Camelia, the Perl 6 bug

IRC log for #parrot, 2008-03-27

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:00 * ewilhelm looks around for the parrot sysadmin
00:00 PerlJam Hmm. From Configure.pl ...
00:00 PerlJam Generating C headers...value for 'revision' in config/gen/config_h/config_h.in is undef at lib/Parrot/Configure/Compiler.pm line 393, <$in> line 19.
00:01 ewilhelm http://smoke.parrotcode.org/smoke/
00:01 purl http://smoke.parrotcode.org/smoke/ is, like, too short
00:02 ewilhelm looks like 99.5% ok is as good as it gets
00:02 chromatic That's not awful.
00:02 chromatic What fails, math stuff?
00:02 ewilhelm it's par for the course on win32 right?  2.2 nines!
00:03 dalek joined #parrot
00:03 * Limbic_Region doesn't run the test suite under Win32/MinGW very often
00:03 ewilhelm pg.t
00:03 chromatic The Postgres bindings?
00:04 Limbic_Region actually - unless requested, I only run it monthly (before release)
00:04 Coke that's enough to keep us honest.
00:04 Limbic_Region PerlJam - the problem is local to you, working fine for me
00:04 Limbic_Region PerlJam - parrot is making now
00:04 Coke cygwin does build for me if on this new machine if I preset the PATH. It *never* worked on my last windows machine.
00:04 Coke I need to respond to that while near the machine that I tested it on.
00:09 * PerlJam does a fresh checkout
00:09 Limbic_Region ok, build failed
00:10 PerlJam Limbic_Region: complaining about MD5 and SHA stuff?
00:10 Limbic_Region yes
00:10 Limbic_Region MD2, MD4, MD5, and SHA
00:10 chromatic does "make parrot" succeed?
00:11 Limbic_Region just a sec
00:11 Limbic_Region nothing to be done for parrot
00:11 chromatic Okay, so it's just the dynpmcs that fail to build.
00:11 Limbic_Region looks like it is something later on in the build that is failing
00:11 chromatic That's a plus.
00:11 Limbic_Region purl paste
00:11 purl i heard paste was (see: nopaste) or like glue but a little safer to sniff. or http://nopaste.snit.ch:8001/ or http://scsys.co.uk:8001/ anywhere shadowpaste is or mmm, sticky paste or You there! Eating the paste. or <see> 2 girls, 1 paste
00:12 chromatic That means the configuration check for the crypto libs is somehow wrong on Cygwin.
00:12 nopaste "Limbic_Region" at 68.49.236.220 pasted "make output on Win32/Cygwin" (987 lines) at http://nopaste.snit.ch/12579
00:12 Limbic_Region well, that's what PerlJam was indicating
00:13 Limbic_Region but I didn't see anything out of the ordinary from Configure.pl
00:13 Limbic_Region Generating CPU specific stuff.....................​....................done.
00:13 Limbic_Region Verifying that the compiler supports function pointer casts............yes.
00:13 Limbic_Region Determining whether your compiler supports computed goto...............yes.
00:13 Limbic_Region Determining if your compiler supports inline...........................yes.
00:13 Limbic_Region Determining what allocator to use.....................................done.
00:13 Limbic_Region Determining if your C library supports memalign........................yes.
00:13 Limbic_Region Determining some signal stuff.....................​....................done.
00:14 Limbic_Region Determining whether there is socklen_t.................................yes.
00:14 Limbic_Region Determining if your C library has setenv / unsetenv...................both.
00:14 Limbic_Region Determining if your platform supports AIO...............................no.
00:14 Limbic_Region shit
00:14 Limbic_Region sorry
00:14 Limbic_Region Determining if your platform supports crypto...................yes, 0.9.8g.
00:14 Limbic_Region that's what I meant to paste
00:14 * Limbic_Region wanders off feeling like a major idiot
00:15 Coke you can stay here and do that! =-)
00:15 ewilhelm hey, that's all I ever do here
00:15 chromatic Where are the crypto libs installed?
00:18 Limbic_Region my ssl libs are in /usr/lib
00:19 chromatic There's no -L/usr/lib in the link line.
00:19 chromatic What if you add it and try to link the dynpmc group manually?
00:19 Limbic_Region presumably they /usr/lib is in the LIB_PATH
00:19 Limbic_Region but I will try that
00:19 chromatic gcc -o digest_group.dll lib-digest_group.o md2.o md4.o md5.o ripemd160.o sha.o sha1.o   -s -L/usr/local/lib  -L/usr/lib -shared -L/home/Gat/parrot/blib/lib -lparrot
00:21 spez joined #parrot
00:21 Limbic_Region well, after cd'ing to src/dynpmc
00:22 Limbic_Region it fails with exact same errors
00:22 Limbic_Region with -L/usr/lib
00:22 chromatic does it need -s or -shared in front of -L/usr/lib ?
00:22 Limbic_Region I will try all variations
00:22 Limbic_Region no difference
00:23 chromatic -lcrypt?
00:23 Limbic_Region just a sec
00:24 ewilhelm by any chance do you have mingw tools in your cygwin path?
00:24 Limbic_Region ewilhelm - no
00:24 Limbic_Region chromatic - no change
00:25 chromatic Interesting.  Mine link against libcrypt and libcrypto.
00:25 chromatic And libcurses.  I thought I removed the link against curses on Linux.
00:25 Limbic_Region heh, that fixed it
00:26 Limbic_Region -lcrypto
00:26 Andy joined #parrot
00:26 Limbic_Region now let me remove things piece by piece and see if it was a combination of things or just the -lcrypto
00:27 Limbic_Region yeah, actually I was right about the -L/usr/lib (not necessary - part of the default path)
00:27 Limbic_Region and the -s / -shared becomes superfolous
00:27 Limbic_Region and the -lcrypt is also not necessary
00:27 Limbic_Region just adding -lcrypto makes it work
00:28 PerlJam next time I mention something about a fresh checkout, you guys remind me that it takes *for* *ever*
00:28 chromatic aptitude install ccache
00:28 Limbic_Region PerlJam - it could be worse, ICU could still be part of the checkout
00:29 chromatic L~R, can you file a bug?  Francois should be able to figure out how to fix things knowing what's missing.
00:30 Limbic_Region I can just send an email to parrotbug@ ... right?
00:30 chromatic yes
00:30 Limbic_Region interestingly enough, after getting past that one failure by hand and going back and doing make again
00:30 Limbic_Region it finishes without a problem
00:30 Limbic_Region and
00:30 Limbic_Region the final link
00:30 Limbic_Region has -lcrypto
00:31 chromatic Strange.
00:31 purl But true.
00:32 PerlJam when I ran Configure.pl just now, before it died I did notice that it found openssl (which should trigger -lcrypto)
00:34 Khisanth PerlJam: at least you aren't doing it on my connection!
00:38 Limbic_Region chromatic - email sent
00:38 chromatic Thanks.
00:38 Limbic_Region no problem
00:39 Limbic_Region so - it seems to me that despite not having built it since Nov 24th
00:39 Limbic_Region not a lot is needed to make it work
00:39 Limbic_Region :D
00:40 * jonathan is realizing that sangria is stronger than cerveza...
00:40 Limbic_Region ticket #52150
00:41 jonathan I've got a first-cut implementation of subset ... of ... where ... though.
00:45 jonathan In as r26570. Have fun. Sleep time.
00:48 svnbotl r26570 | jonathan++ | trunk:
00:48 svnbotl : [rakudo] First-cut implementation of subset ... of ... where ...; likely has issues, but it's better than nothing.
00:48 svnbotl diff: http://parrotvm.org/svn/parrot/revision/?rev=26570
00:57 Coke the file_metadata test appears to be spuriously using Parrot::Revision
01:03 Theory joined #parrot
01:04 rdice joined #parrot
01:11 * Coke finally cuts that email.
01:12 davidfetter joined #parrot
01:33 jeffreykegler joined #parrot
01:40 chromatic http://majek4.blogspot.com/2008/03/u​seful-c-extensions-gcc-specific.html
01:40 shorten chromatic's url is at http://xrl.us/bibap
01:40 chromatic likely() and unlikely() look interesting.
01:42 * Coke tries to find more low hanging deprecation.
01:52 Coke when removing something deprecated, should I feel guilty about not expunging it from the book?
01:52 chromatic Very.
01:53 Coke Ok. but I think I'm still going to skip that step, as the book needs a major rewrite and this is more involved than simply removing something from a list.
01:53 chromatic File a bug against the book perhaps?
01:54 Coke hurm. if I did that for everything in the book, we'd have another 400 tickets. If I file *one* ticket it's too vague.
01:54 chromatic I dunno then.
01:54 chromatic I'll forget too.
01:54 Coke I suppose I could just file a single ticket regarding the one thing I'm removing.
01:54 Coke ... Or I could add an "errata" section.
01:55 ewilhelm there's a book?
01:55 Coke Yup. Somewhat out of date, but we got a copy of the print book "parrot and perl6 essentials" given to us by o'reilly.
01:55 Coke (it's checked into the repo. docs/book)
01:57 Coke chromatic: there is more classoffset goodness in runtime/parrot/library/SDL/*.pir
01:57 chromatic Really?
01:57 chromatic Oh, I fixed examples/sdl/
01:58 chromatic I'll work on it.
01:58 Coke Danke.
02:03 Coke chromatic: does adding docs/book/appendix.pod with ERRATA seem reasonable? It smacks of having a TODO list checked into the repo, but as it's docs, that seems less offensive to me.
02:03 chromatic Works for me.
02:13 * Coke tests a build sans classoffset, expecting no issues...
02:14 ewilhelm left #parrot
02:22 * Coke gets a hang in stm/runtime for the first time in ages.
02:31 Khisanth joined #parrot
02:42 kid51 joined #parrot
02:42 svnbotl r26571 | coke++ | trunk:
02:42 svnbotl : [deprecated]
02:42 svnbotl : -remove classoffset opcode
02:42 svnbotl : -inc. from docs and other locations, except:
02:42 svnbotl : -in the book, where it's tied to examples. There, add an ERRATA that we need to
03:01 svnbotl joined #parrot
03:05 Andy joined #parrot
04:00 cotto_home joined #parrot
04:16 AndyA joined #parrot
05:05 svnbotl r26574 | chromatic++ | trunk:
05:05 svnbotl : [SDL] Refactored away deprecated classoffset attribute access.
05:05 svnbotl diff: http://parrotvm.org/svn/parrot/revision/?rev=26574
05:09 jrockway joined #parrot
06:24 jrt4 joined #parrot
06:52 Ademan joined #parrot
07:16 cotto_home ok 21 - freeze/thaw - shallow
07:17 cotto_home yay!
08:08 iblechbot joined #parrot
08:30 darbelo left #parrot
08:35 wknight-away joined #parrot
09:06 svnbotl r26575 | fperrad++ | trunk:
09:06 svnbotl : [pct]
09:06 svnbotl : add method 'clone' (needed by Lua)
09:06 svnbotl diff: http://parrotvm.org/svn/parrot/revision/?rev=26575
09:13 svnbotl r26576 | fperrad++ | trunk:
09:13 svnbotl : [Lua]
09:13 svnbotl : - fix where & traceback
09:13 svnbotl : since r26495, Parrot_Context_info.file is a STRING* (previously a char*)
09:13 svnbotl diff: http://parrotvm.org/svn/parrot/revision/?rev=26576
09:18 ruoso joined #parrot
09:52 Ademan joined #parrot
10:37 purl joined #parrot
10:46 svnbotl r26577 | fperrad++ | trunk:
10:46 svnbotl : [Lua] PAST-pm deprecation
10:46 svnbotl : - replace PAST-pm/PAST by PCT/PAST
10:46 svnbotl : - replace PAST-pm/POST by Lua/POST (a private copy of PAST-pm/POST)
10:46 svnbotl diff: http://parrotvm.org/svn/parrot/revision/?rev=26577
10:53 svnbotl r26578 | fperrad++ | trunk:
10:53 svnbotl : [Lua]
10:53 svnbotl : update MANIFEST
10:53 svnbotl diff: http://parrotvm.org/svn/parrot/revision/?rev=26578
10:54 lidden joined #parrot
11:10 iblechbot_ joined #parrot
11:21 particle joined #parrot
11:35 wknight8111 joined #parrot
11:36 rdice joined #parrot
11:42 particl1 joined #parrot
12:00 ruoso davidfetter, hi... have some time? I'm starting to work on a new project and I'm willing to try some things I'm not so used in db modelling... and I was wondering if I could exchange some ideas with you about it...
12:01 ruoso althought it's certainly offtopic for #parrot and we should probably move it to somewhere else...
12:08 mj41 joined #parrot
12:38 svnbotl r26579 | fperrad++ | trunk:
12:38 svnbotl : [Lua]
12:38 svnbotl : - fix POD
12:38 svnbotl diff: http://parrotvm.org/svn/parrot/revision/?rev=26579
13:14 jrt4 joined #parrot
13:18 Limbic_Region joined #parrot
13:22 gryphon joined #parrot
13:42 Coke I have made a google doc using gadgets to show pie charts of the bug summary data. If only I could figure out how to say "let anyone see this..."
13:43 Coke ah, it's in the document, not the document manager.
13:47 pmichaud I can see the summary data but not the pie chargs
13:47 pmichaud *charts
13:47 Coke the pie charts are on the next tab.
13:47 Coke no?
13:47 purl i guess no is is <reply>
13:47 Coke no, no is <reply>well, maybe.
13:47 purl okay, Coke.
13:47 pmichaud when I click on "Graphs" I get a page full of boxes with "404 Not Found"
13:48 pmichaud perhaps I need to be signed in?
13:48 Coke possibly. I just "published" the document, try that url:
13:48 Coke http://spreadsheets.google.com/pub?key=​pvdt32cKG2wmfq74DWgGwFw&amp;output=html
13:48 shorten Coke's url is at http://xrl.us/bigwt
13:49 pmichaud it works after I sign in
13:49 pmichaud (the url received via mail)
13:49 pmichaud the url you just gave here seems to work even if I'm not signed in
13:49 Coke one is for collaborators, one is for everybody.
13:50 pmichaud the one for collaborators requires sign-in to view the graphs.  the one for everybody seems to 'just work'
13:50 Coke no clue if this will be useful, but it was easy. =-)
13:51 pmichaud this is open tickets, or open+new, or ...?
13:51 Coke it's whatever is coming from the bug summary. =-)
13:51 pmichaud ah
13:52 Coke probably just new+open (it doesn't mention stalled)
13:52 pmichaud overview.html already has pie charts?
13:52 pmichaud or is that new?
13:52 Coke The html may have always had them.
13:52 Coke the email version does not.
13:53 pmichaud particl1: ping
13:58 TonyC joined #parrot
13:59 nopaste joined #parrot
14:05 Andy joined #parrot
14:17 * Coke yawns.
14:29 GeJ_ joined #parrot
14:43 kid51 joined #parrot
14:46 Coke he's busy. =-)
14:46 Coke (particle).
14:46 pmichaud np, it's not urgent
14:55 svnbotl r26580 | jkeenan++ | norevision:
14:55 svnbotl : Further work on http://rt.perl.org/rt3/Tic​ket/Display.html?id=52154.
14:55 svnbotl : Eliminate repository-revision-number related code post-configuration.
14:55 svnbotl : Check revision number only if SVN is available.
14:55 svnbotl diff: http://parrotvm.org/svn/parrot/revision/?rev=26580
15:09 Coke Wow. only 16 days to get my passport; was told 4-6 weeks.
15:12 Limbic_Region that reminds me, I have to renew my passport - hard to believe it has been 10 years now
15:13 * kid51 had an 11-week wait for passport renewal last year
15:14 Coke I wonder if they've just cut through the backlog.
15:14 kid51 Yes, but only because they've outsourced to private contractors who are reading your passport file at this very moment.
15:15 kid51 Coke:  feel free to look into the norevision branch re RT 52154.
15:20 Coke kid51: roger.
15:25 pmichaud when I got my passport in dec, it only took 10 days
15:26 pmichaud I was totally shocked -- I had been expecting a multi-week wait
15:26 pmichaud of course, I went ahead and went with "expedited service"
15:26 Coke they're now selling expedited as 3 weeks, but regular is 4... crazy.
15:27 pmichaud well, expedited "guarantees" 3 weeks, but it's often much faster than that.
15:28 Coke How can I tell configure "use this version of make preferentially"?
15:28 Coke (cygwin is grabbing mingw32-make instead of the 'make' I type on the command line.)
15:28 avar MAKE=path ./configure ?
15:28 Coke ... doh. --make=; duh.
15:40 skids joined #parrot
15:40 Coke building parrot on cygwin, I get the following error:
15:40 Coke runtime/parrot/library/Crow.pir
15:41 Coke ./parrot.exe -o runtime/parrot/library/Crow.pbc runtime/parrot/library/Crow.pir
15:41 Coke error:imcc:syntax error, unexpected $end in file 'runtime/parrot/library/Crow.pir' line 146
15:41 Coke QueryHash.pir builds fine, though. I don't see anything obvious in that file: it has 146 lines, though.
15:44 Coke it's the HEREDOC
16:03 cognominal joined #parrot
16:10 cotto_work Is it a Bad Idea to store an INTVAL in visit_info->extra during thawing?
16:10 marmic joined #parrot
16:11 cotto_work it works, but it strikes me as questionable
16:18 davidfetter ruoso, it was a little early here when you asked, but i do now :)
16:20 ruoso davidfetter, shall we move to private, or some other channel?
16:25 gryphon joined #parrot
16:28 IllvilJa joined #parrot
16:32 peeps[work] joined #parrot
16:35 kid51 joined #parrot
16:39 rdice joined #parrot
17:12 cotto_work feather?
17:12 purl it has been said that feather is http://feather.perl6.nl/ or a perl6 community development server
17:38 cognominal joined #parrot
17:48 Infinoid there.  I've got a svnbotl replacement that does everything svnbotl does except for identifying the branch it was in (the rss doesn't seem to contain that info), and except for the hanging/timeouts.
17:49 Infinoid does anyone know who maintains the rss?
17:50 kid51 I don't know about the rss. diakopter is knowledgeable.  Personally, since so much of my work is in branches, I would hate to lose that notation.
17:53 svnbotl joined #parrot
17:54 ewilhelm joined #parrot
17:59 Infinoid yeah, that's why I haven't already replaced svnbotl :)
18:00 Infinoid without the branch tag, the output is really confusing.
18:01 Tene Huh, looks like my laptop isn't broken after all.
18:01 Tene Thanks, Coke!
18:02 kid51 Hey, Coke, what are your hourly rates for laptop repair?
18:02 Coke 100/hour, 4 hour minimum.
18:03 Tene What's that measured in commits?
18:07 Coke Tene: so I was half joking, but lolcode didn't seem to survive the recent merging, if you now have a working laptop... =-)
18:07 * Coke opens a ticket.
18:09 * Tene svn up
18:13 * Coke hates to keep opening tickets!
18:33 kid51 joined #parrot
18:40 Coke zzzzz what?
18:40 purl \ gregor
18:56 Psyche^ joined #parrot
19:16 lidden joined #parrot
19:17 sjansen joined #parrot
19:20 svnbotl r26585 | chromatic++ | trunk:
19:20 svnbotl : [GC] Fixed a memory leak introduced in r26379.  In particular, the
19:20 svnbotl : consolidation of Parrot_forall_header_pools() meant that the non-const PMC pool
19:20 svnbotl : was never swept or freed during global destruction.
19:20 svnbotl diff: http://parrotvm.org/svn/parrot/revision/?rev=26585
19:27 Coke seen tewk?
19:27 purl tewk was last seen on #parrot 1 day and 2 hours ago, saying: particl1: Made a couple of minor changes to SoC application.  [Mar 26 10:19:18 2008]
19:36 barney joined #parrot
19:49 Coke msg tewk commented on your soc app.
19:49 purl Message for tewk stored.
19:52 tewk Coke: Cool, do I need to read the comments and revise or is it just part of the SoC process?
19:55 particle joined #parrot
19:58 Coke you need a comment to edit your proposal.
19:58 Coke my comment asks you to edit your proposal. =-)
19:58 Coke basically, looks good, just needs a little polish.
19:58 particle hi-o
19:58 Coke yours is the second parrot application I see.
19:59 Coke ah, there's particle. particle, see #soc. =-)
19:59 * Coke knew I should have just waited for him to show...
20:08 b2gills joined #parrot
20:20 pmichaud particle: ping (if convenient)
20:20 purl I can't find (if in the DNS.
20:21 pmichaud I'll be glad to review proposals also :)
20:24 anna30 joined #parrot
20:25 particle wazzup?
20:27 pmichaud obra asked me if it would be possible to get a nice red/green display of rakudo's test suite
20:28 pmichaud I think it would be great, but don't know what we need to do to make it happen.  Was curious if you wanted to head it up, or if I should throw it open for someone else to hack on...?
20:28 particle i'm swamped. it's not hard, especially with TAP::Parser
20:28 pmichaud okay
20:28 particle just need to find somebody with tuits
20:28 pmichaud I'll file an RT ticket and advertise on list
20:29 particle fab. i still owe you a ticket
20:30 pmichaud np, I have it in my head.  Actually, I'll file the ticket now :-)
20:31 pmichaud I have about 15 mins to burn
20:31 jrt4 joined #parrot
20:32 particle oh, hey, i got a sigplan email for a conference i thought was interesting for you or larry
20:33 pmichaud I'll take a look
20:33 Coke red/green.
20:33 pmichaud that could be fun, depending on where sigplan is located this year
20:33 particle CFP: SIGPLAN Workshop on Undergraduate Programming Language Curriculum
20:33 Coke er, for red/green that would "jfw" if it was run by default.
20:33 Coke (in languages/smoke)
20:33 pmichaud coke:  there's a ticket for that, yes.
20:33 particle coke: not with fudge it wouldn't
20:34 pmichaud iirc we haven't been able to get languages/smoke to work with the pct-based languages
20:34 Coke that's not "the test suite". that's the *spec* test suite. =-)
20:34 Coke and I don't see why not, if it's running things locally.
20:34 pmichaud oh.  I think obra's request was more tuned to the *spec* test suite :-)
20:34 pmichaud so I was imprecise earlier :)
20:35 Coke pmichaud: Ok, that's two hurdles, yes.
20:35 particle ewilhelm and i spoke briefly about having a day where he, colin and i (and mebbe allison and chromatic) get together to bust out the consolidated test harness
20:36 pmichaud obra was thinking that having a graphic chart of what works and what doesn't would be very useful to show progress
20:36 Coke sure1
20:37 Coke particle: if nothing else, we can tackle that at yapc::na.
20:37 pmichaud seems like there ought to be a standalone tool that could take TAP output and feed it through the HTML generator
20:37 Coke ... be nice if there was a list of things to hack on at yapc, actually.
20:37 ruoso joined #parrot
20:37 pmichaud (list of things) -- whatever happened to the "high-priority ticket" display?
20:37 particle it'd be nice to tie that red/green into the spec
20:38 Coke as i recall, the problem with perl6 specifically is that the tests look like perl5, neh?
20:38 Coke (and therefore don't get run with the proper harness?)
20:38 pmichaud I want to step back from that a second
20:38 * Coke addes perl6 to the language testing to see what breaks.
20:39 pmichaud I think that the harness is too "integrated" at the moment
20:39 pmichaud instead I'd like to have a way of capturing the TAP output from the tests and passing it through another tool to format/collate it however we want
20:39 pmichaud as opposed to always trying to do it all within a single process
20:40 daxelrod joined #parrot
20:40 Coke there's 3 different things going on here: making it worth with the existing system, upgrading the existing system, and rewriting it, and I think you, particle and I are each talking about one of those 3 items.
20:40 Coke If I can do #1 while we're waiting for #2 or #3, then that's still a win.
20:41 pmichaud correct
20:41 pmichaud if we can do #1, then #2 and #3 become much less important (and I'd drop the issue altogether)
20:41 pmichaud however, it seems that #1 hasn't been easily do-able, which is why I'm exploring #2 and #3 :-)
20:41 Coke izzit me, or is building something based on PCT slower than the old tools?
20:41 pmichaud it may be slower than the older tools
20:42 Coke I mean, better to spend the time at compile time than runtime (like tcl does...)
20:42 pmichaud partially because NQP is using utf-8 when compiling things (c.f. yesterday's conversation on strings)
20:42 pmichaud so, NQP is far slower than it ought to be
20:43 pmichaud or were you referring to a different aspect of "building something"?
20:45 particle pmichaud: here's a link: http://www.sigplan.org/pl-workshop/
20:47 particle i forgot to bring that up at sixperl yesterday
20:47 jonathan Would be interesting to be a fly on the wall at that conference.
20:47 pmichaud particle:  interesting.  however, I probably don't want to spend my time writing a white paper :-)
20:47 particle definitely.
20:47 purl Absolutely!
20:47 ewilhelm pmichaud: we can easily spool the test output to a directory tree for that sort of post-mortem html-ifying
20:47 particle they want three pages. it's not a lot.
20:48 particle you mean post-morifying html?
20:48 ewilhelm fudge is a bit trickier I guess, what is it exactly?
20:49 ewilhelm like a pass that fails?
20:49 particle yes
20:49 particle fudge sets the return value to 1
20:49 particle it's a preprocessor
20:50 particle basically,  all tests pass, but the result is non-zero
20:50 particle you still get a failure with the default tap parser
20:50 ewilhelm is this as the function-call level, process level, or TAP stream?
20:50 particle we need a parser that understands when tests have been fudged
20:50 pmichaud particle:  we do?  why?
20:51 pmichaud alternatively, we can have a parameter (or wrapper) to fudge that tells it how to fudge the return code
20:51 ewilhelm I think we probably want to do that with proper TAP
20:51 ewilhelm purl: fudge?
20:51 purl i think fudge is better :) or for fat fuckers or http://compsoc.dur.ac.uk/~tsp
20:51 particle i want to know which tests pass without fudge, which pass with fudge, and which fail
20:51 Coke I thought fudge simply generated a different return code, causing the test to always pass #TODO, but I admit I didn't read it. =-)
20:51 Coke no, fudge is http://compsoc.dur.ac.uk/~tsp
20:51 purl okay, Coke.
20:51 particle the goal is for all tests to pass without fudge
20:52 ewilhelm 403
20:52 pmichaud ewilhelm:  fudge does preprocessing on the tests so that the compiler (rakudo) is able to skip tests it doesn't know how to parse yet
20:52 particle http://svn.pugscode.org/t/fudge
20:52 pmichaud i.e., fudge is a test file preprocessor
20:52 ewilhelm 404
20:53 pmichaud http://svn.pugscode.org/pugs/t/spec/fudge
20:53 particle oops, t/spec/fudge
20:53 * ewilhelm having hard time gleaning docs from error pages :-D
20:53 particle argh.
20:54 particle http://svn.pugscode.org/pugs/t/spec/fudge
20:54 ewilhelm so it basically does on-the-fly TODO and SKIP insertion?
20:54 particle yep
20:54 pmichaud a bit more than that, actually
20:54 particle based on the comments in the source
20:54 pmichaud the 'skip' insertion actually comments out the tests
20:55 particle #?rakudo skip 1 <-- does nothing in pugs
20:55 ewilhelm well, if running the result gives valid TAP, then it is just a matter of having your ./prove do the prep work
20:55 ewilhelm hmm, how does the comment-out work WRT the plan()?
20:55 pmichaud skip substitutes 'skip' lines in the TAP output
20:55 ewilhelm AndyA: you following any of this?
20:56 pmichaud i.e., it comments out the tests and effectively replaces them with 'skip' results
20:56 ewilhelm ok, so it does leave placeholders in the TAP
20:57 ewilhelm yeah, that should work fine
20:57 particle it makes a new file
20:58 particle fudge rakudo foo.t # creates foo.rakudo
20:58 ewilhelm we still need to form, fire, and glaze the Test::Harness API, but this all sounds workable
20:58 particle then foo.rakudo is passed to perl
20:58 ewilhelm perl?
20:58 purl rumour has it perl is or a rather nice compromise between lisp, c and my native spoken language or a 4 letter word
20:58 particle *perl6
20:58 TimToady generally just use t/spec/fudgeall preprocess all the files that prove wants to run
20:58 TimToady automatically maps the fudged filenames in the list for you
20:59 ewilhelm right, parrot's existing prove was built to read the shebang and pass to the appropriate parrot call
20:59 TimToady then whatever calls *that* program :)
20:59 Coke TimToady: should it be possible to insert a source filter in perl5 so that if it sees a "use v6;" in the perl source it exec's parrot instead?
21:00 wknight8111 joined #parrot
21:00 particle actually, rakudo's tests are passed to perl t/harness
21:00 particle perl t/harness t/spec/S03-operators/foo.rakudo
21:00 pmichaud actually
21:00 * ewilhelm needs to break you all of some of those habits
21:00 pmichaud perl t/harness --fudge t/spec/S02-literals
21:01 TimToady Coke: well, a better question is whether you can do it without source filtering...
21:01 Coke (seems like it runs the use before it does the source filter.)
21:01 ewilhelm ./prove --fudge t/spec/
21:01 Coke TimToady: perhaps by somehow overriding 'use v6;' - except that isn't a normal pragma, izzit?
21:01 pmichaud does/can 'prove' understand a --fudge option?  (I'm showing my testing system ignorance here)
21:02 Coke oooh, it does work.
21:02 ewilhelm pmichaud: we have control of the option handling -- it is ./prove not `prove`
21:03 ewilhelm Coke: the prove implementation can exec parrot
21:03 nopaste "coke" at 72.228.52.192 pasted "itle" (23 lines) at http://nopaste.snit.ch/12584
21:03 pmichaud why is "./prove" better than "perl t/harness"?
21:03 ewilhelm pmichaud: same deal, but only one harness for all of parrot
21:03 ewilhelm not languages/foo/t/harness, etc
21:04 pmichaud so you're saying we need to come up with a better t/harness
21:04 pmichaud that can handle all of our languages
21:04 ewilhelm yes.  one
21:04 purl yes. one is sendmail, and the other is smail.
21:04 ewilhelm purl: one is also less than n
21:04 purl okay, ewilhelm.
21:04 Coke as an aside, if I run 'perl5 -I. -Msixperl path/to/rakduo/test.t' with that nopaste, it dies with "run parrot".
21:05 ewilhelm https://svn.perl.org/parrot/branches​/unified_testing/tools/util/runtests
21:05 Coke so we could shim that in as a stopgap to exec parrot perl6.pbc
21:05 shorten ewilhelm's url is at http://xrl.us/bihof
21:05 ewilhelm Coke: see line 113 --^
21:05 ewilhelm we get the pbc from the shebang
21:06 Coke not for the spec tests.
21:06 Coke neh?
21:06 pmichaud ...the shebang doesn't have a pbc
21:06 ewilhelm why not?
21:06 pmichaud because the spec tests are for all of the perl6 implementations, not just rakudo
21:07 ewilhelm well, are they valid perl 6?
21:07 pmichaud however, the unified t/harness could provide the pbc
21:07 Coke ewilhelm: yes, but how do you tell perl5 that?
21:07 ewilhelm yeah, just add a special-case for detecting '^use v6-alpha' and pretend it is like a shebang
21:08 particle some languages don't allow a shebang
21:08 ewilhelm like which?
21:08 purl it has been said that like which is easier and which works faster/more efficient/better?
21:08 particle pheme
21:08 Coke particle: presumably those won't look remarkably like perl5 and use the same file extension.
21:08 particle (scheme)
21:08 kj joined #parrot
21:08 pmichaud if we have a unified t/harness that understands parrot's language's tree, then we don't need the test files to specify the pbc or language
21:09 Coke if you can extend the shebang there, that's great: what's stopping us from rolling this back into trunk? =-)
21:09 ewilhelm perldreamer got fed-up waiting for changes to Test::Harness
21:09 ewilhelm or at least that's my guess judging from the irc/svn logs
21:10 ewilhelm pmichaud: yeah, I think we probably just need something like a config of globs mapped to pbc's
21:10 ewilhelm or dirs not globs
21:11 particle some languages have mixed language tests
21:11 particle it would suck if those languages also couldn't use shebangs
21:11 Coke particle: any other than tcl?
21:11 particle parrot's t/op/ uses perl and pir
21:11 Coke (why would shebangs stop working?)
21:11 ewilhelm AndyA: I can haz API?
21:11 particle scheme doesn't see # as a comment char
21:12 particle and doesn't accept #! as valid input
21:12 Coke particle: who are you talking to?
21:12 particle i mean, doesn't parse #! as a shebang
21:12 particle the peanut gallery. am i in the right channel?
21:13 Coke I was trying to figure out if you were answer my question, or still beating the scheme thread from 4m ago. or something else.
21:13 ewilhelm Coke: shebangs could keep working wherever a dir was not config'd
21:13 particle as long as a single harness for all languages is flexible enough for us, let's do it
21:13 ewilhelm line 107 s/unless/elsif/
21:13 Coke then we'd only need config'd dirs where shebangs didn't work... which is the small set of places we're already at.
21:14 ewilhelm yep, or whatever sort of magic identifier you want
21:14 Coke we need this unified because we can't test our compiler toolkit at the moment except by building languages... which we can't smoke. so we're flying blind.
21:14 Coke er, "sooner than later".
21:15 Coke ewilhelm: are you the defacto "champion" for the unified branch?
21:15 ewilhelm If I can get api changes into T::H without them getting reverted, yes
21:15 AndyA :P
21:15 ewilhelm ah, there he is
21:15 AndyA What do you need ?
21:15 purl i heard do you need was rw?
21:16 ewilhelm AndyA: the parrot's 'prove' has never gotten into a form that would be stable WRT T::H
21:16 AndyA OK
21:16 ewilhelm e.g. overrides of _get_parser_args(), trouble with @ARGV handback, etc
21:16 ewilhelm https://svn.perl.org/parrot/branches​/unified_testing/tools/util/runtests
21:16 shorten ewilhelm's url is at http://xrl.us/bihoy
21:17 ewilhelm pmichaud: I think it would also benefit from some executive decisions in cleanup, layout, and config policy on the parrot side
21:18 ewilhelm which is where the PNW locals might be able to help
21:18 AndyA OK, my head's elsewhere at the moment. Is there a mailing list we could move it to? I'm in async mode right now :)
21:18 Coke ewilhelm: if you need a decision on something on the parrot side, I can drive that.
21:22 pmichaud I don't have much to add to the cleanup/layout/config policy on the parrot side, other than for it to remain as sane as possible
21:22 ewilhelm AndyA: your pipermail's are 403  http://www.hexten.net/pipermail/tapx-dev/
21:23 AndyA Ah - thanks. Willfix.
21:23 pmichaud I'll revisit this a bit later -- gotta take daugher to soccer
21:23 pmichaud *daughter
21:24 ewilhelm http://www.google.com/search?q=p​arrot+runtests+site%3Ahexten.net
21:24 shorten ewilhelm's url is at http://xrl.us/bihpc
21:28 * diakopter watches as Infinoid works magic with SVN::Web and botnix
21:35 * Infinoid FAILS to work magic with SVN::Web and botnix
21:35 * Coke wonders how to get a glob into an array.
21:39 nopaste "Coke" at 72.228.52.192 pasted "instead of perl path/to/rakudo.t, use perl -Msixperl path/to/rakudo.t." (26 lines) at http://nopaste.snit.ch/12586
21:39 * Coke neverminds.
21:40 Coke with sixperl.pm and adding -M to the invocation of the script, this works.
21:40 Coke this may be enough of a hack to get us moving forwardin trunk.
21:44 cotto_work where could I find some info on setting up a development environment for Parrot on OSX?
21:47 Coke cotto_work: Install the developer tools. This gives you all the comand line tools you need.
21:47 Coke (as well as XCode, if you want the GUI; parrot's build doesn't use xcode, just gcc)
21:49 cotto_work thanks
21:49 ewilhelm you can install gcc *without* xcode?
21:54 wknight8111 joined #parrot
21:55 Ademan_ joined #parrot
22:13 wknight8111 how do I go about calling a function, using PIR, that's in a different namespace then where I am currently?
22:14 kj get_hll_global, or maybe get_root_global? Retrieve the object into a register and then invoke that
22:14 kj object=sub
22:14 kj not entirely sure which one you should use.. but at least you can specify the namespace.
22:15 wknight8111 okay, thanks. The docs don't seem to have this piece of info, and it seems like it should be absurdly common
22:15 kj it should be added to compiler_faq I think
22:15 kj it's a good question
22:17 wknight8111 find_global, i found it. Thanks
22:29 particle it's in the namespaces pdd iirc
22:30 wknight8111 some kind of syntactic sugar would be nice for that
22:30 wknight8111 like 'namespace':'function'(params)
22:30 wknight8111 or ['name';'spaces']'function'(params)
22:33 wknight8111 but that's just my opinion on it
22:33 particle feel free to submit a ticket or ping the list
22:35 particl2 joined #parrot
22:36 wknight8111 I dont like complaining about things that I'm not prepared to fix myself
22:49 Khisanth joined #parrot
23:02 dalek joined #parrot
23:03 * diakopter waits for a commit
23:14 skids joined #parrot
23:23 dalek joined #parrot
23:24 peepsalot joined #parrot
23:31 wknight8111 does pdb still exist? "make pdb" doesnt seem to work on my system
23:33 jonathan wknight8111: try make world
23:33 jonathan If you're on Win32, you may need make pdb.exe
23:33 wknight8111 that works, thanks
23:33 jonathan de nada
23:59 Coke ewilhelm:

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

Parrot | source cross referenced