Camelia, the Perl 6 bug

IRC log for #parrot, 2008-04-25

Parrot | source cross referenced

| Channels | #parrot index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:26 wolverian joined #parrot
00:36 DarkWolf84 joined #parrot
00:37 Theory joined #parrot
01:17 cotto_work joined #parrot
01:37 DarkWolf84 joined #parrot
01:51 Theory joined #parrot
02:28 Ademan joined #parrot
02:37 guru joined #parrot
02:42 tetragon joined #parrot
03:32 AndyA joined #parrot
03:51 contingencyplan joined #parrot
04:19 Psyche^ joined #parrot
04:54 Infinoid japhb++ # very elegant code
04:54 Infinoid but what's the darwin case for?  I haven't read kid51's patch in detail, but your addition seems to suggest some additional changes elsewhere
04:56 iblechbot joined #parrot
04:58 japhb Infinoid: thank you!
05:00 japhb Infinoid: auto::opengl needs to handle darwin separately from other *nix variants because of the '-framework' flags that are darwin-specific.  Most other library detection steps seem to be able to treat darwin as a generic *nix (or so I gather)
05:01 Theory joined #parrot
05:02 Infinoid it never seemed very clear to me whether the -framework flags were actually necessary
05:02 Infinoid it sounded like it worked without any linker flags at all, and decided at runtime which library to use
05:03 Infinoid ~the -framework option was just a way of forcing the issue
05:03 Infinoid -~
05:03 zarchne joined #parrot
05:03 japhb Last I heard, they weren't ... but I never saw a patch or commit from an OS X person confirming that.
05:03 Infinoid tetragon would know, I wouldn't.
05:04 japhb But even so, we still have to treat darwin as an exception, because I remember saying that the normal *nix -l flags *wouldn't* work.
05:04 Infinoid anyway, if what we have now works, I don't see any real reason not to stick with it
05:04 japhb so darwin would just get nothing added, I suppose.
05:05 japhb I don't treat it as high priority, but I'd like to confirm if the -framework flags can be dropped across all OS X & XCode releases, for two reasons:
05:06 japhb 1. because I don't like having flags that are known to serve no purpose; it obfuscates the truth about what is happening
05:07 japhb 2. The next time we have a library that gets treated as a framework in OS X, we will want to do the right thing (and save ourselves some effort trying to find the right flags)
05:07 Infinoid I wonder if your modification means the config plugins which don't care about darwin have to copy their "default" stuff to "darwin" now
05:07 japhb Nope, that's why I had the || in the my $libs line
05:08 Infinoid but I'm too lazy to read the patch and find out. :)
05:08 Infinoid great!
05:08 Infinoid goodnight
05:08 japhb You never have to specify anything other than default, actually.  :-)
05:08 japhb you too!
05:08 purl ohhh oh oh, the sweetest thing...
06:06 Zaba_ joined #parrot
06:15 uniejo joined #parrot
06:50 Zaba joined #parrot
08:36 IllvilJa joined #parrot
09:57 dakkar joined #parrot
10:02 contingencyplan joined #parrot
10:23 kid51 joined #parrot
10:32 wknight8111 joined #parrot
10:58 DarkWolf84 joined #parrot
11:37 rotty` joined #parrot
11:50 rdice joined #parrot
11:54 tetragon joined #parrot
11:55 skids joined #parrot
12:03 ruoso joined #parrot
12:25 Zaba joined #parrot
12:40 ambs joined #parrot
12:59 Zaba my %hash = (test => "moep"); <-- this is not valid perl6, right?
12:59 * ambs hopes it is
13:00 ambs but I have *no* idea
13:00 ambs :D
13:00 Zaba it causes rakudo to segfault
13:00 Zaba or well parrot, more like
13:00 jonathan Zaba: It's valid Perl 6, but use the colonpair notation for paris at the moment.
13:00 jonathan Segfault is...pretty extreme.
13:00 Zaba jonathan, hm?
13:00 Zaba lemme bt full it
13:00 jonathan my %hash = (:test("moep"));
13:01 Zaba oh.
13:01 ambs yuck
13:01 Zaba http://rafb.net/p/htOdEt53.html
13:01 ambs purl, seen coke
13:01 Zaba here's bt full
13:01 purl coke was last seen on #parrot 17 hours and 6 minutes ago, saying: pmichaud: new ticket with a see also is probably best, since you're asking for something else. =-)
13:01 ambs purl, seen merlyn
13:01 purl merlyn was last seen on #moose 35 days and 15 hours ago, saying: ... http://methodsandmessages.vox.com/lib​rary/post/the-year-of-smalltalk.html  [Mar 20 15:09:20 2008]
13:02 Zaba code: my %hash = (test => 'moep');
13:02 Zaba jonathan, ^
13:04 pmichaud I don't think rakudo understands list assignment yet
13:04 pmichaud and I'm not sure that hash knows how to be assigned a list
13:04 pmichaud or, in this case, a Pair
13:04 cognominal not for hashes
13:05 pmichaud I still need to apply the patch that adds the hash composer, but the code needs to be cleaned up a bit
13:05 pmichaud (it doesn't follow any recognizable coding standard)
13:05 cognominal what I need to do to comply?
13:05 Zaba so the code it segfaults in will be changed anyway?
13:05 pmichaud consistent indentation, primarily
13:06 cognominal oh
13:06 cognominal Zaba : problemes is the list operator when using with 0 and 1 operands
13:06 cognominal I have filed a bug
13:06 Zaba okay.
13:07 pmichaud s/list operator/list context
13:07 pmichaud in    my %hash = ( foo => 'bar' );    there is no "list operator"  :-)
13:07 cognominal I meant the coma operator
13:07 jonathan pmichaud: fatarrow doesn't work because it doesn't parse to item assignment...
13:07 jonathan So the RHS of the => swallows up everything, and if there's commas, it makes a list.
13:07 Zaba I suppose that's wrong?
13:08 pmichaud jonathan: okay, makes sense
13:08 pmichaud is it correct in STD.pm ?
13:08 jonathan It's not an issue, in theory, if you're just having one element.
13:08 jonathan pmichaud: STD.pm is correct, but we can't do what it does yet, I don't think.
13:08 jonathan I left a comment in the grammar.
13:08 jonathan Saying what it should be.
13:08 Zaba why can't STD.pm be used?..
13:09 pmichaud we can do it with a custom parse routine
13:09 pmichaud Zaba: there isn't anything that can execute STD.pm yet
13:09 Zaba oh... right
13:09 pmichaud STD.pm uses constructs that we haven't implemented (and aren't likely to implement until the summer sometime)
13:09 askie joined #parrot
13:10 Zaba pmichaud, like which?
13:10 pmichaud proto regexes, primarily
13:10 Zaba I see
13:11 gryphon joined #parrot
13:12 Zaba what is used for character classes in perl6 regexes?
13:12 pmichaud enumerated character classes?    <[aeiou]>
13:12 cognominal pmichaud, do you know how to fix the namespace for  named rules in grammar?
13:12 Zaba oh, makes sense
13:12 pmichaud cognominal: I don't understand the question
13:13 cognominal grammar A {   token a { }  } ;  # the pir code generated has an empty namespace insead of A
13:14 pmichaud have to pass a 'grammar' option to the PGE rule that compiles the token
13:15 cognominal you mean that the change does not involve PGE but the way it is called?
13:15 pmichaud correct.
13:15 purl no, it's not!
13:15 pmichaud so I should rephrase it as  "have to pass a 'grammar' option to the PGE::Perl6Regex compiler"
13:16 pmichaud I think I toyed with the idea of changing the option name to 'namespace' instead, for consistency with other components
13:16 nopaste "cognominal" at 82.67.232.89 pasted "hash composer" (96 lines) at http://nopaste.snit.ch/12813
13:17 cognominal is that better?
13:17 pmichaud a little.  still doesn't completely line up (at the bottom)
13:18 cognominal arf
13:18 pmichaud need quotes around 'Iterator'
13:18 cognominal is there a tool to check it complies?
13:18 pmichaud there might be a pir coding stds test
13:18 pmichaud might look in t/codingstd/
13:20 pmichaud in many cases I think it would be better to use the Perl 6 functions rather than the Parrot opcodes, though
13:21 pmichaud thus,   'die'(errstr)   instead of   die errstr
13:21 pmichaud and   the equivalent of    Hash.new()  instead of   $P0 = new 'Hash'
13:22 cognominal indeed.
13:22 particle there's no pir codingstd test yet
13:22 ambs All tests successful
13:22 purl Guess your work here is done, ambs. Your notice of intestination will arrive shortly.
13:23 pmichaud particle++ # thanks
13:23 particle i really wanna write a hlltidy based on parse trees generated by pct, and written in nqp
13:23 particle but nqp might not be good enough
13:23 jonathan particle: Write it in Rakudo. ;-)
13:23 Zaba what is hll?
13:23 purl i heard hll was High Level Language
13:24 cognominal how far rakudo is from compiling itself?
13:24 particle jonathan: could do. suppose it'd be a good test
13:24 particle *very*
13:24 particle far, that is.
13:24 pmichaud depends on what one means by "compiling itself"
13:24 Zaba umm, rakudo isn't written in perl6, is it?
13:24 particle many parts of it are
13:24 jonathan Well, not really...rakudo can already do everything NQP can, almost?
13:24 pmichaud the grammar and actions for rakudo are written in Perl 6
13:25 pmichaud there there are some support functions and runtime builtins that are written in PIR
13:25 pmichaud Rakudo doesn't have an equivalent of NQP's   PIR q<...>;
13:26 cognominal here-document is missing too.
13:26 Zaba pmichaud, is it possible to write all builtins in perl6?
13:26 pmichaud Zaba:  my theory is no
13:26 pmichaud Zaba:  if you can tell me how to write infix:+ in Perl 6, then maybe.
13:26 rdice joined #parrot
13:27 Zaba pmichaud, well, maybe some are?
13:27 pmichaud sure, I think that many are.  That's a definite goal for us.
13:27 pmichaud But lots of them aren't.
13:27 Zaba for example, say can be implemented through print..
13:27 pmichaud Even something like 'substr' seems like it would be difficult to write in Perl 6
13:28 Zaba pmichaud, even with regexes?
13:28 pmichaud zaba:  well, here's a canonical example
13:28 pmichaud just a sec, nopasting
13:29 gryphon joined #parrot
13:29 nopaste "pmichaud" at 76.183.97.54 pasted "code for 'substr' in PIR" (16 lines) at http://nopaste.snit.ch/12815
13:30 pmichaud so, if there's a Perl 6 version that can be written more understandably and efficiently than that, we could use it.
13:30 pmichaud but I'm doubtful.
13:31 Zaba hmm
13:31 pmichaud and even the "new 'Perl6Str'"  portion will go away soon
13:31 pmichaud actually, it should go away now and be replaced with prefix:<~>
13:32 pmichaud In any (efficient) language implementation, I think there are always some primitives that are written in some language other than itself
13:34 Zaba maybe..
13:34 cognominal my modified hash composer works with item, array, hashes but still has a problem with pairs
13:35 pmichaud cognominal: fat arrow pairs, or all pairs ?
13:36 cognominal all pairs
13:36 cognominal > say :a.perl
13:36 cognominal Segmentation fault
13:36 purl (Core dumped)
13:36 cognominal ouch
13:36 pmichaud :a.perl  doesn't use the hash composer function, does it?
13:37 pmichaud Zaba: anyway, things like substr and infix:+ are why I haven't been on a strong push to try to get "every builtin written in Perl 6".  For some builtins, that's actually more work (and less efficient) than just writing the PIR.
13:38 Zaba hmm
13:38 cognominal no
13:38 pmichaud that said, it would be very interesting to see someone come up with a pure Perl 6 definition of every builtin.  :-)
13:39 cognominal and it will oblige us to optimize the compiler to avoid deal with pads everywhere
13:41 jonathan cognominal: Having seen the backtrace for the pairs issue, can kinda see what's going on.
13:42 particle src\gc\dod.c(338) : warning C4098: 'pobject_lives' : 'void' function returning a value
13:42 jonathan Might do us well to just write a Pair class ourselve though. And not use the Parrot PMC. It's not really what we want.
13:42 pmichaud Is there an important advantage to using Parrot's built-in Pair class?
13:42 jonathan Not that I can think of.
13:42 pmichaud I was a little surprised that we did that in the first place.
13:42 pmichaud My plan was always to create a Pair class in PIR, if not Perl 6
13:43 particle who broke my gc? aargh. time to go hunting.
13:43 cognominal say :a.value  crashed too
13:43 mj41 joined #parrot
13:43 pmichaud I'm strongly in favor of ripping Parrot's Pair PMC out of Rakudo entirely
13:43 jonathan pmichaud: Do it.
13:43 Zaba brb
13:43 jonathan Or let someone beat you to it.
13:44 pmichaud (for that matter, I'm strongly in favor of ripping Parrot's Pair PMC out of Parrot entirely, but that's a different call)
13:44 jonathan :-)
13:44 * particle agrees with pmichaud
13:44 pmichaud I'm travelling to a conference for the weekend, so I won't get to it until early next week
13:44 pmichaud at least, I doubt I'll get to it until Sunday at the earliest.
13:44 cognominal thx anyway for commenting my code.
13:45 cognominal I need feddback to make progress
13:45 pmichaud I am eager to apply the hash composer patch, because it will help us out in a lot of areas
13:45 cognominal I can't test if for pairs now
13:45 pmichaud I might just apply it and add comments about things I think need to be changed
13:46 jan joined #parrot
13:46 nopaste "cognominal" at 82.67.232.89 pasted "hash composer" (97 lines) at http://nopaste.snit.ch/12817
13:46 cognominal yes please
13:46 pmichaud I think the hash composer probably also wants a   .param pmc pairs :slurpy :named    argument
13:47 * particle fixes a poorly placed #endif and recompiles
13:48 cognominal jonathan: I tried to fix    my @a = [ 1 ].   I expected +$<statement> to give me the length of the statement
13:48 pmichaud so that it can pick up     hash(:a(5), :b, :!c)      (assuming that's a valid way to pass pairs to the hash composer -- need to check synopses)
13:49 pmichaud +$<statement>  will return the numified value of $<statement>'s result object
13:50 cognominal how can I get to $<statements>  length
13:50 pmichaud "length" isn't a valid concept in Perl 6.  Do you mean "number of characters"?
13:50 cognominal I meant it the number of kids matches
13:50 pmichaud ah, number of elements
13:51 pmichaud $<statements>.elems()
13:51 cognominal ho
13:51 pmichaud (I'm not sure that's implemented, but that's the correct way to do it)
13:51 pmichaud could also do
13:51 Andy joined #parrot
13:51 pmichaud +@($<statement>)
13:52 pmichaud (numify the array component of $<statement>)
13:52 pmichaud I know that +@($<statement>)  should work in NQP
13:53 pmichaud have to do $job for a bit
13:53 dalek r27161 | particle++ | trunk:
13:53 dalek : [gc] fix poorly placed '}' preventing compilation
13:53 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=27161
13:54 cognominal my @a = [ 1 ]  and my @a = ( 1)  are supposed to be the same thing?
13:55 particle jonathan: any progress with tapx?
13:55 cognominal tapx?
13:55 ambs tcl failing
13:55 ambs t/cmd_lsort.t    1   256    21    0  ??
13:55 ambs 1 subtest skipped.
13:56 particle ambs: known, iirc. coke?
13:56 particle cognominal: TAP::Parser, Test::Harness 3.x
13:56 ambs particle, good
13:56 particle we've really got to improve our spectest failure reports
13:57 jonathan particle: Not had chance to look yet, sorry.
13:57 Coke technically, it passed all the tests. =-)
13:57 Coke it's the inferior runloop problem.
13:57 Coke (I'm throwing an exception from PIR through C back into PIR)
13:58 particle cognominal: i asked AndyA to give us some love from those modules. in tapx's svn repo (http://svn.hexten.net/tapx/trunk) he added an option allowing us to ignore the exit code from the test files
13:58 particle using that enables us to mark as passing those fudged tests which pass
13:58 particle (as all fudged tests return exit code 1 atm)
13:59 particle however, i wonder if there's a better way to mark which test files have been fudged
13:59 particle perhaps their file extension is enough
14:04 Tene Mmm... new laptop... can't wait to compile parrot on it.
14:08 particle tene: what kind? i got my t61 tuedsay.
14:12 Tene t61p
14:14 pmichaud what's the reason for fudged test returning exit code 1 ?
14:14 particle that's nvidia gfx, right?
14:14 particle pmichaud: to distinguish it from non-fudged tests
14:14 Tene yeah, nvidia gfx
14:14 particle i got internal gfx to save battery
14:14 pmichaud particle:  seems like that should be a command-line option to *fudge* then, not to the harness
14:14 pmichaud i.e., we should be able to tell fudge  "don't report fudged tests as failing"
14:14 particle pmichaud: that's just what i was thinking about
14:15 Tene I considered that, but I like having this many pixels.
14:15 Tene I'm thinking about getting the ultrabay battery, though.
14:15 pmichaud the default invocation can return a failing code -- that makes sense.  But if someone passes an option to fudge saying "pass fudged tests", then I think that's in the realm of "all's fair if you predeclare"
14:15 particle pmichaud: is there a mode into which i can put pge so it will capture whitespace?
14:15 particle pmichaud: i'll code it up
14:16 pmichaud particle:  I think you're asking for :keepall
14:16 particle is that implemented?
14:16 pmichaud (PGE doesn't implement :keepall yet)
14:16 particle didn't think so.
14:16 pmichaud but I think that's what you're asking for.  :-)
14:16 particle ok.
14:16 particle yes, i think so, too.
14:16 particle really, i just want :keep(<ws>)
14:16 pmichaud essentially it says to go ahead and capture non-capturing subrules
14:16 particle or whatever
14:17 pmichaud what about <ws> that might occur inside of some other non-captured subrule?
14:17 particle iunno. probably that too
14:17 particle anyway, i can deal without any of it.
14:18 particle however, i'd like to know what's best to use to manipulate the parse tree... nqp or tge
14:19 pmichaud yes.  The start positions for each parse node are there, though, so you could infer the <ws> from that.
14:19 particle yes, kinda
14:19 particle \n is lost
14:19 pmichaud personally, I'd use nqp, but only because I'm more familiar with it
14:19 pmichaud ...\n is lost?
14:19 particle \r\n
14:19 pmichaud not really
14:20 particle darn it.
14:20 particle i really need to finish that yaml dumper
14:21 pmichaud if you just keep the original source string around (and it's there, in the match nodes themselves), then you can easily use the positions to see the whitespace
14:28 rdice joined #parrot
14:32 apple-gunkies joined #parrot
14:32 turrepurre joined #parrot
14:37 wknight8111 joined #parrot
14:37 Coke particle: (yaml dumper) I was expecting an email or something some time ago with what you thought you had left.
14:38 particle ah, right.
14:39 Coke even if I don't get to it, open a ticket or something with the remaining tasks.
14:40 particle what's the syntax for patch? patch -p0 . <file.patch ??
14:41 pmichaud patch -p0 <file.patch
14:41 pmichaud assuming that the paths in file.patch correspond to the current directory
14:41 particle doesn't seem to be working here. hrmm.
14:41 particle yep
14:41 PerlJam try -p1
14:42 pmichaud if the paths are    a/languages/perl6/...   and b/languages/perl6/...   then use -p1
14:42 pmichaud if the paths are languages/perl6/... in both cases then use -p0
14:42 pmichaud (-p indicates the number of directory prefixes to strip before applying the patch)
14:52 * particle used the visual patch util since cmdline wasn't working
14:55 * jonathan has had some windows patch programs getting horribly confused over line endings before now
15:15 mire joined #parrot
15:15 Theory joined #parrot
15:25 ambs joined #parrot
15:31 ambs purl, squaak?
15:31 purl wish i knew, ambs
15:33 particle squaak is languages/squaak or the language in kj's tutorial at parrotblog.org
15:33 particle squaak?
15:33 purl it has been said that squaak is languages/squaak or the language in kj's tutorial at parrotblog.org
15:34 ambs particle, that didn't help much
15:34 particle what are you looking for?
15:34 ambs what kind of language? Just another strange one? :)
15:34 pmichaud squaak is also http://svn.perl.rog/parrot/trunk/languages/squaak/
15:34 purl okay, pmichaud.
15:34 wknight8111 it's based on Lua, if i recall
15:34 particle i have no idea, i haven't read the tutorial
15:35 pmichaud iwbni squaak had a README file there
15:35 Infinoid squaak?
15:35 purl i think squaak is languages/squaak or the language in kj's tutorial at parrotblog.org or http://svn.perl.rog/parrot/trunk/languages/squaak/
15:35 pmichaud and the README could point to the current tutorial location, whatever it is
15:35 wknight8111 but it's a toy language, used for demonstration only
15:35 ambs wknight8111, ok, that was what I was looking for
15:35 Infinoid no, squaak is languages/squaak or the language in kj's tutorial at parrotblog.org or http://svn.perl.org/parrot/trunk/languages/squaak/
15:35 purl okay, Infinoid.
15:35 pmichaud maybe I'll grab a README from the parrotblog text
15:35 Infinoid (fix typo in perl.rog :))
15:35 pmichaud Infinoid: thanks
15:35 pmichaud my fingers aren't in the irght place thsi morning
15:36 Infinoid strangely, I type better when I haven't really woken up yet
15:36 ambs :D
15:37 ambs compiling parrot makes my mbp hot
15:38 * particle hopes nobody added the hcf op to parrot's mac compile
15:38 particle hcf?
15:38 purl hcf is halt and catch fire or (: foldoc hcf) or http://en.wikipedia.org/wiki/Halt_and_Catch_Fire
15:38 ambs :D
15:39 particle hcf is also an experimental parrot opcode
15:39 purl okay, particle.
15:40 ambs mbp is landing
15:51 ambs Usual test failing on mac
15:51 ambs t/examples/library.t                       (Wstat: 256 Tests: 4 Failed: 1)
15:51 ambs Failed test:  3
15:52 ambs pmichaud++ # readme for squaak
15:55 dalek r27162 | coke++ | trunk:
15:55 dalek : [codingstd]
15:55 dalek : update to new ops syntax to avoid generating code that will never
15:55 dalek : be executed.
15:55 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=27162
16:17 iblechbot joined #parrot
16:33 mire joined #parrot
16:34 particl1 joined #parrot
16:55 ambs l8r
17:02 * Tene svn co parrot
17:03 pmichaud (squaak readme)   note that all I did was write the message to the list -- I didn't add it to the repo yet.  Hopefully someone else can do that, since I'm kinda pre-occupied today.
17:04 Debolaz Tene: You can also git clo it now. :-)
17:04 Debolaz (Read-only of course)
17:04 Tene Debolaz: I usually use git-svn, actually.
17:04 Tene real0m55.664s
17:04 Tene user1m23.690s
17:04 Tene sys0m12.970s
17:04 Tene for make -j 3
17:05 * particl1 needs to partition his hdd and install hearty
17:06 Zaba hearty?
17:06 Tene new ubuntu release
17:06 particle yep, came out yesterday
17:06 particle hearty heron
17:07 * Debolaz needs a perl6 based project to play with.. A small web framework perhaps.
17:07 Debolaz Does rakudo support sockets?
17:07 Tene Not yet, I think
17:07 particle no, just $*IN, $*OUT, $*ERR
17:07 Debolaz :~(
17:07 particle .open, .readline
17:08 Debolaz Yeah, readline was the result of my nagging. :-)
17:08 Zaba is it standard?
17:08 particle keep nagging!
17:08 pmichaud "hardy"
17:08 pmichaud installed the new release last night :-)
17:09 * Zaba doesn't use ubuntu
17:09 pmichaud kubuntu, here
17:09 * Zaba doesn't use .*ubuntu
17:09 Zaba :>
17:10 Tene rakudo is real 0m36s
17:15 particle er, right. hardy.
17:15 davidfetter joined #parrot
17:15 particle tene: you get 2.6GHz?
17:20 Tene particle: 2.5
17:21 particle i'm faster :P
17:21 Coke Our compiler warnings should probably follow the same path as our codingstd warnings: only enable those we expect to pass, and have everything else be checked if you verbosely ask for it (in this case with --cage at the config step) - then once something passes we can move it from --cage to the default. make sense?
17:21 Tene particle: how much ram?
17:21 particle 4GB
17:21 Tene Same.  HD?
17:21 pmichaud coke:  makes sense to me
17:21 particle 200GB 7200kps
17:22 particle coke: we're already doing that, no?
17:22 particle msvc compiles with warnings at the highest level below 'all'
17:23 Coke particle: ISTR I'm getting build warnings, and cannot remember the last time I saw something move from --cage to teh default.
17:23 Coke (which may be cause none were fixed)
17:23 particle i haven't seen any new warnings flags added lately.
17:23 particle likely your warnings are due to new code
17:24 Coke define "new"
17:24 Coke since the last time someone touched the warnings flags?
17:24 Coke yah, default build on feather has warnings.
17:26 Zaba feather?
17:26 purl feather is probably http://feather.perl6.nl/ or a perl6 community development server
17:26 Zaba oh.
17:26 nopaste "coke" at 72.228.52.192 pasted "warnings from build on feather." (28 lines) at http://nopaste.snit.ch/12818
17:28 particle yes, coke. chromatic has been playing with jit code in the past week
17:28 particle ...or two.
17:29 Coke those are not warnings from the last week. =-)
17:29 * particle has had build warnings for 5 years
17:29 Coke ... nevermind.
17:29 particle those are good warnings, from what i can see
17:29 particle the code should be fixed
17:32 pmichaud particle: I agree, but coke is saying that short of fixing them (which seems to be taking a long time), they should appear only in response to a --cage flag
17:32 pmichaud otherwise we're in the habit of not looking at warnings at all, and never see when new warnings appear
17:32 pmichaud (because the new ones get drowned out by the existing ones)
17:33 particle i guess i missed that part.
17:34 pmichaud it took me a second read to completely understand it as well
17:34 particle compile a list of known files with known warnings, and disable them in the regular build?
17:34 particle we do have a CFLAGS file
17:34 particle similar to that
17:34 pmichaud particle: something like that, yes.  But keep the regular build as free of warnings as we can, and use --cage to do a more rigorous test for warnings
17:35 pmichaud much like we have some tests in codingstds that aren't run by default
17:35 pmichaud when specific warning conditions are cleared, they can be moved from --cage to the default
17:35 particle iirc there is a --cage build or configure flag
17:35 particle maybe i'm wrong
17:37 particle i think i'm getting confused with the splint work andy was doing with ptc
17:41 cognominal pct?
17:41 purl it has been said that pct is a good solution, but heavyweight or the Parrot Compiler Toolkit
17:41 cognominal ptc?
17:41 purl ptc is paul t cochrane or mailto:paultcochrane@gmail.com
17:43 particle svn st
17:43 Infinoid splint is a pain
17:43 particle grr
17:43 Coke there. opened two tickets which cover the majority of the extant warnings I see.
17:44 dalek r27163 | particle++ | trunk:
17:44 dalek : [rakudo] ws and formatting fixups in actions.pm
17:44 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=27163
17:45 particle rakudo's action file has way too many subs over 50 lines
17:45 particle many are ~150
17:46 contingencyplan joined #parrot
17:48 Infinoid is that a codingstd issue?
17:49 particle we don't have coding standards for perl 6 code yet :)
17:49 particle it's certainly a maintenance issue
17:49 Infinoid yes... I was just wondering if we had standards yet :)
17:49 particle mccabe complexity is way too high
17:50 Coke I don't think file size is necessarily an indicator of complexity.
17:50 Coke particle: are you referring to "cyclomatic complexity" ?
17:51 particle yep
17:52 particle coke: subroutine size is an indicator. it's strongly correllated
17:52 particle but, these are complex subs, with many nested if/else
17:52 Coke but that's not a function of *file size*.
17:52 Coke that's a function of subroutine size.
17:52 Coke neh?
17:53 particle what did i say about file size?
17:53 Coke ... nothing, apparently. whoops.
17:53 particle :)
17:56 * Coke opens a final ticket for the remaining current warnings.
17:56 particle coke++
18:27 sjansen joined #parrot
18:36 b2gills joined #parrot
18:37 Ivatar joined #parrot
18:45 Debolaz Gah, not being able to write Foo.bar ("argument"); is weird.
18:45 particle try Foo.bar\ ("argument");
18:46 Debolaz Still weird. :-)
18:46 ambs joined #parrot
18:47 Debolaz But since I don't have to enclose if, while, etc, in () anymore, I guess I can get used to hugging the ()'s on method calls.
18:47 * particle thinks putting spaces between the method and the params is weird
18:48 Debolaz I used to do it, since I wanted if () and function () to look the same. Ie, my brain parses if as a type of function even though I am aware it isn't.
18:48 teknomunk joined #parrot
18:59 particl1 joined #parrot
19:02 ambs particl*: added the signature you requested :)
19:02 particl1 thanks!
19:05 TimToady cultural sensitivity is the new cultural imperialism
19:05 TimToady perl 6's whitespace rules are a form of cultural imperialism that enforces cultural sensitivity  :)
19:05 particle long live the king!
19:05 pmichaud TimToady: btw, if you haven't received it yet, my package should be arriving today or tomorrow
19:06 TimToady 'kay
19:06 pmichaud inside is a postage return label -- I think you will be able to reuse the box
19:06 TimToady probably can't return it till Tue at the earliest, depending on how my tests Monday go
19:06 pmichaud that's fine
19:10 Debolaz Is there a way to get coercion for arguments in perl6 similar to how the Moose type system does it?
19:11 Debolaz And I suppose even more important: Is it implemented if so?
19:21 TimToady Debolaz: given that Moose is based on the p6 OO design, I'd like to know where it diverges...
19:22 TimToady and whether those divergences are just making up for having to run on p5...
19:24 Debolaz Well, what I'm specifically looking for is this: http://search.cpan.org/~stevan/Moose​-0.40/lib/Moose/Cookbook/Recipe5.pod
19:24 shorten Debolaz's url is at http://xrl.us/bjtqd
19:24 Debolaz I'm looking through the p6 synopsis, but can't really find anything similar..
19:25 Debolaz So I was hoping for some pointers in the right direction.
19:25 jan joined #parrot
19:26 particle port Params::Coerce to perl 6?
19:26 particle ;)
19:27 Debolaz particle: Ie, I'm not crazy or just reading the wrong documents, there really is nothing directly equivalent for p6? It's not a disaster, I'd just like to know that I'm not missing something obvious.. :)
19:30 barney joined #parrot
19:33 * particle *almost* has fudge* and rakudo test code reporting errors how we want them
19:38 Tene particle++
19:38 PerlJam particle++
19:41 TimToady you want them to report errors as "passed" don't you?  :)
19:44 PerlJam Debolaz: are you saying that you think perl 6 doesn't do type coercion?   I seem to recall that it does if you specify the appropriate transformation routines (otherwise how would you coerce a Dog into a String?)
19:44 TimToady Debolaz: looks to me like it's mostly just trying to make up for a lack of MMD in p5
19:45 Debolaz Multi method dispatching is something I hate with a passion.
19:45 TimToady are there also any rational reasons?
19:46 Debolaz Well, I just feel it's very unelegant. It works, but it has always just screamed "workaround for lack of elegant solutions" to me.
19:46 TimToady I think that, by definition, the least inelegant solution is elegant.  :)
19:47 Zaba_ joined #parrot
19:47 Debolaz class Foo; coerce from Bar with $bar -> { Foo.new($bar) }; method something (Foo $a is coercible) {}; # Something like this would be more elegant imho.
19:49 Debolaz The method is no longer required to deeply understand how Foo is constructed, and there doesn't need to be 10+ multimethods for every real method that wants to use coercible parameters.
19:50 Debolaz PerlJam: I'm not saying I think that, I'm asking.
19:50 TimToady this can probably be solved with "does Coercible" or some such
19:50 ambs joined #parrot
19:51 TimToady in any case, the fact that there is a set of types that should be coercible probably means you haven't factored the roles right yet.
19:52 TimToady mmd with roles is a much different beast than mmd with classes
19:53 dalek r27164 | particle++ | trunk:
19:53 dalek : [rakudo] generalize command-line argument processing in the test harness
19:53 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=27164
19:54 Debolaz TimToady: Are there any specific examples I can look at? Keep in mind, my background is primarily with Moose, p6 is relatively new to me. :)
19:56 Debolaz TimToady: Because this doesn't really make sense in my head, it reads to me that you recommend requiring a specific interface (As defined by the role) instead of a specific class, but what I want is to handle arguments which does not support the required interface at all.
19:56 PerlJam Debolaz: if you keep watching Jonathan Worthington, he's adding oop stuff to rakudo  :)
19:56 Debolaz TimToady: I may be reading you very wrong here, which is why I'm looking for examples.
19:57 dalek r27165 | particle++ | trunk:
19:57 dalek : [rakudo] "improve" test failure reporting
19:57 dalek : ~ add new --keep-exit-code option to fudge in the spectest target
19:57 dalek : ~ this makes passing fudged tests report that they pass, instead of fail
19:57 dalek : ~ also added note that eventually this option must be ripped out
19:57 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=27165
19:58 particle ok, as of r27165, passing fudged test files are reported as passing in rakudo.
19:58 Coke even if they fail? awesome! <duck>
19:58 * Coke has trouble reading particle's sends today, and keeps up the trend.
19:58 particle :)
19:58 PerlJam Debolaz: Could you give an example?  Because it keeps sounding to me like you're expecting a Dog to transform into a Cat without either having knowledge of the other.
19:58 Coke why are you frowning!
19:59 particle :P
20:00 Debolaz PerlJam: In the example I used, The class Dog knows how to construct a Dog object from a Cat object.
20:02 * particle prefers a more real-world example
20:02 particle like, say, coercing a Date object from a String
20:03 Debolaz particle: Or a HTTP::Request object from an url string.
20:03 particle there ya go.
20:04 Debolaz However, that makes it more difficult to understand why mmd seems like such a horrible solution to me, so I tried to make the example less tied to one particular situation.
20:05 PerlJam Debolaz: yet single dispatch is fine?
20:05 Debolaz PerlJam: ...?
20:05 purl Yada yada yada hasn't been implemented yet!
20:05 dalek r27166 | bernhard++ | trunk:
20:05 dalek : Remove lib/Digest/Perl/MD5.pm.
20:05 dalek : There is Digest::MD5 in core of Perl 5.
20:05 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=27166
20:06 Debolaz PerlJam: It's not at all implied what the problem is in that statement...
20:35 barney Is there a ~~ op in Perl 5 ?   Encountered in RT#53362
20:36 Tene 5.10
20:36 barney Ah, smart match
20:36 ambs :)
20:41 Coke barney: do we need lib/Pod/Simple.pod ?
20:41 Coke or lib/Pod/Simple/*.pod ?
20:41 Coke (... or Pod::Simple... yah, yah, I know it's not core in 5.8)
20:43 dalek r27167 | bernhard++ | trunk:
20:43 dalek : #53362: [NEW] Add a test for the sysinfo op.
20:43 dalek : Courtesy of Brad Gilbert
20:43 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=27167
20:47 barney It's used in a couple of places, and in Perl 5 core only since Perl 5.9.3
20:47 Coke right. my initial quesiton was regarding the .pod files we appear to have included when we sucked it in.
20:48 barney Ah. Let's get rid of it.
20:55 spinclad (re squaak): huh.  here i was thinking it was a toy impl of squeak.  (which would be cool too)
20:58 cotto_work that seems to be a common misconception
21:02 cognominal > sub hi { say "hi" }; hi
21:02 cognominal ResizablePMCArray: Can't shift from an empty array!
21:02 cognominal > sub hi { say "hi" }; hi()
21:02 cognominal invoke() not implemented in class 'Perl6Str'
21:02 cognominal > sub hi { say( "hi") }; hi()
21:02 cognominal hi
21:02 purl hi, cognominal.
21:02 cognominal comments?
21:02 purl Butt out.  I'll repeat that sentiment stronger if you haven't figured out the message yet.
21:03 cognominal dealing with bots is even more hopeless tan dealing with code
21:04 dalek r27168 | bernhard++ | trunk:
21:04 dalek : Get rid of unneeded *.pod files in lib/Pod.
21:04 dalek : coke++ for suggesting this.
21:04 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=27168
21:12 Coke thanks, barney.
21:14 ambs Coke, did you see my email?
21:14 ambs :)
21:17 Coke yes. was trying to determine a reply. Only folks I know in .eu are smash and leo
21:17 Coke both of whom haven't done much recent hacking.
21:17 Coke I'll go if you fly me out. ^_^
21:17 Infinoid does .uk count as .eu?
21:17 dalek r27169 | kjs++ | trunk:
21:17 dalek : [squaak] add readme file. fix my name in maintainer file. update manifest.
21:18 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=27169
21:19 dalek r27170 | kjs++ | trunk:
21:19 dalek : [squaak] add svn:mime-type property to file.
21:19 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=27170
21:20 ambs Coke, hehehe. Smash will be there, of course
21:20 ambs Infinoid, doesn't it?
21:22 Coke ambs: he doesn't know what your original question was.
21:22 ambs Coke, I know :D
21:22 Coke *thwap*
21:23 Coke in any case, your best bet is probably to just ask on list.
21:23 ambs yes, probably
21:23 purl Really? Probably? Are you Certain it's not certain? Are you sure it's unsure? I think you need to look harder.
21:23 * ambs slaps purl
21:23 purl o/` Hit me baby, one more time o/`
21:23 Coke coke is also just a figurehead.
21:23 purl okay, Coke.
21:23 Infinoid ambs: if so, I'm pretty sure jonathan counts too, I think he's in .uk :)
21:24 Infinoid (I could be wrong about that.)
21:25 particle jonathan is in spain, moving to slovakia
21:25 ambs spain? nice :D
21:25 ambs I can always ask him to get to Portugal before :)
21:26 Coke jonathan has some commitment problems. =-)
21:26 * ambs coughs
21:26 Coke about his place of residence. :P
21:27 ambs :D
21:27 dalek r27171 | kjs++ | trunk:
21:28 dalek : [squaak] add a note to readme file about squaak not being Squeak.
21:28 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=27171
21:29 Coke seen merlyn
21:29 purl merlyn was last seen on #moose 35 days and 23 hours ago, saying: ... http://methodsandmessages.vox.com/lib​rary/post/the-year-of-smalltalk.html  [Mar 20 15:09:20 2008]
21:30 Coke yapc?
21:30 Coke yapc::na?
21:30 purl it has been said that yapc::na is June 16-18 at IIT in Chicago (again! bring your blankets!) or
21:30 Coke yapc::na is also Coke
21:30 purl okay, Coke.
21:30 particle going to yapc::na?
21:30 purl going to yapc::na is sungo
21:30 particle going to yapc::na is also particle
21:30 purl okay, particle.
21:31 Coke going to yapc::na is also Coke
21:31 Andy joined #parrot
21:31 purl okay, Coke.
21:31 Coke no, yapc::na is June 16-18 at IIT in Chicago  (again! bring your blankets!)
21:31 purl okay, Coke.
21:32 ambs yapc::eu?
21:32 purl somebody said yapc::eu was see YAPC::Europe
21:32 ambs yapc::Europe?
21:32 purl yapc::Europe is Copenhagen, Denmark - 13-15AUG08 and http://www.yapceurope2008.org/ye2008/
21:32 ambs purl, ptpw?
21:32 purl wish i knew, ambs
21:32 ambs purl, PtPW is the Portuguese Perl Workshop (6-7 June, 2008)
21:32 purl OK, ambs.
21:33 cotto_work going to yapc::na is also cotto
21:33 ambs damn Pittsburg for stealing PPW.
21:33 purl okay, cotto_work.
21:33 ambs going to yapc::europe?
21:33 purl going to yapc::europe is (: going to YAPC::EU 2007)
21:33 ambs goint to yapc::eu 2008
21:33 ambs going to yapc::eu 2008?
21:33 ambs purl, going to yapc::eu 2008?
21:33 purl ambs: bugger all, i dunno
21:33 ambs d'uh
21:34 ambs purl, going to yapc::eu 2008 is ambs, smash
21:34 purl OK, ambs.
21:38 Coke no going to yapc::europe is See: going to yapc::eu 2008
21:39 Coke no, going to yapc::europe is See: going to yapc::eu 2008
21:39 purl okay, Coke.
21:39 Coke whee.
21:39 ambs Coke, thanks
21:42 skids joined #parrot
21:54 ambs and, night, folks
21:55 tewk joined #parrot
22:14 grim_fandango joined #parrot
22:16 AndyA_ joined #parrot
22:56 tetragon joined #parrot
23:10 japhb joined #parrot

| Channels | #parrot index | Today | | Search | Google Search | Plain-Text | summary

Parrot | source cross referenced