Pugs t-shirts www.cafepress.com/pugscode | Pugs 6.2.9 released | pugscode.org | pugs.kwiki.org | paste: sial.org/pbot/perl6 | www.geeksunite.net
Set by stevan on 15 August 2005.
autrijus migrating from p4, that's the first command we found missing 06:24
so clkao implemented it right away :)
it'd be "svk desc YYYY"
svnbot r6886 | autrijus++ | * fix kind error for ghc 6.5
luqui ghc 6.5 you say? 06:25
autrijus -devel
GeJ hello gentlemen, good morning
autrijus greetings GeJ.
gaal i heard good things about p4. might try to use it in a setting where no "scary OSS" is allowed one day.
luqui autrijus, making pugs depend on 6.5 is not allowed :-)
autrijus luqui: sure.
luqui come to think of it
gaal giggles
luqui what are the 6.4 features that pugs uses?
autrijus luqui: but breaking it for 6.5 would be unfortunate 06:26
luqui of course
autrijus luqui: GADT, STM, TH
luqui oh wow, those are all new to 6.4?
autrijus yeah.
well, TH was seriously revamped
the other two were new 06:27
luqui likes GADT
gaal have to go for a while. see you later!
luqui the proposal I sent to sixperl had a vague proposal of GADT embedded in the middle of it
autrijus cool. dependent types is indeed handy
luqui so, when you say "dependent types"... 06:28
what exactly are you referring to
(making sure I didn't miss an important feature or something)
autrijus ok. in untyped lambda calculus, or in "dynamic languages" 06:29
we have anonymous functions
and function application 06:30
which is terms depending on other terms
(1+2) is a term that the value depends on the value of terms 1, + and 2
luqui gotcha
spinclad (recognizes the other two, forgets what TH is)
autrijus spinclad: Template HAskell
spinclad ah, thx 06:31
autrijus luqui: and in polymorphism, we have term depending on types
that is, the type of arguments dictates at runtime which function variant gets fired
including MMD etc
then we also have type operators
--> (+) (-)
which is types depending on types
the value of (Int --> Str) depends on the value of Int and Str
luqui comprendo 06:32
autrijus finally, completing the cube
we have types depending on terms!
which is called dependent types
luqui eg
autrijus consider a "case" expression over a GADT
data Term a where 06:33
Cond :: Term Bool -> Term a -> Term a -> Term a
Succ :: Term Int -> Term Int
IsZero :: Term Int -> Term Bool
the term -- or rather the constructor of the term -- at runtime
dictates the types of the type in "a" position
which is why we can write typesafe
eval :: Term a -> a
luqui okay good 06:34
autrijus here's a helpful link: en.wikipedia.org/wiki/Lambda_cube
luqui I grokked all that, but I didn't see that nice symmetry or have any vocabulary for it
hmm, would you like me to bounce you the proposal so you can sanity check it, amend important stuff before @Larry butcher it, etc? 06:35
autrijus sure thing
luqui it's mostly an incomplete brain dump of the stuff i'm thinking about "theories" and whatnot
but it outlines the ideas
autrijus nod
do get a copy of TaPL and ATTaPL if you have the chance :) 06:36
luqui uh, @acronyms>>.expand please
autrijus Types and Programming Languages 06:37
luqui n/m
oh, just out of time
autrijus Advanced Topics on $&
I mean $/
GeJ !$
luqui ? 06:38
??!!
GeJ oops, sorry, that was shell... wrong channel. My mistake
luqui autrijus, will do 06:39
it's not too heavily academic is it? 06:40
luqui loses interest in dry academic prose quickly
autrijus nope, it's extremely hands on
with concrete implementations interleaved with discussion
gaal it's been known to get people to implement interpreters that turned into compilers, though.
autrijus indeed.
luqui well, we want a little of both right? 06:41
autrijus luqui:
gaal goes off for real &
autrijus sub min(::T *@values --> ::T | Ord[::T]) {
mm junctive types ;)
luqui that's not a junction
it's like => in Haskell
autrijus oh wow
luqui but on the other side where it belongs
autrijus more overloading
of symbols
luqui | doesn't exist anymore
autrijus sub min(::T *@values --> ::T where Ord[::T]) { 06:42
autrijus ponders
spinclad gaal: well, that's natural enough, do enough partial evaluation and you may end up with a compiler without noticing
luqui oh, I mean as a junctive type constructor
autrijus nvm -- I just thought | is packed with 'or' in perl6
luqui gotcha
that "where" seems to go with the ::T, not with the whole signature
anyway, it's not a syntax proposal 06:43
autrijus indeed
& 06:47
dduncan hello campers! 07:02
s/hello/good morning/
svnbot r6887 | Darren_Duncan++ | /ext/Locale-KeyedText : resynced with the final Perl 5 version 1.6.0, just released on CPAN 08:16
dduncan fyi, I have now just uploaded the first complete set of my Perl 5 modules on CPAN that have 3-part version numbers, indent with spaces rather than tabs, and have various other non-functionality-changing (they are fully backwards compatible) improvements ... this is the first batch of changes that is exclusively/mainly for applying Perl Best Practices (tm)(r)(etc) to my Perl 5 work, for the community's benefit ... also, any of these comp 08:23
for those who care, but don't know what I'm referring to, it is these new distros, soon to reach your favorite CPAN mirror: Locale-KeyedText-1.6.0, SQL-Routine-0.70.0, Rosetta-0.48.0, SQL-Routine-SQLBuilder-0.21.0, SQL-Routine-SQLParser-0.2.0, Rosetta-Engine-Generic-0.21.0 08:42
gaal autrijus, ping 08:49
integral why do they need three part version number?
dduncan those are a lot more expressive 08:50
acme__ sure all version numbers are evil?
integral dduncan: Have you changed the way you do releases of bugfixes to use the new level of numbering? 08:51
dduncan its easier to tell major from intermediate from minor releases
under the new system, I will release bug fixes as changes to the minor-most digit, and large feature adding changes as changes to the middle digit
before now, nearly every release was a change to a single digit, whether major or minor 08:52
luqui uses a revision-like version number scheme
just because it's easier for me that way
and nobody really cares
dduncan there's an additional advantage
luqui three-part is better though... if I did care
dduncan and that is if I want to release more than 29 more versions of SQL::Routine before I tick over to version 1-point-oh 08:53
under the new system, it is perfectly valid to have 0.114.0, for example
and they'll sequence correctly with 0.98.0 etc
spinclad re: ping: ssh! he's sleeping! 08:54
gaal oh, is he?
dduncan fyi, my very next planned set of releases are mainly just for running these through Perltidy (didn't do that yet), and the version nums for those will be an increment of the minor-most digit
gaal i didn't think he was because i didn't see any commits from him.
dduncan I may perhaps rewrap my POD at 78 chars rather than the current 80 chars as well, for the same reasons that is suggested for code 08:55
one reason is, you get a level of quoting in email without wrapping 08:56
and another is, apparently, some terminals are effectively just under 80 chars wide, not fully 80 chars
spinclad or, well, maybe not. thinking home-centrically, but US East != .tw. but i think you've thoroughly refuted me.
gaal dduncan, may as well make it 75 and have quoted patches not wrap. :-) 08:57
dduncan thanks for bringing that up now ... I prefer to get the new width right/best on the first attempt ... and 75 is a rounder number anyway ... thanks! 08:59
my next update round will involve this re-wrapping, and my first run ever of Perltidy 09:01
luqui ?eval sub infix:<@@> ($x, $y) { $x + $y } 3 @@ 4 09:02
evalbot_6887 7
gaal 75 is a rounder number than 78? 64, that's a round number. 09:03
luqui I say 78 is rounder than 75, but 88 is rounder than both 09:04
dduncan yes, but 88 is also wider than 80
I'm going to go with 75, unless someone convinces me that is too *wide* 09:05
gaal i say it's a pity code can't be made to wrap beautifully automatically, and then there would be no need for a limit anyway.
luqui ?eval sub infix:<!@^%> ($x, $y) { $x + $y } 3 !@^% 4 09:06
evalbot_6887 7
dduncan maybe this can be accomplished, with Perl programs anyway, if Perltidy is tied into the email program, so it can auto-wrap *correctly*
gaal in our age of wide terminals 75 cols for code is *usually* too little, *but* wrapped code is so horrible that it's reasonable to make a limit, and 75 is as good a limit as they come.
Khisanth hates horizontal scrolling 09:07
gaal i'm pretty sure it can be made correctly, but also pretty sure it can't be made beautifully.
integral tries to remember what the line length is for handwriting
dduncan there's also the advantage in regards to *printing* ... if one wants to print, they know the code won't wrap on paper, regardless of how wide your screen is
gaal sure, same problem.
only with a higher price on horizontal scrolling.
:) 09:08
dduncan anyway, its 2am here, so I'm going to bed in 10 mins or so, good night
kakos ?eval sub infix:<a> ($x, $y) { $x + $y } 3 a 4
evalbot_6887 7
gaal night. 09:09
luqui ?eval 3 eqv 3 09:10
evalbot_6887 bool::true
luqui ?eval 3.0 eqv 3
evalbot_6887 bool::false
luqui darn
?eval sub infix:<@@> is tighter(&infix:<*>) ($x, $y) { $x + $y } 3 * 4 @@ 5 09:12
evalbot_6887 Error: unexpected "(" expecting trait or block
luqui hmm... darn
luqui jots that down on his dog todo 09:13
gaal are you making a dog?
10:42 Aankh|Clone is now known as Aankhen``
rindolf Hi all! 11:39
autrijus: here?
autrijus rindolf: yes? 12:06
gaal ugh, three ghc versions, none of them build pugs on my box any more :-( 12:07
autrijus, hi. have time to talk about lexical pragmas?
autrijus gaal: I don't think I have brain for that yet, sorry. maybe after a nap or three -- say, 5hrs from now 12:08
<- totally consumed by errands today
but feel free to type ahead 12:09
gaal okay, i'll be arond in ~7-8hrs myself, i hope my own brain holds.
autrijus okay, deal
nap &
rafl gaal: The debian package with perl5 embedding works quite well, btw. 12:12
gaal: The only problem I have is that the pugs::run and pugs::hack manpages are simply empty. I can't reproduce that outside the chroot, though. Any ideas?
gaal typing ahead: i read mjd's proposal, patch, and followup discussions on p5p, and i think i more or less understand this. i first need help mapping this to pugs concepts (eg: an op/COP is what in pugs, an Exp?), possibly modify the design if there's any need, and eventually no doubt with some practical haskell. 12:13
rafl++ # great news!
re the empty manpages, absolutely no idea. can you strace and see what files are being accessed? 12:14
rafl Yes. I think I'll try that today. I'm currently not at my development box. 12:15
gaal a fully-featured pugs binary in debian would be awesome. It's much easier for people to 'apt-get install pugs' than 'apt-get install ghc-cvs ; cpan Perl6::Pugs' and start tweaking. this will surely bring new hackers, or even just curious p5 users to look at p6. 12:17
rindolf autrijus: hi! 12:32
autrijus: we discussed Perl 6 and pugs in #perl.
autrijus: why does $a,$b,$c=1,2,3 assigns only to $c while my @array=4,10,20 12:34
assigns to @array? How do the precedences of "=" and "," relate to each other?
gaal rindolf, he's taking a nap 12:35
rindolf gaal: I see.
gaal (hi)
rindolf gaal: hi, what's up?
gaal all's well :) 12:36
rindolf gaal: that's good.
Juerd keeps wondering why rindolf continually misses basic things like, in this case, precedence 12:42
integral I thought at one point p6's , was going to have difference precedence to p5's?
Juerd It was considered for a short while 12:45
autrijus agreat! the home server that serves autrijus.org was just up in smokes 13:03
apparently the power supply burned itself up for fun :-/
Juerd Why does so much of your hardware fail? 13:04
Bad power lines?
autrijus I wish I know why
:-(
rafl autrijus: Hello, I'd like to know which sources and object files are needed to be installed for pugs to be working. I think the object files are needed to be linked to the generated haskell code to form the pugs haskell runtime (integral told me that). Are the source files also used in some way? 13:07
svnbot r6888 | fglock++ | * lazy_lists.t - more tests, TODO list 13:08
autrijus rafl: theoretically only interfaces (.hi) and object files (.o) are needed, but since pugscc currently uses -make instead of -lpugs 13:09
that means the .hs files needs to be around as ewll.
if we manage to build a libpugs.a
then all the .hi .o and .hs would be unneccesary
rafl That would be much nicer. Is someone working on that already? 13:11
autrijus no.
not as far as I'm aware, that is 13:12
rafl Thanks for clarification! 13:13
svnbot r6889 | fglock++ | * PIL-Run - Test.pm is more verbose 13:57
r6890 | fglock++ | * lazy_lists.t - svn properties, plan()
vkon was recent change ??:: to ??!! intentional? not only much test fail, but that is harder to remember IMHO... 14:05
... I mean ternary operator that was '?' in P5
QtPlatypus vkon: Yes it was. 14:06
The idea is so that :; can be used unambiously for other things.
14:09 mml_ is now known as mml
vkon QtPlatypus: that's fine... you mean there were difficulties with '::' when dealing with package names separator? 14:15
QtPlatypus Nore when people where going ::<$varname> 14:16
integral vkon: much test fail? I tried to change as many as possible, but if you've found ones that I've missed, feel free to fix them 14:17
PerlJam integral: What's blib6 for exactly? None of the files in there appear to have changed ??:: to ??!! 14:20
(perhaps that's what vkon is talking about)
integral PerlJam: o_O I thought it was a perl6 version of blib
vkon integral: sorry for I didn't analyzed that matter deeply, because I run tests and go sleep, and only checked results briefly, and I can look deeper later. 14:21
integral I very possibly missed lots of ??::s
integral sets up to run the test suite
PerlJam decides to look at a clean pugs checkout for the first time in a loooong time. 14:22
integral ugh, tests are slow :-(
vkon t/rules/rules.t alone takes many minutes to run... I think may be it's reasonable groups within that file that are anyway fail, into one single expandable group, and this way try to reduce compiling time for t/rules/rules.t 14:23
t/rules/rules.t compiles slowly because it is 100+Kb in size... I think no-one will be against if I will reorganize that file so it will at least be faster in compiling time, with probably using groups of evals, or somesuch... 14:25
integral only failure is in t/subroutines/lvalue_subroutines so far 14:29
xerox Hey! 14:30
Do you know where is the "Powered by Ph.D" image? :)
vkon integral: my pugs,ghc are at home, so I can't share my information, and, of course, my speculations on ??!! could perfectly be false alarms...
integral vkon: Well, it's really my fault. I should have run the tests myself last night 14:31
vkon ok, I'll go now and will return with faster t/rules/rules.t file... 14:33
svnbot r6891 | Stevan++ | Perl6::MetaModel 2.0 - just commiting because I am going to upgrade my OS 15:26
r6892 | putter++ | PIL-Run - improved timeout. But smoke-perl5 still requires 3 manual interventions. 15:39
gc_mnc yo 15:45
is there way to do something like `` with pugs 6.2.9 without doing it all by hand? 15:46
fglock putter: ping 15:48
svnbot r6893 | fglock++ | * PIL-Run - reverse() works for Str and Array 16:09
putter fglock: pong 16:11
fglock I had a bit of trouble finding out the problem in reverse() - but not it's ok, thanks
putter ok 16:12
are you going to be around this weekend? i'm doing a no-jobsearch weekend, so I thought if we, steven, iblech, et al, were around, we might push PIL-Run to plausibility. 16:13
fglock sure - it's raining here :)
putter its 30% now, without lexicals. +10% for rx:perl5, which should only require mm'izing Match, and $0...$9 aliases. 16:14
:)
if mm support for packages and modules exists (does it?), it would be nice to get that nailed down. 16:15
fglock how about start migrating to mm2?
putter if a simple "use p5 lexicals work" (more in a moment), doing lexicals should be straightforward. if not, more work.
is mm2 ready for it?
fglock mm2 has packages and modules, afaik. 16:17
using p5 lexicals would be nice
putter re lexicals, autrijus, anyone, - does anyone know what exactly the changes in lexical scoping are between p5 and p6? It would be nice to use p5-with-hacks. but if there are gotchas, we might need to do something more indirect.
stevan: any cries of "no! dont use mm2 yet!"? 16:19
fglock: sure, let's try mm2. do you want to look into it? 16:20
fglock I wrote some tests in t/data_types/lazy_lists.t - I'm trying to make some more to pass - I'll take a look in mm2 in a moment 16:21
putter autrijus, anyone: similarly, are there any known gotchas in how p6 does control flow (next, last, return, exceptions) which make it non-straightforward to use p5 equivalents with hacks.
k
fglock re: control flow - there are coroutines and continuations 16:22
putter my weekend is still 5-ish hours away from beginning, so I wont be intense until then.
full-(non-escape-)continuations I believe are optional. but regardless, doing them would require either CPS transform, with really bad speed implications, good compilation which wont exist for a while, or maybe we can get lucky with a hack. The last would need a working non-full system anyway, so I thought we'd focus on that first. 16:24
re coro... any of the p5 Coro modules do something plausible? 16:26
fglock autrijus said it would be slow to use p5 Coro - he didn't say it doesn't work :) 16:27
putter :)
integral Coro has the issue that you can't clone a continuation
wolverian Coro also has the issue that it has never worked for me!!! 16:28
integral it's only worked for me on MacOSX
wolverian I've only tested Linux 2.[46].
fglock btw, it would be nice to be able to clone closures (for lazy lists) 16:29
putter fglock: do we have closures yet? ;)
we sort of create one hanging off of Sub and Code 16:30
svnbot r6894 | iblech++ | * Usual svn props.
r6894 | iblech++ | * Perl6-MetaModel2.0: docs/misc_drawings.txt: s/\s*$//
r6894 | iblech++ | * Test improvements:
r6894 | iblech++ | * t/01-sanity/07-try.t: Added the most basic try{} test.
r6894 | iblech++ | * t/operators/binding/{arrays,hashes}.t: @foo := @bar does not create new
r6894 | iblech++ | containers (and same for hashes).
r6894 | iblech++ | * t/builtins/numify.t: +"Inf ", +"3 ", +"NaN\t", etc.
putter iblech: btw, thanks for your 07-mumble tests. they were very useful.
ok, clone-able closures for lazy lists. 16:31
fglock well, map() works if you don't look too closely
putter ;) 16:32
wolverian ?eval do if 1 { "a" } else { "b" } 16:33
evalbot_6894 'a'
wolverian ?eval my $foo = do if 1 { "a" } else { "b" }
evalbot_6894 \'a'
putter notes make smoke still requires a manual intervention to get by MIME. 16:34
svnbot r6895 | iblech++ | PIL2JS: 16:36
r6895 | iblech++ | * &prefix:<~> stringifies Inf and -Inf to "Inf" and "-Inf" now (instead of JS's
r6895 | iblech++ | "Infinity" and "-Infinity").
r6895 | iblech++ | * &prefix:<+> numifies "Inf", "-Inf", "Inf " etc. correctly and returns 0
r6895 | iblech++ | (instead of NaN) on bogus numbers (+"abc").
putter pauses to appeciate svnbot 16:37
putter pauses to appeciate iblech :)
fglock: any thoughts on what order things should be approached in? mm2 transition, naming (mm2 packages and lexicals), control flow (return, next/last, ...?). 16:41
? 16:42
anyone who isnt usually part of the PIL-Run insanity want to help out for the weekend? "PIL-Run hackathon"? 16:43
:)
fglock do you have an idea how much each one of these impacts the smoke result? 16:44
putter no
putter tries to massage smoke log into something informative... 16:46
fglock making multi subs work properly is important for Prelude; it would be nice to have lexicals and file access too; flow control is trivial if we are in p5 space
we can add p5-regexes to the list too 16:47
yesterday's Hash slice hacks have broken a few Array tests :( 16:49
putter I'm afraid even a distilled snoke log is 300k. I dont think pastbot would appreciate that. ;) 16:52
does the mm have multi subs?
putter is very unclear on what the mm has working and what not... 16:53
fglock Value/Code.pm has; I think mm2 has too
putter k
pasteling "putter" at 66.30.119.55 pasted "smoke-perl5 error log - trimmed a great deal" (2955 lines, 197K) at sial.org/pbot/13012 16:56
fglock it looks like anything we do will improve it :) 17:00
putter :) 17:01
pasteling "putter" at 66.30.119.55 pasted "smoke-perl5 error log - trimmed slightly less (more informative, but huge)" (26046 lines, 467K) at sial.org/pbot/13013 17:04
svnbot r6896 | fglock++ | * perl5/Array - fixed @a[10] := $b
putter ok, on this last paste, the # lines show all the symbols occuring in the code. An application of undef means one of them, a sub, was not defined. 17:05
putter should really add an "autovivifying &foo" warning. that would greatly clarify things. todo list.
and I'm off. I expect to be back in ~5 hrs, but that may slip. 17:06
fglock bye
putter bye &
svnbot r6897 | iblech++ | PIL2JS: Prelude::JS::Operators: int(Inf), int(-Inf), int(NaN) return Inf, -Inf 17:09
r6897 | iblech++ | or NaN now (instead of NaN). t/builtins/math/infinite.t passes 8/8 now.
r6898 | iblech++ | * t/builtins/lists/map.t: Minor cosmetical fixes. 17:21
r6898 | iblech++ | * PIL2JS: Prelude::JS::Str: &substr can be used as an method now.
r6899 | fglock++ | * perl5/Scalar - auto-dereferencing in Scalar.store(Scalar) 17:33
r6900 | iblech++ | * Splitted t/builtins/lists/reduce.t into reduce.t and new 17:57
r6900 | iblech++ | t/operators/reduce-metaop.t.
r6900 | iblech++ | * t/builtins/io/time.t: skip the times() test on PIL2JS.
r6900 | iblech++ | * t/builtins/hashes/slice.t: =for discussion'ed test testing whether
r6900 | iblech++ | (a => 1, b => 2)<a b> works. Will ping p6l in a minute.
r6900 | iblech++ | * t/builtins/arrays_and_hashes/exists.t: Cleanup and made the test not
r6900 | iblech++ | use the (mostly unspecced) stringification of pairs.
r6900 | iblech++ | * t/builtins/perl.t: According to S02, a standard Perl should be able to
r6900 | iblech++ | reparse .perl's return value, added tests for this. Also added tests for
r6900 | iblech++ | recursive data structures (but had to skip the actual tests because they
r6900 | iblech++ | infloop). 17:58
18:06 typester is now known as typeout 18:32 typeout is now known as typester
gaal hey there! 20:08
how is everyone? 20:09
dudley Quite well. How about you? 20:12
gaal not bad, not bad.
Aankhen`` I'm happy... it rained here and it's cooled down for the first time in a long while, and I finally found a place to download Condition Zero 1.2. 20:13
s/\.$/ from./
gaal i finished GTA SA recently. games are fun when you have more important things to do :) 20:14
Aankhen`` Heh.
gaal A isn't around, is he? 20:23
Aankhen`` goes to sleep. 20:24
G'night.
gaal .sleep for %::.keys:grep{ /^A/ } 20:25
putter hellllooo pugs night shift ;) 21:08
gaal hey putter :) 21:09
fglock_ welcome back
putter hi gaal, fglock
unless anyone has better ideas, I'm going to give things a chance to be simple and add return, last, next. then clean up thunks and try a simply my(). 21:10
fglock_ ok 21:11
putter great 21:12
more for fun, how difficult would it be to make the Match at the bottom of PrimP5.pm into something the mm can live with?
fglock_ looking... 21:14
putter putter can't quite remember the rationale for the $xxFoo variable names in PrimP5, but keeps writing them out of inertia. perhaps that should change. especially if they meant "an autogenerated name which will get changed when a human touches it". ;) 21:15
^conner_ how is the perl6 grammar being specified in pugs?
svnbot r6901 | fglock++ | * lazy_lists.t - more tests
putter ^conner_: Parser.hs, haskell... err, I forget the name of the approach. 21:18
fglock_ putter: re Match - I'll try to move it to mm (I'm saving it to a new file) 21:20
stevan_ ^conner_: it uses a Haskell parser combinator library called parsec IIRC 21:21
dudley Parsec?
putter fglock_: :-) ! 21:23
^conner_ is it ebnf with attributes or something more exotic?
putter www.cs.uu.nl/~daan/parsec.html 21:24
stevan_ ^conner_: much more exotic
^conner_: have a look here too www.cs.uu.nl/~daan/download/parsec/parsec.html
^conner_ it's a bit more theory then I understand... so this is a top down parser 21:28
the 'semantics' look just like attribute grammars thou
the token associativity stuff looks neat 21:30
putter fglock_: was there a vision of how return() in Code's... return? a last LABEL hack? something else? 21:33
svnbot r6902 | putter++ | PIL-Run - added non-working return, next, last
putter stevan_: back! :)
stevan_ putter: yes!!
putter lol
stevan_ just finally upgraded to Tiger today
putter ahh 21:34
stevan_ and I keep realizing how many things i forgot to install :)
important stuff,.. like "make" 21:35
putter again mentions the idea of someone putting together a Bundle of pugs-required CPAN modules. (apropos forget and install:)
fglock_ putter: a simple return should work - the block is called by the 'do' method in Code.pm 21:37
stevan_ putter: IIRC, iblech implemented all those using exceptions 21:38
putter err, does return() in a Code return from the code or from the enclosing mumble?
stevan_ has to run and do some real world things & 21:39
fglock_ it returns to the do() method, which will take care of checking the returned type
putter ah. oh well. &
sub f() { for (3,4) -> { return 5 }; return 6; } f() 21:41
5
the {} is a code, yes? (putter goes to check)
stevan_: will there be any pattern to your presence over 21:42
the next two days or so?
fglock_ 'sub xx(){}' is a Sub
'{return 5}' is a Block 21:43
putter { return 5 } is showing up as a PCode of type SubPointy 21:44
stevan_ putter: I will likely be doing the changelog work on sunday morning,.. but other than that,.. maybe some late night hacking (U.S. EST)
fglock_ we don't have SubPointy yet, but Sub should work
stevan_ really goes & 21:45
putter ok, bye &
fglock_ by stevan
stevan_ :)
putter fglock_: but Sub will (correctly) absorb the return(), yes? 21:46
ok, backing up....
fglock_ yes, it should 21:47
putter plain return() is insufficient (final, etc, and we're using subs internally). two posibilities come to mind: Exceptions (stevan++), and hand threading through labels (goto et al). 21:48
fglock_ putter - I created a Match.pm skeleton 21:49
svnbot r6903 | fglock++ | * PIL-Run - added preliminary mm version of Match class
fglock_ still needs moving methods to the right places
putter awesome! putter goes to look...
fglock_ I think the overload can be kept where it is 21:50
putter feel free to gut the overload, this being solely to support the mm'ed pilrun. 21:52
neat. though I afraid I'm not entirely clear on the next step either. among the mm docs, is there a quickref "bootstrap is something you dont have to worry about. for class C{} say foo, for a rw field, say bar to created it, and hee to access it. for methods ... etc". 21:54
let's see, where was the other thread... 21:55
fglock_ there are the t/ files
putter oh yes. Exceptions are simpler. Labels are perhaps faster. Simple wins. ;) Exceptions, unless someone can think of a third option, or disagrees... 21:56
t/ files?
oh yes, to show use.
righ
right even
Is Error.pm still the best place to get exceptions from? 21:57
fglock_ sorry - I don't know enough about Error.pm 21:58
putter fglock_: is there a way to "just in time" make an object into an mm object? with Match it didnt matter. But if exceptions are going to be used... never mind. control flow exceptions and die exceptions can just be different decendants of a common ancesstor. 21:59
there is a hack currently in PrimP5 to give Exception::Simple a ->bit method, so unless $! {...} will work. 22:00
putter contemplates asking #perl
fglock_ do you mean to make objects of an existing class? just call new()
putter the original thought was to deal with (more efficiently created) native exceptions, and then coerce them if/when they needed to be touched as p6 objects. but with return()-implementing exceptions not needing exposure to p6, die() exceptions and return() exceptions can simply be (very) different classes, one mm and one not. 22:02
fglock_ ok 22:03
putter re Error.pm, I'll see if it cleanly wraps in ApiX, so we can easily change it later if needed. 22:04
putter goes to check Error.pm's cpan test, but my fuzzy recollection is it's now part of the standard distribution.
fglock_ finished with Match.pm - but it is not tested at all 22:09
brentdax Okay, I've finished my Log::Selective module, including 21 passing tests. Should I put it in the Pugs repo? On FreePAN? CPAN? 22:12
svnbot r6904 | fglock++ | * PIL-Run - Match.pm - added remaining methods
fglock_ brentdax: how about pugs ext/ ? 22:15
brentdax I can do that, certainly.
putter ahh. thank you fglock_. fglock_++
ext/ has the advantage that if ?? !! changes to something else... 22:16
(it will get handled by bulk fixing) 22:17
brentdax Heh. I actually wasn't using the ternary operator anywhere, but I can certainly see the benefit.
putter p6 is still in flux enough that archiving code elsewhere seems an invitation to obsolescence 22:18
svnbot r6905 | brentdax++ | Add Log::Selective module.
Juerd Are "item context" and "slurpy context" official yet?
putter no idea
"foo" ~~ /(.)(.)/ $/[1] #=> Fyi: Can't locate object method "fetch" via package "Match" at /home/net1/perl6/pugsxpl2/perl5/PIL-Run/../Perl6-Container/lib/Perl6/Container/Scalar.pm line 105 22:20
?
fglock_ mmm
putter sorry fglock_, i didnt mean to suck you into regex land from wherever you were working on... 22:21
fglock_ np
I know what is happenning, but I don't know how to fix it
Juerd What was the reason for calling list context slurpy?
What confusion is avoided? 22:22
(item context is because the difference between "scalar" and "Scalar" is annoyingly small)
svnbot r6906 | putter++ | PIL-Run - from() and to() implemented as methods on Match. Dont work yet. 22:24
fglock_ putter: it is fetching the match from a scalar, which is ok, but it may be using extra parameters - in this case the fetch() is recursive
putter I've yanked from() and to() from PrimP5. They return undef, but perhaps they are not being set... 22:25
hmm, shouldn't package Match be disappeared from PrimP5? 22:26
luqui Juerd, to distinguish from the context that expects a real array (or ref)
fglock_ oh yes, it should
Juerd Why is "list context" not good enough, given that we still have difference between lists and arrays?
luqui well, it was good enough to confuse p6l folks 22:27
putter re fetching match... so what next?
Juerd luqui: Do you know if these names are official yet?
luqui isn't sure, but thinks so
but it's still perl 6
so nothing is official until 6.0
fglock_ putter: there is a commented message in line 104, Code.pm - you can uncomment it and see what is happening 22:28
Juerd If they are, I wonder why the keywords for context forcing. are scalar and list
s/\.//
luqui Juerd, good question
Juerd Please, don't give me that crap. I know very well that everything can change at any point. However, it is important to have some sense of what is official. Even if you don't like that name.
putter fglock_: err, with Match gone, "foo" ~~ /(.)(.)/ stops working... 22:29
luqui thought that you were asking if it was okay to propose a change
please calm down
Juerd Oh, no.
I'm writing a PM article
svnbot r6907 | putter++ | PrimP5: removed obsoleted package Match definition.
luqui anyway, that keyword inconsistency is a little irritating now that I think about it
Juerd I'm explaining the most basic Perl 6 things in a series of articles (though nobody knows that it is a series yet)
luqui maybe you ought to bring it up
Juerd Sorry, I was rude.
perlmonks.org/index.pl?node_id=490468 is the first 22:30
I try to pick one subject at a time, which is very hard.
And: Perl 5 is past tense, Perl 6 is present
luqui sure
putter just saw a talk where the email(?) tool had a little face on the side of the window, and some emotional content estimation heuristics, to give people feedback on the emotive content of their messages while they were written. since in email, one lacks the feedback loops one has in face to face conversation.
luqui reads 22:31
putter goes to look at message in Code
fglock_ sorry - it is Scalar.pm 22:32
putter ah 22:33
luqui Juerd, cool, keep it up 22:34
luqui considers getting a perlmonks account
Juerd Why not - they're free :)
It's the perfect way to keep in touch with the average user
luqui by "considers", I meant "begins" 22:35
fglock_ putter: Match->new() should be expecting named arguments 22:36
in PrimP5, line 636
buu Perlmonks is awesome. All hail.
fglock_ haven't written a line in Perlmonks for a month now 22:37
buu I haven't written much either 22:38
fglock_ I've been spending all my free time here :)
buu I find the questions generally fall in to three or four categories. The first one is a RTFM question that I don't care enough to repeat again. The second types are really complicated ones that would take a fair amount of effort to answer, which I'm too lazy for. The third are the ones that are already answered =/ 22:39
luqui is perlmonks based on e2?
or vice versa?
Juerd No, a heavily patched everything <2
AFAIK, perlmonks was first
Sure I am not
luqui you sure aren't 22:40
:-)
Juerd (Wow, that is a weird (though correct) sentence in English.)
putter fglock: like Match->new('$.val_bool' => $b, ... ? 22:47
fglock_ yes 22:48
putter still doesnt work. will check in
fglock_ new() never complains, but it should 22:49
buu ?eval sub y { (1,2,3) } y()[-1]
evalbot_6907 \3
buu YAY!
?eval sub y { (1,2,3) } y[-1]
evalbot_6907 (1, 2, 3)
buu Oh that's co.. weird?
What happened there? 22:50
putter same behavior in pilrun
luqui ?eval sub y { (1,2,3) }; my @y = y; @y[-] 22:53
evalbot_6907 Error: unexpected "[" expecting word character, "::", term postfix, operator, postfix conditional, postfix loop, postfix iteration, ";" or end of input
luqui ?eval sub y { (1,2,3) }; my @y = y; @y[-1]
svnbot r6908 | putter++ | PrimP5.pm: attempting to get Match to work
evalbot_6907 \3
luqui ?eval sub y { (1,2,3) }; (y)[-1]
evalbot_6907 \3
wolverian y[-1] looks fine to me.
fglock_ putter: try adding some debugging code like 'warn Match->val_bool' to see what's going on
luqui hmm, it's a good question
does subscripting impose a scalar context? 22:54
putter Match.pm wasnt being loaded in MainX
luqui I guess that would make the sub return a list
fglock_ :)
luqui which would then be indexed
er, an array
so it ought to work either way
wolverian Juerd, that's a nice perlmonks article on ~. congrats! 22:55
Juerd wolverian: Thank you. 22:56
svnbot r6909 | putter++ | PIL-Run - add Match.pm to MainX 22:58
r6910 | putter++ | Match.pm: return real values
fglock_ putter: did it work? 23:00
putter ok, have str and bit and int working, and set_as_failed() 23:02
now for array
best guess is still the line in Scalar?
fglock_ I'll try to replicate the problem - what expression you are using to test? 23:03
putter so postcirumfix[] is being called with $/ and 0...
let me sync up 23:04
ok r6911.
"foo" ~~ /(.)(.)/; say $/; $/[0] 23:05
the mangled variable name is $scalar__47x
p5ugs> :5 $scalar__47x->fetch() 23:06
Match=HASH(0x10ab9c0)
p5ugs> :5 $scalar__47x->fetch(0)
Method (fetch) not found for instance (Match=HASH(0x10ab9c0)) at /home/net1/perl6/pugsxpl2/perl5/PIL-Run/../Perl6-MetaModel/lib/Perl6/MetaModel.pm line 248
:5 ref $scalar__47x 23:07
Scalar
fglock_ it looks like $/ is a string - and you are using [0] on it
putter p5ugs> :5 ref $scalar__47x->fetch 23:08
Match
fglock_ in order to say $/[0], you have to define fetch()
putter ok 23:09
svnbot r6911 | putter++ | PrimP5.pm: some Match related fixes
putter (looking ahead, how does one distinguish fetch() for $/[0] and for $/{foo} ? (low priority question))
fglock_ we could redefine [] to call array_fetch and {} to call hash_fetch - currently, both call fetch() 23:10
putter ah. that will likely need to be separated. 23:11
re fetch, how does one fetch the entire array? eg, @$/
fglock_ you call fetch on the array, without parameters - this should be a separate method too 23:12
putter wooot! $/[0] #=> f 23:14
.from and .to work... going to try perl5.t ! :-) 23:15
fglock_ :) great! 23:16
svnbot r6912 | putter++ | Match.pm: added fetch() 23:20
Juerd perlmonks.org/index.pl?node_id=358255 # draft of 2nd article 23:21
fglock_ leaving ... 23:22
svnbot r6913 | putter++ | PrimP6.pm: added not(). not 0 gives an odd warning out of Perl6/Value.pm 23:25
putter oh
wait...
first 50 tests now all passing. now running it on the whole 900. 23:26
moment of "wooot!" :)
just thought you'd like to know before heading off...
thanks very much for your help. 23:27
fglock_ cool - I'll read the irc log later & 23:28
Juerd wolverian: Please proofread the 2nd :) 23:29
wolverian Juerd, sure! where? 23:30
svnbot r6914 | putter++ | PrimP5.pm: Match fix from()/to()
wolverian Juerd, oh, sorry, I'm blind.
Juerd, maybe you could extrapolate on what Ref being subtype of Scalar means. 23:33
(s,extrapolate,write,)
Juerd Like what? 23:36
Oh, I get it, I think 23:37
Done 23:39
I wish I knew what exactly the difference between item context and Scalar context would be
putter woot! PIL-Run passes 98% of perl5.t :) fglock++ 23:53

Powered by Cro and the Raku® Programming Language.

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

Thank you!