Camelia, the Perl 6 bug

IRC log for #parrot, 2008-07-24

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:01 cotto_work some delicious nopasta would be ok too
00:02 Ademan joined #parrot
00:03 davidfetter joined #parrot
00:04 cotto_work perl6: class a is b {}
00:04 polyglotbot OUTPUT[Null PMC access in type()␤current instr.: '_block12' pc 49 (EVAL_13:29)␤called from Sub 'parrot;PCT::HLLCompiler;evalpmc' pc 744 (src/PCT/HLLCompiler.pir:448)␤called from Sub 'parrot;PCT::HLLCompiler;compile' pc 438 (src/PCT/HLLCompiler.pir:303)␤called from Sub 'parrot;PCT::HLLCompiler;eval'
00:04 polyglotbot ..pc 776 (src/PCT/HLLCompiler.pir:473)␤called fro...
00:04 cotto_work perl6: eval 'class a is b {}'
00:05 polyglotbot OUTPUT[[oops; continuation 0x834996c of type 21 is trying to jump from runloop 548 to runloop 110]␤src/inter_call.c:301: failed assertion 'PObj_is_PMC_TEST(sig_pmc)'␤Backtrace - Obtained 19 stack frames (max trace depth is 32).␤  (unknown)␤    Parrot_confess␤      Parrot_init_arg_op␤
00:05 polyglotbot ..Parrot_init_ret_nci␤          (unknown)␤            (un...
00:05 timbunce joined #parrot
00:09 AndyA joined #parrot
00:12 cotto_work DietCoke, nopaste please?
00:16 particl2 joined #parrot
00:17 particl2 opbots, names
00:17 particl2 stupid bots
00:18 cotto_work opbots, names
00:18 particle1 opbots, names
00:18 DietCoke cotto_work: I replied to the ticket. what do you want me to nopaste?
00:18 particle1 op me!
00:19 particle1 trust particle1
00:19 Whiteknight trust particle1
00:19 Whiteknight i guess the bots don't like me either
00:19 particle1 it may be opbots, trust particle1
00:19 cotto_work no.  Thanks for the rt reply.
00:19 Whiteknight opbots, trust particle1
00:19 slavorg Ok
00:19 slavorgn Ok
00:19 particle1 Whiteknight++
00:19 Whiteknight awww...they do like me!
00:19 Topic for #parrotis now #parrot Parrot 0.6.4 "St. Vincent Amazon" Released | http://parrotcode.org/ | 15 new, 648 open, 80 stalled tix | logged in http://irclog.perlgeek.de/parrot/today
00:19 dalek r29711 | Whiteknight++ | gsoc_pdd09:
00:19 dalek : [gsoc_pdd09] update header allocator to flag items differently in different GC states
00:19 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=29711
00:22 dalek r29712 | coke++ | trunk:
00:22 dalek : [tcl] http://code.google.com/p/p​artcl/issues/detail?id=58
00:22 dalek : Eliminate some more __foo style sub names
00:22 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=29712
00:23 cotto_work coke++
00:29 Andy joined #parrot
00:36 s1n left #parrot
00:37 japhb joined #parrot
00:40 dalek r29713 | Whiteknight++ | gsoc_pdd09:
00:40 dalek : [gsoc_pdd09] add additional trace checks in case items are added to the queue post-trace but pre-sweep. doesn't fix any bad tests
00:40 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=29713
00:40 dalek r29714 | coke++ | trunk:
00:40 dalek : [tcl] http://code.google.com/p/p​artcl/issues/detail?id=58
00:40 dalek : Eliminate some more __foo style sub names
00:40 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=29714
00:42 apeiron joined #parrot
00:42 dalek r29715 | Whiteknight++ | gsoc_pdd09:
00:42 dalek : [gsoc_pdd09] codingstd fixes for the garbage i just committed
00:42 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=29715
00:43 ^conner joined #parrot
00:44 bacek joined #parrot
00:46 cognominal joined #parrot
01:55 Andy joined #parrot
02:13 dalek r29716 | coke++ | trunk:
02:13 dalek : [tcl] pass another spec test in dict.test (and elsewhere, no doubt.) The
02:13 dalek : error message for an invalid subcommand is slightly different than that
02:13 dalek : for the options.
02:13 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=29716
02:20 dalek r29717 | coke++ | trunk:
02:20 dalek : [tcl] the force skip is no longer necessary, but many of these need to be
02:20 dalek : TODO'd. Also, tclsh8.5 disagreed with us on the return value of the last test.
02:20 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=29717
02:55 particl1 joined #parrot
02:56 Psyche^ joined #parrot
03:10 Theory joined #parrot
03:15 dalek r29718 | coke++ | trunk:
03:15 dalek : [tcl] reclaim 47 spec tests in dict.test;
03:15 dalek : - the recently reactivated TclDict vtable get_string had a segfault.
03:15 dalek : - avoid it entirely by translating the PIR sub it invoked into C
03:15 dalek : - delete the now-unused PIR sub
03:15 dalek : - Fix the algorithm to use the right container PMC type.
03:15 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=29718
03:18 rhr joined #parrot
03:44 dalek r29719 | coke++ | trunk:
03:44 dalek : [tcl] reclaim another 122 spec tests from dict.test
03:44 dalek : Avoiding the original segfault is nice, but actually having a sane
03:44 dalek : VTABLE_get_string avoids another crash and gets us further through
03:44 dalek : the test file.
03:44 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=29719
03:56 teknomunk__ joined #parrot
04:04 Theory joined #parrot
04:29 pmichaud japhb++ mj41++ # corrections to my slides, now updated on website, thanks!
04:45 Infinoid pmichaud: in "The next few months will see substantial improvements", by "substantial improvements", I think you mean "us taking over the world, ushering in a new era of peace, prosperity, and free waffles and ice cream".
04:45 Infinoid hope this helps, have a nice day.
04:47 davidfetter free waffles *and* ice cream?!? i'm so there
04:58 pmichaud I'll go for free waffles and ice cream.
04:58 * pmichaud wonders how he can stage a free waffle/ice cream event
05:07 masak joined #parrot
05:16 pmichaud afk # sleep
05:16 teknomunk_ joined #parrot
05:19 Psyche^ joined #parrot
05:44 cotto_home (tickets++)--
05:59 TiMBuS joined #parrot
06:10 masak cotto_home: you don't like the way people appreciate tickets?
06:14 uniejo joined #parrot
06:22 cotto_home I don't like how the number keeps increasing
06:23 cotto_home (although it shows that people care)
06:25 verve joined #parrot
06:32 masak cotto_home: given that tickets represent known issues, which would not be non-issues if not for the tickets, but simply unknown issues...
06:35 cotto_home I don't think it's a bad thing to have more tickets; it's just annoying to close one, only to have the number not change or increase.
06:35 masak now, _that_ I can understand
06:36 masak I was going to say "maybe you need more devs?", but that just made me think of Fred Brooks
06:37 masak but maybe more devs are in fact needed
06:38 cotto_home fortunately, there's no formal definition of "late"
06:38 cotto_home in Parrot's case
06:40 masak except having your niche filled for you
06:40 masak hm, that's what you meant by "no formal". right.
06:42 cotto_home I meant there's no hard deadline, but I'll pretend like I was smart enough to anticipate that.
06:43 masak heh :)
06:43 masak re tickets, I usually make an attempt to patch my own bug reports
06:43 masak so far, I have something like a 15% success rate, and often my patches need curating of some sort
06:44 masak but I believe my understanding of rakudo and parrot is improving
06:45 cotto_home me too, but it's disappointingly easy for me to get in over my head.
06:46 masak lately, I've found it surprisingly easy to grok a new aspect of parrot. but I'm still picking low-hanging fruit.
06:47 masak and some things still baffle me
06:48 masak for example, I haven't been able to find comment out the thing that makes this hang:
06:48 masak ./perl6 -e '"$"'
06:58 masak it practically has to be somewhere in src/parser/quote_expression.pir, but I haven't found it
07:02 iblechbot joined #parrot
07:07 bacek_ joined #parrot
07:08 bacek__ joined #parrot
07:35 timbunce joined #parrot
07:36 _timbunce joined #parrot
07:49 spinclad joined #parrot
09:32 donaldh joined #parrot
10:58 ruoso joined #parrot
11:10 Psyche^ joined #parrot
12:11 masak how do I trace things in rakudo?
12:11 masak `./perl6 -t perl6file` doesn't seem to cut it
12:13 moritz I think you can only trace at the parrot level
12:13 masak ./perl6 -h tells me that there is a -t flag
12:14 masak anyhow, I didn't succeed on the parrot level either
12:14 masak I'm trying to debug the "$" error, and would be greatly helped by knowing the exact call chain
12:15 masak something like this:
12:15 masak ../../parrot --runcore trace perl6.pbc -e '"$"'
12:15 masak ...except the above hangs and doesn't trace
12:17 moritz ../../parrot -t 1 perl6.pir -e '"$"'
12:18 moritz with -t 2 you get # find_method class 'CodeString' method 'escape': NCI
12:18 moritz as the last line of non-hanging output
12:18 masak moritz: what's the difference between -t 1 and -t 2?
12:19 moritz the verbosity
12:19 purl the verbosity is good.
12:20 masak indeed.
12:20 masak -t 1 gave me some additional respect for parrot
12:21 moritz because it produces so much output? ;-)
12:23 masak in effect, yes
12:23 masak I respect CPAN for the same reason :/
12:23 moritz and /dev/urandom? ;-)
12:24 iblechbot joined #parrot
12:24 masak moritz: who doesn't? :)
12:24 Psyche^ joined #parrot
12:25 masak moritz: ok, so I know we get stuck in src/parser/quote_expression.pir
12:26 masak what do the last few calls tell me?
12:26 masak emit, unique, escape, unique, emit, escape
12:26 masak and what's a CodeString?
12:26 purl rumour has it a CodeString is one of those that I think could give a nice performance win if it were written in C instead of PIR :-)
12:27 masak purl: I wasn't asking you.
12:27 purl masak: i'm not following you...
12:27 moritz purl: masak means "SHUT THE FUCK UP", only more polite ;-)
12:27 purl moritz: huh?
12:27 masak purl: what moritz said.
12:27 purl masak: wish i knew
12:27 masak I wish you knew too.
12:28 moritz quote_expression.pir doesn't call a method named escape
12:29 masak there's a hash key named 'escape', though
12:29 masak and a local variable named escape
12:29 moritz and I wonder if the NCI at the end of line means it's a C function that's being called
12:29 masak ah :/
12:30 moritz masak: are you sure it hangs in quote_expression.pir?
12:31 moritz the methods in that file seem to quote expression, not parsing quoted expressions
12:32 masak moritz: I have commented out a line in that file, and after that it worked
12:32 masak i.e. didn't hang
12:32 moritz ok
12:32 masak I'm not sure you're right about it quoting expressions
12:32 masak look at the comments in grammar.pg about it
12:33 masak moritz: try commenting out line 109 in that file and you'll see
12:33 moritz I see what you mean
12:34 masak that's as close to a solution as I've got
12:35 masak quote_expression calls quote_concat, which defines a local variable escapes, but never uses it
12:36 masak quote_concat calls quote_term, which uses escapes
12:37 masak but the changes I make there seem to have no effect, so there the trail gets cold
12:38 * moritz wonders what happens if you delete/comment out lines 221 and 222
12:39 moritz sanity tests still pass
12:40 masak weird :)
12:41 moritz the one in line 109 has an effect because escapes is used in line 116 again
12:42 moritz spaghetti assembler--
12:42 masak moritz: yes
12:43 masak and it's that options hash that's passed on to the other methods
12:43 masak something tells me the hangup is not in quote_expression
12:44 masak but lower down, when we actually start parsing things
13:02 dalek r29720 | infinoid++ | trunk:
13:02 dalek : [darwin] Apply patch from RT #57224.  kasei++, rurban++.
13:02 dalek : * Reverse the order of @arches in the darwin hints file.  Since /ppc/ will
13:02 dalek :   match both "ppc" and "ppc64", it is important to do search/replaces for
13:02 dalek :   "ppc64" before doing them for "ppc".
13:02 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=29720
13:05 Infinoid kasei: so after that patch, does parrot build to completion and test out ok?
13:12 justin joined #parrot
13:15 dalek g.roehrbein@gmx.de | Inter-HLL Mapping Notes:
13:15 dalek link: http://www.perlfoundation.org/parro​t/index.cgi?inter_hll_mapping_notes
13:21 dalek r29721 | coke++ | trunk:
13:21 dalek : [tcl] get past the last dying test in dict.test; we can now run the entire
13:21 dalek : test file and get the trailing TAP output. Still some failures/skips.
13:21 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=29721
13:22 skv joined #parrot
13:33 gryphon__ joined #parrot
13:43 Tene perl6: say "zomgl"
13:43 polyglotbot OUTPUT[zomgl␤]
13:43 Tene infinoid++
13:44 cognominal joined #parrot
13:47 donaldh Infinoid: I updated rt #56828 with a new patch that includes a test for the new NCI parameter type.
14:01 timbunce joined #parrot
14:01 dalek r29722 | coke++ | trunk:
14:01 dalek : [tcl] In addition to providing a list of spec tests to skip, also provide
14:01 dalek : a list of spec tests run as TODO. This gives us a chance to actually notice
14:01 dalek : when things start working. Need to go through the current list of skips and
14:01 dalek : change those that can be run over to TODOs. Setup list.test with the new
14:01 dalek : TODO-ability.
14:01 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=29722
14:04 jan joined #parrot
14:09 cognominal joined #parrot
14:17 Infinoid donaldh: thanks!  I'll look at it later this morning.
14:17 uniejo joined #parrot
14:26 cognominal joined #parrot
14:54 donaldh Infinoid++
15:05 Infinoid I haven't actually _done_ anything yet... :)
15:06 moritz Infinoid: that's a subtle way to build pressure on you ;-)
15:06 Infinoid oh YEAH?  well, moritz++ !
15:07 moritz Infinoid: as long as the ++ isn't connected to a particular task I don't feel any obligation ;-)
15:10 Infinoid ++'s need :attributes
15:11 Infinoid # Parrot VM: PANIC: vVi is an unknown signature type.
15:11 Infinoid guess it requires a realclean
15:31 dalek r29723 | infinoid++ | trunk:
15:31 dalek : [nci] Apply patch from donaldh++ in RT #56828, with whitespace fixes.
15:31 dalek : * Add "V" NCI signature, for opaque void** writeback pointers.
15:31 dalek : * Add a test for it.
15:31 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=29723
15:31 donaldh Infinoid: and a Configure.pl
15:31 purl i think a configure.pl is not a spammy domain name. it's not even registered.
15:32 donaldh doh, that's obvious.
15:33 Infinoid without Configure.pl, "make test" runs *really* fast
15:33 donaldh :D
15:34 donaldh that's a nice idea purl. Anyone thought of registering configure.pl ?
15:34 japhb Clearly *someone* did ... :-)
15:34 japhb ping configure.pl
15:34 purl I can't find configure.pl in the DNS.
15:35 dalek r29724 | coke++ | trunk:
15:35 dalek : [tcl] fix stub of [trace] so it actually does nothing instead of fail.
15:35 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=29724
15:36 dalek r29725 | coke++ | trunk:
15:36 dalek : [tcl] Convert a few spec test SKIPs into TODOs (and add some more TODOs)
15:36 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=29725
15:38 donaldh I've been thinking about inter HLL mapping some more.
15:39 donaldh Perhaps we need to nail down what people actually _want to do_ with it.
15:39 donaldh E.g. I suspect that nobody actually needs to pass really complex datastructures by value between different HLLs
15:41 japhb Perhaps not now -- but if the border is arbitrary, then some poor user is going to run up against it and wonder what the heck is going wrong
15:41 donaldh true
15:42 japhb Converse of that is that if for deep reasons there does need to be a border, we need to draw it in glowing neon paint.
15:43 donaldh I'm just thinking aloud that apis tend to take primitive values, shallow collections and opaque structures.
15:44 japhb I think one way to find that border is to work our way up from the bottom -- what can we guarantee will work with a simple scheme, and see if we can do better, iteratively.
15:44 donaldh That might be a good approach.
15:45 donaldh Take JDBC for example; relatively simple API calls that return complex, opaque objects with accessor methods.
15:45 donaldh Even a ResultSet is opaque until you actuall start realizing rows out of it.
15:45 donaldh s/actuall/actually/
15:45 japhb IMHO the two simplest schemes are 1) YAML serialization and 2) limiting ourselves to mapping to/from core PMCs.  We should spec out what each of those gives us, and see if we can do better.
15:45 japhb nod
15:46 moritz what does serialization buy us?
15:46 donaldh Not sure.
15:47 japhb While the 'opaque object with methods' concept seems simple, I think it's more difficult than it seems -- and means we have to figure out how to handle it in languages that aren't natively OO
15:47 timbunce joined #parrot
15:47 japhb moritz: it's dead simple, with a very limited set of types.
15:48 Tene japhb: how would that be any better than mapping to/from core types, then?
15:48 moritz japhb: so no user defined types at all?
15:48 japhb donaldh: self-referentially, might want to add a log link to your first comment 10 minutes ago to the wiki page ....
15:48 donaldh nod
15:49 japhb Tene: because YAML doesn't have more complex data structures like OrderedHash to worry about
15:49 Tene japhb: so what will yaml do with an orderedhash?
15:49 japhb moritz: I meant it not as a "best choice", but as a baseline that we could improve on.
15:49 donaldh We need a very explicit list of problems. More than the bullet points in the wiki.
15:50 donaldh 1. direct mapping, e.g. Perl 6 Int to Tcl Integer
15:51 japhb Tene: probably the best it could do is serialize out to a hash, and put the keys in the right order.  Which actually come to think of it, works fine.  BUT: If the destination language doesn't know what an OrderedHash is, it will probably treat it as just a normal hash, and I meant that to be explicitely OK.
15:51 donaldh 2. indirect mapping, e.g. Perl 6 Int, Str, Num to Perl 5 Scalar.
15:53 japhb Tene: note that IIRC YAML can mark the desired actual data type to "bless into" in Perl 5 speak, so YAML could perfectly serialize an OrderedHash.  *de*serializing it would be where the info gets lost
15:53 Theory joined #parrot
15:55 Tene The problem, though, is with custom types or extended types, right?
15:55 Tene If we return an orderedhash, the other language can deal with it fine, right?
15:55 Tene In what situations would returning an orderedhash be a problem that serialization could fix?
15:56 moritz I don't quite understand what problem this is trying to solve that couldn't be solved by just letting the foreign language handle the method calls
15:57 moritz if ruby returns an object of type Foo to my perl 6 script, all method calls on that should follow ruby semantics, right?
15:57 Tene moritz: the problem is, for example, if I return a perl6 into to ruby, you can't call ruby's methods on it, 'cos it's the wrong type.
15:57 Tene moritz: Right, exactly.
15:57 japhb sorry, was afk
15:57 timbunce joined #parrot
15:57 moritz why not? you have to let the ruby backend handle the semantics of the method call
15:58 moritz I parsed that the wrong way, sorry
15:58 moritz forget my last remark
15:58 purl moritz, I didn't have anything matching my last remark
15:59 japhb moritz: What if you have a deep class heirarchy, and hand an object of the deepest class to the other language.  Assume the languages have different MROs.  What happens?
16:00 japhb Tene: Because Parrot implements more "base types" than really are base types in many languages.
16:00 Tene japhb: In what situations would returning an orderedhash be a problem that serialization could fix?
16:00 moritz japhb: ruby's types don't appear in Perl 6's type hierarchy, so you can't do MMD - and MRO is handled by the language which handles the method call semantics
16:01 japhb Tene: Forth has ints, arrays, and sometimes floats.  It has one master hash (the dictionary).  What is an Ordered Hash to Forth?
16:01 moritz well, if a language can't handle a data type, there's nothing you can really do
16:02 japhb moritz: So if you say that the dest language handles the method call, then you're saying that it will do the wrong thing.
16:02 moritz it's like returning a closure to basic - useless, because basic has no  syntax for calling them
16:02 Tene japhb: What is some forth code that would not work if the object it's using is an OrderedHash?
16:02 moritz japhb: wrong in what way? would you care to provide an example?
16:03 Tene In what case would you even ask for an orderedhash from another language?
16:03 Tene What would you be trying to do?
16:03 japhb OK, hold on, I'm getting bombarded.
16:03 Tene Also, if the language is capable of calling eval to get weird objects, it's certainly capable of calling eval to do otherlanguagestuff to those objects.
16:03 * Tene waits.
16:04 donaldh Tene: that was my point. We might be trying to solve things that don't make a lot of sense. E.g. passing an ordered hash to a language that can't do anything with it. Pointless.
16:04 japhb moritz: You just said that the same language that handles the method call handles MRO.  Fine, I'll buy that.  But then you must choose the source language to make the method call, because the dest language will be wrong if the MROs don't match.
16:04 donaldh MRO?
16:05 japhb Method Resolution Order
16:05 japhb Tene: your point about eval is well taken, but seems like a heavy-handed way of doing things. "You must eval every op on a foreign data type"
16:06 moritz japhb: it's not "wrong", it's just not what native objects would do
16:06 Tene japhb: The only reason you would *ask* for a weird datatype is if you needed to do weird things with it, which is the purpose of using the other language.
16:06 moritz japhb: at least I can't see how it's wrong. That's why I asked for an example
16:07 dalek Donald Hunter | Inter-HLL Mapping Notes:
16:07 dalek link: http://www.perlfoundation.org/parro​t/index.cgi?inter_hll_mapping_notes
16:07 Tene I do have an idea, though.
16:07 moritz if I have a type that a ruby function returns, and I call a method on it - how on earth could I expect the method call to be done with Perl semantics?
16:07 Tene It seems to me that the types we really care about are the core data types.
16:07 donaldh nod
16:07 Tene List, Hash, Int, Float, String.
16:08 Tene We're already doing some handling of those with .HLL_map
16:08 donaldh Yes, but .HLL_map only works for any one HLL at a time.
16:08 Tene If we return a Point or a Square or a WhoTheHellKnows, we don't *care* about translating it.
16:08 japhb Tene: Your other point about Forth code that would not work if an object is an ordered hash.  It was a well-intentioned but badly-worded example.  But it does bring up an interesting point.  Forth has no native reason to support an OrderedHash.  But if we insisted that every language support the basic Parrot types, it could usefully decide to turn an OrderedHash into an array, or perhaps an array of pointer pairs.  Hmmm.
16:08 Tene so we should perhaps just try to ensure that the .HLL_map type stuff works right.
16:08 Tene Perhaps a .HLL_unmap
16:09 Tene .HLL_unmap 'List', 'ResizablePMCArray'
16:09 Tene Then the calling language can re-map it.
16:09 moritz does forth support user defined data types?
16:09 donaldh I'd really like to avoid copying the core data types when we map.
16:09 moritz if not I see little chances for a bridge at all
16:10 japhb Tene: You are choosing the "map source to parrot to dest" system.  (Not that it's wrong, I'm saying, you appear to be voting for that.)
16:10 donaldh But we have this problem of the asymmetric mapping
16:11 donaldh Tene: that case almost isn't necessary since List isa ResizablePMCArray
16:12 donaldh No need to unmap, but it is necessary to map to the target language.
16:12 donaldh Unfortunately that's not what .HLL_map means AFAIK.
16:12 japhb moritz: yes, it does support user defined types.  But A) you have to build them up from very primitive ops, and B) OO in Forth is even more screwy than Perl 5 -- not only does every Forth (with built in OO) define the structure of a class and an object differently -- they all use different syntax, too.
16:13 japhb And it's not unusual for people to define their own OO system.  Like, from scratch.
16:16 japhb donaldh: Just to be clear, what do you do if the HLL class is a subclass of a Parrot type that is itself a subclass of another Parrot type?  Just choose the deepest child class that is still in the Parrot core?
16:16 donaldh japhb: I don't know.
16:17 iblechbot joined #parrot
16:17 Tene Also, to be clear, it's not just about returning, it's also about calling.
16:17 Tene It's not even always about eval.
16:18 Tene use ruby:Something::Whatever;
16:18 donaldh Tene: agreed. My comment at the start of this thread was that method parameters tend to be simpler than return values.
16:19 donaldh Although Perl tends to buck that trend.
16:19 Tene I'm still not persuaded that we need to do anything at all.
16:19 japhb We also have the fun of deciding what to do when language A calls language B, takes the return value and hands it off to language C.  And heavens forfend that language C call was to register a callback for the object from language B ....
16:19 donaldh heh
16:20 Tene Which is why I think we need to either do nothing or keep it *very* simple and limited in scope.
16:21 donaldh Okay, so here's where I started. I can write PIR to call sqlite and construct a result that is a ResizablePMCArray of Integer, Number, String, etc.
16:21 donaldh In Perl 6, the .HLL_map rules ensure that parrot actually produces a List of Int, Num, Str, etc. from my PIR code.
16:21 japhb OK, I'll go on record as believing that this problem is way more complex than it appears at the surface, and whatever we decide to try, we need to start NOW, so that when our first attempt runs up against a big ugly problem, we have time to work things out before 1.0.  Because after 1.0, we'll have a REALLY hard time making changes in such a core feature.
16:21 donaldh That's great, in a single HLL world.
16:23 donaldh But if I then try to call a Tcl or Python method with that List of Int, Num, Str, etc. then the target HLL will not get what it expects and .HLL_map plays no part. AFAIK
16:24 moritz somehow I don't think that you can always get DWIM'miness with HLL interop
16:24 Tene OBTW
16:24 Tene About yaml.
16:24 moritz not as a restriction of the parrot system, but in general
16:25 japhb moritz: we're not expecting to get full DWIM.  We're trying to suck as little as possible.
16:25 Tene What if I have an array of ZOMGLOL objects?  Can you serialize the array while keeping those objects opaque?
16:25 donaldh japhb: I think I will hack in some additional 'use' capability to Rakudo so that I can experiment.
16:25 japhb YAML pretty much by definition is not opaque.
16:25 Tene donaldh: I'm curious about whether we can use .HLL_map as hints on when do conversion.
16:26 donaldh Tene: maybe we can.
16:26 japhb donaldh: THANK YOU.  I've only been asking for that for ... oh ... half a year.
16:26 Tene japhb: pmichaud says that it's on his "finish by saturday" list, or some such.
16:26 Infinoid donaldh++ # p6 "use"
16:26 japhb (I trust pmichaud that he's got bigger fish to fry, but damn, I *want* this!)
16:27 donaldh japhb: I'm sure pmichaud and jonathan will tear it up and replace it when they do it right, but it would help in the short term.
16:27 sjansen joined #parrot
16:27 Infinoid moritz: did I do it right this time? :P
16:27 japhb Tene: really?  Cool!
16:27 japhb donaldh: :-)
16:27 moritz Infinoid: yes, Infinoid++ ;-)
16:27 donaldh Infinoid: oooh I feel the pressure mounting.
16:28 dalek r29726 | coke++ | trunk:
16:28 dalek : [tcl] Provide the ability to TODO spec tests via globs instead of exact matches
16:28 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=29726
16:29 donaldh Keep up the discussion. I need to drop off; children to feed.
16:48 MagNET joined #parrot
17:05 cjfields joined #parrot
17:08 MagNET joined #parrot
17:09 Andy joined #parrot
17:16 timbunce joined #parrot
17:19 dngor joined #parrot
17:31 Andy joined #parrot
17:35 timbunce joined #parrot
17:52 cognominal joined #parrot
18:18 Theory joined #parrot
18:31 rurban joined #parrot
18:31 rurban moritz?
18:31 moritz yes?
18:31 pmichaud yay!  I haz an oscon network connection!
18:31 pmichaud yay!  I haz an oscon network connection!(finally!)
18:32 rurban About my pugs installation for your blog: I found my blog in my journal http://use.perl.org/~rurban/journal/36897
18:32 Theory joined #parrot
18:32 moritz pmichaud: cool ;)
18:32 moritz rurban: looking
18:32 rurban title: pugs installation on cygwin
18:33 moritz we should link that in pugs' INSTALL file
18:37 moritz rurban++
18:38 rurban I'm just writing in my journal about my perl6 packaging desaster :)
18:38 moritz ;)
18:40 Theory joined #parrot
18:40 timbunce joined #parrot
18:56 jhorwitz joined #parrot
19:06 grim_fandango joined #parrot
19:13 dalek r29727 | moritz++ | trunk:
19:13 dalek : [rakudo] add S12-class/ineheritance.t to spectest_regression
19:13 dalek :  +10 pass, +7 todo, +15 skip - still much work for jonathan ;-)
19:13 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=29727
19:17 pmichaud oscon network connection: FAIL
19:18 Infinoid ohnoes
19:18 pmichaud apparently I can use oscon wifi only from D139/140
19:18 pmichaud anywhere else it doesn't work.
19:18 Infinoid ohnoescon?
19:22 Infinoid is it just one AP?  or a bunch of them?
19:24 pmichaud I don't know.
19:28 timbunce joined #parrot
19:31 pmichaud afk for a while, lunch
19:46 Ron joined #parrot
19:52 dalek r29728 | moritz++ | trunk:
19:52 dalek : [rakudo] two more S06 tests for spectest_regression
19:52 dalek :  +6 pass, +3 skip
19:52 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=29728
19:52 * moritz decides to call it a day
20:02 Ron joined #parrot
20:02 Ron_ joined #parrot
20:11 dalek r29729 | tewk++ | gsoc_nci:
20:11 dalek : [gsoc_nci] dynamically jitted nci stubs pass all tests on i386
20:11 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=29729
20:15 mj41 hi, I updated parrot-lgp and http://pr.perl6.cz/parrot-​lgp/wiki/First_prototypes (feel free to correct my English :-)
20:19 iblechbot joined #parrot
20:19 davidfetter joined #parrot
20:23 jonathan mj41: pises dobre po anglicky. :-) ale s/algorith/algorithm/
20:27 mj41 jonathan: Dakujem. Ty pises dobre po Slovensky. Ucis sa rychlo. :-)
20:29 timbunce joined #parrot
20:32 jonathan mj41: Dakujem...studujem a ucim sa...slovencia je tazka!
20:36 donaldh joined #parrot
20:39 cotto_work coke, ping
20:42 dalek r29730 | Whiteknight++ | gsoc_pdd09:
20:42 dalek : [gsoc_pdd09] updating to trunk r29729
20:42 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=29730
20:44 cbrandtbuffalo joined #parrot
20:49 DietCoke joined #parrot
20:49 DietCoke cotto_home: yes?
20:49 DietCoke you'll get better response times via email, you know. =-)
20:52 cotto_work how's your win32 build environment set up?
20:53 DietCoke ... "fine" ?
20:53 DietCoke I basically followed the directions in the win32 readme to get the free ms stuff.
20:53 cotto_work I mean which tools do you use and what did you have to do to get them to play nice with msvc.
20:53 cotto_work ok
20:53 DietCoke note that those directions refer to the perl5 directions.
20:54 DietCoke but if you want gcc on win32, you can just install strawberry perl, and use the gcc/make that come with it.
20:54 cotto_work I'm trying to get msvc working
20:54 DietCoke hokay. I just followed the directions. I never use msvc for anything other than parrot.
20:54 DietCoke I also yelled at particle a lot.
20:56 cotto_work also, how does it work that you can be summoned to irc?
20:56 Tene Yelling at particle is both productive and fun.
20:57 moritz heh ;-)
20:57 DietCoke cotto_home: moritz wrote some enabling technology.
20:57 * moritz ?
20:57 DietCoke (I have a perl script that trolls his log.)
20:58 DietCoke or, magic.
20:59 Tene or, shouting at particle.
20:59 cotto_work simple, but still a cool trick
20:59 DietCoke yelling at particle?
20:59 DietCoke stupid purl.
20:59 Tene purl: yelling at particle is both productive and fun.
20:59 purl OK, Tene.
20:59 Tene yelling at particle?
21:00 Tene Bah.
21:00 DietCoke Did someone change purl so you have to address him? ah well.
21:00 cotto_work purl, status?
21:00 purl Since Wed Jul 23 09:06:34 2008, there have been 297 modifications and 171 questions.  I have been awake for 1 day, 11 hours, 53 minutes, 46 seconds this session, and currently reference 711166 factoids. Addressing is in optional mode.
21:00 DietCoke purl: a bot says what?
21:00 purl i haven't a clue, dietcoke
21:00 Whiteknight joined #parrot
21:00 Tene coke?
21:00 Tene laaaaaaaaaame
21:00 Tene purl--
21:00 purl Tene: sorry...
21:01 Tene purl++
21:01 * DietCoke ~~
21:02 DietCoke joined #parrot
21:02 DietCoke ... yes?
21:02 DietCoke purl, Coke is also Kibo-esque.
21:02 purl OK, DietCoke.
21:02 DietCoke tene: yes?
21:03 Tene Huh?
21:03 * DietCoke smacks tene
21:03 * DietCoke smacks purl!
21:03 Tene That was before you left!
21:03 DietCoke Yes, I just realized that. =-)
21:03 Tene purl: coke is also FULL OF REVENGE
21:03 purl OK, Tene.
21:04 * DietCoke ~~ again. =-)
21:04 DietCoke tene++
21:05 moritz so, how's OSCON?
21:05 mj41 jonathan: Dobru noc.
21:06 pmichaud oscon would be better if I could get on the wireless.  :-|
21:07 pmichaud other than that, oscon is going well
21:07 moritz pmichaud: you should recruite more hackers for parrot/rakudo, not hang out here ;-)
21:07 dalek r29731 | coke++ | trunk:
21:07 dalek : [tcl] cleanup set.test todos.
21:07 dalek : also, whitespace cleanup
21:07 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=29731
21:07 dalek Donald Hunter | Inter-HLL Mapping Notes:
21:07 dalek link: http://www.perlfoundation.org/parro​t/index.cgi?inter_hll_mapping_notes
21:07 jonathan pmichaud: Will likely hack Rakudo tomorrow.
21:07 pmichaud well, it's not just irc -- I'd like to be able to svn up and respond to mail
21:08 jonathan pmichaud: Anything you'd like me to do/not do?
21:08 pmichaud jonathan: okay.  I don't know what my status will be (see above note re: wireless)
21:08 jonathan Sure.
21:08 jonathan I'm not 100% I will hack either. I had to have a (minor, don't worry too much) surgery on my leg on Monday.
21:08 pmichaud okay.  We'll definitely be hacking on rakudo on saturday
21:09 jonathan And still not 100% with it.
21:09 jonathan Ah?
21:09 jonathan With wireless?
21:09 jonathan Should I save my Rakudo day for Saturday?
21:09 jonathan And relax etc tomorrow?
21:09 jonathan I can do that.
21:09 moritz jonathan: I moved an inheritance test to spec/, lot's of stuff to clean up there, I think
21:09 pmichaud oscon finishes tomorrow -- we're having a hackathon on saturday (at a different location)
21:09 Tene I've got over a week of free time starting saturday.
21:09 pmichaud so me, particle, allison, chromatic, timtoady, others will be hackathoning on saturday
21:10 Tene Can I hackathon from far away?
21:10 pmichaud sure!
21:10 Tene Yay!
21:11 jonathan pmichaud: OK, I have to go back to doctors tomorrow anyway, hopefully it'll be fine, but who knows...I think I'll have tomorrow off and join everyone hacking on Sautrday.
21:13 pmichaud that might be better.  I'll have had a chance to regroup and will have wireless then
21:13 pmichaud as it is I've basically been off-the-net since last Thursday
21:13 Tene Anyone have any parrot-related tasks for me to work on tonight?
21:14 pmichaud my inbox runneth over.  I really need to find a good off-line mail solution.
21:14 mj41 jonathan: I hope I will have some free time to work again on TapTinder  ( http://tt.ro.vutbr.cz/report/pr-Parrot/rp-trunk ) next week. On TODO list are "clients driven by server" and then "Rakudo spec-tests".
21:14 mj41 Good night.
21:14 cbrandtbuffalo joined #parrot
21:14 jonathan mj41: Great! Dobru noc!
21:17 jonathan pmichaud: OK, Saturday it is.
21:22 teknomunk joined #parrot
21:24 Theory joined #parrot
21:36 cjfields joined #parrot
21:38 Theory joined #parrot
21:41 cotto_work gah.  I just typo'd INTERP for INSERT
21:42 moritz ETOOMUCHPARROT
21:43 donaldh moritz: EBADPARAM
21:44 moritz /kick donaldh ENOOP ;-)
21:45 * moritz should go to bed know, but as he knows himself he'll stay up at least another hour
21:45 * donaldh agrees
21:46 * donaldh also thinks child will need fed in 4 hours - why bother going to bed.
21:46 timbunce joined #parrot
21:47 moritz how old is he/she?
21:47 donaldh he is three months.
21:48 moritz your first child?
21:48 donaldh second. My daughter is 20 months old.
21:49 moritz nice
21:50 donaldh The thing that I miss most is not being able to find more than about two hours of keyboard time.
21:50 donaldh And that is usually during antisocial hours.
21:50 moritz nobody said it'd be easy ;-)
21:51 donaldh yup.
21:51 donaldh do you have children?
21:51 moritz no
21:52 moritz not even married (yet)
21:52 donaldh (yet) sounds hopeful tho.
21:53 moritz well, after being together four years you start making plans... ;-)
21:53 donaldh just so
21:57 ilbot2 joined #parrot
21:57 Topic for #parrotis now #parrot Parrot 0.6.4 "St. Vincent Amazon" Released | http://parrotcode.org/ | 15 new, 648 open, 80 stalled tix | logged in http://irclog.perlgeek.de/parrot/today
21:57 Tene Some might disagree.
21:57 Whiteknight plans, tell me about it. I'm up to my armpits in plans
21:57 Whiteknight and I tell you what, it's not me making them :)
21:58 moritz .oO( in sovjet russia plans make you )
21:59 * Infinoid plans to buy mint chocolate oreo cookies on his way home
21:59 Infinoid I try not to plan anything more complicated than that.
21:59 * davidfetter wonders how many people here were old enough to drink legally when there was a soviet russia
21:59 bacek joined #parrot
22:00 donaldh what's the cutoff age?
22:00 Tene A what kind of russia?
22:00 Tene ;)
22:01 * moritz can barely remeber Germany's reunion
22:01 moritz it's about the first political event that I can remeber
22:01 * jonathan was very young when Soviet Russia was still around
22:02 jonathan Like, I hadn't even started programming.
22:02 * jonathan is taking a 3-4 week holiday in Russian in the [autumn|fall]
22:02 jonathan s/Russian/Russia/
22:03 Limbic_Region joined #parrot
22:04 donaldh You'd have to be at least 35 to have been old enough to legally drink to the Soviet Union in the UK.
22:05 jonathan Heh. Missed it by over a decade!
22:05 donaldh I scrape in.
22:09 * donaldh rofls and wonders how he has never encountered LOLsheviks before
22:10 davidfetter IM IN UR ALS, CROSSIN' INTO EUROPEZ
22:12 Tene Over a decade.
22:12 donaldh I CAN HAZ REIGN OF TERRORZ
22:13 davidfetter that'd be lolobins
22:15 Infinoid lolcode: VISIBLE "REIGNZ OF TERROR?  DO NOT WANT"
22:15 polyglotbot OUTPUT[REIGNZ OF TERROR?  DO NOT WANT␤]
22:16 timbunce joined #parrot
22:17 cotto_work shouldn't that be "VISABLE"?  it seems like an odd oversight to spell a 7-letter word correctly in the lolcode spec.
22:17 Infinoid http://lolcode.com/examples/hai-world
22:17 Infinoid good point though.  should probably log a bug report about that
22:18 cotto_work or "fix" the wiki
22:18 Infinoid lolcode: VISABLE "REIGNZ OF TERROR?  DO NOT WANT"
22:18 polyglotbot OUTPUT[Scope not found for PAST::Var 'VISABLE'␤current instr.: 'parrot;PCT::HLLCompiler;panic' pc 156 (src/PCT/HLLCompiler.pir:103)␤called from Sub 'parrot;PAST::Compiler;bind' pc 5318 (src/PAST/Compiler.pir:1472)␤called from Sub 'parrot;PAST::Compiler;post_children' pc 1696
22:18 polyglotbot ..(src/PAST/Compiler.pir:365)␤called from Sub 'parrot;PAST::Compiler;as_p...
22:18 Infinoid aaw.
22:19 Infinoid I do love how it requires keywords in all caps
22:19 timbunce joined #parrot
22:24 Tene It's not required, it's just that all the builtin ones are, and it's case-sensitive.
22:25 Tene Just, y'know, coincidence.
22:36 Infinoid is there a lolbug@parrotcode.org?
22:37 cotto_work there's an easy way to find out
22:40 gmansi joined #parrot
22:43 pmichaud I was.  :-)
22:43 _timbunce joined #parrot
22:44 pmichaud (old enough to drink when ussr existed)
22:45 Whiteknight I'm barely old enough to drink now
22:45 Whiteknight actually, that's not true, but I still get carded when I go out
22:45 Tene perl6: say 'infinoid++'
22:45 polyglotbot OUTPUT[infinoid++␤]
22:46 Tene Infinoid: you misspelled VISIBLE
22:46 Tene lolcode: VISIBLE "REIGN OF TERROR?  DO NOT WANT"
22:46 polyglotbot OUTPUT[REIGN OF TERROR?  DO NOT WANT␤]
22:46 Infinoid that's the point.
22:46 Infinoid [15:17] <@cotto_work> shouldn't that be "VISABLE"?  it seems like an odd oversight to spell a 7-letter word correctly in the lolcode spec.
22:46 Limbic_Region I thought it was great to get into bars when I was 15 and looked 25.  Now that I am about to be 2 ** 5 and look like 42, not so much fun
22:47 Limbic_Region salutations all btw
22:47 Tene I've been told several times that I appear to be 10 years older than I am.
22:48 pmichaud I have it on judicial record that at one point I looked 14 years younger than I was.  :-)
22:49 Tene Both of my girlfriends look quite a bit younger than they are.
22:49 Tene I've been accused of dating jailbait more than once.
22:49 pmichaud ..."both"?
22:49 Limbic_Region well, before my wife started bearing our children - she looked 10 years younger than her age (we are 3 years apart)
22:49 Limbic_Region so it looked like a 40 year old was dating a 17 year old
22:49 * Limbic_Region got plenty of disapproving looks
22:50 Tene pmichaud: monogamy doesn't agree with me.
22:52 Limbic_Region Tene - I am too lazy not to be monogamous.  Heck, before I got married I was too lazy even to date 1 at a time and often settled with being alone
22:52 Limbic_Region Tene - so you have far more energy than me
22:53 Tene chitchat: Transcript show: 'Limbic_Region++'
22:53 polyglotbot OUTPUT[Limbic_Region++␤]
22:53 moritz pmichaud: take a look at t/spec/S29-array/elems.t when you have tuits, it looks a bit borked but otherwise mostly passes ;-)
22:56 Tene Limbic_Region: The inexperienced sysadmin says "Another box?  Where will I find the time to play nethack?" and does not understand.  The master sysadmin manages hundreds of boxes and understands that automation is true laziness.
22:57 cotto_work (automatic girlfriend)--
22:58 * Tene is reminded of a Dresden Dolls song.
23:02 kid51 joined #parrot
23:20 Theory joined #parrot
23:26 teknomunk_ joined #parrot
23:31 bacek joined #parrot
23:35 Andy joined #parrot
23:35 timbunce joined #parrot
23:48 TiMBuS joined #parrot
23:49 cbrandtbuffalo joined #parrot
23:54 Theory joined #parrot

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

Parrot | source cross referenced