pugscode.org <Overview Journal Logs> | pugs.kwiki.org
Set by autrijus on 22 May 2005.
mugwump are you saying duck typing is not enough for roles? I see a 1:1 equivalence. 00:00
stevan not at all, I think duck typing (as I understand it) is enough 00:01
I think the opposite is not true
that there is more to duck typing that roles can provide
however i think the issue at hand is more a nomenclature issue in the docs 00:02
ijoyce_laptop
stevan does() carries the idea of Roles with it
I think we should use something else when we are talking about duck typing 00:03
mugwump what extra things can duck typing do that we care about? 00:07
stevan let me rephrase that 00:08
duck typing (as I understand it) does not imply roles 00:09
while Roles does imply duck typing
mugwump I see roles as a convenient way of grouping behaviours
stevan you dont need roles to do duck typing
mugwump and grouped behaviours are the essence of duck typing
stevan do you have any resources for duck typing? I would like to read more on it actually 00:10
I think maybe my understanding of it is differnet then yours
svnbot6 r3697, autrijus++ | * Parsing for user-defined nullary functions. 00:13
r3697, autrijus++ | * `try {...}` literal allowed at expression level.
r3698, autrijus++ | * add [=>] to quickref/reduction.
r3699, autrijus++ | * unbreak Parrot compilation in light of the tabulaRasa change
r3699, autrijus++ | that includes compile-time pads.
r3700, autrijus++ | * ($_) is now ok as user-defined hyper operators signature.
r3700, autrijus++ | * don't store envStash into subEnv; that saves some space 00:14
r3700, autrijus++ | * Syn "try" is no more; `try {...}` is always an expression now.
r3701, autrijus++ | * Normalise all Shell commands to begin with `:` and
r3701, autrijus++ | reorganise Help message for `:h` a bit.
r3702, autrijus++ | * try {...} fix in tests.
japhy what is the effect of \c[THIS;THAT]?
is it [\c[THIS] | \c[THAT]]?
or is it \c[THIS]\c[THAT] ?
I haven't found a definitive answer either way :/ 00:15
autrijus it's \c[THIS]\c[THAT] 00:16
same as \x[0a;0d]
japhy ok. 00:17
I didn't find an accurate description of use of the semicolon there
autrijus yeah. if you need exact description, may need to ask on p6l.
japhy I'll treat it as you put it for now. 00:19
svnbot6 r3703, autrijus++ | * print stack trace for errors in separate lines.
r3704, autrijus++ | * implement "try" as a Prim.
japhy I suppose I'll also have to ask on p6l what the exact rules about whitespace are. 00:23
I need t oknow if '\c [NAME]' is legal or not...
autrijus I think not, as a general rule 00:24
but doesn't hurt to ask.
(in Perl 6, bracketlikes are less space tolerant than p5)
eg.
svnbot6 r3705, autrijus++ | * repair Pugs compilation backend.
r3706, autrijus++ | * puny attempt of adding at least some of these to ChangeLog.
autrijus print (1+2)*3
prints 9, not 3.
japhy I really wish, for once, Pepsi and Coca-Cola wouldn't make the exact same variants on their producs. 00:29
autrijus pugs> [+]<<<< ([[1,2], [3,4]], [[5,6], [7,8]]) 00:32
((3, 7), (11, 15))
iblech++ # this is so sick
I mean slick
clkao sick 00:33
stevan I agree with clkao it is sick
but that is not a bad thing :)
at least not around here :)
mugwump: we could do monkey-typing :) (peak.telecommunity.com/DevCenter/MonkeyTyping) 00:34
autrijus whew. journal finally up. 00:37
stevan mugwump: I agree with your entire duck-typing table 00:47
svnbot6 r3707, mugwump++ | filter changes down to rest of document, incorporate notes on class traits
stevan (in light of all the duck typing stuff I just read)
clkao is there any shortcut to filter hash by key in p6? 00:48
mugwump %hash<%hash.keys.grep:{ ... }> ? 00:49
%hash.kv.grep:{ $_.key ~~ rx// } ? 00:50
(to return a list of pairs suitable for assigning to a new hash)
clkao that's not shorter 00:51
mugwump the first is a slice, the second will be making a new hash
(with '%newhash = ' prepended)
clkao "map { $_ => $hash{$_} grep { .. } keys %hash " is an annoying p5 idiom
mugwump yes, so .kv() will return a list of pairs 00:52
autrijus errands & 00:54
stevan mugwump: perl6_meta_model.pod is making more sense to me now 00:57
however I am not sure how Role will interact with Type 00:58
mugwump Roles help define types
stevan ok 00:59
mugwump but a type still always has a class
stevan ok so a Type is defined by a combination of Class and any attached Role
mugwump that class might just be "empty", with only one attached role
the Roles are optional, but yes
this Type is the Class::Compiled we were talking about before. 01:00
stevan so the Type is really the "compiled" Class we were talking about?
LOL
scary, thinking the same thing half a world away
ok, so my model is not that different from yours
matter of fact I think we easily unite them 01:01
my head is more in the implementation details, which I think shows in my doc
clkao appearantly >>+<< is 4 times faster than iterating the array and perform + ourselves. 01:05
mugwump ok. I think it would be good to put down the M3 at this point and instead prototype some M1 and M2 behaviour
as we only need the M3 to be flexible for user-definable traits 01:07
s/M3/M2/
(ie, we only need the M3 for user-definable traits)
clkao *** hyper => <SubRoutine(<anon>)>
at lib/Benchmark.pm line 18, column 9-30
1123.8227956215862 / s
*** normal => <SubRoutine(<anon>)>
at lib/Benchmark.pm line 18, column 9-30
402.0747054802783 / s
mugwump I'd blame haskell++ for that, clkao 01:08
clkao heh ya. 01:09
mugwump maybe the EvalMonad has to take fewer steps for a hyperop...
clkao sure it's one op
stevan mugwump: so is any of what I created suitable for M3 ?
mugwump it's a good starting point, I think. But by developing the M2 and M1 we've got some "test cases" for the M3 to represent and build 01:11
stevan ok, sounds good 01:12
I am going to have to hand over the development reins to you for a while
I am behind at $work as it is 01:13
mugwump me too :)
stevan LOL
damn $work, always getting in the way
mugwump yes, my professional work is more and more interfering with my Real Work 01:14
stevan I have been able to rationalize my Pugs work to my $boss, but not to our $client :)
mugwump Read any Dirk Gently?
stevan yes
not in a while though 01:15
I actually have a first edition of first book which I bought when it came out
mugwump I usually explain my development in terms of the interconnectedness of everything
"This might seem to be nothing to do with releasing a new project, but in fact it is connected by its very essence with the project" 01:16
stevan :)
meppl gute nacht 01:17
stevan night meppl
meppl good night stevan
stevan ok, I am gonna sign off for the night, talk to you later mugwump 01:23
mugwump good talking to you stevan, sleep well 01:24
earthquake! 01:27
mugwump polls www.geonet.org.nz/drums.html to see how far the needles moved 01:29
japhy I'm confused with the concept of \C[THIS;THAT] 01:49
does that match ONE character or TWO?
that is, does it match one character that is not 'THIS' or 'THAT', or does it match two characters, so long as they aren't 'THIS' and 'THAT' respectively?
mugwump www.geonet.org.nz/x2408156g_l.html # richter 4.2, piddly 01:50
japhy richter is old school.
mugwump ain't no skool like the old skool
ihb nothingmuch: hehe, Beatnik is awesome :-) 01:53
japhy: reading up on p6 rules? 01:58
japhy writing Perl6::Rule::Parser (subclass of Regexp::Parser)
ihb ah
japhy and trying to figure out how things actually work.
I don't need to implement them, but I need to create a reasonable object 01:59
\C[THIS;THAT] is not defined at all :(
ihb it seems there are quite a few things that aren't defined yet. 02:00
ihb & 02:01
vcv- perl6 is a big big language 02:15
its not something like..say..php. which defines a whopping 4 operators and 4 datatypes..with 3,900 built in functions all using different naming conventions
and diff. casing as well 02:16
not big as in bloated, but very carefully defined and covers everything youll need
autrijus camels are supposed to be stand-alone and survive in the desert 02:31
:)
svnbot6 r3708, autrijus++ | * Type-to-type smartmatch, e.g. `Int ~~ Num`, now works.
autrijus japhy: I think \C[CR;LF] should mean the opposite of \c[CR;LF] 02:32
eric256 hey autrijus.. i was surpised to see you played my game. ;)
autrijus japhy: namely, that it matches any two chars, as long as they are not crlf
of course I do :)
<- learned C with LPC
eric256 LPC? 02:33
autrijus the langauge of LP MUD
that was 12 years ago :)
eric256 ahhh. yea i like making mud bots.
i made an MSN MUD bot a while ago. that was fun. wandering around chatting with people and solving puzzles ;) 02:34
i want to make monster creation a kind of factory... i was thinking about using anonymous subs but how would i call them multiple times like $bat->() xx 3? 02:35
autrijus map $bat, 1..3; 02:37
like that?
eric256 if bat is a ref to a sub will that call it three times? 02:38
autrijus sure
eric256 oh i didn't catch that. i was thinking map {$x}, 1..3
autrijus won't work...
map {$x()} 1..3
eric256 yea no kidding. lol. 02:39
autrijus but that's weird, because () cancels {}, so you might as well write map $bat, 1..3
eric256 yea i just never would have thought of using map like that to call a code-ref. 02:41
any way to set the one property equal to another in the construction? Monster.new(:life(100),max_life(:life)) obviously doesn't work. lol. or can we override BUILD now? 02:44
svnbot6 r3709, autrijus++ | * `loop (;0;) {...}` will no longer execute the loop body.
autrijus eric256: not yet, probably post 6.2.4 02:46
I'll bbiab... 02:47
&
kelan `map $bat, 1..3` would call a sub ref in $bat three times? i would expect it to return a three-element list with each element being the ref in $bat 02:50
svnbot6 r3710, autrijus++ | * unTODO, unEVAL, and generally clean up tests.
eric256 i guess map is expecting a code block or code ref 02:51
map {$bat} 1..3; would act as you are saying
kelan that probably isnt useful, but `map $bat, 1..3` doesnt look like its doing a call anywhere, so it would surprise me
eric256 me too 02:52
kelan hmm. yes, thats a good explanation
eric256 but it certainly wouldn't be the first thing to surprise me. ;)
kelan its expecting a code block, so it uses the one ref'd in $bat
eric256 yep
kelan but what would it do if $bat was, say, an arrayref? 02:53
die?
doesnt seem very nice
eric256 it dies with an message cannot cast form VList to VCode (well it says alot more than that, but thats the gist) 02:54
kelan heh, well i meant semantically what does it do
its pretty unfriendly, because what if i don't know what's in $bat when i write the code? 02:55
eric256 if you don't know what is in $bat then what could you be hoping it would do in code-ref context?
kelan hmm 02:56
map doesnt take expressions anymore? 02:57
if thats the case, it makes more sense
but people probably wont like it
eric256 expressions? i've only ever used map {} @list 02:59
kelan perldoc gives two forms for map in p5: map BLOCK LIST and map EXPR, LIST 03:00
i hardly ever use the expression form, myself, but i see it used quite often
japhy it offers a slight speed boost in certain circumstances. 03:01
there's no need for the overhead of a new scope
s/^/when/
kelan what do you think, japhy. do you read `map $bat, 1..3` as calling a coderef in $bat three times?
japhy kelan - in Perl 5 or Perl 6? 03:02
kelan heh
well thats the thing~
!
japhy no, seriously.
kelan in perl 6
japhy ah.
well, I'd expect you need to do something more than just '$bat'
kelan me too 03:03
i mean i can get used to it. its just surprising 03:04
oh hmm 03:05
japhy any idea what \C[THIS;THAT] should do in a Perl 6 regex?
kelan reading the definition of map in S29 makes sense, tho
eric256 multi sub infix:<<.?.>> ($low,$high) {int ( rand ($high - $low) + $low ) + 1; };
japhy eric - I like that operator.
eric256 anyone see something horribly wrong with that? it is pulling numbers all out of range
kelan japhy: no, but i think autrijus's assessment was pretty good. opposite of \c[THIS;THAT] 03:06
japhy really?
kelan - I missed it...
kelan oh
japhy does it mean <!before \c[THIS;THAT] ..> ?
kelan < autrijus> japhy: I think \C[CR;LF] should mean the opposite of \c[CR;LF]
< autrijus> japhy: namely, that it matches any two chars, as long as they are not crlf
japhy ah, ok.
that's what I kinda hoped it would mean
eric - I don't see why that should be wigging out. 03:07
that's how the function is written normally
eric256 thats what i thought. something else is going on then.
space after rand....dany meaning full space! 03:08
kelan the thing about map makes sense if you think about it in higher order terms:
it always takes a code ref, and calls it on each argument
japhy oh. well then.
now that I know that, I won't be surprised.
kelan pretty sensible
nod 03:09
japhy is just getting into all this "perl 6" business
kelan but i think esp written with the comma will surprise p5 programmers used to the expression form
oh well
eric256: try writing that randomrange function like this: 03:10
multi sub infix:<<.?.>> ($low,$high) {int( rand($high - $low) + $low ) + 1; };
without space between the function name and opening paren
p6 is all weird about that now
japhy I hope only temporarily.
kelan no 03:11
japhy ick-o.
kelan it might be only for method calls, but
japhy well, I suppose it's not a big deal.
eric256 yea i found it. thanks kelan. i am sure there is a reason for it but it drives me nuts
japhy I don't think I put spaces in between functions and their parens.
eric256 half my p6 bugs have invovled that so far
kelan these are different: `obj.method (1);` and `obj.method(1);`
the first calls the method with no args
i put space inside the parens: `obj.method( 1 );` 03:12
svnbot6 r3711, eric256++ | fixed but with .?. operator
r3712, eric256++ | fixed another bug caused by meaningfull space
kelan so it wont affect me, but it is rather nasty
eric256 anyone know why they added yada operators? ... ??? !!! ? 03:16
kelan ... is nice for pre-declaring
tells the person there is a real definition somewhere
i don't really get the others. they do things like warn and die, but otherwise are equivalent to ... 03:17
japhy for complete-ness
autrijus ??? and !!! are suggested by #perl6 03:26
I think nothingmuch
svnbot6 r3713, eric256++ | fixed some bugs with random numbers, monster factories, some balance added, fixed fleeing
r3714, autrijus++ | * Implement ($type_ref ~~ Type) too.
r3715, autrijus++ | * dump.ast can't take nonces (yet)...
autrijus or somebody else
and larry accepted them
kelan what do they add to just using ...?
im just curios
-u-
autrijus I don't know, it's not my idea
kelan heh ok:) 03:27
autrijus I guess it's kind of like XXX in comments
kelan autrijus: what time is it for you?
autrijus kelan: 11:26am
my uptime is something like 20 hours now
kelan go to sleep:P
20 hour uptime is silly
autrijus that is an excellent idea!
kelan heh
why don't you sleep at night? 03:28
code better at night?
autrijus well, it's like this
I hack all day
and when I'm about to sleep, I was reminded that I need to do this journal thing
but writing journal easiliy takes 1~2hr
kelan ack really?
autrijus and during that, I run into bugs etc that needs my attention
so I hack some more
kelan well see
autrijus etc. :)
kelan you should just note the bugs and not fix them! 03:29
autrijus that is... not my style :)
kelan so you go in cycles like 20 hours awake, 12 hours asleep, ...?
autrijus more like 22 vs 14
but yeah
kelan eep 03:30
autrijus I could do 44 vs 28 often when I'm young ;)
kelan 28 hours sleeping? 03:31
you are young:P
autrijus decides to not dwell on this topic :) 03:33
kelan heh
svnbot6 r3716, eric256++ | minor updates to quickref docs 04:07
mugwump oh, no, I keep wanting to use junctions in Perl5! 04:13
svnbot6 r3717, eric256++ | Rough draft of control structure quickref, feel free to mangle as needed
eric256_ check out www.perlmonks.org/index.pl?node_id=453220 04:14
it provides some concept of junctions to p5
Aankhen`` What is ??? !!! for? 04:18
svnbot6 r3718, eric256++ | Changed suggested 'con' to 'control' because of windows conflics with the name 'con'
gaal !!! dies and ??? warns on unimplemented code 04:19
mugwump so what's ... for if !!! dies ???
gaal so you can do sub calc { do_something; ??? # FIXME corner case }
mugwump !!! is shorthand for die "this should never happen" 04:20
Aankhen`` Ah.
autrijus mugwump: ... only dies when the result is actually used. 04:25
lazy evaluation. 04:26
mugwump aha
a poison value, of sorts
autrijus sleeps. :) & 04:37
Aankhen`` Sleep well, autrijus. 04:39
svnbot6 r3719, eric256++ | Smart match table from S04. Seems like a good QuickRef 04:51
r3720, mugwump++ | Move Set to Set::Hash, and Set::Junction to Set, part 1 05:32
r3721, mugwump++ | Move Set to Set::Hash, and Set::Junction to Set, part 2
r3722, gaal++ | s/nonces/nonce/ (English usage; www.worldwidewords.org/qa/qa-non1.htm .
r3722, gaal++ | I owe an apology to this word, which was already a victim of false splitting.)
r3723, mugwump++ | Sketched beginnings of M1 & M2 prototypes, next to nothing there yet
06:10 chady_ is now known as chady 07:23 castaway_ is now known as castaway
Juerd morning 07:37
webmind morning :) 07:38
masak morning 08:04
nnunley Morning
08:32 Aankh|Clone is now known as Aankhen`` 08:49 nnunley_ is now known as nnunley
robertle any parrot folks around? 11:11
svnbot6 r3724, scook0++ | * Some non-toplevel function type signatures (for readability)
r3724, scook0++ | * Minor tweaks to Haddocks
iblech morning :) 11:14
scook0 iblech: hi
svnbot6 r3725, iblech++ | Usual svn properties, vertical alignment of docs/quickref/reduction. 11:26
r3726, iblech++ | Minor test fix (delegation.t) and minor cosmetical cleanups (ChangeLog, Main, Pugs.Help). 11:47
SM_ax could barely compile pugs on 512M RAM! 12:13
autrijus iblech: they norm of ChangeLog is to have no trailing dot :D 12:16
Releng for 6.2.4 in progress | pugscode.org <Overview Journal Logs> | pugs.kwiki.org 12:16
iblech autrijus: Ok (it was mixed, so I counted and noticed there were more lines with ., so I though this was the standard) 12:17
autrijus sure. no worries
svnbot6 r3727, autrijus++ | * `$obj.method($arg1, $arg2)` can now MMD dispatche over all arguments. 12:18
autrijus the smoke looks quite well
can you help me by inspecting all the from_perl6_rules failures to see if they make sense?
i.e. if the test is plainly broken, fix the test; otherwise mark as :todo<feature>
iblech Sure :) 12:19
autrijus or rather, :todo<pge> :D
iblech ;)
autrijus nah. :todo<feature> is fine :)
iblech anchors.t -- the usual eval_ok thing with lexical vars, fixing 12:20
autrijus I really wonder if there's a way to make that work
duplicating %CALLER:: into lexical space?
shallowly link it?
autrijus ponders
svnbot6 r3728, iblech++ | * Unbreak Net::IRC (try {...} --> try {...};) 12:28
r3728, iblech++ | * BEGIN {...} --> INIT {...} in example bots
nothingmuch btw, i noticed that t/statements/control_blocks.t is still not really doing anything 12:31
i'll uncomment it when I get back home, if no one can beat me to it
wolverian hmm. I thought try is a statement? 12:34
iblech Thought that too, but no, it's like Perl 5's eval {...}, a "normal" sub, which needs trailing ;
wolverian is that specced? which AES? :) 12:35
integral hmm, I thought perl6 had a general rule now for where ; is needed after } ?
iblech S04
wolverian thanks
hmm. that makes me wonder. how does perl5 actually figure out when do { } is used in an expression and when as a statement? it can be both, can't it? 12:36
oh. no. it's just an expression. hmm. I wonder where I got that idea from. 12:37
iblech What was the new syntax for character classes again? <[a..z]>?
wolverian yes.
iblech Thanks
wolverian I really don't like try { } not being statement level. especially if the non-block 'do' syntax is in. (is it?) 12:38
do try { } # expression :)
svnbot6 r3729, iblech++ | config_h.pl -- Bikesheding. 12:42
r3730, iblech++ | First part of t/rules/from_perl6_rules/ fixes (anchors, array_cap, capture).
iblech autrijus: Hm, I wonder if it's problematic to add -j to the external Parrot options (Parrot.hsc:24)? 12:43
(-j -- JIT)
I'll try that in a sec 12:44
autrijus sure 12:45
robertle do you guys know parrot? 12:47
svnbot6 r3731, autrijus++ | * unTODO
robertle i have a small question there:
i am plying around with it and managed to write a *very* small compiler targeting parrot 12:48
but it just uses the built-in types statically
which isn't what i want
in the end i would like to have a dynamically typed lang
now the question: would one just use PMCs for every variable?
even for the simple ones? or would one want to mix PMCs and primitives? 12:49
and if: is there an example somewhere how this is done?
iblech robertle: I think so, yes. You would probably have (say) YourLangInt, YourLangStr, etc. But if your compiler is really smart, it cound optimize some YourLang* PMCs to native I, N or S registers 12:53
svnbot6 r3732, iblech++ | * Run external Parrot with -j (gives small speedup).
r3732, iblech++ | * charset.t fixes.
robertle iblech: thanks. so i'll start with PMCs and see what can be optimized later... 12:54
iblech robertle: Yep
autrijus robertle: irc.perl.org #parrot would probably be helpful :)
robertle autrijus: ah, cool! i thought here would be the "canonical" irc channel... 12:55
Odin- Not for Parrot... :p 12:56
Juerd What do people here think of a Perl 6 development server, with free access for Perl 6 people? 13:00
svnbot6 r3733, iblech++ | Remaining t/rules/from_perl6_rules fixes.
Juerd has already gathered some opinions in private, all positive 13:01
autrijus dinner & 13:03
robertle: you can mix pmc and prims as long as there's opcode for them
robertle: for lang-specific PMCs, see Py* and Perl* 13:04
iblech Juerd: (saying again) Such a server would be useful, methinks 13:12
svnbot6 r3734, iblech++ | t/rules/from_perl6_rules -- Forgot two TODOs and one unTODO.
wolverian Juerd: sounds like a very good idea to me. 13:13
robertle autrijus: i just found tha lua<->parrot compiler, that has a lot of example code that is applicable for me...
kungfuftr robertle: knewt is a big lua fan too 13:14
broquaint lua++ # see. Far Cry 13:16
svnbot6 r3735, iblech++ | Fix "{ BEGIN { 42 } }" (t/pugsbugs/begin_in_strings.t, moved to t/syntax/). 13:23
autrijus iblech: I need to run for dinner now... I'll tackle t/oo/submethods.t and t/operators/arith.t meanwhile 13:24
(just so we don't duplicate work)
iblech autrijus: Ok
Juerd I've ordered the parts for the box 13:28
I expect to have it online by the end of this week
(Depends on the colocation)
wolverian Juerd++ 13:29
Juerd This all started when I registered perl6.nl and needed a purpose :) 13:31
I think in this way I can be more helpful than by donating money
wolverian I wish I had something to donate. 13:32
iblech stevan: FYI, File-Spec -- t/10_unix_test -- hangs
Juerd I have little, but I get parts at wholesale, and made a deal for free colocation this morning
The specs: Pentium 4 2.4 GHz, 1024 MB DDR, 80 GB S-ATA in RAID 13:34
(Intel mainboard) 13:35
iblech Fixing BEGIN { exit }, CHECK { exit }, and INIT { exit }... 13:39
wolverian Juerd: now I'm sad. my only server currently is a 200mhz pentium pro, with 48mb ram. :) 13:45
Juerd thought it'd be best to spend some extra on memory and some less on cpu, given haskell :P 13:46
wolverian heh. :)
iblech BTW, Corion++ # wrote pugsrun/04-dash-ce.t which immediately discovered a pugs -c bug 13:55
Juerd Suggestions for a hostname? :) 14:01
Maybe a contest would be fun?
webmind hostname for? 14:04
svnbot6 r3736, iblech++ | Fixed exit_in_begin.t. 14:05
r3737, iblech++ | * Fixed BEGIN { exit } and CHECK { exit }. 14:10
r3737, iblech++ | * Moved t/pugsbugs/exit_in_begin.t to t/statements/ and added
r3737, iblech++ | t/statements/exit_in_check.t.
Juerd webmind: dedicated perl6 box 14:12
integral hmm, something from LotR maybe to stay in theme?
kolibrie I like that 14:15
webmind Juerd, woody woodpecker? fast, but completely mad ? 14:16
svnbot6 r3738, autrijus++ | * unbreak arith.t and submethods.t
webmind :)
nothingmuch dogloo
autrijus rehi
hm, it looks like BUILD will strategically very helpful 14:17
I'll attempt a quick implementation. :)
Khisanth even earlier than BEGIN? 14:22
autrijus no, BUILD is object initialisation
Khisanth bah! why couldn't it have been CREATE :p 14:23
iblech Khisanth: Sure! submethod CREATE(...) {...} &BUILD ::= &CREATE :) 14:24
Khisanth what is a submethod? 14:26
iblech A method which is not inherited and which is only called if you directly call it 14:27
There's the special submethod BUILDALL, which calls all other BUILDs of all classes involved
So each subclass can have its own initializer without having to do stuff like $self->SUPER::init()
Khisanth so it's like some sort of private method
iblech Well, only that it's not private :) $obj.submethod or $obj.Class::submethod work 14:29
S12 prefers the term (IIRC) infrastructural methods
nothingmuch bah, @Larry have still not commented on lazy context 14:33
*sigh*
autrijus nothingmuch: I got warnocked for nearly 4 days :) 14:34
nothingmuch while we're on the subject of lazy 14:35
i'd like this feature:
that way i could say 14:36
being able to know if you're being executed to satisfy lazyness
my $var = maybelazy call();
and if this code is not lazified by the user, this is strict
if not, it's lazy
and while we're on the topic, be able to lazify AOP style 14:37
at the invocant/method granuality
iblech my $var = want.lazy ?? lazy { call() } :: call();
autrijus right.
nothingmuch iblech: yep, sort of
so lazify(&Class::DBI::get) 14:38
but being able to decide at what instance level
eric256 the only responses on the lazy context had nothing to do with lazy context 14:53
Juerd webmind: Far too long 14:57
webmind: And woody without the woodpecker is confusing with debian woody, which it won't run
svnbot6 r3739, iblech++ | Tentative implementation of nothingmuch's lazy proposal. 14:58
r3740, autrijus++ | * move enterEvalContext to Monads.hs so Prim can use it.
r3740, autrijus++ | * introduce the `maybeM` abstraction that lifts fmapM into monads.
webmind Juerd, yes 14:59
autrijus iblech++ # jfdi on nothingmuch's proposal
iblech :) autrijus, style question: Are tabs in *.hs considered bad? 15:00
autrijus iblech: well, not bad perse, it's just my vim will accidentally retab it form time to time 15:01
I guess we can do something like 15:02
-- vim: expandtab shiftwidth=4:
to style it
iblech sets these two settings in his ~/.vimrc...
autrijus danke 15:03
eric256 iblech so we have a lazy { } construct now? 15:06
iblech eric256: Yep :)
eric256: (It was actually extremely simple to implement, as all Haskell "variables" are lazy per default) 15:07
Forth_ "It is more or less a description of how how Perl 6 programs" in 02Internals.pod 15:09
svnbot6 r3741, iblech++ | ChangeLog -- Added note about `lazy {...}`.
Forth_ has no tool now 15:10
autrijus Forth_: ok, fixed
cognominal is about to switch to vim. it seems easier than to figure utf8 in emacs. comments? 15:15
svnbot6 r3742, autrijus++ | * typo fix from Forth
Limbic_Region greetings and salutations all 15:17
autrijus (or anyone) - got a windows binary of the latest (or at least sometime today) of pugs executable?
I have a few cycles to work on my latest project and Jonathan's build was from last Friday 15:18
some of the bugs I had to work around that were supposedly fixed are still b0rk from my perspective
PerlJam Limbic_Region: start using linux and keeping up won't be a problem ;-)
Limbic_Region PerlJam - I have no problem at home 15:19
PerlJam just had the weird experience of seeing src/warnock.c in a parrot build
Limbic_Region well - I won't as soon as I get net access again (Wednesday) 15:20
PerlJam (at least that's what I thought I saw)
Khisanth PerlJam: that would tell him nothing about whether or not it's working on windows!
PerlJam turns out it was just a juxtaposition of warnings.c and packout.c
Limbic_Region has very limited capabilities on his government workstation and tends to play by the rules
ok - perhaps someone is willing to try a nopaste code snippet and tell me if it works with the latest build? 15:23
svnbot6 r3743, iblech++ | Various test fixes.
r3744, iblech++ | Added clarifying note to t/syntax/parsing_gt.t.
Limbic_Region perlbot nopaste
perlbot Paste your code here and #<channel> will be able to view it: sial.org/pbot/<channel>
iblech Limbic_Region: Sure
ninereasons PerlJam, that reminds me to ask, what does it mean "to Warnock" ? "ignore"? 15:24
PerlJam ninereasons: en.wikipedia.org/wiki/Warnock 15:25
pasteling "Limbic_Region" at 129.33.119.12 pasted "Is this bug fixed?" (127 lines, 3.6K) at sial.org/pbot/10339
ninereasons thank you, PerlJam :) 15:26
Limbic_Region iblech - if that doesn't work then the predecrement/postincrement bug still exists in one form or another
PerlJam ninereasons: sure, no problem.
iblech *** Modification of non-creatable array value attempted: -3
Limbic_Region iblech - so the bug is still there
it goes away if you uncomment the predecrement above 15:27
iblech (The error line: @position[ --$cur ]++)
Limbic_Region and change the array indice from --$curr to just $cur
PerlJam okay ... I've just built a parrot and am confused as to why it still says it's r8114 when I've clearly updated to r8143
Limbic_Region $cur is getting decremented twice for some reason
separating it makes the problem go away
PerlJam I'll just assume it's a parrot bug in the build
Limbic_Region but my simplistic test case apparent works as expected iblech 15:28
Juerd Wow, warnock has its/his own wikipedia page now? :)
Limbic_Region iblech - try just the following 15:29
pugs -e "my $curr=4; my @array=1..5;@array[ --$curr ] ; say $curr"
actually, I would like people on different platforms to try it
on my Win32 build - it prints "2\n" which is WRONG
iblech Limbic_Region: Yeah, you're right, the index is evaluated twice
Limbic_Region but supposedly that bug was already fixed
iblech Limbic_Region: It prints 3 here 15:30
Limbic_Region now you see why I find it odd
as a 1 liner it works for you
but in the full code snippet it does not
iblech Limbic_Region: It seems the index is only evaluated twice, if the whole expression is post-incremented (@array[--$curr]++)
Limbic_Region and as a 1 liner for me (albeit from a Friday build) it doesn't work 15:31
damn
that's my client eating plus symbols
what I pasted indeed had a postincrement
notice how far the semicolon is away?
ok - so that's the test case then 15:32
iblech Limbic_Region: Ah, yes. I'll add a pugsbugs test and investigate 15:33
svnbot6 r3745, iblech++ | More test fixes.
Limbic_Region thanks iblech
mind testing another one?
iblech Sure, next bug :) 15:34
pasteling "Limbic_Region" at 129.33.119.12 pasted "Lines with Potential bugs have been marked" (126 lines, 3.5K) at sial.org/pbot/10340 15:36
Limbic_Region ok - two lines in that one
having to do with 1..$foo - 3
supposedly it was also fixed
iblech Limbic_Region: Yeah, (isolated) they work here 15:37
Limbic_Region ok 15:38
how about one more?
iblech Sure
masak the reduce operator is broken
Limbic_Region this one will require debugging to see what's going on
as it won't spit out an error
I believe it goes into an infinite loop
but my minimal test case works - grrr 15:39
masak [+] (1,2,3) gives an error
unexpected "{"
iblech masak: Works fine here. :)
masak :/
iblech masak: Your revision?
masak I have revision 3745 15:40
err, maybe I didn't compile it, hold on... :)
iblech :)
pasteling "Limbic_Region" at 129.33.119.12 pasted "See the # Possible Bug line" (125 lines, 3.6K) at sial.org/pbot/10341 15:41
PerlJam sial.org does a decent job syntax highlighting perl6
Limbic_Region iblech - so the the last one worked as expected right - no problem removing the parens in 1..$foo - 3?
this one should just sit there - no output
iblech Limbic_Region: Yep 15:42
Limbic_Region not sure how to debug if you get the same behavior though
autrijus eric256: hm, is the warnings in wizards.p6 intentional?
iblech Limbic_Region: Indeed, infinite loop
autrijus eric256: can I take them away?
Limbic_Region my minimal test case didn't exhibit the bug
iblech (it seems)
autrijus also, BUILD is there now -- you can make use of it
iblech autrijus++ :)
Limbic_Region ok - so if you have any spare tuits ;-)
yay - being persistent pays off as he did find at least 1 real bug 15:43
wooo hooo
svnbot6 r3746, autrijus++ | * BUILD.
ingy autrijus: I was thinking that you could give a little Haskell workshop at the start of the Pugs hackathon
autrijus ingy: sure thing
ingy autrijus: will any other Haskell "experts" be there? 15:44
autrijus I lost the list 15:45
nothingmuch will be there
theorbtwo too?
so that's two
Limbic_Region iblech - thanks for your help 15:46
iblech np
Limbic_Region if you find out anything with that infinite loop (which oddly enough works as 2 separate statments) let me know
Limbic_Region is off to find foo
food even
vitaminmoo Hmm, is it a my local problem that's causing current parrot svn to not compile, or is it a repo-side thing? 15:47
masak iblech: everything works; it was just me being un-updated :)
iblech masak: :) 15:48
masak &
autrijus iblech: submethod.t is weird
svnbot6 r3747, iblech++ | Added Limbic_Region++'s test case.
autrijus class Baz { submethod blarb() { $was_in_baz_submethod++ } }
class Grtz is Baz { submethod blarb() { $was_in_grtz_submethod++ } }
and it expects Grtz.new.blarb to call _both_ blarb()s 15:49
is that correct?
there's no blarbALL, here
or am I missing something?
iblech autrijus: You're correct. I even asked on p6l if it's correct that only Grtz::blarb is called, but apparently missed to fix the test 15:50
Only Grtz::blarb should be called
But if one says Grtz.new.Baz::blarb then Baz::blarb should be called
autrijus right.
okay, I fixed the test accordingly
iblech Ok, you may want to put a ref to thread "BUILD and other submethods" of p6l in the test. 15:53
autrijus can you add it if you have it handy?
iblech Sure
autrijus is still deep in relenging
danke
iblech++
svnbot6 r3748, autrijus++ | * fix submethod.t
gaal|work oh, releng! i'm at $work and can't help much :( 15:54
autrijus that's fine :)
enjoy $work
gaal|work thanks :)
iblech Limbic_Region: The infinite loop bug is indeed weird... 15:57
broquaint I've got a patch for map.t, who can/should I bounce it on to? 16:05
svnbot6 r3749, iblech++ | Added p6l ref to submethods.t. 16:06
r3750, autrijus++ | * fix parrot codegen
iblech broquaint: nopaste it here
broquaint Shouldn't failing eval_ok()s not break the whole test unit? 16:07
iblech Yep
broquaint Hrm ... 16:08
Well, I'll paste what I've got anyhow.
autrijus down to 13 fails
broquaint What's the nopaste link? 16:09
iblech perlbot nopaste
perlbot Paste your code here and #<channel> will be able to view it: sial.org/pbot/<channel>
pasteling "broquaint" at 81.187.213.103 pasted "New map.t with the first eval_ok() breaking the whole unit" (100 lines, 3.3K) at sial.org/pbot/10342 16:10
svnbot6 r3751, iblech++ | Updated docs/quickref/README and moved quickref/match to quickref/smartmatch. 16:12
r3752, autrijus++ | * unTODO now BUILD is working
iblech broquaint: the first eval_ok test should fail, as map { $_ => 1 } @list is parsed (correctly) as map (hash { $_ => 1 }) @list 16:13
broquaint: It does parse, but during execution it dies 16:14
broquaint Along with the rest of map.t, iblech.
And I thought the "map { $_ => 1 } @list" was meant to magically return a list of hashes, no? 16:15
iblech broquaint: Yeah, but that's a different problem ;) -- Pugs tries to cast the hash into a code, which of course doesn't work, and then dies untrappable
autrijus down to 6 pugsbugs! 16:16
theorbtwo is not a haskell expert by any stretch.
autrijus but theorbtwo hacked in eval_haskell :)
broquaint Ah, so type errors aren't for the likes of eval_ok() then?
iblech broquaint: No, if you want a LoH you'd have to write map { scalar { $_ => 1 } } @list, I think
broquaint: It depends, most type errors are catched, but, as you see, some are not. 16:17
svnbot6 r3753, autrijus++ | * eval_haskell now sets $! on failure again
r3754, autrijus++ | * initialise $_ in default_scalar.t
r3755, autrijus++ | * TODO and cleanup var.t
iblech autrijus: :)
broquaint I see, thanks, iblech :)
autrijus map { hash($_ => 1) } 16:18
is perhaps more readable
iblech: do you have more tuits to attack pugsbugs? :)
I'm doing t/pugsbugs/unhashify.t and t/pugsbugs/postincrement_in_subscripts.t -- I wonder if you can take a look at t/pugsbugs/and_precedence.t, t/pugsbugs/bare_block_with_dollar_underscore.t and maybe t/pugsbugs/map_constant_list.t 16:19
that's all there's left
Limbic_Region iblech - so other than weird, any idea what's going on?
broquaint Can you see a way of running that test (i.e ensuring breakage) without that untrappable error, iblech (or anyone else for that matter)?
iblech autrijus: map_constant_list -- I don't think ("foo","bar").map(){ $_.substr(1,1) } is valid syntax, is it? 16:20
autrijus of course not 16:21
iblech broquaint: You can try to use "try": is try {...}, 17, ...
broquaint: But some type errors are (currently) really untrappable for the Perl-space
autrijus: Ok, then map_constant_list is fixed :) 16:22
pasteling "broquaint" at 81.187.213.103 pasted "map.t with a few extra tests" (99 lines, 3.3K) at sial.org/pbot/10343 16:24
svnbot6 r3756, iblech++ | Added broquaint++'s map test.
r3757, iblech++ | Fixed map_constant_list.t -- will move to a new dir next.
r3758, iblech++ | Moved t/pugsbugs/map_constant_list.t to t/builtins/lists/map_constant_list.t.
broquaint Thanks, iblech :)
iblech Limbic_Region: By slapping "say 1", "say ", etc. to most lines, I was able to track down the root of the infinite loop, but then I saw you've already marked it :) I haven't yet found out more 16:25
broquaint: np :)
knewt eeeek. ok, not having purchased the plane tickets yesterday i have to travel back on the sunday to get a reasonable price now. the flight leaves at 9am on a sunday. /me cringes at having to get to the airport at 7am on a sunday morning 16:33
pasteling "AtnNn" at 66.36.135.156 pasted "are these bugs?" (37 lines, 440B) at sial.org/pbot/10344 16:37
Limbic_Region iblech - heh
I knew where it was - just not wy 16:38
why even
my isolated test case made the problem go away and I haven't figured out the right combination of things to cause it to reappear
was hoping you might know some tricks I didn't
iblech AtnNn: [|] and [&] are bugs 16:40
AtnNn: my $a, $b is a missing feature
Juerd I thought they were operators.
iblech :) 16:41
AtnNn: $*_=5;sqrt(4); works fine here
Juerd: I referred to the bug report of AtnNn 16:42
AtnNn its confusing, when you declare a variable and pugs prints out the result fo the previous line 16:44
iblech Agreed.
AtnNn: I could reset $*_ to undef before running commands in the interactive shell, but I then it'd be confusing that manually setting $*_ won't work 16:49
gaal|work Juerd, ping 16:51
Juerd pong
gaal|work how about, for the hostname... "hump" ? :-)
Juerd See my rationale against camel :) 16:52
Though hump is much cuter already.
PerlJam Juerd: I like "onion" best btw.
Juerd I quite like onion, but I never received the message that suggests it...
I wonder how that went wrong.
gaal|work [email@hidden.address] Juerd. 16:53
PerlJam Maybe it was only sent to the list and you just haven't gotten it yet.
Juerd Oh, I did receive that message
I even replied to it!
Just completely overlooked that paragraph
gaal|work it was a short paragraph.
While I'm here, can I have either 'gaal' or 'roo' please? 16:54
Juerd You can, but please mail me 16:55
gaal|work sure.
Juerd That eases replying :) 16:56
gaal|work .go(:home) &
Juerd Bye
svnbot6 r3759, iblech++ | Bare blocks with $_ fix -- part 1. 17:05
wolverian featherhump? 17:07
Limbic_Region a sort block can have any code you want in it provided it returns -1, 0, 1 right?
so you could say $^a for instance
iblech Limbic_Region: Yep 17:08
Limbic_Region iblech - will see what I can do about debugging 17:09
I know what I expect and might be able to see what's going wrong if I can determine at what point it stops doing that 17:10
iblech Limbic_Region: Ok, that's a good idea! 17:13
autrijus: ruleStandaloneBlock now works WRT $_ correctly, but there're some blocks which aren't detected as standalone blocks
bbiab & # food 17:14
Limbic_Region iblech - might be a problem with sorting an empty list 17:17
svnbot6 r3760, autrijus++ | * Hash and array sub parameters are now read-only by default, same as scalars.
autrijus iblech++
clkao (is copy)++
Limbic_Region getting somewhere now 17:18
I think I might be able to make a minimal testcase
svnbot6 r3761, autrijus++ | * fix head.p6 golfing. 17:32
r3762, autrijus++ | * Slurpy hash parameters no longer count as nonslurpy during arity matching.
iblech re. Limbic_Region: That'd be excellent! :)
Aankhen`` W00t! 17:33
I'm getting a dual-core Pentium 630!
Or perhaps an AMD dual-core processor, if they're released soon and at a reasonable price... 17:34
Limbic_Region iblech - might be an optimization 17:38
somehow return () is getting changed to return undef
Juerd Why?
undef in list context is 1 element!
While () in list context is 0 elements
Yet undef in scalar conetxt
context too
Limbic_Region Juerd - I don't know why
gaal rehi
Limbic_Region without the sort block it works correctly
iblech Juerd: You're correct, I think, it's a bug in Pugs (it seems) 17:39
Limbic_Region adding a sort block somehow affects it
I am minimizing the test case but as it stands now - it is pretty small
iblech Limbic_Region++, very good 17:40
Limbic_Region it gets even more interesting
if a statement after the return () is present - the bug appears 17:41
without it
it disappears
even though that statement never gets executed
iblech bad...
Limbic_Region perlbot nopaste 17:42
perlbot Paste your code here and #<channel> will be able to view it: sial.org/pbot/<channel>
gaal is everythinhg in src/syck pristine or are local changes ok? i want to look into a warning.
pasteling "Limbic_Region" at 129.33.119.12 pasted "minimized test case" (20 lines, 411B) at sial.org/pbot/10345 17:43
autrijus gaal: feel free to commit local changes as long as you report to why
svnbot6 r3763, autrijus++ | * unTODO.
autrijus I mean, Why
Limbic_Region iblech - that is about as small as I can make it
now let me explain why it is so interesting
without the sort block - the bug goes away 17:44
the bug doesn't manifest itself unless there is a statement after the return () line
so it only shows up with the sort block AND a statement after the return () line
if those two conditions are present, return () seems to change to return undef
so the test case is easy enough 17:45
iblech Now this is *really* weird
Limbic_Region @foo.elems should be 0 but it is 1
why and how to fix it on the other hand - I have no idea 17:46
iblech Maybe it's the sort block which converts the () to undef 17:47
eric256 i've seen the @foo.elems = 1 when the list is empty... not nice. but i couldn't nail it down and make it happen on a regular basis 17:48
autrijus hm 17:50
map({ 3 } 1..10);
should this be allowed? 17:51
eric256 which part of that is in contention? the parens without comma?
autrijus the lack of comma after }
Aankhen`` IMHO, it shouldn't be allowed.
autrijus ok. 17:52
Aankhen`` When you use parens, you're saying "Do What I Say" -- atleast, that's my interpretation of it -- and that means you want to be explicit about the parameters.
autrijus anyone want to write a test for it?
Aankhen`` One sec, let me sync up 17:53
.
I really hope I have a copy of Pugs built and installed.
Limbic_Region sorry - got dc there
Aankhen`` WB.
Limbic_Region hmmm - seems my client is having a problem 17:54
in any case - the empty array isn't really empty - it has a single undef element which it shouldn't
Aankhen`` autrijus >> Does that apply to all sub/method calls? 17:55
autrijus Aankhen``: sure, but just test map{}
is enough
Aankhen`` Yeh, I was wondering about the file to put it in. 17:56
t/subroutines/sub_arg_errors.t?
autrijus sure
Limbic_Region iblech - if you need further help from me you might want to email
Aankhen`` OK.
Limbic_Region not sure what's up with my client 17:57
svnbot6 r3764, autrijus++ | * `map({...} @list)` is no longer valid syntax. 17:58
r3764, autrijus++ | * `foo 3 and foo 4` is now parsed as two separate function calls.
autrijus preflight starting soonish. 17:59
iblech: how's bare_block_with_dollar_underscore.t coming along?
iblech autrijus: All blocks which are correctly detected as lone blocks work, e.g. { $_ + 5 }. But, for example, { 42; { $_ } } does not work, dunno why yet 18:01
autrijus ok, I'm taking a look
t/builtins/lists/uniq.t parsefail -- take a look?
may be because of r3764
iblech It'd be helpful if Parsec had some kind of tracing switch, which'd cause it to print all rules it tries to STDERR 18:02
Sure
autrijus iblech: you can do that with trace calls
ruleName = trcae ("Trying me") $ do
ruleName = trcae ("Trying me") $ rule "foo" $ do
s/trcae/trace/
iblech autrijus: Yeah, doing so, but it'd be cool if Parsec would do that, possibly even with nice indenting 18:03
autrijus with some hacking you may be able to hook it into the "rule" :)
I mean the "rule" function itself
iblech Yep. Will do that probably then, when I find time 18:05
Limbic_Region actually - I think client issues have gotten fixed
I also seem to be reducing it further
autrijus I fixed bareblock 18:07
iblech WOW! How?
FAST! :)
autrijus see diff. committing
Limbic_Region perlbot nopaste 18:08
perlbot Paste your code here and #<channel> will be able to view it: sial.org/pbot/<channel>
autrijus done
gaal build entered smoke a minute ago, shall i restart?
autrijus gaal: no need to... wait till preflight
iblech: when in doubt use brute force 18:09
gaal hm, to get past the "can't locate File/Spec.pm" error, do i need blib/ in perl6lib?
...or something like that? :)
iblech make 18:10
svnbot6 r3765, iblech++ | Fixed uniq.t.
r3766, autrijus++ | * auto-executing bare blocks no longer take any arguments.
gaal hmmm, make smoke doesn't create a blib
iblech autrijus: *understands diff*. It seems a little bit hacky, but works :) 18:11
gaal: Yeah, then make w/o params
gaal: Yeah, then make w/o params first
pasteling "Limbic_Region" at 129.33.119.12 pasted "Weird bug test case - see comments for how to test and what to change to change behavior" (15 lines, 372B) at sial.org/pbot/10346
gaal thanks.
is make =:= make optimized these days? 18:12
Limbic_Region iblech - that is a minimal test case - step 1 introduces the weird behavior, step 2 makes it go away again, I also showed how to test for the behavior .elems
and s/iblech/iblech, autrijus, gaal, et all/
except the first Step 1 should be Step 2
grrr
autrijus gaal: yes. 18:13
iblech gaal: make unoptimized; make; make smoke
pasteling "Limbic_Region" at 129.33.119.12 pasted "Corrected weird problem test case" (15 lines, 372B) at sial.org/pbot/10347
gaal on my machine optimized pays off 18:14
pasteling "iblech" at 80.81.4.110 pasted "more reduction" (12 lines, 330B) at sial.org/pbot/10348 18:15
svnbot6 r3767, autrijus++ | * make the test more straightforward now we don't need the lives_ok.
gaal *** Can't modify constant item: VUndef at t/builtins/undef.t line 222, column 9-89 18:16
Limbic_Region iblech - I am not sure how much sort has to do with it 18:17
I can now get the weird behavior without sort
though it IS how I found it in the first place
in any case - I know nothing of Pugs internals so I have taken it as far as I know how 18:18
pasteling "iblech" at 80.81.4.110 pasted "Even more reduction" (5 lines, 122B) at sial.org/pbot/10349 18:19
gaal what does *sub mean? 18:20
Limbic_Region yay iblech - woo hoo
now it is important to verify that once that gets fixed that the introduction of the sort block doesn't break it again
iblech gaal: run sub etc and then flatten
gaal ah, right. 18:21
autrijus I'm finally done with bugtracing.
Limbic_Region remember the predecrement problem was supposedly fixed but in the context of an overall postincrement it was still there
autrijus iblech: make that nopaste into tests?
iblech autrijus: Yep
autrijus ok. I'll begin preflight anyway
please :todo<bug> it :) 18:22
iblech sure
Limbic_Region hi-fives iblech
autrijus Limbic_Region++
iblech++
gaal forum2.org/gaal/pugs/smoke-cygwin.html # r3637
autrijus iblech: oh, Net::IRC is probably ready to be OOified now (but post-6.2.4) 18:23
gaal: please try 3767 :)
Corion The GHC mailing list pointed me in the general direction of System.Win32 - I seem to have missed that completely %-)
iblech autrijus: :) Very good
gaal errr :) 3676, i can't type :) 18:24
Corion Oh. It also seems to be a separate library and not with stock GHC.
autrijus still try 3767 :)
Corion: huh? I think it's with ghc
gaal hmm, interesting :)
autrijus 6.4 at least
Corion autrijus: Hmm - I always look via www.haskell.org/ghc/docs/latest/htm...index.html , and it's not there
gaal where'd the harness get that bogus r from? 18:25
Corion Is there some other way I should look ?
autrijus Corion: your own system
Corion autrijus: Oh :)
autrijus c:\ghc\ghc-6.4\docs\libraries
or something like that
Corion Still no System.Win32 18:26
revdiablo Limbic_Region: ping
Limbic_Region autrijus - you plan on releasing 6.2.4 sometime today?
gaal full moon tonight! you feel lucky--more--
revdiablo Limbic_Region: checkitout: www.perlmonks.org/index.pl?node_id=459659
autrijus Limbic_Region: in an hour.
or less.
Limbic_Region revdiablo: pong
Jonathan_ 'lo all
Corion We are moving into the general direction of preflight?
autrijus Corion: yeah
Corion So I'll start smoketests ;) 18:27
autrijus hey Jonathan! can you provide icu-less parrot 0.2.0 builds?
if so, I'd be happy to list it in README
Limbic_Region hey Jonathan_ - do you not build builds over the weekend?
revdiablo - I would only add a link or 2 (but not a bunch) for more info
revdiablo Limbic_Region: suggestions on what links? 18:28
Jonathan_ autrijus: Probably can do that, yes - haven't been doing much recently due to lack of time/illness.
Limbic_Region revdiablo - my Get involved with Pugs has several good ones, maybe you just link to that
Aankhen`` autrijus >> I'm really sorry, I got distracted by some major melodrama. 18:29
Jonathan_ Limbic_Region: They should normally happen at weekends, but last few weekends I've had to go away due to family stuff.
autrijus Jonathan_: aw, illness. you got better?
Aankhen`` And my working directory was apparently a mix, so I couldn't even run `nmake clean`.
Limbic_Region autrijus - have found several bugs with my latest project and I am sure more are lurking under the cover - will wait until after I have a 6.2.4 to investigate further
autrijus Limbic_Region: 'kay
Limbic_Region thanks Jonathan_ and no worries - I assumed it was automated 18:30
Jonathan_ autrijus: Long story short, my father passed away recently (cancer), then in worry/stress of that plus exams looming at uni I've pulled myself an anxiety related gastric problem. Am getting better, but not quite there yet.
Limbic_Region Jonathan_ plus plus for taking the time to help those that can't build for themself
svnbot6 r3768, iblech++ | Test for bug found by Limbic_Region++ (return_with_trailing_stuff.t).
r3769, autrijus++ | * fix plan.
Limbic_Region autrijus - it seems to me the best way to find bugs is to write code ;-) 18:31
Limbic_Region finds porting p5 to p6 the easiest way to write that code
gaal i thought it was to release?
Jonathan_ Limbic_Region: It's automated, apart from one problem: the makefile generationg bug I reported on the mailing list ages ago. So I've been having to hack the makefile by hand. Wonder if MakeMaker update would help...
iblech FYI, I'm looking in ext/*/t now
gaal [1] t/builtins/strings/append.t pugs.exe: cannot cast from VList [] to Pugs.AST.Internals.VCode 18:32
Corion And btw, the first response to my post to glasgow-haskell-users is www.haskell.org//pipermail/glasgow-...08495.html - seems like they want BSD3 license. I don't know what the difference between BSD3 and Artistic is.
(yet)
autrijus I'm willing to relicense to BSD3 :) 18:33
for the part that they want, anyway.
gaal sorry, wrong test. it's in t/pugsrun/09-dash-uppercase-c.t
(concurrency :)
revdiablo Limbic_Region: ok, linked to your node. I'm off to lunch
gaal ..on windows. 18:34
autrijus Corion: BSD3 is the advertisement-less BSD license
Limbic_Region revdiablo plus plus - advocacy advocacy advocacy
Corion autrijus: Ah, so there shouldn't be much conflict between Artistic and BSD3.
autrijus Corion: artistic can't be made into bsd3 18:35
Corion autrijus: It can't ? Because Artistic wants to retain the author?
autrijus no, you have to rename the code :)
and document all the deviations
so it's not "can't"
it's "takes quite some work"
Corion Ugh.
pasteling "gaal" at 192.115.25.249 pasted "more test fails" (10 lines, 401B) at sial.org/pbot/10350 18:36
autrijus but the copyright holder can always relicense.
Corion So we should triple-license src/Pugs/Compat.hs
(or Pugs :) )
autrijus nono
we just relicense the relevant chunks in Compat.hs as BSD3
drop those thunks from Compat
revdiablo Limbic_Region: maybe we can take turns posting something about perl6 or pugs at least once a week? ;) (ok, I'm really off to lunch this time)
autrijus and use GHC 6.4.1 :D
s/thunk/chunk/
gaal sqlite-license it
autrijus gaal: I wish.
Corion autrijus: Yeah. I hope they take the stuff in so I have to worry less about it and can fire off angry mails about it :)
autrijus Corion: indeed 18:37
Odin-LAP sqlite is PD, right?
autrijus is fond of the sqlite license
Corion Odin-LAP: Yep
autrijus Odin-LAP: yes, it has this blessing license.
Odin-LAP In the ideal world, there wouldn't be anything but PD, and people would specify the authors of each bit...
:p
gaal sqlite.org/copyright.html # vgrep for "dedicate" 18:38
Odin-LAP (Or at least where it came from if 'stolen'.)
autrijus mmm Xanadu :) 18:39
iblech autrijus: FYI, definition of a sub infix:<ā‰ > works, but then using it does not
autrijus k.
iblech (other Unicode ops work fine)
autrijus weird. 18:40
Corion sub infix:<#> { :-))))
Odin-LAP Corion: That way lies insanity. 18:41
svnbot6 r3770, autrijus++ | * don't be too eager about annuling bare blocks.
Limbic_Region revdiablo - my next post is going to be my p6ified version of the last post
autrijus ...but #perl6 is the very manifest of insanity anyway
Odin-LAP autrijus: Good point. 18:42
gaal smoke matrix up 18:44
oy, i'm finally seeing the borkage you guys mentioned ther other day 18:45
huh? why didn't it test rules :( 18:46
autrijus hm, stevan is not here 18:47
svnbot6 r3771, autrijus++ | * ChangeLog cleanup before adding more entries
autrijus who wants to do the heroic act of reviewing r3656 to r3771 and update ChangeLog? :)
iblech Will do 18:48
autrijus (complicating the fact is that some of the changes are already in ChangeLog)
iblech++
revdiablo did anyone ever figure out whether t/pugsbugs/and_precedence.t was really caused by precedence, or something else? 18:56
svnbot6 r3772, iblech++ | HTML-Entities -- unTODO
r3772, iblech++ | fp -- :todo<bug> weird operator overloading bug.
autrijus revdiablo: yes, and I fixed it 18:57
revdiablo ah, nice.
I love it when I add a test and it gets magically fixed a day or two later. :)
autrijus it's because in parenless form loose operators should not be dispatched.
revdiablo: that's how it works around here :)
revdiablo it's a fun way to develop 18:58
autrijus sure is
gaal 3770 smoke lookin' good so far 19:00
Juerd 10 registrations so far, for the dev server 19:01
svnbot6 r3773, autrijus++ | * unbreak map{} use in 09-dash-uppercase-c.t
r3774, autrijus++ | * unTODO.
r3775, autrijus++ | * bump versions. preflight!
Juerd Imagine that this should increase the productivity of 10 people! 19:02
autrijus I want an account too :)
uploading preflight 19:03
clkao whoot!
autrijus please use "make test" instead of "make smoke" to test preflight
clkao what registration?
autrijus clkao: see p6l/p6c 19:04
Juerd autrijus: mail me :) 19:06
autrijus: That's easier for my administration
metaperl juerd
svnbot6 r3776, iblech++ | Updated ChangeLog to contain the revisions 3657-3771.
Juerd metaperl: That be me
autrijus pugscode.org/tmp/Perl6-Pugs-6.2.4-pre1.tar.gz 19:07
eric256 must be a heck of a change log
autrijus Preflight! pugscode.org/tmp/Perl6-Pugs-6.2.4-pre1.tar.gz | pugscode.org <Overview Journal Logs> | pugs.kwiki.org 19:07
autrijus please test :) 19:08
revdiablo dang, my r3772 build hasn't even cooled off yet 19:09
theorbtwo "tobez" and "colomon" are missing from AUTHORS. 19:10
Juerd metaperl: ? 19:11
autrijus theorbtwo: tobez is gruber is anton 19:12
not sure about colomon
gaal is it just me or has a test run slowed down since two weeks ago? 19:14
autrijus I havn't noticed 19:15
theorbtwo Blink, blink... since when does OSSF have useinfo pages?
gaal "duration: 901" (seconds) on a machine that did 6.2.2's make smoke in about 250 19:16
theorbtwo That makes this /so/ much easier.
autrijus theorbtwo: ~two years ago or so
or one year ago
Juerd metaperl: ?
gaal maybe it's just my machine being sucky? /me goes to look
autrijus iblech: so I gather you're fixing ext/*/t/? need any help?
svnbot6 r3777, iblech++ | * Locale-KeyedText -- Renamed test to lowercase names so they're run
r3777, iblech++ | * Locale-KeyedText -- Skipping most tests
r3778, autrijus++ | * minor cleanup to ChangeLog
autrijus otherwise I'll review changelog for a final time 19:17
iblech Yeah, currently doing Set
autrijus cool
iblech CGI seems to be borked, too
theorbtwo Wow. That was easy. 19:18
gaal nyeh, make test is broken with cygwin. sorry.
clkao good, you will want to use svk on that server heh 19:19
juerd++
Juerd :) 19:20
Jonathan_ autrijus, Limbic_Region: There's a (hopefully) automated Parrot build now available from www.jwcs.net/~jonathan/perl6/parrot-win32.zip - next job is to point the Pugs building at a Parrot tree so we get rules with my Win32 Pugs. Recommend using the 0.2.0 release rather than the current latest SVN tree?
Limbic_Region Jonathan_ plus plus 19:21
thanks
svnbot6 r3779, theorbtwo++ | AUTHORS update 19:22
Limbic_Region don't know about the Parrot version though
Juerd - it sounds like you should make a post at the Monastery regarding the Perl Development Server
Jonathan_ Limbic_Region: If we use 0.2.0, we get stability. If I use the latest Parrot, I guess we get the latest work on PGE, so most up to date rules support.
Juerd Limbic_Region: No. And please don't. 19:23
Limbic_Region oh... ok
Juerd Limbic_Region: It's a mere P4 2.4
Limbic_Region several monks seem like they would get involved if they could build GHC/Pugs/etc
Juerd Yes
And they can
Limbic_Region I /msg'd diotalevi about it already - sorry
Juerd But WAY too many people read PM 19:24
Oh, private messages are okay :)
Limbic_Region Jonathan_ - my recommendation would be to use 0.2.0 as a fall back (if that's possible)
PerlJam Juerd: you aren't ready to suffer the perlmonk effect? ;)
Juerd PerlJam: I am, and have
gaal L~R, who's stuck because they need to build GHC?
Juerd PerlJam: I want to see the load with a limited number of users first
Limbic_Region gaal - apparently diotalevi
gaal what platform is missing binaries? or is he going hardcore? 19:25
Juerd PerlJam: Opening it up to practically the whole perl 5 world now can make this experiment a disaster before it starts
Limbic_Region Juerd - might I also suggest a wiki to discuss what sort of collaborative projects result from the community server?
Juerd Limbic_Region: And I don't think anyone benefits from a torrent of new contributors all at once...
Limbic_Region: Yes, a wiki will be there
autrijus gaal: "make test" is broken? how?
Jonathan_ Limbic_Region: I'll look into it - need to go do some other stuff for now. 19:26
Limbic_Region I ask only because I wouldn't have a need for the shell account (I can build just fine) - but I might want to if there's mischief and mayhem to be made
gaal i'm still getting the error on undef.t with preflight 1
autrijus, need to cygpath stuff
autrijus gaal: ok... you want to fix it before release?
gaal: nopaste the undef.t error again?
Limbic_Region gaal - apparently there are no binaries or he is going hardcore - either way I don't know the particulars
autrijus undef.t worksforme here 19:27
Juerd Limbic_Region: I don't understand.
pasteling "gaal" at 192.115.25.249 pasted "undef.t error" (2 lines, 87B) at sial.org/pbot/10352
Limbic_Region Juerd - what I am saying is that I don't have a NEED. But if there is a project going on (that I would learn about through a wiki) I might WANT to get an account 19:28
theorbtwo wanders off for an Ally McBeal episode.
iblech autrijus: FYI, undef.t works here, too
gaal autrijus, i don't know if it's worth testing, cygwin being a hacky psedeotarget anyway
worth fixing, that is 19:29
autrijus iblech: nod.
Juerd Limbic_Region: There should be no project private to this server, I think.
Limbic_Region: It's connected to the internet
Besides, this server may go down at any moment (in theory only)
metaperl Juerd: what do you want to do with Perl6 and XML parsing? I just read your p6l post
Juerd Let's not introduce single points of failure
metaperl Juerd: what are you registering people for? (I missed the convo earlier in this group). 19:30
Juerd metaperl: I want someone to simply start testing the generated grammar (the grammar generation script needs updates too)
metaperl: perl 6 development server
metaperl oh I see... I might be interested... if your grammer part of pugs now?
Juerd metaperl: That was ordered in parts today and will be online by the end of this week probably
No
The server is the perfect solution for people who have no linux box of their own and think they can put access to a linux box to good use for Perl 6 19:31
autrijus gaal: svn up and run undef.t by hand and see if it works
gaal roger that
metaperl Juerd: what about matts comments "XML can't be parsed using the grammar alone." 19:32
svnbot6 r3780, autrijus++ | * don't test undef.t's rules part when there's no rules support.
r3780, autrijus++ | gaal++ for reporting it.
PerlJam Juerd: It would be nice if it were a freepan mirror too (but that's something else entirely)
gaal svn: Failed to add file 'ext/Locale-KeyedText/t/lkt_00_compile.t': object of the same name already exists
autrijus gaal: kill that dir and svn up again
iblech FYI, almost finished with Set 19:33
gaal also, either t/subroutines/splatty_with_type.t or t/oo/class/nested_use.t is failing (more info when the run completes) 19:34
Juerd metaperl: Note and ignore
metaperl: This has little practical use except stress testing the rule engine with an existing grammar
PerlJam: As some Dutch say, romeo delta
metaperl Juerd: so this would be a way for me to learn about p6 rules?
sounds useful
and also learn about XML 19:35
Juerd Which stands for RD which stands for Regel Dat (literally: regulate that, idiomatically: make it happen if you want it)
:)
metaperl which I only have a layman's understanding of
Juerd metaperl: Yes, it ought to
wolverian sorry, what grammar are you talking about?
metaperl wolverian: perlmonks.org/index.pl?node_id=179755 19:36
wolverian: an XML grammer
gaal autrijus, what's necessary at build time for rules support to go in? because i did have parrot in my path.
not embeddedly perhaps.
metaperl Juerd: I would like to get involved with testing the XML grammar
wolverian metaperl: ah. thanks a lot. 19:37
gaal in any case, your skip fix works.
autrijus gaal: just by having parrot in your path is enough.
cool, gaal++
wolverian Juerd++ # wow, that's an amazing node 19:38
gaal well, it doesn't look to be enough on cygwin :( but it's probably a path issue again.
autrijus gaal: sadly I'm out of tuits to look at cygwin :-/ 19:39
gaal yeah, we all are. i know, only it's the least crummy setup i have to build on :( 19:40
i'll try to play with the free ms toolchain tomorrow to see if i can build perl5+parrot+pugs with it
Juerd wolverian: It's ancient, 2002! :) 19:41
wolverian Juerd: haha. and I wondered why it wasn't in the recent best!
iblech Ok, Set done. Is anybody working on CGI? 19:46
svnbot6 r3781, autrijus++ | * fix try{} usage in CGI.pm
r3782, iblech++ | Fixed and TODOed Set.
gaal heh heh 19:47
wolverian Juerd: is onion going to run parrot too?
PerlJam wolverian: That's probably up to you as an account holder. 19:48
autrijus iblech: only 1 failure. please look at it 19:49
t/header...........ok 1/8# Looks like you planned 8 tests, but ran 11
I'm still ChangeLogging
iblech autrijus: Ok
autrijus probably just a plan fix
wolverian PerlJam: hmm, sure, but if the plan is to have a system-wide, continually rebuilt pugs, why not parrot as well? 19:50
(if that isn't the plan, sorry, I misunderstood.)
gaal util/smoke_loop.sh is a start, folks 19:51
Limbic_Region [14:26] <Limbic_Region> autrijus - you plan on releasing 6.2.4 sometime today? [14:26] <autrijus> Limbic_Region: in an hour. 19:58
svnbot6 r3783, iblech++ | CGI -- Plan fix of header.t.
Limbic_Region did it happen?
or still working on bug fixes?
gaal L~R, see topic
Limbic_Region still preflight - gotcha 19:59
you know - these meetings at $work are really beginning to interfere with my pugs fun
autrijus iblech: I'm fixing File::Spec
feel free to skip to next one
(or are you doing so?)
iblech Ok
IIRC that File::Spec was the last borked module 20:00
Checking...
Yeah, all other ones work 20:04
autrijus good. I fixed that one 20:06
anyone finished running main tests of pre1?
other than gaal that is
obra is that svn head? 20:09
svnbot6 r3784, autrijus++ | * fix the while loop test in File::Spec::Unix
obra will do so
iblech autrijus: FYI, a test was still failing in File::Spec, I force_todoed it. 20:11
autrijus iblech: don't -- I'm fixing it :) 20:13
iblech autrijus: To late :(
autrijus that's fine
Limbic_Region it can be unfixed
or rather untodoed
iblech starts building the live cd
Limbic_Region really shouldn't try conversation on Mondays
clkao autrijus: there was a failure for me 20:14
t/builtins/undef.t 1 256 72 24 33.33% 61-72
was this fixed?
autrijus clkao: yeah
ok. all tests shall now pass. 20:15
svnbot6 r3785, iblech++ | File-Spec -- force_todo the last failing test.
r3786, autrijus++ | * File::Spec::Win32 - fix canonpath()'s misuse of $0
r3786, autrijus++ | * unForceTODO 20_win32_test.t
autrijus rolling pre2!
gaal decides a voodoo reboot is in order. 20:16
clkao oh, larry's giving keynote for yapceu
autrijus yeah
sadly I've exhausted my conference budget :) 20:17
clkao everytime i build pugs optimized the laptop is burning my lap
hate
autrijus: budget of money or time?
autrijus clkao: money _and_ time.
clkao damn you, yapceu will be my birthday 20:18
autrijus time is fixable if I don't stay around and tourism
clkao i should prepare a gift
autrijus money is less fixable :)
clkao stick around europe ? i thought you are going to .ee ?
autrijus Pre2! pugscode.org/tmp/Perl6-Pugs-6.2.4-pre2.tar.gz | pugscode.org <Overview Journal Logs> | pugs.kwiki.org 20:18
autrijus clkao: that's too long a delta 20:18
a month 20:19
Tester wanted for pre2: pugscode.org/tmp/Perl6-Pugs-6.2.4-pre2.tar.gz
obra autrijus I'm running tests on head 20:22
is that the same?
autrijus obra: it's the same.
obra ok.
I'm into t/rules 20:23
autrijus "make smoke" or "make test"?
obra but man. the test suite got slow again
make smoke
bad?
autrijus k.
well, it won't test the modules that way, but that's fine
obra ok
I'l run make test next
autrijus we have ~700 new tests.
obra wow 20:24
_metaperl Juerd, did you want to discuss the p6/xml thing some more?
I started the convo at home, but left for work (where I am now).
_metaperl alt-tabs to work
autrijus and 677 revisions -- compared to 507 for 6.2.3 20:25
(and 316 from 6.2.2)
autrijus mutters something about velocity and singularity :) 20:26
clkao autrijus: if someone from london is going .at can i ask you to bring something for me? 20:27
autrijus clkao: depends on what that thing is
svnbot6 r3787, autrijus++ | * add Jonathan's URL back in the hope that Win32 build for Parrot 0.2.0
r3787, autrijus++ | will be up soon.
clkao i guess mainly tea and/or liquor. 20:28
autrijus tea I can easily handle
not so sure about liquor
clkao ok..
if so i'll get someone to prepare things and get them packed to you 20:29
autrijus 'kay 20:30
Limbic_Region grats autrijus et all on 6.2.4
autrijus Limbic_Region: thank you! you helped a lot you know :)
(hopefully he'll backlog and see this.)
clkao autrijus: did you see thing about bnechmark.pm i mentioned yesterday?
autrijus clkao: about op dispatch? yea 20:31
so what's stopping you from committing Benchmark.pm?
you still have 10 minutes!
clkao i am lame 20:33
obra would make smoke warn if there were failures?
gaal yes. 20:34
autrijus I take that question as a "all clean". 20:35
gaal: are you testing pre2 again?
obra all clear
me is running make test 20:36
autrijus gaal: I can't duplicate nested_use and splatty_with_type.t
gaal i'm making a new smoke now, wel'll see where it was
autrijus okay. 20:37
gaal it was in one of them - but stderr is confusing when running with concurrency (which is why i'm not using it this time, too :/
Juerd wolverian: If you compile parrot, it'll run.
autrijus changelogging done.
ok, I'm ready to fly.
pending obra and gaal's reports. 20:38
Juerd metaperl: I have no special need to discuss the subject, but if you have questions, just ask
autrijus writes announcement etc
iblech autrijus: If you find it's important enough, you can link to the live cd: linide.sf.net/pugs-livecd-6.2.4.iso
svnbot6 r3788, autrijus++ | * cleanup ChangeLog, hopefully for the final time before 6.2.4.
autrijus iblech: sure!
gaal autrijus, i should have a smoke up in about two minutes. 20:39
autrijus gaal: k. I may need you to remind me the url again
iblech ok :)
I've just re-run all ext/ tests, all clean here
autrijus same here too
woot, we made it. and it's not even 5am this time
(famous last words?) ;) 20:40
gaal :)
Darren_Duncan question about r3777 ... 20:42
why must test names be all lowercase to execute?
obra autrijus: still running
autrijus Darren_Duncan: that's just a convention to disabled various "Disabled/" tests
Darren_Duncan: I'm not sure if it even makes sense.
Darren_Duncan so that convention will go away later? 20:43
autrijus sure, as soon as you or somebody else hack in code to only skip Disabled instead of any uppercase
obra: k
obra so many todo tests 20:44
gaal fwiw make smoke should do that already (default --exclude pattern in yaml_harness.pl)
Darren_Duncan in any event, I will download 6.2.4-pre2 now and test it ... I assume it is akin to a release candidate
autrijus obra: the ratio of todo tests is decreasing though :)
Darren_Duncan: yes, it is :)
obra is there a graph?
autrijus obra: mj has one somewhere
forgot the url 20:45
gaal grrr, t/rules is teh slow
eric256 is suprised he hasn't seen a commit graph yet showing the commits / minute|hour|day|year
autrijus eric256: CIA has one
gaal gotta figure out how to embed parrot next time :)
eric256 FBI too?
autrijus heh 20:46
but since we lost the mirror code in wagner it's not updating
gaal svn log | R something
autrijus rrdtool?
gaal no, R the stat package
autrijus ahh.
gaal try googling for that eh? :)
iblech eric256: There's util/svnlog2graph.pl 20:47
obra ok .....ok
4/4 skipped: skipped tests - rules support appears to be missing
iblech eric256: But it only shows total num of commits as of each day
gaal (meaning "R" is not a very catchy name) 20:48
Darren_Duncan now on 'make' for pre2 - I'm using the simplest setup of system perl 5.8.1 and zero Parrot or readline etc 20:49
eric256 i don't have the prereqs for svnlog2graph :( 20:51
obra blew out on filespec tests 20:52
iblech eric256: wait, I'll upload a img
obra Failed Test Stat Wstat Total Fail Failed List of Failed
-------------------------------------------------------------------------------
t/20_win32_test.t 165 1 0.61% 42
Failed 1/4 test scripts, 75.00% okay. 1/275 subtests failed, 99.64% okay.
iblech obra: autrijus fixed that one
obra it then stopped running more tests.
autrijus yeah.
gaal oh, crap, testgraph.pl is coredumping on me :(
autrijus so that's fine :)
gaal i can upload my tests.yml if it helps.
Aankhen`` G'night. 20:53
iblech eric256: m19s28.vlinux.de/iblech/pugs-svngraph.png
obra rerunning just the module tests 20:54
gaal come to think of it, what's the other computer for.
obra it'd be cool if we had a target for that
eric256 whats the green vs red?
iblech eric256: Red is the # of commits, green is the # of developers, but scaled
autrijus iblech: is that url permanent?
iblech autrijus: No, I'll load it to a perm server, sec 20:55
autrijus i.e. may I link to it in my announcement? :)
ah. I'll mirror to pugscode.org
Juerd iblech: scaled?
gaal forum2.org/gaal/pugs/smoke-cygwin.html # r3787
probelm in system :( 20:56
iblech Juerd: Yeah we only have 70 or so commiters (don't know the exact num), not 700
Juerd iblech: Oh, like that, scaled
gaal and i still don't have the p6 rules for some reason
iblech autrijus: linide.sourceforge.net/pugs-svngraph-6.2.4.png
autrijus danke
iblech np
Juerd iblech: Where does the recent explosion in committers come from?
gaal :q 20:57
oops.
iblech Juerd: Dunno, ask the new committers :) 20:58
gaal okay, does builtins/system.t fail on linux too? i need to lives_ok and todo it on windows.
the third subtest.
obra I may have fucked up tests, but I'm seeing Set/Junction.pm missing when running tests
rerunning make test as I go hiome
autrijus obra: if you run them by hand a make install is needed first 20:59
iblech gaal: 3/3 successful here
autrijus gaal: ok, please simply skip it on win32.
obra autrijus: i'll report back
autrijus instead of lives_ok and todo'ing
gaal k
q 21:01
autrijus Darren_Duncan: how's test going? 21:03
Corion system.t dies on the 3rd test on Win32
gaal Corion, fixing,
Corion cool
gaal r3789 21:04
but i'm still worried about why my pugs doesn't have rule support
oh, right, that cygpath thing. darn. :) 21:05
Corion, tomorrow i'll pick your brains a bit about building a native perl on windows if you don't mind.
svnbot6 r3789, gaal++ | skip crashing test on win32
Corion gaal: It's really easy, but that's why I can do it ;) 21:06
gaal ;)
Corion gaal: I think I even have a batch file that downloads everything you need for MSVC7 and building Perl :)
gaal i'll only use the free ms tools
don't own a full VC 21:07
autrijus system.t fix looks good. 21:08
any other problems from people still smoking?
Darren_Duncan going fine 21:11
mugwump no, I haven't smoked since the weekend 21:12
autrijus Okay, fly!
Juerd I quit smoking, sorry :)
gaal cheers
autrijus :)
Juerd never smoked though
Juerd going home 21:13
anyone with thoughts on rules/policies for the development server, please msg me
gaal remind people to be nice about ghc :) 21:14
Juerd private, please
I'm going home, and won't be reading all of the backlog
mugwump heck ghc eats my 512mb RAM athlon compiling pugs
Juerd afk
autrijus hits "submit" 21:17
svnbot6 r3790, autrijus++ | * This be Pugs, version 6.2.4.
autrijus 6.2.4 released! | pugscode.org <Overview Journal Logs> | pugs.kwiki.org 21:17
gaal yay! 21:17
autrijus yay!
#perl6++
obra autrijus: if I make; make test, should I get 21:18
t/junction...........pugs: *** Can't locate Set/Junction.pm in @INC (@INC contains: /home/jesse/svn/pugs/blib6/lib /usr/lib/perl6/5.8 /usr/share/perl6/5.8 /usr/local/lib/perl6 /usr/local/share/perl6 .). at t/junction.t line 1, column 1
It was my understandng that that wasn't the case that needed make install
autrijus obra: it's known that making test in the ext/*/ won't work
as it's tied to the toplevel maketest.
it sucks -- as clkao pointed out -- but it's the best we've got
obra I did the toplevel make test 21:19
and got that
autrijus hm.
obra me goes looking for /Set::Junction
I have nothing named Junction.pm in my checkout 21:20
iblech Just rerun ext/Set (make test directly in ext/Set), works fine here
All tests successful, 7 subtests skipped.
autrijus no, obra is right.
we fubar'ed ;) 21:21
get ready for 6.2.5 :)
iblech hrm? But all fine here?
autrijus mugwump--
# renaming Set::Junction to Set _without_ fixing the test.
obra did try to warn ;)
Corion It's all fine here too, I think
autrijus obra: I know, thank you :)
Corion weird ...
autrijus it's fine probably because you all have Junction.pm in your perl6/lib/ installed. 21:22
gaal Corion, do you make smoke or make test? (I can only make smoke atm)
autrijus if you remove them, it will fail.
Corion nmake test
... but I may be missing an nmake clean
iblech autrijus: Right... will rm /usr/lib/perl6 and make install again 21:23
Corion autrijus: Ah ... yeah, that could be...
autrijus I fixed the tests for Set.pm 21:26
poll: Does this warrant a 6.2.5? :)
knewt a 6.2.4.1 maybe 21:27
autrijus that's another possibility
gaal _pl1 :)
obra _ is bad from pause persepctive 21:28
and I'd suggest not making it quadversioned
gaal i was just joking, remembering early linux vers
Corion I'm just noticing - GHC throws some warnings at me for my FFI stuff - I should fix them, tomorrow. :) 21:29
svnbot6 r3791, autrijus++ | * remove ::Junction from junction.t.
r3792, autrijus++ | * rename basic.t to hash.t
r3793, autrijus++ | * rename junction.t to basic.t.
iblech autrijus: Ok, svn up, rm -r $PERL6LIB, make, make install, cd ext/Set, make test works
autrijus iblech: right, and all other modules works as well
gaal does make install refuse to install w/o force even when one test fails?
autrijus gaal: yes.
gaal er, doesn't PERL6LIB say ext/Test/lib ? or is the recommendation now blib6/lib ? 21:31
autrijus blib6/lib/ I think, not sure
gaal ...becasue if it is the wiki needs updating
autrijus please do. 21:32
rolling 6.2.5. no preflights this time. :)
Corion I've always used blib6/lib
gaal updated 21:33
Corion I have a "make test" running for whatever is/was latest... I should update to r3793 I guess :)
autrijus Corion: nah, just let it run
it should fail on junction.t
how far along is it?
Corion Still compiling
autrijus ah. I won't wait then :)
Corion make optimized takes long now, but make unoptimized takes far too long to test :) 21:34
gaal then you can svn up underneath it, no? :)
Juerd re
autrijus obra++ # saving us much possible trouble
obra heh
Corion Welcome home, Juerd 21:36
iblech autrijus: link for the live cd is the same, but with -6.2.5 at the end
(uploading)
autrijus iblech: good
rename the svn-graph as well?
:D
iblech ok 21:37
autrijus danke schƶn
iblech kein problem :)
Juerd Corion: Thank you 21:42
svnbot6 r3794, autrijus++ | * This be Pugs, version 6.2.5. 21:43
r3795, autrijus++ | * come to think about it, that change belongs to
r3795, autrijus++ | `Bundled Modules`, not to `Tests, Examples and Documentations`.
gaal does that mean 6.2.6? *ducks*
wolverian 6.2.6? :)
gah, faster than me.
autrijus nah. we'll jump to 6.2.7 ;)
kidding, of couse :)
gaal well, like i said, there's no faster way to find bugs than to release... 21:44
autrijus 6.2.5 released - talk about velocity! | pugscode.org <Overview Journal Logs> | pugs.kwiki.org 21:44
obra heh 21:44
autrijus announcement sent, journal updated :) 21:45
gaal Corion, are you using NTFS on the volume you're building pugs on? 21:46
autrijus thanks to all. I'll go grabbing some food.
Corion gaal: Yeah, I think so
gaal you never had any permission problems did you? maybe it's a cygwin thing. i needed to chmod +x . ; and the temp files aren't getting erased 21:47
Corion temp files aren't getting erased here either. It's a Win32 API problem, because Pugs keeps the filehandles open too long, and Win32 doesn't erase open files 21:48
svnbot6 r3796, Juerd++ | fix layout for quickref/control
r3797, Juerd++ | It's no longer "@foo" but "@foo[]"
r3798, Juerd++ | fix long lines (>80 char)
r3799, Juerd++ | fix layout
Corion Hmmm. Coros have arrived. I should revisit LWP::Simple then... 21:49
gaal hmmm, does it also happen when we explicitly close? or is it an instance of the destruction timeliness problem?
Corion gaal: In some last revamp (1 week ago or 2 ?), explicit close was also broken. 21:50
gaal this is more incentive for me to expose pipeOpen already.
Corion gaal: Oh yes - pipes would come in handy.
gaal we've had them for a month now :(
Corion And BTW, I think I also have an idea how to make system() still run async{} blocks - just run the system() in a separate OS thread 21:51
gaal only there was no agreement about perl6 s29-style spec
Corion We have them? Ah. They are there, but are not reachable in Perl6 ...
Juerd Are there any ideas on making haskell use much less memory?
gaal hey, the compiler d/l finished
Corion Maybe make a module/procedural/oo interface instead of using open "|" ?
Juerd Except the obvious make unoptimized
gaal Corion, they are. use the fully qualified name
see Prim 21:52
Juerd Haskell using 500 MB is a problem on a shared shell server with "only" 1 GB
Corion gaal: Oh - I should look through Prim again ...
Juerd: I think you can set the mem size for Haskell through some switch
gaal oh, maybe the function was removed from the data at the bottom
but it's trivial to add
add back
integral does it use less if you don't use --make, and compile the pugs sources individually?
Juerd I have no idea 21:53
integral Juerd: one idea would be to have a wrapper to invoke ghc that serialised actual invokations of it
Juerd Corion: There is no match for /mem/ in ghc's manpage
Corion Like, a Perl script :)
gaal wasn't there a -M option or someinthg once?
Corion Juerd: The manpage doesn't list all options if I remember right.
integral there's a -H option for the RTS
Juerd What is rts?
integral: Yes, but that'd suck :) 21:54
gaal runtime
svnbot6 r3800, Juerd++ | update README, add references to smartmatch in op
Juerd integral: Idualy, several users should be able to compile at the same time. The problem is that while CPU power can be time sliced, memory cannot with such ease.
Swapping makes it much slower for everyone
And putting 4 GB of memory in the box isn't an option at this point
gaal oh, Corion, the problem with the pipe implementation si tat it doesn't bless the FHs into perl style, so they're slurpy
integral how long does a full build take on that hardware anyway? 21:55
gaal but that can be fixed
Corion gaal: Ah well - slurpy pipes are better than no pipes :)
gaal i'll work on it tomorrow (unless you beat me to it) but i've found the lack of specs on this to be rather exasperating :( 21:56
Juerd integral: I don't know, but it'll surely take more than 10 minutes when optimized
integral yeah, hmm.
gaal anyway, i'm off. night all.
integral hmm, I should run a compile on my 1GB box and see how much it uses 21:58
Corion Anyway, good night all! 21:59
nnunley Juerd: One of the techinques that Darcs uses is to use antimemoization
Donno how much it would help with pugs.
clkao happy fubar! 22:00
iblech As Corion (from *.de, too), I'll sleep now :) 22:02
Night :)
Darren_Duncan okay, 'make test' with pre2 is done
integral hmm, the build is up to Pugs.AST.Internals, and it's only 85M 22:14
ah, now it's shooting up 22:15
mugwump argh, sorry for the Set::Junction thing guys
the hash-based version still had a bug, and for some reason I lost my pugs executable and had to run by the time it was going to finish building 22:16
kungfuftr mugwump: UNACCEPTABLE! 22:17
*blink*
mugwump so I thought I'd just rename it .. didn't think about it being near release time
mugwump accepts a flogging from kungfuftr
kungfuftr mugwump: hhhmmm... you might like that too much... nielo would anyways 22:18
Juerd tnx.nl/3947NPFZ 22:28
draft motd
comments please
integral hmm, ghc 169M, cc1 >200M 22:32
eric256 33 tests unexpectedly succeeded. 22:35
stevan damn, I missed all the fun :) 22:36
stevan went and saw revenge of the sith :) 22:37
integral the simplest way to improve memory usage is just for someone to go and fix Pugs.AST.Internals ;-)
Juerd: from compile I've just done, it seems that gcc's memory usage can be higher than GHC's, but for a very short time. GHC uses <200M, but for ages 22:40
Juerd optimized?
integral yep
but this is a 500MHz box, so maybe the GC is running more because of the extended time? 22:41
eric256 why does nmake test copy all the module files? 22:46
svnbot6 r3801, ninereasons++ | check in a few cookbook changes, especially examples/cookbook/01strings/01-02default-variables.p6, and some other minor changes in the examples directory; and remove the motd directory
autrijus so, I'm writing a new meditation on perlmonks 22:47
pasteling "autrijus" at 220.132.132.105 pasted "draft perlmonks meditation" (14 lines, 1K) at sial.org/pbot/10359 22:48
"autrijus" at 220.132.132.105 pasted "draft perlmonks meditation" (25 lines, 1.1K) at sial.org/pbot/10360
autrijus (with better formatting) 22:49
now I'm beginning to fill in the LIST HERE part. I think I'll think aloud in #perl6 and see what other people think :)
(the timeline link is pugscode.org/perl6-timeline.png )
Juerd whoop, no account for chromatic or abigail should they want any :) 22:50
autrijus Juerd: why? :)
Juerd autrijus: I've been advised to require users to make their full names known to me
For juridical purposes
autrijus lol 22:51
Juerd I normally provide access to paying customers - they can be tracked by bank account if needed
autrijus * the most important thing to me is to mix Perl 6 code with existing Perl 5 applications, sharing each other's modules and objects. an improved Inline::Pugs can do that
Juerd I don't think the box will be used for illegal activity 22:52
mugwump eg, calling DBI and TT from Perl 6
autrijus mugwump: right.
Juerd But just in the very unlikely case it will be, I have to make sure I can't be held responsible.
See also my draft motd
mugwump gets inspired to write PSA::Cache::Entry::Pugs, to let you code your site's logic/business rules in Perl 6 22:53
Juerd Which I wrote instead of the advised 16 page user agreement, which nobody will read anyway.
autrijus * the next most important thing is to make the Perl 6 part run at a competitive speed. improved pugs codegen can do that.
Juerd Funnily, this single screenful of text contains as much information as those 16 dead tree pages.
autrijus: Speed and DBI are my major issues at this point. 22:54
autrijus * close to that is to make the bridging as lightweight as possible, preferably with no object cloning overhead. ponie does just that.
Juerd: my priorities are DBI, then Speed. do you have it reversed? :)
clkao bwahaha 22:55
ijoyce should write a Perl6::Hibernate 22:56
eric256 DBI in perl6 will be great
autrijus eric256: TIMB says he's starting to relook it.
eric256 i can't barely bring myself to read or write perl5 code anymore.....it just feels so lame. lol 22:57
autrijus * next to those, it'd be neat to have a stable-ish language core, with online/offline documentation that won't be invalidated in a few weeks.
eric256 any reason it would skip rules tests when parrot is in the path?
autrijus eric256: not sure. does it skip all rules tests? 22:58
eric256 t/builtins/undef................................ok 2/72 skipped: skipped tests - rules support appears to be missing
i'm still waiting for ti to finish. maybe the undef tests are just using an outdated logic to discover if rules are available
clkao *grin* if major xs users are looking at perl6 directly what do we need ponie for?
autrijus clkao: because it takes ages ;) 22:59
to port from xs, that is
eric256 i should have timed the testing...it is taking a while
nnunley Mmm. Ponie has some doing before it can be ridden. 23:01
autrijus hm, come to think about it, stable-ish language with docs is more important than same-memory-space sharing aka ponie.
I can live with marshalling and semi-rpc.
clkao nnunley! bedtime!
ijoyce no docs scares people away
autrijus verily 23:02
quickly obsoleted docs is better than no docs
but is still scary
ijoyce agrees
autrijus * after all above, I'd like to see Perl 6 modules that offers something neat that cannot be done in Perl 5 without much pain. 23:03
ijoyce such as?
mugwump proper operator overloading! 23:04
eric256 i was thinking about that two autrijus but i didn't come up with anything yet
autrijus I'd like Seaside.
a Seaside-Halo workalike.
eric256 ?
integral would like Text.PrettyPrint in p6, that'd show off operators
clkao mmmmm
autrijus eric256: blogs.inextenso.com/seaside/blog/le...e/jjGALGja 23:05
theorbtw1 wonders if 6.2.5 has something in purticular to do with velocity.
eric256 so did the whole Data::Dumper module get replaced with a simple .perl call?
autrijus yeah
integral &
Jonathan_ autrijus: It appears embedding Parrot into Pugs isn't supported on Win32 with MSVC++ compiler, so I guess I need to ship a built Parrot?
autrijus Halo has this very killer-appish feel
Jonathan_: right
vcv-- How do i try embedding parrot into pugs? since im on windows
autrijus and like integral, I'd like to see a formatting combinator library that can _double_ as a parsing library. 23:06
i.e. generating Form and Rule from a single set of combinators.
it's like a TT2 that also does Parsec for you. 23:07
eric256 just got to the operators/ tests....man this takes forever. ;)
autrijus <- done it in perl5, and it's purely perl 5's limitation that prevents this idea from reaching the full potential
* and after that, it'd be neat to have a perl 5 -> perl 6 translator ;) 23:08
revdiablo I wonder how larry's perl5->perl5 translator is coming 23:09
autrijus hm. what am I missing.
mugwump autrijus: how about an iterator for the Tree module that uses coro's
mauke perl5->perl5 is done
theorbtw1 tries compiling jhc again.
23:09 theorbtw1 is now known as theorbtwo
revdiablo mauke: was there a post about it somewhere that I didn't see? 23:09
mauke it's called "cat"
clkao e/win28 23:10
revdiablo mauke: well, no, he was doing a perl5->AST->perl5
mauke: as a first step to a perl5->perl6 translator
clkao karma vcv
jabbot clkao: vcv has neutral karma
mauke yeah, that's a bit harder
theorbtwo OTOH, the input p5 and the output p5 were supposed to be identical... but only in order to make sure that it was working and saving everything. 23:11
vcv-- karma? :/
Jonathan_ autrijus: How should I best lay this out in what I distribute? Pugs needs to find parrot.exe, so ideally we'd want parrot.exe in the same folder as pugs.exe?
theorbtwo Sadly, only perl5 can parse perl5, so that's the only reasonable way to do it.
mugwump method traverse_iter($self: Str ?$traversal_order) returns Code {
return { $self.traverse({ yield $^node }, $traversal_order); };
}
something like that, anyway. 23:12
theorbtwo Hm, I almost feel like "this sub/method may yeild insted of returning" should be an attrib of the code obj. 23:13
(For similar reasons that you should say try { ... CATCH {} ...}.)
mugwump but you don't know which until it runs!
theorbtwo Which what? 23:14
mugwump whether it returns or yields 23:15
autrijus Jonathan_: sure, that can work. just parrot.exe itself should be enough
Jonathan_: test it to be sure, though
theorbtwo Hm? If a yield statement is in there, it might yeild.
If not, it won't.
(And error at runtime if it tries to use crafty tricks to do so anyway.) 23:16
.oO(Wow, I just said "yield" and "yeild" in the same sentance.)
Jonathan_ autrijus: Have parrot.exe in same folder as pugs.exe, running from that folder and it says it can't find parrot. Hmmm... 23:18
svnbot6 r3802, mugwump++ | Add coro-based iterator to Tree module... untested
autrijus Jonathan_: hm, adding it to PATH won't work? 23:21
Jonathan_ autrijus: Adding it to path does work, just doesn't look in the CWD.
theorbtwo Hm, I feel like 02Internals should give more details about just what C<reduce exp@(Syn name exps)> actually means. 23:23
...in part because the point of that bit seems to partly be to introduce some Haskell, and in part because I don't actually know what it means.
svnbot6 r3803, ninereasons++ | examples/cookbook/01strings/01-02default-variables.p6: smaller correction/adjustment (add a more standard solution)
Jonathan_ autrijus: Even when it does find it in the path, I get this: pugs> if "hello" ~~ /h(\w+)/ { say $1; } 23:25
*** Cannot parse PGE: h(\w+)
*** Error: end of file
undef
autrijus try 23:26
if "a" ~~ /a/ { say $1 }
clkao e/win28
autrijus er
if "a" ~~ /(a)/ { say $0 }
revdiablo wouldn't it also be $0?
autrijus if that doesn't work, cd into src/pge and run
..\..\parrot run_pge.pir 1 1
Jonathan_: I know!
Jonathan_: you want parrot.exe to be placed in src\pge\
Jonathan_ Oh. :)
autrijus the same dir as run_pge.pir
if that still doesn't work, then I'd like to see error msg from run_pge.pir 23:27
eric256 is STILL running tests. geez. lol
theorbtwo Odd... jhc doesn't like dumpGlob = (doParseWith $ \env _ -> do {glob <- liftSTM $ readTVar $ envGlobal env ... -- it gives a syntax error on the g in glob. (Unformatted by hand.) 23:29
Jonathan_ autrijus: I guess I need to distribute src\pge directory too then. :)
autrijus Jonathan_: most definitely so :)
I think you want the entire src/.
but src/pge/ alone is minimal req 23:30
Jonathan_ OK, I'm testing this in the pugs tree I just built, so everything is there. I put parrot.exe inside src\pge and it says it can't find the parrot executable. I put it at the same level as pugs.exe and it finds it and gives the end of line error. 23:35
svnbot6 r3804, theorbtwo++ | PA02 -- reflow the bits just a bit wider then 80 chars, and some minor
r3804, theorbtwo++ | English corrections.
Jonathan_ "..\..\parrot run_pge.pir 1 1" gives no errors. 23:36
autrijus but is there any output? 23:37
it should say
PGE_Match 0 1 [] []
Jonathan_ No, no output.
autrijus that's bad.
so maybe we need a parrot tree. gasp
Jonathan_ Wonder if we need parrot runtime folder... 23:38
autrijus possible. but I really need to sleep :-/
catch you in a few hours
Jonathan_ OK, sure. See you tomorrow, maybe I'll fathom this. :S 23:39
Sleep well. :)
autrijus 6.2.5 released! | Your opinions wanted: perlmonks.org/?node_id=459750 | pugscode.org <Overview Journal Logs> | pugs.kwiki.org 23:39
autrijus ( perlmonks.org/?node_id=459750 is the wishlist thing. :-) ) 23:40
clkao autrijus: i want proper nci bridged 23:46
autrijus clkao: reply to that thread :D 23:48
clkao i am no monk
theorbtwo clkao: That's easy to fix. 23:50
clkao ok i am anonymous monk 23:51
theorbtwo perlmonks.org/?node=create+a+new+user 23:52