Camelia, the Perl 6 bug

IRC log for #parrot, 2009-04-01

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:04 bacek_ joined #parrot
00:09 AndyA joined #parrot
00:18 mikehh Happy All Fools Days - for those that have passed into April - later for those still in March
00:21 mikehh on a more serious note I  am failing (well Parrot, actually) t/op/trans.t (test 13 atan2) on the jit core
00:23 rg which platform?
00:23 purl I'm running on OS/2 on an Atari, can you help?
00:26 mikehh_ joined #parrot
00:27 rg could be related to TT #38 which was considered fixed 6 days ago
00:28 mikehh_ there have been some ,ods to atan recently - I wonder if they might be relevant
00:28 Coke purl++
00:28 mikehh_ s/,ods/mods/
00:30 GeJ Good morning everyone
00:32 mikehh_ I have recently been working with Ubuntu Intrepid Amd64 where make fulltest passes but obviously it doesn't us the jit core
00:32 tetragon joined #parrot
00:33 mikehh_ Gej: Good morning
00:37 mikehh_ rg: sorry got disconnected - this is oon Kubuntu Intrepid i386
00:38 rg hmm kid51 didn't say which linux he tested, but it worked for him.
00:39 mikehh joined #parrot
00:40 rg i take it it's passing without jit?
00:40 mikehh All the other cores pass
00:41 mikehh including smolder 19505 and 19530
00:41 rg i guess you'll want to reopen TT #38 with details about your platform
00:41 mikehh ok I will
00:42 cotto tewk, is that patch a wip or ready to apply (in addition to the changes in the patch I posted)?
00:44 japhb Infinoid: it seems to me the fundamental problem with a transparent r/w mirror is that the trust system generally used for git and SVN is different -- and the SVN password system is not forwardable like git ssh keys are; the git mirror would have to know everyone's SVN passwords, which just wouldn't fly.
00:45 japhb The alternative is to have the git -> SVN commits be made by a single aggregating user that marks commit logs with the real commiter's username.
00:45 Coke (that idea)--
00:45 japhb Which pretty much makes svn blame useless.
00:45 cotto my svn password is hunter2, if it helps
00:45 Infinoid japhb: Yes, that's a problem.
00:45 Coke if he set up that scheme, I'd push to revoke his cla. =-)
00:46 japhb heh
00:46 * Coke hopes cotto is lying.
00:46 * cotto is
00:46 japhb Coke: my point was that basically both alternatives are unusable
00:47 japhb ... but I sense you already knew that.  :-)
00:49 japhb The trees side of my personality says, "Well, can we get the SVN repo to authenticate with ssh keys?"  The forest side says, "If we just bit the bullet and switched to git, this wouldn't be a problem at all ...."
00:51 japhb How are the Rakudo core doing with their git switch?  Getting more comfortable now?
00:52 japhb (I ask because I recall the last time switching Parrot to git came up, someone (allison?) raised the idea of waiting to see how the Rakudo guys fared ....)
00:53 Infinoid The problem is, I think parrot branches a lot more than rakudo does, so the benefit of git isn't as noticable there
00:53 japhb there == ?
00:54 Infinoid rakudo
00:54 japhb yes, I agree
00:54 Infinoid They still don't seem to have learned why you'd rebase rather than pull.  I suspect that annoys the heck out of them.
00:55 japhb You know, the more I think about it, the more it seems that Parrot shouldn't just move to git, it should move to a social git system like github (or whatever the self-hosted version is).
00:56 Infinoid I do think the web-based forking and pull request stuff are interesting ideas.  Haven't had much chance to try them out tho
00:57 japhb There's a strong desire to have feature branches be public, but I'm not sure we can convince people to simultaneously learn git from scratch, and learn how to manage publicly-visible branches
00:57 Infinoid When I was starting out with git, just getting a commit pushed to the server took me several days of head-scratching
00:57 Infinoid (that was on a repo.or.cz project)
00:57 japhb Infinoid: very nice.  Tried github recently to work a patch for Web.pm, and it was very self-obvious
00:58 Infinoid awesome
00:58 japhb Fork, clone, hack, hack, hack, push, send pull request, done
00:58 Infinoid Do you have to clean up the fork afterwards?
00:59 japhb You can nuke it, or leave it around.  Free accounts have a certain base amount of disk space for forks, and if you were bumping up against it, you might care ... but even then, they're very clear that users working on open source projects need only send an email, and they can get their repo quota raised.
01:01 Infinoid Cool.  I was kinda wondering if they'd auto expire old forks after some period of inactivity, or something
01:01 Infinoid Sounds like they want to use gobs and gobs of disk space... more power to them
01:02 japhb Infinoid: mmm, good question ... I didn't delve that deeply into the docs.  But I doubt they care that much; bare git repos are, all things considered, tiny.
01:05 Infinoid Try saying that after the 600th clone of the parrot repo :)
01:06 japhb Infinoid: only people who fork use disk space.  r/o cloners are free.  :-)
01:06 Infinoid true
01:06 mikehh I tried to re-open tt #38 - but iTrac says I do not have permission to change fields
01:07 mikehh should I open a new ticket
01:07 japhb Infinoid: besides which, I'd bet github would see 600 Parrot forkers as 600 opportunities to upsell to pay accounts.  :-)
01:08 Infinoid mikehh: Yes please.
01:08 allison mikehh: is it exactly the same failure, or a different failure in the same test?
01:09 mikehh I think it is the same - let me check
01:09 Infinoid Hmm, that is the same test
01:09 Infinoid [17:21] < mikehh> on a more serious note I  am failing (well Parrot, actually) t/op/trans.t (test 13 atan2) on the jit core
01:10 mikehh I get the same results as the original report
01:11 kid51 joined #parrot
01:11 mikehh fails with noy ok 1 through 12 ok 13 etc
01:11 mikehh s/noy/not/
01:11 allison mikehh: read the followup, and especially the reason jkeenan closed the ticket,
01:12 Infinoid mikehh: Are you saying everything before atan2 fails, but atan2 passes?
01:13 allison mikehh: if it still seems the issue isn't covered, suggest a new ticket linking back to the closed ticket as possibly related
01:13 kid51 Infinoid:  IIRC, that is correct.  atan2 started to pass; lots of other stuff did not
01:13 dalek parrot: r37832 | jkeenan++ | branches/dir_simplify (2 files):
01:13 mikehh no it is test 13 (atan2) that gives those results
01:13 dalek parrot: Remove long-deprecated attributes in init::install: bin_dir, include_dir, lib_dir.  See https://trac.parrot.org/pa​rrot/ticket/524#comment:5.
01:13 dalek parrot: review: https://trac.parrot.org/parrot/changeset/37832/
01:14 rg kid51: which linux were you using to test tt #38?
01:15 kid51 rg:  Yes:  plain-jane Linux i386.  (I also have Darwin PPC, but that's much slower.)
01:15 Infinoid kid51: Fedora, by any chance?
01:16 kid51 No.  Debian stable.
01:16 Infinoid Ok, thanks.  (I was thinking of a different bug in the same hash bucket)
01:16 mikehh kid51: does it still pass that test now on the jit core
01:17 kid51 Don't know.  Just got back from doing laundry, and have been working on a different issue.
01:17 kid51 Have to go to dinner.
01:18 kid51 Can you post any issue in ticket?  I will try to respond within a day or so.
01:18 mikehh I have seen some work done on atan (supposedly on Solaris) recently - that might have had an effect?
01:18 rg mikehh: there was nothing committed
01:19 rg the atan2 problem on solaris is completely unrelated.
01:19 kid51 mikehh:  Possibly.  I don't claim to understand these issues.  I merely report test results and try to monger tickets.
01:23 timbus joined #parrot
01:23 dalek tracwiki: v144 | allison++ | ParrotRoadmap
01:23 dalek tracwiki: https://trac.parrot.org/parrot/wiki/Par​rotRoadmap?version=144&amp;action=diff
01:24 shorten dalek's url is at http://xrl.us/benbkv
01:26 rg i can tell you that the test still succeeds on freebsd. no linux here ;)
01:27 mikehh ok opened TT #530
01:28 TiMBuS joined #parrot
01:32 mikehh the test passes on all the other core tests
01:32 mikehh just fails on the jit core
01:33 mikehh all the other jit core tests pass
01:34 mikehh I haven't run the rest of fulltest but did test the remaining cores r and S
01:34 rg mikehh: yes, jit is very special. on freebsd 3 other tests in trans.t fail for no reason that i can think of
01:36 mikehh there were a few failures on the jit core before the 1.0 release, but most of those were TOTOed including this one
01:37 mikehh well I am not sure when this one was TODOed
01:38 mikehh but at the moment all the other jit core tests PASS or are SKIP or TODO
01:39 mikehh that is on Kubunto Intrepid i386
01:40 mikehh on Ubunto Intrepid Amd64 all the tests pass but that does not test the jit core
01:40 mikehh s/Ubunto/Ubuntu/
01:43 allison mikehh: the test failure you're reporting is marked as TODO?
01:44 mikehh allison: No - it was at r37704m but NOT now
01:45 rg no, it was un-todoed 6 days ago when we considered tt #38 fixed
01:47 allison so, it boils down to "we've found a platform where this problem we thought was fixed isn't fixed"
01:47 mikehh r37704 was the last time I ran make fulltest on this platform - I have been working on Amd64 the last week or so
01:48 allison mikehh: I'll reopen the ticket, you can add details
01:48 mikehh I just switched back today (well yesterday now)
01:49 allison mikehh: okay, reopened, please submit an explanation to the ticket
01:49 rg allison: <mikehh> ok opened TT #530
01:50 davidfetter allison, could you send me whatever you've got--even the stuff you'd planned to throw away--re: sandboxing? :)
01:52 allison rg: okay, adding link to TT #530 from TT #38 and reclosing it.
01:52 allison davidfetter: I'd have to get it working before it would do you any good
01:53 davidfetter Somebody(TM) might get inspired to fix it, or might take it as a big red warning sign. either one's good
01:56 allison davidfetter: looking at it now... no, I don't want to give anyone ideas. It was important to try it out, but isn't the way I want it implemented.
01:57 allison davidfetter: I don't want to inspire anyone to do it that way. :)
01:57 davidfetter then it falls back to "warning to others"
01:57 * davidfetter pictures a sign hung on the code reading, "used light-weight threads improperly" or such
01:59 allison davidfetter: oh, that's easy, sandboxes shouldn't be implemented as threads
02:00 davidfetter heh
02:00 allison davidfetter: because they have to be usable in unthreaded contexts too
02:00 davidfetter ok
02:00 davidfetter so apart from broken code, what do you have?
02:01 allison davidfetter: the design document
02:01 davidfetter where's that?
02:01 purl That's the money you could be saving with Geico.
02:01 allison http://docs.parrot.org/parrot/latest/h​tml/docs/pdds/pdd18_security.pod.html
02:01 shorten allison's url is at http://xrl.us/benbq2
02:02 allison or docs/pdds/pdd18_security.pod in the repository
02:02 davidfetter ok, thanks :)
02:02 davidfetter anything else?
02:02 purl i heard anything else was violating dry. and we're in the wrong channel.
02:06 allison what else would you like?
02:09 allison davidfetter: the PDD is very general, partly because we're poking into untested waters here
02:10 allison davidfetter: if this were a company, what I'd do next is start interviewing programmers with a strong security background, specifically ones who had implemented similar systems in the past
02:10 allison davidfetter: then talk with them about the ways parrot is different, and how the security model will be appropriately different to fit
02:11 allison davidfetter: if they get it, and come back with a few good suggestions for improving the idea based on their experience, then they're hired
02:11 davidfetter heh
02:11 davidfetter ok
02:12 allison davidfetter: since this isn't a company, I can't hire anyone, but I'm looking for volunteers in a similar direction :)
02:12 rg how many of those do you think are out there?
02:12 allison rg: I know a few, but they're all working insane hours
02:13 allison rg: so, I'd say they're relatively rare
02:14 allison but then, so are virtual machine inventors with an interest in pushing the state of the art
02:17 Infinoid ...which are probably more plentiful than virtual machine inventors with their heads stuck in the 70's
02:18 kid51 Anyone here familiar with tools/dev/mk_language_shell.pl ?
02:19 kid51 I ask because it's one of the few files potentially affected by https://trac.parrot.org/parrot/ticket/524
02:21 Infinoid I'm not, but it looks relatively easy to test
02:22 kid51 After eliminating/transforming some deprecated configure attributes, I ran it (for the 1st time) in my sandbox of a branch.
02:22 kid51 I got these error messages:
02:23 kid51 Creating src/pmc/Makefile
02:23 kid51 Can't open perl script "/usr/local/lib/parrot/1.0.0-de​vel/tools/dev/gen_makefile.pl": No such file or directory
02:23 kid51 Creating Makefile
02:23 kid51 Can't open perl script "/usr/local/lib/parrot/1.0.0-de​vel/tools/dev/gen_makefile.pl": No such file or directory
02:23 kid51 Creating src/ops/Makefile
02:23 kid51 Can't open perl script "/usr/local/lib/parrot/1.0.0-de​vel/tools/dev/gen_makefile.pl": No such file or directory
02:23 allison kid51: the versions with the underscores have been listed as deprecated for a while now in init/install.pm
02:23 kid51 Now, I don't have an installed parrot.  But these messages suggest that the expectation is that this will only be run with an installed parrot.  Is that correct?
02:23 allison why are the versions without the underscores being deprecated?
02:24 kid51 Err, I think it's the other way around.
02:24 kid51 fperrad reversed himself after Andy D objected.
02:24 allison kid51: yes, the mk_language_shell is for an installed parrot
02:25 allison kid51: ah, okay, I had a similar confusion reading the first part of the ticket
02:25 kid51 I will post patch
02:30 kid51 patch posted in https://trac.parrot.org/parrot/ticket/524
02:35 janus joined #parrot
02:37 cognominal joined #parrot
02:51 Andy joined #parrot
03:07 darbelo joined #parrot
03:15 davidfetter joined #parrot
04:07 diakopter joined #parrot
04:07 TimToady joined #parrot
04:09 dukeleto can anybody flesh out the parrot project ideas on the gsoc project ideas page: http://www.perlfoundation.org/pe​rl5/index.cgi?gsoc_2009_projects
04:09 shorten dukeleto's url is at http://xrl.us/bekq26
04:10 dukeleto they are a bit light on details
04:11 dukeleto also, does anyone know of any students working on a parrot proposal for gsoc? I don't see one yet
04:11 dukeleto the deadline is this Friday, so this has me slightly concerned
04:11 tewk dukeleto: I've got one that will be in tomorrow
04:12 dukeleto tewk: great! feel free to send it the tpf-gsoc-students list for pre-review
04:13 jimmy joined #parrot
04:14 tewk cotto, its a complete patch and fixes t/pmc/nci_66.pir but your patch has other nci.t failures.
04:21 darbelo dukeleto: I'm also working on a proposal, should be mostly-ready in a few hours.
04:21 dukeleto darbelo: good to hear!
05:00 dukeleto i am reading about Parrot_callback_* now, that is where the fun begins
05:02 dukeleto i must say, the callback section of pdd16_native_call.pod is a lot less scarier that reading the perlcall manpage for the first time
05:03 tetragon joined #parrot
05:04 dalek parrot: r37833 | Infinoid++ | trunk (4 files):
05:04 dalek parrot: Apply patches from bacek++ in TT #527:
05:04 dalek parrot: * Implement lookup table for PIO_SOCK_* constants
05:04 dalek parrot: * Update examples/io/http.pir to use constants.
05:04 dalek parrot: review: https://trac.parrot.org/parrot/changeset/37833/
05:08 Patterner joined #parrot
05:11 darbelo I believe that having internals "Less scary than Perl 5" is one of the goals that got parrot started.
05:19 Infinoid I, for one, like pmcs quite a lot more than I like XS.
05:19 dukeleto me too :)
05:20 dalek tracwiki: v82 | gaurav++ | Languages
05:20 dalek tracwiki: Added "porcupine" (Pascal)
05:20 dalek tracwiki: https://trac.parrot.org/parrot/wiki/​Languages?version=82&amp;action=diff
05:20 shorten dalek's url is at http://xrl.us/bencbp
05:24 dalek tracwiki: v83 | gaurav++ | Languages
05:24 dalek tracwiki: Gave Smolder links names (so they take up less space)
05:24 dalek tracwiki: https://trac.parrot.org/parrot/wiki/​Languages?version=83&amp;action=diff
05:24 shorten dalek's url is at http://xrl.us/benccd
05:24 dalek parrot: r37834 | Infinoid++ | trunk/src/io (3 files):
05:24 dalek parrot: [io] Consolidate socket constant mappings
05:24 dalek parrot: * The mappings from parrot socket constants to system socket constants are identical between the unix and win32 implementations.  Consolidate them into the api source file.
05:24 dalek parrot: review: https://trac.parrot.org/parrot/changeset/37834/
05:29 chromatic joined #parrot
06:02 bacek_ Infinoid: your last commit was probably wrong...
06:08 uniejo joined #parrot
06:14 Infinoid Can you explain why?
06:14 Infinoid I can revert if needed, but I don't really like having two copies of the same thing if we can help it
06:16 Infinoid I do see that I missed removing the lookup parts from Parrot_io_socket_win32
06:23 dalek parrot: r37835 | Infinoid++ | trunk/src/io/socket_win32.c:
06:23 dalek parrot: [io] Remove an extra bit of code I missed in r37834.
06:23 dalek parrot: review: https://trac.parrot.org/parrot/changeset/37835/
06:37 bacek_ Infinoid: socket_api doesn't #include <sys/socket.h> or win32 version. So all this PF_* ans SOCK_* are actually undefined
06:37 bacek_ (Or at least can be undefined)
06:44 flh joined #parrot
06:57 Tene_ joined #parrot
07:03 tuxdna joined #parrot
07:24 uniejo joined #parrot
07:26 mikehh I left male fulltest_all running when I decided to have some sleep earlier - it failed a second test at r37831 - t/manifest/02-regenerate_file.t
07:29 mikehh t/manifest/02-regenerate_file.t - Failed 1/12 subtests -  Failed test 'No need to regenerate MANIFEST' - at line 43
07:36 mikehh as well as the t/op/trans.t - failed test 13 (atan2) in the jit-core I reported earlier - TT #530
07:37 mikehh that is on Kubuntu Intrepid i386
07:38 iblechbot joined #parrot
07:52 jq python april fools joke: they are going to target parrot as their sole implementation (http://www.python.org/dev/peps/pep-0401/)
07:53 jq sad to hear that only for an april fools day
07:55 moritz aye
07:59 cotto I like it.  It means more exposure for Parrot.
08:00 cotto Plus it's not like Parrot 1.0 is at the stage where it's ready to host a production-ready language implementation.
08:01 cotto I of course look forward to the time when it is, but there's work to be done between now and then.
08:02 moritz naiively asked, why does it take so long? there seem to be other open source VMs that were developed in much shorter time frames
08:02 moritz is parrot simply much mor ambitious?
08:03 cotto I think it's a combination of ambition and poor initial management, but I'm not certain.
08:03 * cotto notes that he seems unusually negative atm.
08:04 darbelo joined #parrot
08:04 Chile is there a particular area that parrot needs the most work on?
08:05 moritz calling conventions, garbage collector (IMHO)
08:05 cotto Chile, how ambitious are you?
08:05 cotto ;)
08:05 Chile cotto, not commiting to anything, I'm just curious if the areas where stuff needs to get done are areas that I'd be interested in working on :)
08:06 cognominal joined #parrot
08:06 cotto Take a look at https://trac.parrot.org/parrot/wiki/ParrotRoadmap and see if any unclaimed items strike your fancy.
08:07 cotto You could also help kj with pirc, although I don't know how well work on that project parallelizes atm.
08:07 Tene_ cotto: I haven't seen any work on pirc in quite a while...
08:08 cotto Last I heard, he was trying to increase its bus number.
08:09 cotto which was 1
08:09 Chile ouch
08:09 Tene_ I started digging iinto it, but I'm having trouble figuring out what some of it is doing wrong
08:10 Tene_ going to harass tewk about it tomorrow
08:10 Chile what's the motivation for pirc over imcc?
08:10 moritz imcc is said to be fairly unmaintainable
08:10 cotto It's one of the less pleasant parts of Parrot, along with Keys and a couple others.
08:11 Chile alright
08:12 cotto like moritz said, maintainability is the main concern
08:23 cotto Chile, also https://trac.parrot.org/parro​t/wiki/ParrotLongTermRoadmap
08:24 Chile yeah, looking there
08:35 Chile what needs to be done for the GC?
08:38 cotto ping wknight8111 about that
08:38 purl I can't find wknight8111 in the DNS.
08:38 Chile also, on the long-term map, there's a reference to an LLVM JIT backend.. what's the deal with that?
08:40 cotto The current jit code needs a bunch of cpu-specific stuff.  I think the idea there is to generate LLVM's intermediate representation.  Then LLVM can generate JIT'd code.
08:41 cotto and Parrot doesn't have to worry about specific cpus
08:41 Chile is anyone working on that, or is it a future thing? I don't see anything in the CVS tree referencing llvm at the moment
08:42 moritz it's a future thing
08:42 moritz I think tewk has a GSOC proposal for that
08:42 moritz (google summer of code)
08:42 uniejo joined #parrot
08:44 Ademan joined #parrot
08:44 Chile ah, alright. when do the SOC decisions get made?
08:45 moritz the proposals have to be handed in this or next week, I think
08:45 moritz I don't know what's the timeline for the discussion, but I'd guess it's decided in three weeks
08:46 darbelo Deadline for proposals is Friday.
08:47 darbelo Accepted proposals are announced April 20.
08:47 Chile alright
09:05 riffraff joined #parrot
09:29 uniejo joined #parrot
09:42 flh joined #parrot
09:51 mikehh joined #parrot
09:53 flh hi everyone! I would need some help about (I guess) parrot calling conventions: I do not understand how I can fetch arguments inside the invoke vtable function
09:55 flh I'm trying to write a pmc which does curried subs (it doesn't seem to exist already)
09:57 cotto flh, from C or PIR?
09:58 flh from C
09:59 flh unfortunately, overriding invoke from PIR is useless due to ticket #103
09:59 cotto Is src/pmc/sub.pmc helpful?
10:00 flh I'm not sure, invoke in sub.pmc doesn't seem to do anything with arguments
10:01 bacek flh: why you trying to do it from C?
10:02 flh because invoke from PIR gets no "self", so I cannot access previously stored arguments
10:03 flh which, I agree, is quite unfortunate since I know how to write a curried-sub class in PIR
10:05 flh it looks like my answer is in src/call/pcc.c, but I don't know how I am supposed to use these functions
10:06 moritz flh: have you looked at pdd03 already?
10:08 flh pdd03 is mainly about PIR (but I've learned about signatures, which also appear in C)
10:08 moritz ok; it was worth a try :-)
10:08 moritz maybe you'll have better luck when our US friends are awake again :-)
10:10 flh ok, that's what I thought, but... it was worth a try :-) (maybe I should simply send an email to parrot-dev)
10:11 moritz flh: that would be easier, yes
10:12 cotto Pressure to fix known bugs doesn't hurt. ;)
10:17 cotto flh++ #doing stuff that runs into bugs
10:17 cotto night
10:19 flh cotto, by the way, I may have a quick fix for #103 if I understand how arguments work from C...
10:21 mikehh joined #parrot
10:36 Casan joined #parrot
11:46 ruoso joined #parrot
11:56 cognominal joined #parrot
12:06 msmatsko joined #parrot
12:28 cognominal joined #parrot
12:35 rg1 joined #parrot
12:38 iblechbot joined #parrot
12:45 Infinoid msg bacek socket_api.c includes io_private.h, which does include sys/socket.h or winsock.h, so this should work (and tests ok on linux).  Please lemme know if you see failures anywhere else
12:45 purl Message for bacek stored.
13:41 Coke if something has UNUSED() but not SHIM() is that slightly wrong?
13:41 ruoso joined #parrot
13:45 Khisanth joined #parrot
13:59 Tene joined #parrot
14:02 Infinoid according to docs/dev/c_functions.pod, the distinction is that SHIM() will never use it, but UNUSED() doesn't use it yet.
14:02 Infinoid how you would test for something like that... I dunno.
14:21 pmichaud Okay, I have a problem that is really stumping me.
14:29 rg care to elaborate?
14:29 pmichaud yes, sorry -- got distracted.
14:29 pmichaud nopaste coming.
14:30 amoc joined #parrot
14:30 nopaste "pmichaud" at 72.181.176.220 pasted "Code to determine codepoint from unicode character name" (9 lines) at http://nopaste.snit.ch/16033
14:30 Andy joined #parrot
14:31 pmichaud I'm using ICU to be able to translate things like   'LATIN CAPITAL LETTER A'  into codepoints like 65
14:32 nopaste "pmichaud" at 72.181.176.220 pasted "sample usage of charFromName method" (15 lines) at http://nopaste.snit.ch/16034
14:32 pmichaud In the METHOD code, see that (unused) declaration of buffer?
14:32 pmichaud If I get rid of it, the code stops working.
14:33 particle joined #parrot
14:34 pmichaud If I change the size of the buffer, the code stops working in different ways.
14:35 shucho_ joined #parrot
14:35 nopaste "pmichaud" at 72.181.176.220 pasted "results when buffer commented out" (5 lines) at http://nopaste.snit.ch/16035
14:37 nopaste "pmichaud" at 72.181.176.220 pasted "results when buffer set to 25 instead of 256" (5 lines) at http://nopaste.snit.ch/16037
14:38 rg i'm not so sure about the c METHOD syntax. where does it take the return type from?
14:38 pmichaud oh.... wait
14:39 pmichaud I might know the problem.  err is uninitialized.
14:39 rg err looks like (the equivalent of) an out parameter to me
14:39 pmichaud it's actually an inout
14:40 pmichaud that's likely what I forgot.
14:40 pmichaud ICU functions check err before doing anything and immediately abort if there was an error from a previous step.
14:40 pmichaud that seems to do it.
14:41 pmichaud Excellent.
14:41 pmichaud Thanks.  Sometimes just describing the problem helps to track it down.
14:41 rg i know. that's the main reason i asked you to elaborate ;)
14:41 Coke that happened to me yesterday for $work. killed at least an hour, spent 2m explaining it, boom.
14:41 pmichaud at least I only lost about 30-45 mins.
14:46 particle they're not lost, they're in the 'character' and 'experience' bin
14:48 pmichaud my character bin often feels like /dev/null :-)
14:53 dalek rakudo: e5f4a34 | pmichaud++ | docs/spectest-progress.csv:
14:53 dalek rakudo: spectest-progress.csv update: 340 files, 8049 passing, 0 failing
14:53 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/e​5f4a34b273511f5c3c5a0ad650943fcccc4827f
14:54 shorten dalek's url is at http://xrl.us/bendqd
15:05 Psyche^ joined #parrot
15:23 AndyA joined #parrot
15:25 alvar joined #parrot
15:33 davidfetter joined #parrot
15:34 Theory joined #parrot
15:34 ilia joined #parrot
15:51 dalek parrot: r37836 | pmichaud++ | branches/pge1:
15:51 dalek parrot: [pge]:  Remove obsolete branch.
15:51 dalek parrot: review: https://trac.parrot.org/parrot/changeset/37836/
15:51 dalek parrot: r37837 | pmichaud++ | branches/p6strings:
15:51 dalek parrot: Branch to refactor perl6-string parsing in pge and pct.
15:51 dalek parrot: review: https://trac.parrot.org/parrot/changeset/37837/
15:53 * Coke wonders if ack can do m//sm
15:54 Andy no
15:54 Andy it is strictly line-oriented.
15:54 Andy nor will it do s/// as I had to turn down last night.
15:57 Coke does it respect $/ or is the line bit hardcoded?
15:57 Coke (can I abuse it, basically.)
15:57 Coke ... nevermind.
16:29 barney joined #parrot
16:51 ilia_ joined #parrot
16:52 darbelo joined #parrot
16:54 dalek parrot: r37838 | jkeenan++ | trunk/MANIFEST:
16:54 dalek parrot: Update MANIFEST to correct failure to add file.  (mikehh:  This should fix 'make manifest_tests'.)
16:54 dalek parrot: review: https://trac.parrot.org/parrot/changeset/37838/
17:05 chromatic joined #parrot
17:13 barney Why does nobody use hllmacros.pir ?
17:17 dalek parrot: r37839 | barney++ | trunk/runtime/parrot/include/README:
17:17 dalek parrot: [docs] mention hllmacros.pir
17:17 dalek parrot: review: https://trac.parrot.org/parrot/changeset/37839/
17:19 Coke barney: ISTR particle thought it would be useful; no one else agreed.
17:20 dalek parrot: r37840 | barney++ | trunk/t/library/hllmacros.t:
17:20 dalek parrot: no need for "include" in .include
17:21 dalek parrot: review: https://trac.parrot.org/parrot/changeset/37840/
17:22 barney I like the conditionals, when writing 'if' in PIR I always have to readjust my way of thinking
17:34 dalek parrot: r37841 | barney++ | trunk/runtime/parrot/include/README:
17:34 dalek parrot: [docs] mention test_more.pir
17:34 dalek parrot: review: https://trac.parrot.org/parrot/changeset/37841/
17:45 Coke I don't think anyone is planning on removing it. enjoy.
17:48 barney I might use it in Pipp
17:48 flh joined #parrot
17:51 Coke anyone know if headerizer barfs on pirc ?
17:51 Coke I'm seeing odd results testing for function docs in there, guessing a headerizer issue.
17:54 darbelo joined #parrot
17:54 Infinoid I don't think headerizer covers pirc by default
17:56 Coke maybe not; c_function_docs.t calls a headerizer function to get the function docs, though.
17:57 Infinoid "make headerizer" completely ignores pirc.  Trying to run headerizer on pirc by hand results in a "can't find HEADERIZER HFILE directive"... I'm not sure anyone's ever tried to run headerizer on pirc before
17:57 dalek parrot: r37842 | barney++ | trunk (25 files):
17:57 dalek parrot: Do not specify 'include' when including test_more.pir.
17:57 dalek parrot: review: https://trac.parrot.org/parrot/changeset/37842/
17:58 Infinoid is the test just using the parser end of headerizer?
17:58 Coke my @function_decls = $headerizer->extract_function_declarations($buf);
17:59 Coke the extracted function declarations include junk for compilers/pirc/src/bcgen.c, e.g
17:59 Coke (you can see by running t/codingstd/c_function_docs.t compilers/pirc/src/bcgen.c
18:00 dalek parrot: r37843 | coke++ | trunk/tools/build/headerizer.pl:
18:00 dalek parrot: [t/docs] these todos already have tests in t/codingstd
18:00 dalek parrot: review: https://trac.parrot.org/parrot/changeset/37843/
18:03 amoc joined #parrot
18:04 Infinoid looks pretty picky about the prototype.  does headerizer update the pod prototypes?
18:05 Infinoid I managed to fix one of the errors with:
18:05 Infinoid -=item C<PMC *
18:05 Infinoid -generate_multi_signature(bytecode * const bc)>
18:05 Infinoid +=item C<static PMC *generate_multi_signature(bytecode * const bc, multi_type * const types, unsigned type_count)>
18:05 Coke what looks picky? headerizer, or the function docs test?
18:05 Infinoid the function docs test
18:05 Coke yes. it's picky.
18:05 Infinoid (which it absolutely should be)
18:05 Coke I have no idea how it ever passed anything. =-0
18:10 Coke Infinoid: can you run t/codingstd/c_indent.c ?
18:10 Coke (looks like it might be your fault. =-)
18:11 dalek parrot: r37844 | coke++ | trunk/examples/io/http.pir:
18:11 dalek parrot: [t] pass file_metadata.t
18:11 dalek parrot: review: https://trac.parrot.org/parrot/changeset/37844/
18:11 Infinoid "Cannot determine source for t/codingstd/c_indent.c"
18:11 Infinoid That's new.  It worked last night
18:11 Coke er, .t
18:11 Coke my bad.
18:12 Infinoid oh, that looks better.
18:12 Infinoid Hmm, I fixed all of those last night.  What changed?
18:12 Coke did you not commit?
18:12 Infinoid I don't know.  I'll fix it now
18:12 alvar joined #parrot
18:12 Coke danke.
18:13 Coke *ACHOOOOOOOOOOOOOOOOOOOOOOO*
18:21 dalek parrot: r37845 | coke++ | trunk (15 files):
18:21 dalek parrot: [t/docs] fixup more function docs (mostly signature fixups)
18:21 dalek parrot: some new boilerplate. 1075 functions left
18:21 shucho joined #parrot
18:21 dalek parrot: review: https://trac.parrot.org/parrot/changeset/37845/
18:22 Coke man is that boring.
18:24 chromatic We could find you some exciting JIT bugs instead.
18:24 cotto It's good to do if you don't have the brainpower for anything else.
18:25 dalek parrot: r37846 | Infinoid++ | trunk (56 files):
18:25 dalek parrot: [cage] Headerizer updates.
18:25 dalek parrot: review: https://trac.parrot.org/parrot/changeset/37846/
18:26 Infinoid uck
18:26 Infinoid I need to revert r37846, it just undid all the shiny documentation updates Coke committed
18:28 cotto ha ha (although I'm not sure at whom)
18:28 dalek parrot: r37847 | Infinoid++ | trunk/src/io/socket_api.c:
18:28 dalek parrot: [cage] c_indent.t fixes.
18:28 dalek parrot: review: https://trac.parrot.org/parrot/changeset/37847/
18:28 cotto </nelson>
18:32 Coke Infinoid: how could you have reverted my updates in the first place?
18:32 dalek parrot: r37848 | Infinoid++ | failed to fetch changeset:
18:32 dalek parrot: Revert r37846 for now, it undid all of Coke++'s hard work in r37845.
18:32 dalek parrot: review: https://trac.parrot.org/parrot/changeset/37848/
18:33 Coke (without having to do an svn update first and getting stuck with that?)
18:33 Infinoid I didn't revert your updates, "make headerizer" did
18:34 Coke OH!
18:34 Infinoid Apparently something's broken there
18:34 Coke that would explain why the docs are the way they are.
18:35 Coke so my commits by hand are useless; as is the c_function_docs test in general. whee!
18:37 Infinoid I'm still not sure I got all your function docs back... moment
18:37 Coke I'm not sure it matters.
18:38 Coke clearly I'll need to fixup headerizer to DTRT. If I do that, it's going to change the func docs test.
18:38 Infinoid Or make it suddenly pass. :)
18:38 * Coke wonders why extract_function_declarations seems to be in both the Parrot module and the perl script for headerizer.
18:42 Coke yah, that's wierd. looks like a half-done refactor.
18:43 dalek parrot: r37849 | Infinoid++ | trunk (11 files):
18:43 dalek parrot: Get the rest of Coke's function documentation back.
18:43 dalek parrot: Infinoid--
18:43 dalek parrot: review: https://trac.parrot.org/parrot/changeset/37849/
18:43 Infinoid ok, I *think* r37849 undoes the rest of my idiocy.  Sorry about that, not having a good day here
18:43 Coke not your fault.
18:44 bsdz joined #parrot
18:44 PerlJam It's Apr 1, just claim it was a poor joke  :)
18:46 dalek website: autarch++ | Final Grant Report for NLNet
18:46 dalek website: http://www.parrot.org/conte​nt/final-grant-report-nlnet
18:48 * Coke got an advance copy of that but apparently didn't read it, as he didn't realize some of those tasks were considered complete.
18:54 bacek good morning
18:55 bacek Infinoid: ah. I've missed include of io_private.h
18:56 Tene joined #parrot
18:58 moritz tewk: ping
19:00 moritz tewk: nice gsoc proposal... there's just one thing I didn't quite understand... you mention that you want to built it on top of @things, including clang - how do you want to use that?
19:00 moritz tewk: I hope you don't want to build parrot with clang, because last I tried that didn't work :/
19:02 PerlJam Where's tewk's proposal?
19:03 moritz PerlJam: http://socghop.appspot.com/student_proposal​/review/google/gsoc2009/tewk/t123860766727 (probably not pulic)
19:03 shorten moritz's url is at http://xrl.us/benexj
19:04 PerlJam cool
19:04 moritz is it public?
19:05 PerlJam oh, no, you have to be signed in.
19:05 PerlJam But I have no idea what "clang" is
19:06 moritz clang is a C compiler that targets llvm
19:18 darbelo Wait, clang doesn't build parrot?
19:20 Coke Infinoid: I think i have a fix to headerizer. needs a little tweaking, but should fixup the function docs nicely.
19:20 mikehh joined #parrot
19:24 Infinoid Coke++
19:25 Infinoid Will this enforce (one way or another) the question you had previously about whether function prototypes should include the argument names as well as their types?
19:25 Infinoid I'm just curious.
19:27 Coke Yes. adding a method to P::Headerizer that generates the expected function doc prototype and uses that when headerizing.
19:27 Coke the new method matches pretty much exactly the logic that was in the t/coding*/*func*
19:28 Coke (mostly a copy of that, with a little extra logic to split lines automagically.
19:28 Coke there we go. seems to be working. make headerizer mostly worked...
19:28 Coke ... and 17 TODO'd tests unexpectedly succeeded.
19:29 Coke ... if only I had realized this sooner. :|
19:32 Coke running headerizer now and doing a make test to make sure i didn't break anything...
19:33 dalek parrot: r37850 | coke++ | trunk (3 files):
19:33 dalek parrot: [t/docs] have 'make headerizer' fix function signatures to match codingstd.
19:33 dalek parrot: factor out logic from c_function_docs.t to a common method, replace logic
19:33 dalek parrot: in headerizer that stripped off args to use this instead.
19:33 dalek parrot: review: https://trac.parrot.org/parrot/changeset/37850/
19:35 Coke Infinoid: thanks for pointing this out. =-)
19:36 dalek parrot: r37851 | pmichaud++ | branches/p6strings/src/pmc/codestring.pmc:
19:36 dalek parrot: [codestring]:  Add 'ord_from_name' to obtain codepoints from character names
19:36 dalek parrot: (e.g., "LATIN CAPITAL LETTER A" and "MUSICAL FLAT SIGN").
19:36 dalek parrot: review: https://trac.parrot.org/parrot/changeset/37851/
19:37 bsdz hi, anyone know if there's an easy way to get an interactive PIR shell (perhaps with readline support)?
19:37 pmichaud there's not really a way to do that.  PIR always requires complete subs.
19:37 pmichaud oh, I suppose one could read lines until they see .end
19:37 pmichaud and the compile whatever they read.
19:38 bsdz ah that's a good point
19:39 bsdz i think i'll stick to my mini pir test scripts. thanks
19:41 moritz s
19:41 moritz (sorry)
19:41 Infinoid Coke: Next time I'll try to find a less destructive way of discovering these things :)
19:42 Infinoid (source control)++
19:46 dalek parrot: r37852 | pmichaud++ | branches/p6strings/t/pmc/codestring.t:
19:46 dalek parrot: [codestring]:  Tests for the charname_to_ord method in CodeString.
19:46 dalek parrot: review: https://trac.parrot.org/parrot/changeset/37852/
19:46 moritz whoever wrote http://www.parrot.org/conte​nt/final-grant-report-nlnet should fix a year in the first paragraph
19:47 cxreg happy birthday, parrot
19:47 pmichaud Parrot 2.0 has temporal logic built in.
19:47 pmichaud That's how we're able to start in November 2009 and finish in March 2009.  :-)
19:47 pmichaud See prior art from DCONWAY
19:48 cxreg for some value of prior
19:48 pmichaud heh.
19:48 contingencyplan joined #parrot
19:48 moritz ;-)
19:48 pmichaud touche'
19:48 PerlJam now that could have been a good Apr 1 joke.
19:49 pmichaud I thought of a terrific April 1 joke last year (in mid April) that I wanted to do for this year.   But then I forgot it.
19:49 cxreg perl 6 is in all possible states of completion simultaneously?
19:49 pmichaud cxreg:  close.  *in*completion
19:50 pmichaud we're just searching for the state where the level of incompletion is epsilon close to zero.
19:50 moritz pmichaud: I had a neat one planned. I wanted to exchange all firefox installations in our institute with IE6 (over wine) ;-)
19:50 moritz but it was too much work
19:50 pmichaud I think my Apr 1 joke was to generate lolcode from past.  But I ended up doing that in august.
19:51 pmichaud (and hence generate lolcode from Perl 6)
19:52 moritz that lightning talk on YAPC::EU impressed many of my fellow Erlangen.pm hackers ;-)
19:53 dalek parrot: r37853 | coke++ | trunk/lib/Parrot/Headerizer.pm:
19:53 dalek parrot: [docs] Fix headerizer to properly strip out function modifiers.
19:53 dalek parrot: review: https://trac.parrot.org/parrot/changeset/37853/
19:57 pmichaud I still can't believe I wrote that talk (code + slides) in under four hours.
19:58 pmichaud speaking of which, I need to get started on my NPW talks.
19:58 pmichaud I'm actually planning to break a personal tradition and have my talks ready _before_ leaving for the conference
19:58 pmichaud as opposed to my tradition which is to write the talks after arriving for the conference
19:58 pmichaud :-)
20:01 pmichaud phone
20:03 dalek parrot: r37854 | coke++ | trunk (84 files):
20:03 dalek parrot: [cage] run 'make headerizer'
20:03 dalek parrot: big change, as headerizer now includes function signatures in docs.
20:03 dalek parrot: adds this to a bunch of files where it wasn't already, and may
20:03 dalek parrot: change due to whitespace in others done manually earlier.
20:03 purl dalek: that doesn't look right
20:03 dalek parrot: review: https://trac.parrot.org/parrot/changeset/37854/
20:20 dalek parrot: r37855 | coke++ | trunk/t/codingstd/c_function_docs.t:
20:20 dalek parrot: [t/docs] update list of files with missing docs after latest headerizer run.
20:20 dalek parrot: review: https://trac.parrot.org/parrot/changeset/37855/
20:21 GeJ Good morning everyone
20:25 Coke Infinoid: done. Danke.
20:26 Infinoid Oh, nice
20:29 rdice joined #parrot
20:30 Infinoid Pulling r37854 caused git to do a gc run
20:30 * Infinoid waits patiently.
20:30 Coke next up is filling in remaining missing boilerplate, then actually writing missing docs.
20:31 cognominal joined #parrot
20:32 Infinoid Coke: Down to 540, from where?
20:32 Coke originally, the test was reporting over 2000 misses.
20:33 Infinoid Wow.  Nice!
20:33 Coke so that 1500 is all from fixing signatures.
20:33 Infinoid Coke++ # cleaning cages with fire hoses
20:34 Coke because that test was not in 'codetest', it was almost never run; probably been broken since 'make headerizer' was added.
20:34 Coke hopefully less fragile now.
20:40 dalek parrot: r37856 | coke++ | trunk (2 files):
20:40 dalek parrot: [t/docs] remove some non-documentation
20:40 dalek parrot: review: https://trac.parrot.org/parrot/changeset/37856/
20:41 gravity joined #parrot
20:51 ruoso joined #parrot
20:52 alvar joined #parrot
21:07 dalek parrot: r37857 | pmichaud++ | branches/p6strings/compile​rs/pge/PGE/Perl6Regex.pir:
21:07 dalek parrot: [pge]:  Refactor handling of perl 6 character escapes
21:07 dalek parrot: * easier to re-use in other perl6-like languages
21:07 dalek parrot: * handles \c[YOUR CHARACTER NAME HERE]
21:07 dalek parrot: * handles \cC
21:07 dalek parrot: review: https://trac.parrot.org/parrot/changeset/37857/
21:09 japhb joined #parrot
21:40 dalek parrot: r37858 | pmichaud++ | branches/p6strings/compile​rs/pge/PGE/Perl6Regex.pir:
21:40 dalek parrot: [pge]:  Remove extraneous debugging 'say' from p6escapes.
21:40 dalek parrot: review: https://trac.parrot.org/parrot/changeset/37858/
22:00 Searle joined #parrot
22:02 Searle Someone awake?
22:02 jonathan For some definition of awake.
22:03 Searle right :-)
22:04 Searle i've a question. i've written a cross referencing tool for parrot and was wondering who could be interested
22:04 Searle someone responsible for docs or so
22:04 PerlJam Searle: are you a student?
22:04 Searle no
22:05 Searle it's here: http://www.raisin.de/raxref-parrot/
22:05 moritz Searle: you can send it to the list, parrot-dev@lists.parrot.org
22:05 Searle moritz: any suggestion for a catchy subject?
22:06 jonathan oooh, shiny! :-)
22:06 PerlJam Searle: cool.
22:06 Searle thanks :-)
22:06 bsdz Searle++ that's pretty cool
22:06 moritz cool indeed
22:06 moritz Searle: "crossing the parrot"?
22:07 dalek parrot: r37859 | pmichaud++ | branches/p6strings (2 files):
22:07 dalek parrot: [pge]:  Fix bug in \c[digits] form of escapes, add tests.
22:07 dalek parrot: review: https://trac.parrot.org/parrot/changeset/37859/
22:07 Searle It's like lxr or mxr, only much faster and only needs vanilla perl5
22:07 Searle moritz: thanks, i'll go with that
22:08 moritz Searle: it would even be cooler if it would render the Pod, not display it raw
22:09 Searle moriz: it's on the TODO: http://github.com/Searle/raxref/tree/master
22:10 moritz Searle: good. I'm very impressed.
22:11 jsut joined #parrot
22:11 cotto Searle: very shiny
22:12 cotto IWBN if it were a little more selective about what it downloaded and if it worked better on a narrow screen, but that's a really cool tool.
22:12 cotto Searle++
22:12 Tene allison: did I get privs on pynie?  I don't recall.
22:13 Searle cotto: the files are actually split in smaller parts
22:13 iblechbot joined #parrot
22:13 Searle cotto: you can resize the slots. the default is splitting by fixed %
22:14 Searle cotto: my screen's 1920 px wide ;-)
22:14 bacek_ joined #parrot
22:14 cotto I mean a different arrangement of slots (which are horizontal and vertical).
22:14 cotto That explains it.
22:14 purl Damn right it does.
22:15 cotto purl++
22:15 cotto But it's definitely at the stage where I want to dig in and implement features myself.
22:15 Tene allison: I'd love to help with pynie, if you could give me commit privs.
22:15 Searle cotto: that's pretty hard with pure css witch it is now
22:16 jsut joined #parrot
22:16 cotto I'm sure it'd get easier with a nice js library.
22:17 cotto Cool.  It looks like he's working on that using jQuery.
22:17 Searle joined #parrot
22:18 Searle cotto: it uses jQuery now
22:19 Tene purl: msg allison where should I discuss pynie issues?
22:19 purl Message for allison stored.
22:24 * cotto has the git repo cloned and is playing with it now
22:24 cotto s/repo/tree/
22:24 * cotto is also getting used to git lingo
22:30 dukeleto cotto: git repo is pretty common parlance as well
22:31 moritz the "repo" is all the information, the "tree" is the checkout (ie the source files at one point in time), no?
22:36 dukeleto moritz: yes, so you clone a repo, but checkout a tree ?
22:37 moritz yes
22:41 pmichaud Tene: I think I can give you commit privs.
22:42 pmichaud actually, I know that I can :-)
22:42 Tene thx
22:42 pmichaud gmail account?
22:42 purl gmail account is super phishy
22:42 * Tene @ allalone.org
22:42 pmichaud sorry, google account
22:42 pmichaud added.  You're listed as a project owner, so you can also grant others commit privs.
22:43 Tene sweet
22:44 pmichaud afk # kid soccer practice
22:46 Limbic_Region joined #parrot
22:51 Tene purl msg allison rumors of pushaction going away?  someone seems to think you told bob rogers that it was?
22:51 purl Message for allison stored.
23:07 Whiteknight joined #parrot
23:53 davidfetter joined #parrot
23:59 TiMBuS joined #parrot

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

Parrot | source cross referenced