handling in the Parrot sub, and the high-level Perl 6 introspection handling in the Code object
22:39 tedv joined
pmichaud and if one doesn't do much introspection, then the Code object does little more than provide a handle to $!do (and handle closure cloning, which becomes trivial) 22:40
jnthn wfm
pmichaud and we preserve the lazy build of llsigs and lazier build of Signatures
jnthn *nod*
pmichaud okay. I think we should get some speed wins. 22:41
we'll certainly build much less.
and closures will work. :-)
22:42 timbunce left
jnthn \o/ 22:42
pmichaud if I haven't mentioned it lately, your code continues to be easy to read and update :)
okay, time for a walk, then I'll start tearing out stuff and refactoring (in a branch, of course)
22:43 hercynium left 22:45 amkrankruleuen joined
jnthn pmichaud: That's good to know. At least if it's not perfect code, it's easy for others to improve. :-) 22:46
22:49 cono left, cono joined 22:59 hercynium joined, rgrau left 23:05 cdarroch left, rgrau joined 23:07 Ross left, sftp left
pmichaud jnthn: would it be painful if I were to rename existing instance of "signature" to "llsig" or something? 23:09
to make it easier to distinguish when we're talking about a lowlevelsignature versus a Perl 6 Signature ?
jnthn pmichaud: I hope not. 23:10
(I hope it won't be painful, that is.)
pmichaud okay
I think that'll be a bit easier to keep straight.
jnthn pmichaud: Until you got confused over it today, I honestly never saw it as confusing. Now I can't see how it wouldn't be confusing.
:-)
pmichaud right now when I see "signature" it's hard for me to know if it's a lowlev sig or a ... right :)
dalek ecza: 4c90cd6 | sorear++ | (3 files):
Start Niecza::Grammar.

installable STD would help a lot.		 23:11
pmichaud .... and Parrot_pcc_get_signature actually obtains a Parrot signature, which is different from lowlevelsignature and Rakudo's Signature, yes?
jnthn And it's naming is *really* f**ked up.
pmichaud Yes, that too. :-)
jnthn Because it actually gets the equivalent of a *capture*. 23:12
But call captures and contexts were merged so today it probably gets a CallContext
lue hai again 23:13
jnthn hellue!
pmichaud and Perl6::Compiler::Signature actually generates llsigs, right?
jnthn Correct 23:14
pmichaud okay, got it. Thanks.
lue [ I spy complaints about Parrot ... so what's new? :) ]
jnthn lue: It's OK, there were complaints about my confusing code too for good balance. ;-)
pmichaud sub_signature is also a llsig? 23:17
jnthn Yes
lue
.oO(I'll avoid tipping the scale again by not mentioning) PMCs. .oO(Oops.)
 23:19
23:21 macdaddy joined, macdaddy is now known as Guest87395
sorear PMC = evil 23:31
under what circumstances does a class have an associated namespace?
cxreg i wonder if you can use ccache with rakudo
23:37 rgrau left 23:45 ash___ joined 23:47 ash_ left, ash___ is now known as ash_
tylercurtis tell masak If someone develops a native-code compiler for Perl 6, it would hardly be the first language with both eval and native-code compilation. Various Lisps are the most obvious examples. 23:53
23:54 Psyche^ joined
sorear phenny: tell tylercurtis to address phenny 23:54
phenny sorear: I'll pass that on when tylercurtis is around.
pmichaud tylercurtis: I'm pretty certain masak++ knows that already. :) 23:55
23:58 Patterner left, Psyche^ is now known as Patterner

Powered by Cro and the Raku® Programming Language.

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

Thank you!