Camelia, the Perl 6 bug

IRC log for #parrot, 2010-05-14

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:16 TiMBuS joined #parrot
00:20 ruoso joined #parrot
00:26 sorear My idea isn't working so I'll need pmichaud.
00:39 jsut joined #parrot
00:41 sorear How do I run just one test from the parrot testsuite?
00:41 Themeruta joined #parrot
00:46 darbelo sorear: prove path/to/the/test
00:47 darbelo prove -v if you want more verbose output.
00:47 sorear nod
00:49 rurban_ joined #parrot
00:50 sorear a slightly excessive amount of recompilation occurs when I remove a single ! from src/pmc/sub.pmc
00:51 darbelo That's a core pmc. A lot of stuff depends on the core pmcs.
00:52 darbelo We accept patches to make our makefiles better :-)
00:53 sorear hurray
00:53 sorear I'm getting an error from a function that's not textually mentioned in Parrot
00:53 cotto_work what function?
00:53 purl i think function is declared in debug.h and defined in debug.c but is only used in imcc/pbc.c , and the const_table is used in a lot of places unrelated to pdb.
00:53 sorear oh, it's called in a macro
00:53 sorear Parrot_get_sub_pmc_from_subclass
00:54 cotto_work looks like a perfectly normal function to me
00:54 sorear is there a good way to run parrottests under gdb?
00:54 cotto_work src/sub.c has it
00:55 cotto_work gdb --args ./parrrot --whatever --else --you --pass --to --parrot
00:56 sorear Single stepping until exit from function Parrot_Sub_nci_set_outer,
00:56 sorear which has no line number information.
00:56 sorear I thought default builds had -g?
00:56 darbelo You optimized? That removes -g
00:56 cotto_work they should
00:56 sorear cotto_work: I ... don't pass args to parrot
00:56 cotto_work optimized don't
00:56 sorear I "prove t/pmc/sub.t"
00:56 sorear cotto_work: not optimized.
00:57 sorear I have line numbers everywhere except src/pmc
00:57 cotto_work in that case I'd recommend manually running the generated pir or pasm file that corresponds to the test you're interested in.
01:02 NotFound If someone is bored, take a look at TT #1635
01:05 dalek TT #1635 created by NotFound++: Runloop unrolling for exception handling
01:05 dalek TT #1635: http://trac.parrot.org/parrot/ticket/1635
01:09 snarkyboojum joined #parrot
01:10 NotFound 'night.
01:12 Util darbelo: From `man gcc`: GCC allows you to use -g with -O.  The shortcuts taken by optimized code may occasionally produce surprising results --snip-- Nevertheless it proves possible to debug optimized output.  This makes it reasonable to use the optimizer for programs that might have bugs.
01:28 plobsing joined #parrot
01:31 plobsing hi #parrot
01:32 sorear yo
01:35 sorear > my $x := 15
01:35 sorear > say($x)
01:35 sorear 15
01:35 sorear \o/ I have NQP-level lex persistence working
01:36 sorear requires a small parrot patch
01:36 davidfetter sorear++ :)
01:36 dalek plparrot: ad5e7e9 | dukeleto++ | PLATFORMS:
01:36 dalek plparrot: Update PLATFORMS
01:36 dalek plparrot: review: http://github.com/leto/plparrot/commit/a​d5e7e9ecb8f7413158ad4c6e5989bc0bdbc3d7c
01:37 LoganLK sorear: share your code for noobs like me? ;p
01:49 cotto HELO
01:52 kurahaupo 451 Service unavailable
01:54 kurahaupo joined #parrot
01:57 cotto 402 Payment Required
01:57 purl 402 Payment Required is the best HTTP status code
01:57 cotto purl++
01:57 cotto 406
02:01 JimmyZ joined #parrot
02:02 abqar joined #parrot
02:07 plobsing how do I add to iglobals from an hll?
02:15 plobsing looks like you don't. :-(
02:51 cotto http://trac.parrot.org/parrot/wiki/LoritoRoadmap
02:51 JimmyZ cotto++
02:53 cotto clock?
02:53 purl cotto: LAX: Thu 7:53pm PDT / CHI: Thu 9:53pm CDT / NYC: Thu 10:53pm EDT / LON: Fri 3:53am BST / BER: Fri 4:53am CEST / IND: Fri 8:23am IST / TOK: Fri 11:53am JST / SYD: Fri 12:53pm EST /
02:59 dalek parrot: r46595 | jkeenan++ | branches/ops_pct/runtime/par​rot/library/Getopt/Obj.pir:
02:59 dalek parrot: [codingstd] Eliminate trailing whitespace.
02:59 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46595/
03:03 dalek tracwiki: v1 | cotto++ | LoritoRoadmap
03:03 dalek tracwiki: initial version, non-authoritative
03:03 dalek tracwiki: http://trac.parrot.org/parrot/wiki/Lo​ritoRoadmap?version=1&action=diff
03:04 JimmyZ_ joined #parrot
03:19 sorear LoganLK: you will need the lexical-persistence-2 branch of NQP-rx, and...
03:20 nopaste "sorear" at 192.168.1.3 pasted "Parrot patch for topic/lexical-persistence-2 NQP-rx" (94 lines) at http://nopaste.snit.ch/20548
03:20 sorear (but half of that is a new test)
03:21 plobsing cotto: wrt PBC moving from PIR to Lorito, I thought Lorito was a local optimization. Is Lorito high-level, or are PBCs becoming low-level (more than they are now)?
03:24 plobsing btw, cotto++ on a clear roadmap
03:24 cotto Lorito is low-level.  My understanding is that pir code will continue to function unmodified but will be implemented in Lorito and that Lorito will be the bytecode format.
03:26 sorear Lorito has a binary form now?
03:26 cotto s/has/will have/
03:26 sorear hmm. npq needs to exist
03:26 cotto It doesn't even have a well-defined form atm ;)
03:27 plobsing Lorito is a binary form for the most part. you're not supposed to write it (unless you're insane or absolutely need to)
03:27 plobsing npq?
03:28 sorear a particularly nice typo in your loritoroadmap
03:29 dukeleto 'ello
03:29 * dukeleto is hanging out at the pdx hackathon
03:29 plobsing My concern with PBC moving to Lorito is that I'm not so sure low-level ops + bytecode portability necessarily mesh well.
03:29 sorear According to design documents I've read in $past, Lorito is trusted
03:29 sorear the security system runs on PIR before it's turned into Lorito
03:30 plobsing trusted how?
03:30 sorear which means we won't be able to accept Lorito PBC from other sources
03:30 plobsing so you'll trust some random lorito you get off the net? good luck with that.
03:30 cotto fix't
03:31 sorear plobsing: this is Parrot, our goal is to be better than everyone at everything
03:31 sorear in particular, we're supposed to run in browsers and take bytecode off the net
03:31 plobsing which is why I'd like that bytecode to not be too low-level
03:32 cotto security will clearly need to an important part of the discussion
03:32 cotto sounds like something that could get hashed out a bunch at yapc::na
03:33 plobsing if the network level bytecode contains (for exampls) arbitrary pointer ops, it'll be hard to tell if someone is trying to smash your stack or not.
03:34 dukeleto has anybody attempted to write Date/DateTime PMC's ?
03:36 dalek tracwiki: v2 | cotto++ | LoritoRoadmap
03:36 dalek tracwiki: npq?  *sigh*
03:36 dalek tracwiki: http://trac.parrot.org/parrot/wiki/Lo​ritoRoadmap?version=2&action=diff
03:36 cotto plobsing and sorear, I'd encourage you to start a LoritoQuestions page.
03:44 cotto seen allison
03:44 purl allison was last seen on #parrot 1 days, 9 hours, 15 minutes and 22 seconds ago, saying: hah :)  [May 12 18:28:47 2010]
03:50 janus joined #parrot
03:51 darbelo JimmyZ: ping
03:51 JimmyZ hello
03:51 JimmyZ darbelo: pong
03:52 darbelo JimmyZ: I think I know why your build fails.
03:52 JimmyZ why?
03:53 darbelo It looks like the files are mis-encoded (from parrot's perspective)
03:54 darbelo When you open config_lib.pir in a text editor you see all of the characters fine. Right?
03:54 JimmyZ yes
03:55 darbelo You shouldn't. IMCC assumes that strings are ASCII, or when prefixed with unicode: that they are UTF8. Your windows is unlikely to be using that.
03:55 JimmyZ config_lib.pir is saved as ascii.
03:56 JimmyZ by default
03:56 darbelo JimmyZ: Yes. That is part of the problem.
03:56 JimmyZ darbelo: I can't save it as UTF-8 too.
03:57 darbelo You shouldn't have any non-ascii characters in there or IMCC will get confused. All non-ascii characters should be \uXXXX escapes.
03:57 darbelo Otherwise IMCC assumes UTF8 (Which might or might not be the encoding of those chars) and fails later.
03:58 darbelo Which is why I think the patches worked for people on linux, utf8 would be the default there.
03:58 cotto darbelo++
03:59 JimmyZ Could it be fixed in IMCC?
04:00 darbelo The simpler fix would be for Configure to check if the string has non-ascii characters on it and properly escape those.
04:00 darbelo I'm not sure IMCC can be made uncode-friendly at that level.
04:01 plobsing the IMCC part wouldn't be too hard (I could do it in a couple hours), but I'd need supprot for UTF-16 encoding in parrot. do we have that?
04:02 darbelo In parrot strings? Sure, UTF-16 and UCS-2 are supported. I'll be adding UCS-4 in the next week too.
04:03 darbelo But they depend on ICU being available.
04:03 plobsing which is less likely on windows, the platform which needs this most, correct?
04:03 darbelo Most likely. Yeah.
04:03 darbelo Er. Less likely?
04:04 darbelo What you said is right :)
04:04 JimmyZ I don't think it a good fix by escaping there
04:04 JimmyZ ;)
04:05 darbelo JimmyZ: Escaping is the easy fix. The right fix is the plobsing is suggesting.
04:05 JimmyZ plobsing++
04:05 darbelo plobsing: We have been skirting the issue of making ICU a hard dep for ages.  This might be the time to finally take the plunge.
04:06 cotto I thought it already was.
04:06 plobsing I'm not sure it even needs fixing. Does unicode:utf-16:"chars" or utf-16:unicode:"chars" or utf-16:"chars" work?
04:06 darbelo cotto: Nopes, you can still build without ICU. You just get less functionality.
04:07 * JimmyZ doesn't want ICU
04:08 cotto I can imagine.  dependencies on windows are a pain.
04:11 * JimmyZ has an ICU on windows, but he doesn't like it
04:11 darbelo JimmyZ: ICU is the backbone of our unicode capabilities, avoiding it isn't much of a long-term option.
04:12 darbelo Rakudo, for example, needs ICU.
04:13 JimmyZ I thought there was a plan to eliminate ICU.
04:14 cotto Maybe you're thinking about Perl 5.
04:15 darbelo ICU is a lot of code to replace. I think we've reinvented enough wheels for now.
04:17 cotto we can reimplement it for the 5.3 release.
04:18 JimmyZ 5.3 , do you mean perl 5.3?
04:18 darbelo Given enough time I'm sure we'll reinvent all wheels. Several times over.
04:19 darbelo Hey, and maybe we'll get a few to be round or somethin ;)
04:19 plobsing JimmyZ: we'll be releasing Parrot 5.3 in 2013
04:20 JimmyZ wow, I can't image the version is so high
04:20 darbelo plobsing: The mayan people disagree with that prediction ;)
04:21 cotto JimmyZ, no.  i was just mentioning a very far-off parrot version
04:56 kurahaupo joined #parrot
04:57 dalek TT #1636 created by dukeleto++: Functions need to be added to the embed/extend API headers
04:57 dalek TT #1636: http://trac.parrot.org/parrot/ticket/1636
05:06 dalek plparrot: ebdb710 | dukeleto++ |  (2 files):
05:06 dalek plparrot: Get rid of embed_string.h
05:06 dalek plparrot: review: http://github.com/leto/plparrot/commit/e​bdb710118dbd2775cc256cc298a791a863fdeb8
05:17 dalek plparrot: b68004a | dukeleto++ |  (2 files):
05:17 dalek plparrot: Make the max subroutine length compile-time configurable constant, default = 64KB
05:17 dalek plparrot: review: http://github.com/leto/plparrot/commit/b​68004ad3f4f44d5f22d0ebaa815b4965f738695
05:18 kurahaupo joined #parrot
05:37 ttbot Parrot trunk/ r46596 MSWin32-x86-multi-thread make error http://tt.taptinder.org/file/cmdout/310249.txt ( http://tt.taptinder.org//bui​ldstatus/pr-Parrot/rp-trunk/ )
05:41 ttbot Parrot trunk/ r46597 MSWin32-x86-multi-thread make error http://tt.taptinder.org/file/cmdout/310262.txt ( http://tt.taptinder.org//bui​ldstatus/pr-Parrot/rp-trunk/ )
05:44 dalek parrot: r46596 | plobsing++ | trunk/src/pmc/nci.pmc:
05:44 dalek parrot: update NCI signature parsing to emit new (half year old) PCC param and return signature
05:44 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46596/
05:44 dalek parrot: r46597 | plobsing++ | trunk (2 files):
05:44 dalek parrot: Give interpreter access to nci frame builders and eliminate requirement of nci thunks being unmanagedstructs.
05:44 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46597/
05:58 sorear PCT quasiquotes would be an awesome addition to NQP
06:00 sorear plobsing: Is that really only a 3-line change or did you forget to stage a file?
06:01 plobsing sorear: yes, it is a legitimate 3 line change
06:01 plobsing although I really can't see why ttbot is hating on my commits
06:15 darbelo plobsing: ./src/pmc/nci.pmc(62) : error C2057: expected constant expression
06:15 darbelo ./src/pmc/nci.pmc(62) : error C2466: cannot allocate an array of constant size 0
06:15 darbelo ./src/pmc/nci.pmc(62) : error C2133: 'sig_buf' : unknown size./src/pmc/nci.pmc(62) : error C2057: expected constant expression
06:15 darbelo ./src/pmc/nci.pmc(62) : error C2466: cannot allocate an array of constant size 0
06:16 plobsing constant size 0?
06:16 darbelo That's whats causing ttbot to whine.
06:17 plobsing I'm not up on my crummy-ms-compilerese
06:17 plobsing what does that actually mean?
06:17 sorear [0] is invalid C
06:17 sorear arrays need to have at least one element
06:18 sorear likewise, var[] = {}; is wrong
06:18 plobsing really? seems like a silly restriction.
06:18 plobsing but also that array will never be of zero length
06:19 darbelo ./src/pmc/nci.pmc(62) : error C2057: expected constant expression
06:19 sorear well, it comes back to the fact that & is into
06:19 darbelo That's the real error. The rest is msvc being dumb.
06:20 sorear an array with zero elements cannot have a unique address
06:20 darbelo size_t  buf_length = sig_length + 2 + 1; is not a constant expression.
06:20 sorear (this is why C++ interprets class whatever { /* methods only */ } to have an unnamed int field)
06:21 darbelo plobsing: you need to allocate sig_buf dynamically.
06:21 darbelo buf_length depends on values not known at compile time.
06:22 darbelo So sig_buf[buf_length] can't be on the stack.
06:24 plobsing C++ doesn't let me stack-allocate arrays?
06:24 * plobsing almost misses fortran
06:25 sorear you can stack allocate arrays
06:25 sorear but in C89, they have to be statically known size
06:25 darbelo MSVC in C89 mode doesn't allow stack allocations that might differ in size from one call to another.
06:25 sorear Parrot isn't C++
06:25 sorear GCC in C89 mode doesn't either :0
06:25 sorear :)
06:26 darbelo plobsing: Yes, a c++ compiler is failing due to c89 non-compliance.  Working with MSVC is like that.
06:27 sorear syntax error: unable to find final } ... in a 19 line statement
06:27 plobsing I got rid of the malloc there because that's the hottest path in NCI. I'd rather not churn the allocator. Can I just throw it in an ifdef?
06:28 darbelo #ifdef WHAT ?
06:28 plobsing ifdef YOU_HAVE_AN_OUTDATED_CC
06:28 plobsing whatever msvc defines
06:29 darbelo I think you want YOU_ARE_NOT_ON_GCC, Sun cc is also strict on c89 comliance.
06:31 darbelo And parrot is, amusingly enough, supposed to be written in c89
06:31 darbelo More like the intersection of c89 and c++ really. Since we have people who compile with g++
06:31 sorear I hear we're in the process of rewriting it in NQP-rx
06:32 * darbelo pulls sorear's bootstraps.
06:34 sorear trying to debug syntax errors in NQP drives me insane
06:34 * plobsing is painfully aware of ppl building on g++. Sometimes I want to call my variables 'new'.
06:36 darbelo Heh, my first week after getting a commit bit I broke the c++ and msvc builds.
06:37 bacek_at_work plobsing, namespace and class are perfect names too
06:41 cotto bacek_at_work, when you have the time I'd appreciate improvements or clarifications to http://trac.parrot.org/parrot/wiki/LoritoRoadmap
06:42 bacek_at_work cotto, you forgot "world domination" at least
06:42 sorear The error, by the way, was forgetting that return was a prefixop.
06:42 cotto That's the same thing we plan every night.
06:44 sorear msg pmichaud I just dropped my wishlist on your wiki page
06:44 purl Message for pmichaud stored.
07:06 dalek parrot: r46598 | plobsing++ | trunk/src/pmc/nci.pmc:
07:06 dalek parrot: c89 conformance
07:06 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46598/
07:12 plobsing yay, no more broken build
07:13 cotto yay
07:16 iblechbot joined #parrot
07:17 fperrad joined #parrot
07:21 pmichaud good evening
07:21 purl Ah, evening. The tumultuous mind tarries and contemplates, reveling in the silence afforded by the diurnal proletariat. Good evening, indeed.
07:22 pmichaud NQP _can_ be a subset of Perl 6 forever.  :-)
07:22 sorear Hello pmichaud.
07:23 dalek parrot: r46599 | pmichaud++ | trunk/src/pmc/sub.pmc:
07:23 sorear (Don't look at my nqp-rx branch; it's wrong and I'm fixing it)
07:23 dalek parrot: [core]: Make sure Sub.set_outer() clears any existing outer_ctx.
07:23 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46599/
07:23 dalek parrot: r46600 | pmichaud++ | trunk (3 files):
07:23 dalek parrot: [past]:  Add .closure() flag to PAST::Block to return closures with
07:23 dalek parrot: captured outer context.
07:23 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46600/
07:23 cotto hi pmichaud
07:23 pmichaud pir:: isn't incompatible with Perl 6.  :-)
07:24 sorear pmichaud: I also patched sub.pmc in a different way...
07:24 * sorear reads your patches
07:24 sorear pmichaud: What's the point of r46599?
07:24 pmichaud my patch removes a buglet from Sub PMC and enables re-attaching the outer.
07:25 sorear re-attaching?
07:27 pmichaud so that a newly compiled sub can have a previously-executed sub as its outer context
07:27 pmichaud more to the point, if we change a sub's notion of its outer lexical sub, we should clear any outer context that we had previously set for the sub.
07:27 pmichaud anyway, it enables the lexical chaining we need for interactive loops
07:28 sorear I'm using a diferrent patch which has Occam-advantages
07:28 sorear hmm.
07:29 sorear pmichaud: I don't see anything in your patch which forces parrot to retain lexpads after the sub returns.
07:29 pmichaud parrot already does that.
07:29 pmichaud nopaste proof coming up :)
07:30 sorear nopaste disproof coming up
07:30 sorear http://pastie.org/959943
07:30 sorear look.  set_outer is never called on a previously executed sub
07:30 sorear set_outer is called before the first call
07:30 sorear yet it doesn't work.
07:31 sorear anyways, I'm very interested to see your working code
07:32 sorear can't make a reasoned comparison without it.
07:34 nopaste "pmichaud" at 192.168.1.3 pasted "example attaching a sub to a new outer (that has already finished execution)" (27 lines) at http://nopaste.snit.ch/20550
07:35 pmichaud the reason why your code isn't working is that your "b" sub doesn't realize it's an outer scope.  So Parrot throws away the context at the end.
07:35 pmichaud i.e., the sub is missing the OUTER flag
07:36 sorear aha
07:36 sorear I don't actually need the ability to bind subs after the inner sub finishes executing.  What made you think I did?
07:36 pmichaud ?
07:36 pmichaud the challenge is that you need it for when the *outer* sub finishes executing.
07:37 sorear I don't need to call B() before B.set_outer
07:37 pmichaud right
07:37 pmichaud that isn't my poing
07:37 sorear So why do I need the B(); B.set_outer bugfix?
07:37 pmichaud if I take that first B() out it still works as expected.
07:37 sorear (I get your point on set_outer.  Just not this point.)
07:37 pmichaud oh.
07:37 sorear and OUTER flags
07:38 pmichaud all I'm doing in my example is demonstrating that even after I've executed B(), I can re-set its outer and then have B() see the new outer scope.
07:38 sorear oh.
07:38 pmichaud before my bugfix, Parrot wouldn't allow that.  Once B() had a outer context set, you couldn't reset it except from an already-executing sub.
07:38 sorear for informative purposes only, here's how my working copy works
07:39 sorear 1. set_outer is modified to optionally take a CallContext instead of a Sub (so there are no lexical lossage issues)
07:39 sorear 2. the PAST mangling stage inserts an op to capture the current context
07:40 sorear 3. outer_ctx is used as is, with no need for outer_sub (Occam)
07:40 pmichaud yes, I thought of that approach as well.  I didn't want to mangle Parrot's Sub PMC API quite that much without checking with the Parrot gods first
07:41 pmichaud I'm not sure that overloading .set_outer in this way is correct; it's possible that there should be a separate .set_outer_ctx method for it.
07:41 snarkyboojum joined #parrot
07:41 sorear in any case, I have a working patch complete with test - http://nopaste.snit.ch/20548
07:42 sorear (pmichaud actually liked the decisions I made while he was unavailable?  *faints*)
07:43 pmichaud sure, I suspected we'd ultimately end up providing a mechanism to set a context for a sub directly, rather than indirectly through other subs.  But I wasn't quite convinced that this is the approach Parrot will ultimately want to take on this issue.
07:44 * sorear has working, generic code and is currently on a refactor
07:44 sorear actually
07:44 sorear this is a very small scale issue
07:44 sorear basically, I can adapt to your fix by changing .set_outer($ctx) to .set_outer($ctx<current_sub>)
07:45 sorear then I can put this patch up on parrot-trac, say it fixes a memory leak
07:45 pmichaud ...memory leak?
07:45 pmichaud not really
07:45 sorear are you one of those people whose definition of "memory leak" cannot exist in a GC system?
07:45 pmichaud no
07:46 pmichaud it would just be imprecise to say that this is resolving any sort of leak
07:46 sorear sub foo(@arg) { for @arg { #`(do stuff) }; };  foo(BIG-ARRAY);  # foo is an outer sub, so BIG-ARRAY is being retained needlessly
07:47 pmichaud right
07:47 pmichaud this doesn't fix that.
07:47 sorear er, right
07:47 pmichaud it fixes it only if something later does   somesub.set_outer(&foo)
07:47 pmichaud and that will happen fairly rarely
07:48 sorear what I'm trying to say is that a variation on my approach will allow parrot to get away from retaining OUTER contexts
07:48 sorear (assuming there are no inner subs retaining via outer_ctx, of course)
07:48 pmichaud we have to keep the OUTER contexts around in order to continue to support autoclose semantics on subs
07:49 pmichaud as soon as we eliminate autoclose, then it's easy to get rid of the leaked contexts.
07:50 pmichaud the existing capture_lex and newclosure opcodes already handle the case where an inner sub needs to maintain a connection to an outer_ctx
07:50 sorear how about I add an entry to the performance tasklist for this
07:50 pmichaud wfm
07:51 sorear irc gets lost easily
07:51 pmichaud ultimately we'd have to deprecate Parrot's autoclose semantics (and run through a deprecation cycle for it)
07:52 pmichaud as a midway point we might be able to allow autoclose for inner subs being called directly from their outer sub, but not for inner subs called after the outer has exited.  that would also eliminate the leaks.
07:54 pmichaud (it still gets a bit nasty if we have the case like Perl 5 allows where one can call lexically nested subs before the outer scope has ever executed)
07:55 sorear pmichaud: lexically nested named subs are not closures in perl 5
07:55 dalek parrot: r46601 | fperrad++ | trunk (5 files):
07:55 dalek parrot: add an URI library
07:55 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46601/
07:56 sorear only anonymous subs close
07:56 * sorear just checked with -e
07:56 pmichaud f();   { my $x;  sub f() { $x = 5; print "$x\n"; };  print "$x\n"; }
07:57 aztack joined #parrot
07:57 sorear ah, I was only checking the named-sub-in-named-sub case
07:57 pmichaud right.
07:58 sorear I love Perl
08:00 pmichaud anyway, I like your patch very much.  I'm still not certain that overloading set_outer for both contexts and subs is the way to do it here... I'd have to think a bit about how other methods have typically been set up.
08:00 pmichaud I agree that ultimately we'll want to be able to set an outer context directly.
08:00 pmichaud I'm just not sure how that will show up.
08:01 sorear a sub is just a prototypical context
08:05 pmichaud incoming nqp bonus :-)
08:06 sorear uh oh.
08:06 sorear I hope you haven't touched HLL::Compiler ;)
08:06 pmichaud Not yet.
08:06 pmichaud It will be touched, but not tonight or tomorrow (likely)
08:06 sorear Good.  I've heavily modified that
08:07 sorear I hope to have my lexical work reviewable in a couple hours
08:07 pmichaud all of it?  or just the interactive part?
08:08 sorear Just the interactive part
08:08 pmichaud wfm
08:08 sorear Which amounts to about 200 lines of modified decompiled code insertions
08:08 pmichaud I'm looking forward to reviewing it :-)
08:08 sorear I'm looking forward to the Rakudo side of the patch.
08:09 pmichaud probably won't get to it until late morning tomorrow
08:09 sorear I'll get to rip out 50-100 lines of code
08:09 sorear Because NQP will handle much more of the eval process
08:09 pmichaud right
08:09 dalek nqp-rx: f427d29 | pmichaud++ |  (3 files):
08:09 dalek nqp-rx: Closures now properly capture outer context.  Also clean up
08:09 sorear What are people's favorite compilation options for a 'fast' parrot?
08:09 dalek nqp-rx: handling of bare blocks a bit, and bump PARROT_REVISION.
08:09 dalek nqp-rx: review: http://github.com/perl6/nqp-rx/commit/f​427d2978c438475ca5c340d1a8220d8601f5c3e
08:09 pmichaud I just use --optimize, myself.
08:10 pmichaud I suspect there may be faster options, but I'm more interested in stablity than speed at the moment.
08:10 pmichaud egads, it's 3:10am already, and I have to be up at 6:40am  :-|
08:11 JimmyZ sleep well
08:11 sorear sleep well
08:11 pmichaud I'll have to do nqp settings tomorrow morning -- need sleep tonight (and I wouldn't trust my work on it anyway)
08:11 cotto night
08:11 pmichaud so, that'll be my priority one tomorrow
08:12 cotto better working than now
08:12 pmichaud should be fairly straightforward
08:12 dalek parrot: r46602 | gerd++ | trunk/docs/book/pct/ch05_nqp.pod:
08:12 dalek parrot: add some missing semicolons; add a my declaration at the first example to illustrate the need of a declaration before using a variable
08:12 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46602/
08:12 cotto I'll make a list of what opsc needs
08:12 cotto that'll cut down your time
08:12 pmichaud yes, a bunch
08:12 pmichaud that will be a big help
08:12 pmichaud oh, I need to update the wiki...
08:13 sorear pmichaud: did you see the nqp requests update?
08:14 pmichaud looking at it now
08:15 pmichaud I agree we need better error messages... not sure why nqp is getting things so far away.  I suspect it's backtracking too quickly where it should instead be generating errors
08:15 pmichaud so the error message always shows up at the beginning of the block or statement that contains the error
08:16 pmichaud having some examples of poor error messages would help
08:16 sorear they're all interrelated to a certain extent
08:17 sorear the worst error messages come from missing commas in the 20 line slabs of NQP we use to fake quasiquoting
08:17 sorear I'll start collecting the bad error messages I get
08:19 pmichaud doesn't nqp already allow returning of multiple values?
08:19 pmichaud it seems to work for me.
08:20 sorear return (1,2) just builds a RPA
08:20 sorear and returns it as an item, not :flat
08:20 pmichaud that's not much different from Perl 6
08:20 pmichaud or Rakudo, at least.
08:21 pmichaud it's just that in Rakudo it's called a Parcel instead of an RPA
08:22 fperrad_ joined #parrot
08:23 sorear pmichaud: the fact that NQP-rx cannot express :flat returns, does it bother you?
08:24 pmichaud not at the moment
08:24 * sorear 's vision of NQP can express all Parrot APIs
08:24 pmichaud that's not NQP's vision at the moment.
08:24 sorear a while back I tried to rewrite blizkost's PIR in NQP
08:24 sorear half of my wishlist is "things I would have needed"
08:24 pmichaud NQP isn't supposed to be a super-Parrot.  In particular, it wants to be able to someday port to non-Parrot backends.
08:26 pmichaud in deciding whether to be closure to what Perl 6 does versus what Parrot provides, NQP tends to want to stay closer to Perl 6
08:26 sorear oh?
08:26 sorear mm
08:26 snarkyboojum joined #parrot
08:26 pmichaud NQP is happy to expose Parrot features as long as they somewhat conform to a P6 syntax
08:27 pmichaud or, more precisely, as long as we can come up with a P6-compatible syntax for the feature
08:28 pmichaud s/closure/closer/  # above, typo
08:29 pmichaud at any rate,  NQP's return statement never translates into a .return PIR instruction anyway.
08:29 pmichaud so there's nothing to :flat
08:32 pmichaud anyway, time for sleep here
08:32 pmichaud be back in a few hrs
08:37 cotto pmichaud, when you feel like hacking:
08:37 nopaste "cotto" at 192.168.1.3 pasted "list of what's used and unused from nqp settings by opsc" (28 lines) at http://nopaste.snit.ch/20551
08:37 cotto opsc builds fine with that subset
08:49 rurban_ joined #parrot
08:49 bakkdoor joined #parrot
09:15 sorear How can a CallContext ever have a current_sub of PMCNULL :/
09:29 sorear aha, ENOSLEEP
09:29 sorear I was reading the "current_sub" named argument.
09:29 sorear not the "current_sub" attr.
09:48 darbelo joined #parrot
09:49 clinton joined #parrot
10:17 joeri joined #parrot
10:36 Whiteknight joined #parrot
10:38 fperrad_ joined #parrot
10:44 dalek rakudo: da421b4 | (Solomon Foster)++ | src/core/ (4 files):
10:44 dalek rakudo: Rework / write trig functions for Any, Numeric, Real, and Num.
10:44 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/d​a421b4307731c07a601034e9c080029ccec4af0
10:44 dalek rakudo: 150eecb | (Solomon Foster)++ | src/core/ (2 files):
10:45 dalek rakudo: Delete Int.sec and Rat.sec.  Apparently Real.sec is enough to handle these cases now.
10:45 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/1​50eecbdf616da8341b7b1ff67775963d9f4428d
10:59 Whiteknight good morning, #parrot
11:02 darbelo Good morning, Whiteknight
11:02 Whiteknight hello darbelo
11:04 darbelo How does the 2.4 release look, management-wise?
11:06 Whiteknight I've been tracking commits and smoke reports, things are looking good-to-go
11:06 Whiteknight I don't think we need much of a feature freeze, so long as the weekend doesn't go crazy
11:07 darbelo ops_pct landing crazy?
11:07 darbelo That one is just around the corner.
11:07 Whiteknight I'm expecting that one. so long as it happens today or tomorrow it shouldn't be a big deal
11:08 Whiteknight I don't think the codestring branch is ready to land either, so that's a load off
11:09 darbelo codestring got half-abandoned after bacek and pmichaud retooled some bits in trunk. So other than some cherry-picking I wouldn't expect anything from that branch.
11:10 darbelo kid51 did raise an interesting point on the list regarding perlcritic failures on ops_pct
11:11 mikehh just got to get perlcritic to avoid some nqp on files with .pm
11:11 mikehh working on it at the moment
11:13 darbelo We could theoretically just 'whitelist' those files on the test, but I'd rather make the test smart enough to stay away from files meant for other versions of perl.
11:14 mikehh ok - it is now filtering out "perl" files with #! nqp shebang line
11:16 mikehh let me do some more tests and will commit it
11:16 darbelo mikehh++
11:30 khairul joined #parrot
11:31 JimmyZ joined #parrot
11:39 dalek rakudo: 6c97fe1 | (Solomon Foster)++ | src/core/Cool-num.pm:
11:39 dalek rakudo: Delete multi sub versions of trig functions in favor of just having the proto versions make the appropriate method call directly.
11:39 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/6​c97fe1d1b237c9a713ff850e9a36a5fc629371c
11:41 iblechbot joined #parrot
11:45 dalek parrot: r46603 | mikehh++ | branches/ops_pct/t/codingstd/perlcritic.t:
11:46 dalek parrot: add filter to perlcritic test to avoid "perl" files with #! nqp shebang line
11:46 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46603/
11:46 dalek parrot: r46604 | mikehh++ | branches/ops_pct/compilers/ops​c/src/Ops/Compiler/Grammar.pm:
11:46 dalek parrot: add shebang line - #! nqp
11:46 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46604/
11:46 dalek parrot: r46605 | mikehh++ | branches/ops_pct/compilers/opsc/src/Ops/OpLib.pm:
11:46 dalek parrot: add shebang line - #! nqp
11:46 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46605/
12:02 dalek parrot: r46606 | mikehh++ | branches/ops_pct/tools/build/ops2c.pl:
12:02 dalek parrot: fix perlcritic failure - missing use strict/use warnings
12:02 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46606/
12:03 pmichaud good morning, #parrot
12:08 JimmyZ good morning, pmichaud, Do you sleep well?
12:11 pmichaud yes, but not enough
12:12 mikehh hi pmichaud
12:13 * mikehh sleep is often elusive
12:23 darbelo Sleep? I think I used to do that... Can't remember what it felt like, though.
12:24 Coke Whiteknight: the codestring branch is now irrelevant.
12:25 Coke I didn't see this mentioned anywhere: http://www.python.org/psf/m​embers/#board-of-directors
12:25 Coke msg allison congrats on the PSF board membership.
12:25 purl Message for allison stored.
12:30 Whiteknight Coke: why is the branch irrelevant?
12:31 fperrad_ joined #parrot
12:32 Coke Whiteknight: Hopefully the email I just sent will explain that.
12:35 dalek parrot: r46607 | coke++ | branches/codestring:
12:35 dalek parrot: All the interesting bits here are now in trunk.
12:35 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46607/
12:35 bluescreen joined #parrot
12:38 iblechbot joined #parrot
12:39 Coke I see we have yet another library being added.
12:39 tetragon joined #parrot
12:42 Coke msg fperrad - when adding a brand new component with no list or IRC discussion, please add an experimental notice for the library. Thanks.
12:42 purl Message for fperrad stored.
12:47 Whiteknight Coke: okay, I see. One less branch to worry about
12:47 kjeldahl joined #parrot
12:51 dalek parrot: r46608 | coke++ | trunk/DEPRECATED.pod:
12:51 dalek parrot: Add notice for another new library.
12:51 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46608/
12:56 dalek TT #1637 created by coke++: experimental library: URI
12:56 dalek TT #1637: http://trac.parrot.org/parrot/ticket/1637
13:04 Whiteknight how does one get perldoc?
13:05 Coke it comes with perl?
13:06 darbelo Whiteknight: Your distribution's usual chanels?
13:06 Whiteknight make smoke is telling me "perldoc required but not detected"
13:06 Coke Whiteknight: are you running as root or something?
13:06 Whiteknight nope
13:06 Coke (or some other thing that might mess with your PATH)
13:06 darbelo Whiteknight: What (OS) are you on?
13:06 darbelo ;)
13:06 Coke where did you get your perl?
13:07 Whiteknight ubuntu 9.04
13:07 Whiteknight ah, "apt-get perl-doc"
13:07 Coke ... why on earth would they split that out. wierd.
13:07 darbelo Coke: debian splits *everything*.
13:07 NotFound $ dpkg-query -S `which perldoc`
13:07 NotFound diversion by perl-doc from: /usr/bin/perldoc
13:07 NotFound diversion by perl-doc to: /usr/bin/perldoc.stub
13:08 NotFound perl, perl-doc: /usr/bin/perldoc
13:08 NotFound Strange :?
13:08 dalek parrot: r46609 | NotFound++ | trunk/src/pmc/nci.pmc:
13:08 dalek parrot: c++ conformance
13:08 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46609/
13:08 darbelo Coke: IIRC it has something to do with shipping only the minimum perl needed to bootstrap the package tools in the base install.
13:09 Coke ah.
13:09 Coke (note to self, immediately install a copy of perl by hand when I get my ubuntu box.)
13:09 darbelo Or maybe not. My knowledge of debian is about as outdated as the software they ship.
13:11 atrodo joined #parrot
13:11 darbelo Also, my last serious use of Ubuntu was back when they were still shipping debian unstable with a different color scheme.
13:23 mikehh ops_pct branch - distro_tests FAIL - Failed test 'MANIFEST.generated is sorted' (something wierd there)
13:23 mikehh all other tests PASS (pre/post-config, make corevm/make coretest, smoke (#33825), fulltest) at r46608- Ubuntu 10.04 amd64 (g++)
13:24 NotFound Can someone take a look at TT #1635 and give some opinion?
13:25 mikehh I tried editing MANIFEST.generated, then I ran it (the data) through sort and it still gave problems with the test
13:29 Whiteknight I'm going to do some Ubuntu 8.04 testing as soon as VirtualBox updates
13:30 mikehh NotFound - I had a good look - looks ok to me but, you might want a more informed opinion
13:33 mikehh Coke: I don't like the way Ubuntu (or debian maybe) handles perl - I always install by hand, so to speak, using 12.0 wheras Ubuntu has 10.1
13:36 Coke any time there's a system perl, that's a good idea. (I do it on OS X too)
13:37 darbelo OpenBSD has a fairly stock 5.10.1 as the system perl. I've had no issues there.
13:38 mikehh they put the directories all over the place which seems to confuse the hell out of cpanp sometimes in /usr, /usr/local /usr/share with perl and perl5
13:38 darbelo A refreshing change from the horribly outdated gcc that I have to put up with as the system compiler...
13:39 plobsing joined #parrot
13:40 mikehh darbelo: I tried installing 4.5 but had a couple of problems, apart from taking forever
13:40 tetragon joined #parrot
13:42 dalek rakudo: 4afd745 | jonathan++ | src/Perl6/ (2 files):
13:42 dalek rakudo: Don't emit variable declarations for placeholders.
13:42 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/4​afd745fc1be40f9d7cb640f7b22c20022793a9b
13:42 dalek rakudo: cddbf30 | jonathan++ | build/PARROT_REVISION:
13:42 dalek rakudo: Bump to Parrot revision 46606.
13:42 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/c​ddbf307ff1dc941cea70371675b66f3b795d267
13:48 Whiteknight All tests pass on Ubuntu 8.04-x86
13:48 Whiteknight so there's one legacy supported environment off the checklist
13:49 Whiteknight If Microsoft doesn
13:49 Whiteknight t even support WinXP SP2 anymore, do we have to support it?
13:49 mikehh Whiteknight: I don't have 8.04 any more - 9.04, 9.10 and 10.04
13:50 Whiteknight mikehh: I have 9.04. 9.10 breaks my computer because of bad drivers so I haven't updated. I do everything else in VirtualBox
13:51 pmichaud so far 10.04 has been super solid for me.
13:51 GodFather joined #parrot
13:51 mikehh bah - I can't get VirtualBox or kvm to work with my wireless setup - would help a lot if I could
13:52 Whiteknight pmichaud: that's a very good endorsement. I may upgrade for no other reason than because you said that
13:53 pmichaud other than having to figure out how to re-enable hotkeys (kubuntu), I haven't run into any other issues yet.
13:53 darbelo mikehh: Installing into a vm?
13:53 mikehh Whiteknight - I have been using it since beta 1 and have had no major problems
13:54 NotFound pmichaud: please take a quick look at TT #1635
13:54 Whiteknight 9.04 had some weird driver error, I think it was my video card
13:54 Whiteknight and the computer would freeze and become unresponsive at random intervals
13:55 mikehh darbelo: I can install but just can't get my wireless connection to work in any of the VM's
13:56 darbelo OpenBSD and VM's don't always get along. The dev's stand is that if it works on hardware but not on a VM, then the VM is buggy.
13:59 mikehh I have a associate who has got pretty much the same hardware as me but uses an ethernet connection to his wireless router, I tried that with a pair of ethernet plugs, but mit failed and managed to blow my wireless card but fortunately had a spare
14:02 pmichaud NotFound: "control returns to the runloop that established the handler..."   ... and then continues executing the PIR code from the point of the address passed as a parameter to rewind?
14:02 pmichaud s/from the point of the/at/
14:03 Whiteknight t/profiling/profiling.t appears to fail on winxp-x86
14:03 NotFound pmichaud: the address is needed because of using a method, an opcode can just use the next address.
14:03 pmichaud ah
14:03 JimmyZ joined #parrot
14:03 pmichaud but yes, that looks very promising
14:04 NotFound The opcode can also have a variant with an address.
14:04 pmichaud I'm not sure I need the "with an address" portion for an opcode -- I can just as easily put a goto following the opcode :-)
14:05 NotFound Yeah
14:05 pmichaud (unless the address can be in a different sub... which sounds evilish)
14:05 NotFound Uh, no. Allowing a different sub requires more changes.
14:05 pmichaud right
14:06 pmichaud so, I'd say "goto after opcode" is plenty sufficient.
14:06 pmichaud if the caller needs to get to another sub, we have plenty of continuations and other mechanisms for that
14:06 NotFound We can always add a variant later if is wanted for some reason.
14:07 pmichaud any chance that .'rewind' should be a method on Continuation PMC and not just Exception?
14:07 pmichaud because there are similar issues with continuations
14:08 NotFound pmichaud: rewind was just the quickest to implement and test mechanism. I don't whink we need it, getting the runloop_id attribute is enough.
14:09 NotFound So allowing continuations is easy.
14:09 pmichaud good enough
14:09 pmichaud iirc, Exception isa Continuation, and we have the same sorts of issues with those.  (http://lists.parrot.org/pipermail/pa​rrot-dev/2009-December/003493.html)
14:10 pmichaud might be a good idea to cross-link TT #1635 with TT #1091, too :)
14:12 pmichaud looks like Bob Rogers also had a candidate patch in that thread:  http://lists.parrot.org/pipermail/p​arrot-dev/2009-December/003548.html
14:13 pmichaud anyway, I'm in favor of whatever works.  :-)
14:15 NotFound pmichaud: yes, seems to be the same idea.
14:16 * moritz would love to see Whatever properly working
14:17 dalek TT #1638 created by whiteknight++: t/profiling/profiling.t fails on WinXP
14:17 dalek TT #1638: http://trac.parrot.org/parrot/ticket/1638
14:18 plobsing libjit framebuilder lives: git://github.com/plobsing/parrot-libjit-fb.git
14:18 plobsing now all I need is testing and documentation
14:22 patspam joined #parrot
14:31 darbelo setup.winxed? That's new.
14:32 NotFound darbelo: is unfinished
14:34 darbelo NotFound: I'm looking at the one on plobsing's parrot-libjit-fb and it looks nice.
14:35 NotFound darbelo: Oh, I was thinking you talked about winxed.
14:35 NotFound Winxed itself is not yet ready for plumage install.
14:37 GodFather joined #parrot
14:39 NotFound pmichaud: the patch from Bob Rogers seems to do the unrolling at ExceptionHandler invoke time, and that way may disallow resume in lots of cases.
14:39 NotFound Personally, I don't like resumes, but a lot of people here do.
14:41 pmichaud well, we have to have them.
14:42 pmichaud it has to be possible to transfer control to an exception handler and then allow it to say "okay, handled, go back to what you were doing"
14:42 NotFound We need to keep the runloop to do that. That point has been discused several times. So the "handling done" opcode is the reasonable solution,
14:43 pmichaud agreed.
14:43 NotFound BTW, "done_eh" is a good name?
14:45 NotFound Mmm... if we plan to use it with continuations, no.
14:55 tcurtis joined #parrot
14:55 JimmyZ joined #parrot
14:57 snarkyboojum joined #parrot
14:57 fperrad_ joined #parrot
14:57 ruoso joined #parrot
14:59 NotFound New patch with the opcode version attached to TT #1635
15:00 GodFather joined #parrot
15:04 dalek parrot: r46610 | mikehh++ | trunk/src/pmc/imageio.pmc:
15:04 dalek parrot: fix codetest failure - tabs in leading whitespace
15:04 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46610/
15:07 dalek rakudo: f204532 | jonathan++ | src/Perl6/Actions.pm:
15:07 dalek rakudo: Avoid emitting broken PIR for things like has ($!x, $!y).
15:07 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/f​204532cb36649521a6b43718197ff87eca0882c
15:08 Coke Jeff Goff?
15:09 Coke jgoff?
15:09 purl jgoff is speaking.
15:09 Coke seen jgoff?
15:09 purl jgoff was last seen on #perl 6 years, 164 days, 50 minutes and 51 seconds ago, saying: seen Elthek  [Dec  3 14:18:16 2003]
15:20 dalek parrot: r46611 | mikehh++ | trunk/lib/Parrot/Docs/Section/PCT_Tutorial.pm:
15:20 dalek parrot: add svn properties
15:20 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46611/
15:22 Coke going to yapc::na?
15:22 purl See "going to YAPC::NA $year"
15:22 Coke going to yapc::na 2010?
15:22 purl well, going to yapc::na 2010 is qw(jhannah rbuels cfedde apeiron dha nacmac dhoss mst chargrill kyriel triddle DrForr coke packy dukeleto kolibrie colomon jasonmay arcanez dylan doy ology waltman)
15:22 Coke ARGH.
15:22 Coke oh wait, there I IS!
15:28 PerlJam joined #parrot
15:37 dalek parrot: r46612 | coke++ | trunk/compilers/pct/README.pod:
15:37 dalek parrot: Remove references to missing languages, previous versions, and the future yet to be.
15:37 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46612/
15:38 mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#33829), fulltest) at r46611- Ubuntu 10.04 amd64 (g++)
15:57 bakkdoor joined #parrot
16:01 atrodo Do the tags from svn not get copied to the git repo on github?
16:04 Coke no clue; I'd ask dukeleto
16:10 dalek parrot: r46613 | fperrad++ | trunk (2 files):
16:10 dalek parrot: [URI] add scheme 'file'
16:10 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46613/
16:13 dalek rakudo: a7af4f0 | (Solomon Foster)++ | src/core/Complex.pm:
16:13 dalek rakudo: Reformat Complex trig functions, and delete unneeded (and incorrect) duplicates.
16:13 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/a​7af4f04d6a91a3b16a56da4625c8a54ebb929e1
16:14 bakkdoor joined #parrot
16:18 PerlJam allison++   python software foundation?  Is she trying to be a human bridge between languages that begin with P?   :-)
16:19 theory joined #parrot
16:24 dalek rakudo: f3998c4 | moritz++ | docs/running.pod:
16:24 dalek rakudo: [docs] document --stagestats, isBEKaml++
16:24 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/f​3998c460d998731b4e71bf549371710e34a80e6
16:26 dalek parrot: r46614 | fperrad++ | trunk (6 files):
16:26 dalek parrot: add an LWP library (first step: protocole file)
16:26 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46614/
16:26 dalek parrot: r46615 | fperrad++ | trunk/DEPRECATED.pod:
16:26 dalek parrot: update notice for experimental libraries
16:26 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46615/
16:38 cotto pmichaud, did you see my nopaste?
16:40 dukeleto atrodo: svn tags do not get copied to github, correct. it might be possible
16:40 atrodo dukeleto> Okay, thanks.  I just wanted to make sure I wasn't missing something
16:42 atrodo I was going to try and setup parrot as a git submodule, and that it'd be useful to pull the version based on a tagged release
16:48 Coke I wonder if Adobe's Warnock and Perl's Warnock are related.
16:50 fperrad_ joined #parrot
16:54 Andy joined #parrot
17:07 Coke ... I am amused that that send was warnocked.
17:27 davidfetter joined #parrot
17:28 LoganLK joined #parrot
17:43 iblechbot joined #parrot
17:44 pmichaud cotto: with the list of methods?  yes.  I'm changing the signatures slightly.
17:44 pmichaud I'm about to commit a first draft
18:10 * sorear returns
18:54 pmichaud pushing nqp-rx initial setting files.  these don't do anything yet, they're just for people to start looking at and commenting on.
18:54 pmichaud now I'll work on adding tests and making them active.
18:57 dalek nqp-rx: 8e501a3 | pmichaud++ |  (2 files):
18:57 dalek nqp-rx: Allow pod comments to be indented by whitespace.
18:57 dalek nqp-rx: review: http://github.com/perl6/nqp-rx/commit/8​e501a3c08bb48503dce33b695c79f09369d223f
18:57 dalek nqp-rx: 04f26c7 | pmichaud++ | src/setting/ (4 files):
18:57 dalek nqp-rx: Add some initial drafts and versions of setting files for nqp
18:57 dalek nqp-rx: users, based on versions cherry picked from bacek++, who probably
18:57 dalek nqp-rx: got them from japhb++, and others.
18:57 dalek nqp-rx: review: http://github.com/perl6/nqp-rx/commit/0​4f26c7c03dd6ed5a6872d369006c227f5e8860a
18:59 dukeleto joined #parrot
19:02 cotto_work pmichaud: If you're just including the minimum set needed by opsc, spew can go.
19:03 pmichaud I figured spew can go in for now
19:03 pmichaud I was just trying to put things where I thought they reasonably fit for the moment
19:03 pmichaud the biggest change is the signature on map()
19:03 pmichaud it now takes a slurpy list instead of a single list
19:04 pmichaud so either call it with   map({...}, |@list)   or as  @list.map({...})
19:04 pmichaud (I did that to be consistent with the other functions that use slurpies.)
19:05 pmichaud anyway, next step is to add some basic tests and get it into parrot
19:07 sorear /almost/ there
19:12 * sorear wonders if it's possible to sanely test the repl
19:13 sorear svn up; make; make install; ... rakudo: perl Configure.PL  YOUR PARROT IS TOO OLD
19:13 sorear I'm guessing it's snapshotting the revision at parrot conf time
19:15 pmichaud I'm guessing rakudo is using a different parrot
19:15 pmichaud rakudo's build process gets the parrot revision number from whatever parrot_config reports
19:16 sorear pmichaud: no, it's the same parrot.  if I just remember to reconfigure parrot every time I up it works
19:16 pmichaud ah, yes.
19:17 pmichaud I can believe that.
19:17 sorear luckily parrot rebuilds decently fast
19:17 sorear even with -O2
19:17 pmichaud if parrot's config process is setting the revision number in pconfig, then it'll be out of date.
19:17 sorear I've got a slightly strange setup
19:17 sorear I have two parrot checkouts side by side
19:18 sorear one of them is configured with --debugging=0 --optimize, the other not
19:18 sorear so I can switch my system parrot between "fast" and "gdb-friendly" with a 15-second sudo make install
19:31 Coke sorear: for "even faster" use ccache to build parrot
19:31 Coke sorear: just use --optimize --ccflags="-g"
19:31 Coke (unless that is too slow. I unno. it's fast enough for me and gdb likes it)
19:33 Coke we need someone to make friends with companies willing to donate money for grants etc.
19:34 sorear Coke: I
19:34 sorear Coke: I've had bad experiences with gdb on -O -g code
19:34 sorear I don't trust it
19:34 Coke hokay.
19:34 cotto_work 4.5 should be better for that
19:34 * Coke asks someone who works at google.
19:36 dalek nqp-rx: b9a6851 | pmichaud++ |  (4 files):
19:36 dalek nqp-rx: Automatically build nqp setting.  Add setting-test target, and first batch
19:36 dalek nqp-rx: of tests.  (Code and configuration shamelessly copied from bacek++.)
19:36 dalek nqp-rx: review: http://github.com/perl6/nqp-rx/commit/b​9a68517b085d0b8f4965b683017933a7f1b4304
19:39 davidfetter joined #parrot
19:45 dalek parrot: r46616 | coke++ | branches/rm_doc_makefile:
19:45 dalek parrot: Remove docs/Makefile, last recursive make in default build.
19:45 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46616/
19:46 cotto_work2 joined #parrot
19:46 pmichaud I have a question about ext/nqp-rx/Defines.mak ....   why are P6object.pbc, PCT/HLLCompiler.pbc, and PCT/PAST.pbc listed there?
19:47 dalek nqp-rx: d0789dc | pmichaud++ | build/Makefile.in:
19:47 dalek nqp-rx: Clean up makefile dependency for setting-test.
19:47 dalek nqp-rx: review: http://github.com/perl6/nqp-rx/commit/d​0789dc99fd33e15f5251d5335df2b992c721c60
19:47 dalek nqp-rx: e7af481 | pmichaud++ |  (2 files):
19:47 dalek nqp-rx: Add .keys, .kv, .values to Hash, fix .delete.  Add 02-hash.t tests.
19:47 dalek nqp-rx: review: http://github.com/perl6/nqp-rx/commit/e​7af481a6a7172b9c014407ddaa10309ee17e8fa
19:47 dalek nqp-rx: 8bc24ed | pmichaud++ | .gitignore:
19:47 dalek nqp-rx: More files for git to ignore.
19:47 dalek nqp-rx: review: http://github.com/perl6/nqp-rx/commit/8​bc24eda96a38a713d99841c2e13151b4f0a8d17
19:49 Coke pmichaud: checking.
19:49 cotto_work2 pmichaud, subst doesn't look like it should compile.  $replace_string is used without being defined
19:50 pmichaud cotto_work2: looking.
19:50 pmichaud it's not compiling at all.. the comments aren't matching up
19:50 Coke pmichaud: looks like they're just dependencies.
19:51 pmichaud is it normal to put dependencies there like that?
19:51 Coke ... yes?
19:51 pmichaud okay
19:52 pmichaud cotto_work: yeah, comments were causing the whole file to be ignored.  need tests :-)
19:52 Coke a lot of the .mak files are split into two parts; one for definitions that is included earlier in the main makefile, and then another for rules, which happens later.
19:52 pmichaud right, nqp is this way also
19:52 pmichaud I somewhat expected it to be pbcs that this component generated, not what it also needed
19:52 pmichaud I'll look at the rules.mak in more detail shortly
19:53 Coke it's only used in the Rules.mk - it's the deps for parrot-nqp
19:53 cotto_work an ops_pct merge tonight is starting to look promising
19:53 sorear \o/ I have lexical persistence in Rakudo now
19:53 pmichaud \o/ ~
19:53 Coke (er, and also in a cleanup step in root.in)
19:53 sorear (Again.)
19:53 sorear One tiny tiny problem though.
19:53 sorear 'say' is a my sub.
19:54 sorear I can't access it from Perl6::Compiler
19:54 Coke (looks like it's slightly abused, as the cleanup for nqp-rx shouldn't cleanup its deps too, but given how we do cleanups now, doesn't really bother anything)
19:54 pmichaud I'll see if I can disabuse it
19:56 pmichaud short break, then I'll add nqp-setting to parrot.
19:58 dalek nqp-rx: 8e3df2d | pmichaud++ | src/setting/Regex.pm:
19:58 dalek nqp-rx: Fix comment errors and $replace_string error in Regex.pm, noticed by cotto++ .
19:58 dalek nqp-rx: review: http://github.com/perl6/nqp-rx/commit/8​e3df2dc2c713e465cb433896a964fd067c328a0
19:58 dalek nqp-rx: 237fdfb | pmichaud++ |  (5 files):
19:58 dalek nqp-rx: Update nqp bootstrap-files target, add nqp-setting.pm to stage0 files.
19:58 dalek nqp-rx: review: http://github.com/perl6/nqp-rx/commit/2​37fdfbc7c69440de9b0ce924d5d20ef7447a0cd
20:06 dalek rakudo: e988665 | (Solomon Foster)++ | src/core/Num.pm:
20:06 dalek rakudo: Special case ceiling / floor for NaN, Inf, and -Inf.
20:06 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/e​98866574ddf9c517c7060bc19d3a3d481bd4040
20:23 sorear pmichaud: topic/lexical-persistence-2 in nqp-rx and rakudo is ready to be looked at
20:23 * sorear is spectesting in parallel
20:46 bluescreen joined #parrot
20:54 sorear pmichaud: pushed a corrected version which passes >0% of the spectests :)
20:55 pmichaud sorear: I'll look at them soon; might be tomorrow-ish
20:56 fperrad_ joined #parrot
20:58 pmichaud cotto_work: nqp-setting.pbc now pushed into parrot trunk (r46617).  Let me know if/when you run into any difficulties.
20:59 pmichaud bacek++, japhb++, and others++ who helped on this
20:59 pmichaud revisions to the settings files in nqp-rx itself welcomed.  tests are especially welcomed.
20:59 pmichaud afk, dinner
20:59 japhb You're quite welcome.  I noticed some definite improvements from my originals, so kudos to those who were responsible.
21:05 sorear and "eval" is broken.  unsuprising.
21:07 ttbot Parrot trunk/ r46617 MSWin32-x86-multi-thread make error http://tt.taptinder.org/file/cmdout/311500.txt ( http://tt.taptinder.org//bui​ldstatus/pr-Parrot/rp-trunk/ )
21:08 dalek parrot: r46617 | pmichaud++ | trunk (8 files):
21:08 dalek parrot: [nqp]: Add initial nqp-setting library to Parrot.
21:08 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46617/
21:18 Whiteknight nothing in parrot repo better be broken
21:24 sorear eval fixed
21:27 pmichaud I did make test before committing, should be okay.  and the stuff I was doing was all in the ext/nqp-rx dir
21:37 rurban joined #parrot
21:44 cotto_work pmichaud, thanks!  I'll test when I can, likely later this evening.
21:49 cotto_work If everything works, how does a merge tonight sound?
21:49 cotto_work Whiteknight ^
21:50 Whiteknight ?
21:50 Whiteknight merge away
21:51 cotto_work ok.
21:51 mikehh will probably have to add some svn:ignore stuff after the merge - will be around though
21:52 Whiteknight What is the lexical-persistance branch on nqp-rx?
21:54 cotto_work mikehh, thank.  Hopefully I'll be in a state to catch that.
21:54 sorear Whiteknight: my WIP
21:54 sorear for 'my' variables in the REPL
21:54 Whiteknight ah, nice
21:55 Whiteknight I tried to figure out that problem in Matrixy too, did not have much success
21:55 mikehh cotto_work: the only problem I had was with MANIFEST.generated - the test in distro_tests reported that it was not sorted - but it was as far as I could see
21:56 cotto_work I remember sorting it.
21:56 cotto_work maybe it uses a different sort than the unix 'sort' tool
21:56 mikehh I tried fiddling with it but it expected some weird things
21:57 mikehh well it uses perl and maybe theren is something different in 5.12 but it works on trunk
21:57 mikehh there is
21:59 sorear Whiteknight: Matrixy?
21:59 purl i guess Matrixy is in need of a huge rewrite anyway. retargetting to nqp-rx would be a good idea
21:59 sorear retargetting to nqp-rx will get you lexical persistence almost for free
21:59 sorear also, buts
21:59 sorear bugs
21:59 Whiteknight sorear: my (now stalled) attempt to write a Matlab compiler for Parrot
21:59 * sorear has diffuse spectest failures
21:59 * Coke removes "experimental" tickets from the Active Tickets Queue.
21:59 Whiteknight I like buts in my software
22:02 Coke pmichaud: hey, as long as you're hacking on nqp-rx, there's a bunch of tickets. =-)
22:06 Coke NotFound: you were just near #1483, yes? (can you check out that patch?)
22:11 seatek joined #parrot
22:30 dalek parrot: r46618 | coke++ | trunk/DEPRECATED.pod:
22:30 dalek parrot: Add a public notice regarding this item from six months ago.
22:30 dalek parrot: review: http://trac.parrot.org/parrot/changeset/46618/
22:33 ttbot Parrot trunk/ r46618 MSWin32-x86-multi-thread make error http://tt.taptinder.org/file/cmdout/311571.txt ( http://tt.taptinder.org//bui​ldstatus/pr-Parrot/rp-trunk/ )
22:35 dalek TT #994 closed by coke++: Fix SVN properties for git-svn users
22:35 dalek TT #994: http://trac.parrot.org/parrot/ticket/994
22:35 sorear argh I'm still failing random spectests
22:37 sorear oh, I broke eval again while fixing something
22:42 clinton joined #parrot
22:51 dalek TT #629 closed by coke++: exception from :immediate sub bus errors.
22:51 dalek TT #629: http://trac.parrot.org/parrot/ticket/629
22:58 Psyche^ joined #parrot
23:05 pmichaud Coke: (nqp-rx tix)  yes, unfortunately not all of them are tagged with nqp.
23:06 pmichaud but I guess/hope I own most of them so I suppose I can find them that way.  :)
23:12 tetragon joined #parrot
23:13 jan joined #parrot
23:17 bacek ~~
23:17 cotto_work hio bacek.
23:17 cotto_work It's looking like tonight will be a good time for the ops_pct.
23:17 bacek cotto_work, excellent!
23:18 bacek pmichaud, can you cherry-pick this commit http://github.com/bacek/nqp-rx/commit/9​2530624bff2c45a70da3129dfd632c43d0d90da ?
23:29 Whiteknight we have an ETA? I have a LOT of testing I want to do on a few systems after the merge
23:30 cotto_work before I go to sleep (or failing that, after the release)
23:30 cotto_work I'm having some friends over tonight, so it won't be until fairly late.
23:31 cotto_work I'd guess 11:30-ish my time.
23:31 cotto_work does that work?
23:32 Whiteknight can't do it now?
23:32 dalek tracwiki: v169 | jimmy++ | WikiStart
23:32 dalek tracwiki: http://trac.parrot.org/parrot/wiki/W​ikiStart?version=169&amp;action=diff
23:32 dalek tracwiki: v170 | jimmy++ | WikiStart
23:32 dalek tracwiki: removed PMCUnionDeprecationTasklist, it's already done
23:32 dalek tracwiki: http://trac.parrot.org/parrot/wiki/W​ikiStart?version=170&amp;action=diff
23:33 cotto_work I can't.  If bacek can he's welcome to.
23:33 Coke seems like anyone with a desire for karma could merge it.
23:34 bacek It's very hard to merge branches in git-svn if they were merged previously with straight svn...
23:35 cotto_work ic
23:35 cotto_work those days are numbered
23:37 bacek hmmm
23:37 bacek Only few conflicts in MANIFEST(.*) and gen/makefiles/root.in
23:37 bacek Testing now
23:37 cotto_work don't forget to update opsc files to depend on nqp-setting.pbc instead nqp-settings.pbc
23:44 bacek erm.... find_codepoint...
23:46 cotto_work did you bootstrap the ops?
23:47 bacek nope
23:47 cotto_work probably a good idea
23:47 purl Of COURSE it's a good idea!  I thought of it!
23:47 bacek I have to "boostrap" them in trunk.
23:51 bacek Sigh... nqp-rx crashing on compiling setting...
23:59 Whiteknight yay! I finally have Ubuntu 10.04 up and running

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

Parrot | source cross referenced