Perl 6 - the future is here, just unevenly distributed

IRC log for #perl6, 2006-01-31

Perl 6 | Reference Documentation | Rakudo

| Channels | #perl6 index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:00 Juerd Fuck, fuck, fuck. START SSHD.
00:00 Juerd What's it doing? Thinking about the meaning of life?
00:01 Juerd It's up
00:01 Juerd But still slow
00:02 azuroth hi Juerd
00:02 Juerd Hello
00:02 azuroth how are you doing?
00:02 Juerd Guess.
00:03 azuroth is this feather you're fighting with?
00:04 Juerd Yes
00:04 Juerd And SATA connectors
00:05 Juerd disk io is slow on feather
00:05 Juerd And I don't know why
00:06 azuroth ick. I've never used SATA, but everyone I know seems to have problems with it
00:07 Juerd I also have many systems running without any problems.
00:10 azuroth main problem I've heard about is booting to windows
00:11 putter Ok, the 3 clusters of ext/ failures now all appear to be a small set of pugsbugs.  this is now a haskell exercise. ;)
00:11 putter dduncan: see previous line. :)
00:12 Juerd azuroth: Haven't had problems with that either
00:12 * Odin- doesn't have any problems with booting Windows: I just don't do it. :)
00:13 putter oh, it's fun.  just don't do it in your bare feet.
00:16 putter Anyone looking for something to do: the ChangeLog still has quite a ways to go before it is up to date (it's at r8750), and needs a cleanup pass.
00:17 putter just svn log -r8750:8775 to get your feet wet
00:20 Juerd GODDAMNIT
00:20 Juerd Fucking goddamned box won't install lilo with a degraded array
00:20 Juerd Then how the hell am I going to get it to use a new kernel, so it can use the other box, so it can rebuild its array
00:21 Juerd Die, lilo
00:21 * Juerd tries grub.
00:22 Juerd "Probing devices to guess BIOS drives. This may take a long time."
00:22 Juerd How long is a long time, genius programmer?
00:22 Juerd A week? A month? A year?
00:22 Juerd A decade?
00:22 Juerd How long until I can safely assume it's broken and reset the box?
00:22 Juerd Oh yay. I/O errors on the disk.
00:22 Juerd I give up
00:22 rafl Juerd: Is feather already working again? If not, you should probably disable cron.
00:24 vel joined perl6
00:26 Juerd rafl: It works, but not well.
00:26 Juerd I don't know wtf is wrong.
00:26 Juerd I can write to the disk at 10 MB/s, which is full line speed
00:27 Juerd Let's degrade the raid array to see what the resynch speed will be
00:28 Juerd      [>....................]  recovery =  3.0% (742976/24418688) finish=8.4min speed=46436K/sec
00:28 Juerd That's normal speed
00:28 Juerd Then why does it perform so badly :(
00:30 grayson joined perl6
00:31 rafl Juerd: Maybe because some pugs builds are running in parallel?
00:31 stennie joined perl6
00:32 Juerd rafl: Both niced
00:32 xern joined perl6
00:35 gilby joined perl6
00:40 dduncan I be back
00:41 dduncan now checking ext/
00:42 putter hey dduncan
00:42 dduncan hi
00:42 putter I fixed the one ext/ file that looked plausibly like a non-pugs-core problem.  the rest are core exercises.
00:42 dduncan so last I checked, the only red zones were in 4 ext/ items
00:42 putter actually, your last smoke had 3 failed tests in magical/env.t
00:42 dduncan so you think the remaining failures are best fixed in the core and not the ext code itself?
00:43 putter yes.  one is mentioned by iblech as that being audreyt's call, the others seem similar
00:44 putter I have no idea what's going on with env.t.  your last smoke is the only place i've seen it fail.  ???
00:44 dduncan that does make sense to an extent
00:44 dduncan since a lot of the ext/ stuff was written before 6.2.10 and not changed since
00:45 putter on the other hand, I didn't look _behind_ the various ext/ failues, ie, what happens if failing test x is removed, do things then pass.
00:45 dduncan so if they fail now its because either the core's broken, or the ext/ was depending on broken functionality that was fixed
00:45 dduncan whereas, I try to isolate the problems ... but some of them just seem to say the core is at fault
00:46 dduncan even if futile, I'll give them a quick look now ... after all, the ext/ could be wrong
00:46 Limbic_Region joined perl6
00:46 dduncan but I'll ask you a question
00:46 dduncan regarding File-Spec
00:46 putter any idea what happened in env.t?
00:47 dduncan I have to step out for a few minutes, but will look closer at env.t
00:47 putter ok
00:49 rafl Juerd: What did you change on feather? I can't imagine that changing memory makes the box that slow..
00:49 Juerd rafl: Motherboard, CPU, memory.
00:50 Juerd 6677 eric256   35  10 43580  40m 1520 R 15.0  2.0   0:56.66 catalog_tests.p
00:50 Juerd 5537 eric256   35  10 43588  40m 1532 R 14.1  2.0   5:40.94 catalog_tests.p
00:50 Juerd 2371 eric256   35  10 44712  41m 1648 R 13.5  2.1  14:15.35 catalog_tests.p
00:51 Juerd That's a bit overdone.
00:51 Juerd Okay, slayed users
00:51 Juerd Now let's see what a kernel build does
00:52 putter Compiling Pugs.AST         ( src/Pugs/AST.hs, dist/build/src/Pugs/AST.o )   /tmp/ghc24699.hc: In function `PugsziAST_zdwcharInc_entry': /tmp/ghc24699.hc:39: warning: cast to pointer from integer of different size
00:52 rafl Maybe it's that slow that erics task isn't finished when the next one starts..
00:52 Juerd make clean is already slow
00:52 Juerd rafl: Could be.
00:52 putter And in Pugs.Lexer
00:52 Juerd But the disk was fast
00:52 Juerd 45 MB/s is normal.
00:52 Juerd The CPU that's installed now is faster than the previous
00:53 rafl I'd blame the mainboard then.
00:53 weinig joined perl6
00:54 putter *** Compiling PIL to JavaScript...
00:54 putter substr outside of string at /mnt/disk150_30G_5/pugs_kennel​/r8869/perl5/PIL2JS/lib/PIL.pm line 349.
00:54 putter Use of uninitialized value in concatenation (.) or string at /mnt/disk150_30G_5/pugs_kennel​/r8869/perl5/PIL2JS/lib/PIL.pm line 349.
00:54 Juerd But what part?
00:54 Juerd Something has to be *broken* to perform so badly
00:54 Juerd But brokenness leads to error messages
00:56 rafl I once had a rally bad disk performance with a SATA disk when I didn't compile the driver in properly, but I guess you did that..
00:57 Juerd But the disk does perform well.
00:57 Juerd 45 MB/s is good, for these disks.
00:57 Limbic_Region nothing in dmesg ?
00:58 Juerd Limbic_Region: Nothing I interpret as bad.
00:58 rafl Juerd: How did you benchmark that?
00:59 putter dduncan: re F:S question, I'll be heading out to dinner in a few minutes.  gone for perhaps an hour.
01:00 Juerd rafl: for i in f r a; do mdadm -$i /dev/md2 /dev/sda5; done; watch --interval=1 cat /proc/mdstat
01:01 Limbic_Region rafl - do you suspect an IO problem or you can't quantify the slowness?
01:01 Limbic_Region iostat should show if the disks are getting slammed for some reason
01:01 Juerd Do you mean vmstat?
01:02 Limbic_Region no
01:02 Juerd If so, do you know how to interpret its output?
01:02 Limbic_Region I meant iostat
01:02 Juerd Does Debian have that?
01:02 Limbic_Region but vmstat is good too
01:02 * Limbic_Region doesn't know as he isn't really a linux guy
01:02 Juerd 0  0      0 1625020  48344 304488    0    0     0     0  259    13  0  0 100  0
01:02 Juerd 0  0      0 1625020  48344 304488    0    0     0     0  260    12  0  0 100  0
01:02 Juerd 0  0      0 1625020  48344 304488    0    0     0     0  266    11  0  0 100  0
01:02 Juerd 0  0      0 1625020  48344 304488    0    0     0    52  275    24  0  0 100  0
01:02 Juerd during idle
01:02 Limbic_Region iostat?
01:02 Limbic_Region or vmstat?
01:03 Limbic_Region both should be done for a collection interval of at least 5 seconds IMO
01:03 Juerd 2  0      0 1620416  48392 304792    0    0     0   244  304   195 90 10  0  0
01:03 Juerd 1  0      0 1620416  48392 304792    0    0     0     0  267   174 86 14  0  0
01:03 Juerd 1  0      0 1602164  48392 303180    0    0     0    20  315   248 94  6  0  0
01:03 Juerd 1  0      0 1565584  48392 303180    0    0     0     0  258    11 95  5  0  0
01:03 Limbic_Region iostat 5 5
01:03 Juerd During perl Makefile.PL in pugs' tree
01:03 Limbic_Region or vmstat 5 5
01:03 Juerd Limbic_Region: vmstat 1
01:03 Juerd ok, vmstat 5 5 then... wait a sec
01:03 rafl Juerd: iostat is in sysstat.
01:04 Juerd 2;130 root@feather:~/pugs# vmstat 5 5
01:04 Juerd procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu----
01:04 Juerd r  b   swpd   free   buff  cache   si   so    bi    bo   in    cs us sy id wa
01:04 Juerd 1  0      0 1622272  48560 304900    0    0   109   140  351   202 74  2 23  2
01:04 Juerd 1  0      0 1612956  48572 303288    0    0     0    68  276   125 93  7  0  0
01:04 Juerd 1  0      0 1522560  48592 303288    0    0     0    66  278    19 98  2  0  0
01:04 Juerd 1  0      0 1613824  48592 304904    0    0     0     2  262    14 97  3  0  0
01:04 Juerd 1  0      0 1548600  48604 304912    0    0     0   383  288    33 98  2  0  0
01:04 Juerd During perl Makefile.PL
01:05 Limbic_Region Juerd - this is for Solaris, but it should help deciphering http://www.adminschoice.com/docs/iost​at_vmstat_netstat.htm#Example:vmstat
01:05 Juerd If you don't mind, my mind is terribly overloaded. I can't learn now
01:06 Juerd I appreciate the pointer
01:07 Juerd iostat shows close to 100% user cpu load
01:07 Juerd Almost none system
01:07 Juerd That should indicate a non-io problem, right?
01:07 Juerd Also almost none iowait
01:07 Limbic_Region well, it is hard lining up your columns but it looks like you aren't cpu idle
01:07 Limbic_Region does debian have top?
01:08 Juerd It does
01:08 Juerd If you want, you can log in on feather
01:08 Limbic_Region if I read that column correctly - you were 0 percent idle 4 out of the 5 samplings
01:08 Limbic_Region that's not good
01:08 Juerd I can provide you with root access if you want to help fight the problem
01:08 Limbic_Region ok -
01:09 Juerd You have root access through sudo
01:09 Limbic_Region that's scary
01:09 Juerd And indeed, the CPU wasn't idle.
01:09 Limbic_Region sudo bash
01:09 Juerd What is scary?
01:09 Limbic_Region root access through sudo
01:09 Juerd Why?
01:10 Juerd I find it quite useful, because I can revoke the rights without changing the root password.
01:10 Limbic_Region well - I don't like having root access to any box I am not personally responsible for
01:10 Juerd Ah
01:10 Juerd Don't worry, there are no backups
01:10 Limbic_Region what's the hostname?
01:10 Juerd feather.perl6.nl
01:11 Juerd It's really weird to have a very slow box with 2 GB ram :)
01:11 Limbic_Region it doesn't have a very high load ATM
01:11 Juerd Limbic_Region: No, it's been idle because I was chatting with you :)
01:12 Juerd I have started a simple make clean
01:12 Limbic_Region no - it is still crazy cpu
01:12 Limbic_Region oh
01:12 Limbic_Region nevermind
01:12 Juerd And already the cpu load is almost through the roof.
01:12 Limbic_Region yeah - I am just watching top
01:12 Juerd Cpu(s): 98.8% us,  1.2% sy,  0.0% ni,  0.0% id,  0.0% wa,  0.0% hi,  0.0% si
01:12 Juerd Insane
01:12 Limbic_Region perhaps unnice your make?
01:13 Juerd Though weird: sum up the values in the CPU column manually
01:13 Juerd Just estimate
01:13 Juerd I get nowhere near 100
01:13 Juerd Oh, now I do, now that find is busy
01:13 Juerd I started make
01:14 Limbic_Region well, everything points to being cpu bound
01:15 Juerd Yeah, but how does that make sense?
01:15 Juerd It's a 2.8 GHz CPU
01:15 Juerd And how to solve it :|
01:16 Limbic_Region that part I don't know - and I don't have time to help, got to go call my Dad
01:16 Juerd Okay; thanks much for your help so far
01:17 Limbic_Region Juerd - if it is still a problem tomorrow I will try to actually try to isolate the problem
01:17 Limbic_Region good luck
01:18 Cryptic_K joined perl6
01:27 dduncan putter, about env.t ...
01:27 dduncan when I'm in Pugs' root directory (with a copy of Test.pm sitting in my cwd), and run ./pugs t/magicals/env.t ... all 11 tests pass
01:28 dduncan if I'm in t/magicals (with a copy of Test.pm in my cwd), and run ../../pugs env.t, then all but the first 6 tests fail ...
01:29 dduncan in the latter case, it seems that the temp file is being written in pugs' own dir, while it is attempting to be read from the magicals dir (the cwd)
01:29 dduncan when I'm in Pugs' root directory, pugs own dir and the cwd are the same, so all the tests pass
01:29 dduncan neither of these scenarios matches my smoke, so I'm not sure what happened there
01:30 putter lol.  I was so hopeful until that last line. ;)
01:30 Juerd I'm very tired
01:30 dduncan I'm normally in pugs' root dir when running make smoke
01:30 Juerd I don't know if I should drive home :|
01:30 Juerd Can't think clearly
01:31 dduncan sleep where you are?
01:31 putter food, drink ;)
01:31 rhesa sugar
01:32 putter lol
01:32 dduncan putter, I don't know what's causing it as yet, but running the current pugs shell leads to incessant beeping, like an unprintable character keeps getting output
01:32 dduncan this happened a few days ago, then went away when I fixed the non-delete problem, and now it's back.
01:33 wolverian joined perl6
01:33 dduncan its return would coincide with the last 16-20 hours of commits
01:33 putter sigh.  perhaps, as you mentioned earlier,  r8840
01:33 dduncan yes, but I'm not sure its that one at fault
01:34 dduncan is anyone else here not using readline and/or can reproduce that symptom?
01:34 dduncan okay, I'm out for c 15 min
01:35 dduncan er, first I'll undo just that shell change and see if it has an effect ...
01:36 putter sounds good
01:36 putter I'll run a full smoke/js/pir spread over dinner.
01:37 putter oh, what was your F::S question?
01:38 dduncan a minute ...
01:39 dduncan before I rewrote Spec.pm, it was assuming that someone who used File::Spec would automatically see the exported functions of other modules that File::Spec uses, like File::Spec::*
01:39 putter yes
01:39 dduncan as if "is export" put things in the global namespace by default
01:40 dduncan is that supposed to happen?
01:40 dduncan I would have thought it only exported into the context where it was required/used
01:40 dduncan which would be Spec.pm, and not the user of Spec.pm
01:40 putter re  is that supposed to happen?, my understanding is not.
01:41 dduncan my rewrite explicitly re-exported all the functions, which made it work
01:41 dduncan but I'm wondering if there was something in the perl 6 spec that said this was actually supposed to export into global, and my change just covered up a core bug
01:42 putter no, I think you did the right thing.
01:43 putter p6 will need more flexible re/export, but I think that's on the todo list.
01:44 Juerd Feather's problem is a memory problem
01:44 Juerd With other (less) memory, it runs fast.
01:44 Juerd But how to solve it :|
01:45 rafl Don't use any memory at all if it's getting faster with less memory?
01:45 putter I'm double checking, but I'm pretty sure your "export to global" is not intended.
01:45 putter s/your//
01:45 wolverian Juerd: ran a ram check yet?
01:46 Juerd wolverian: Have none
01:46 dduncan okay, I
01:46 dduncan will be away for 15
01:46 Juerd And without bootable CD's, I'm not going to run one either
01:46 rafl Juerd: There are several packaged for Debian.
01:46 wolverian Juerd: memtest86 is designed to run from a bootable floppy/cd
01:47 Juerd wolverian: No floppy, no cd.
01:47 Juerd rafl: I don't dare running one if I have no alternative means of booting.
01:47 rafl Juerd: You can boot if with $your_favourite_bootloader
01:47 wolverian Juerd: heh, install it on the HD then. debian has a package that should do the right thing (tm)
01:47 putter the http://dev.perl.org/perl6/doc/design/syn/S11.html "Importation" example suggests not.
01:48 rafl Juerd: There are also programs that run in userspace like memtester.
01:48 putter google: "is export" site:dev.perl.org  :)
01:48 Juerd rafl: Are those any good?
01:48 putter dinner. bbl.  hopefully with some smokes.
01:48 Juerd The boot loader thing is what I'm scared about. Userspace is fine
01:49 rafl Juerd: memtest works well.
01:49 Juerd Dear ASUS, "Latest beta bios" is NOT a good changelog entry.
01:50 Juerd Especially not if it's the ONLY changelog entry, for the last 5 releases
01:53 Juerd "The CMX1024-3200C2 is a 1024 MByte DDR SDRAM DIMM designed for extreme speed and is one of the
01:53 Juerd fastest 1 GByte modules available."
01:53 Juerd Yeah, that's why it was so expensive, and that's why I bought it for that other box that never happened
01:53 Juerd But the speed isn't quite showing :|
02:01 Amnesiac joined perl6
02:02 Juerd Manually setting memory timings appears to help
02:02 Juerd Last time I ever did that was on my 386/387 machine.
02:02 Juerd Insert lots of profanity here.
02:04 Juerd Hm, no
02:04 Juerd Had the wrong module installed
02:05 autrijus joined perl6
02:05 autrijus is now known as audreyt
02:05 audreyt dduncan: try r8870 for your libreadline (really emulated by libedit on Tiger)?
02:05 vel joined perl6
02:06 audreyt (C language)-- # no module system
02:06 audreyt D++ # sane module system
02:08 wolverian graphviz++ # eternal fun
02:08 wolverian http://chronoa.dy.fi/~wolv​erian/reboot-languages.svg # here is a graph from another channel that maps users to languages. should I run it here? :)
02:12 audreyt go ahead :)
02:12 * rafl votes yes
02:13 dduncan I'm currently running makefile.pl; will run make shortly
02:14 dduncan make
02:15 dduncan make error; rl_catch_signals undeclared
02:15 dduncan I think we were going to disable use of readline for systems like mine
02:17 audreyt mm
02:17 audreyt but Makefile.PL did compile for you?
02:17 dduncan yes
02:18 dduncan and the readline link was not disabled this time
02:19 audreyt okie
02:19 Juerd I'm getting insane
02:19 rafl Juerd: What's up this time?
02:20 Juerd I have one RAM module with which it boots fast and normal
02:20 Juerd Three with which it does not.
02:20 Juerd And I don't know what to do about it
02:20 * Juerd starts to cry
02:21 dduncan audreyt, it may be simplest to simply disable readline on any system whose version doesn't implement those signal things, as was done 24 hours ago
02:21 rafl Juerd: Maybe a RAM slot is broken?
02:21 dduncan and/or if the version fails to have something we know is a new feature
02:21 Juerd Hm, two that run fast, but not together.
02:21 Juerd rafl: Tried different slots :)
02:22 dduncan then readline is used if it exists and is newish; it is not used otherwise
02:23 audreyt dduncan: yeah, I've committed a Makefile.PL
02:23 audreyt please rerun Makefile.PL
02:23 audreyt and see if Makefile still contains -DPUGS_HAVE_READLINE
02:23 audreyt r8871
02:23 dduncan pulling
02:24 dduncan makefilepling
02:25 dduncan the messages did not say that readline was disabled
02:25 dduncan but looking at makefile anyway ...
02:26 dduncan the makefile also does contain -DPUGS_HAVE_READLINE
02:26 dduncan so it wasn't shut off
02:27 audreyt weird.
02:28 dduncan I can try cleaning, if that makes any difference
02:28 audreyt can you trace Makefile.PL a bit?
02:28 audreyt that probably doesn't
02:28 audreyt the line that does this is line 178
02:28 audreyt and 631
02:28 dduncan I'm already looking at 178
02:28 audreyt the code snippet on line 178 should not compile+run on your system
02:29 audreyt but in any case you can save it under test.hs
02:29 audreyt and see if "ghc --make test.hs" works
02:30 audreyt alternatively, remove the -v0 in r644 (change it to -v3 perhaps)
02:30 audreyt s/r644/line 644/
02:30 audreyt and see if some useful debug info can be obtained
02:30 audreyt (I'll brb)
02:31 dduncan I removed the  "-v0",
02:31 putter audreyt: why not go back to the full test snippet which is known to work?
02:31 dduncan I would concur
02:33 dduncan removing the  "-v0", caused a few extra output lines to appear near the top:
02:33 dduncan Chasing modules from: Setup.lhs Compiling Main             ( Setup.lhs, Setup.o ) Linking ... ghc-pkg: cannot find package plugins Chasing modules from: /tmp/pugs-tmp-26787.hs Compiling Main             ( /tmp/pugs-tmp-26787.hs, /tmp/pugs-tmp-26787.o ) Linking ... Chasing modules from: /tmp/pugs-tmp-26787.hs Compiling Main             ( /tmp/pugs-tmp-26787.hs, /tmp/pugs-tmp-26787.o ) Linking ...
02:33 dduncan I will also mention that a haskelish error message did used to appear there when the test was working, yesterday
02:34 dduncan instead of the a-ok we see here
02:34 dduncan I will try making the 178 test longer, like the one from the other day ...
02:35 putter the version in -r8839 should work
02:36 audreyt ok, reverting to it.
02:39 audreyt committed as r8872.
02:39 dduncan I made my own revert and am running it
02:40 dduncan fyi, it seems to be just _rl_catch_sigwinch that it's complaining about here
02:40 audreyt indeed, though that might be a good thing
02:40 dduncan and that seems to have worked, mine that is
02:41 audreyt good. the reason I left it out from the probe was that I found out in systems with new (gnu) readline, sigwinch shouldn't be disabled
02:41 putter smoke of r8869 is up
02:41 audreyt but in Makefile.PL probing it's a good way to separate gnu and nongnu readlines
02:42 dduncan yes indeed
02:42 dduncan my own change, deltad with r8871 fixed the problem
02:42 audreyt dduncan: try trunk r8873?
02:43 dduncan I will now replace it with yours, for synchronizing
02:44 dduncan in my case I just changed the makefile
02:44 dduncan now I have pulled yours
02:44 audreyt putter: ok, one-line summary of the core bug(s) that needs to be addressed for the ext/ failures?
02:45 audreyt I know of the @array_of_spans[$index].stringify problem iblech mentioned in the commit log
02:45 putter the shortened test also didnt have a call to readline, yes?  so control may never have actually hit the library.  perhaps ghc just optimized it all away?
02:46 putter at least one of the two flags was just an extern integer.  if the other one was too...
02:46 audreyt putter: indeed that might be the case.
02:46 audreyt yup
02:46 dduncan with your version, readline is disabled , ... now making
02:47 dduncan the shell also works with your version
02:47 dduncan so that problem's seemingly resolved
02:47 audreyt yay
02:47 Juerd Great
02:48 Juerd In the process, one dimm got fucked up when I dropped it on the floor
02:48 Juerd There goes EUR 80
02:48 Juerd BUT
02:48 Juerd With a lot of manual hacking in the bios, I have now one of the new dimms and both two old dimms together
02:48 Juerd With usable speed
02:49 Juerd (It boots within a minute, instead of in 8 minutes)
02:49 dduncan fyi, the only significant difference between my version and yours is that I just left the 'compile only' function, while you changed it to 'compile and run'; the actual haskell code (from r8839) was identical
02:49 dduncan and I didn't change shell.hs
02:49 Juerd And the total is 2048 now, I just lost the opportunity to upgrade two other boxes with 512 each :|
02:49 Juerd Ah well
02:49 dduncan so the compile alone with sigwinch will fail and result in no readline linking
02:49 audreyt dduncan: yup, and that's what we want
02:50 dduncan right ... I read the diff backwards ... your version was the one not running it
02:50 putter *** Cannot cast into Hash: VRef <Array::Const> as in ext/Set-Infinite/t/set-infinite-functional.t
02:52 putter also ext/Perl6-Container-Array/t/array-lazy.t
02:55 putter and *** Cannot 'shift' scalar  as in ext/Perl6-Container-Array/t/array-operations.t
02:55 putter I think that's the full set.
02:56 audreyt k
02:56 rafl Juerd: Great, thanks!
02:59 Juerd Putting it in the rack now
03:01 putter I'm wrong - there's a fourth failure mode.  curious.  ext/Set-Infinite/t/set-infinite-recurrence.t and ext/Set-Infinite/t/set-infinite.t are all  not ok Infinite Loop.  So that could well be a Set::Infinite bug, not a core bug.
03:03 putter curious.  don't know how I missed that.  just not in my log.  ah well.  sorry.
03:05 audreyt it's fine, as I just addressed it.
03:05 audreyt all Span tests now pass.
03:05 Juerd feather is up
03:06 Juerd I'm going home now
03:06 putter Juerd++
03:06 Bit-Man joined perl6
03:06 Juerd Or, if it turns out I'm unable to drive safely, to the first hotel I encounter.
03:07 audreyt Juerd++
03:07 audreyt 04:01 AM (CET) - Upgrade finished. It took 6 hours and cost a lot; don't ask.
03:07 audreyt Juerd++ Juerd++
03:08 Cryptic_K joined perl6
03:09 audreyt joined perl6
03:09 putter Juerd: caffeine?  cold water on face?
03:09 svnbot6 joined perl6
03:10 putter svnbot6++
03:10 putter welcome back, you were missed.
03:10 putter ?eval 3
03:17 dduncan now pulled 8875
03:17 vell joined perl6
03:21 dduncan make complete
03:26 dduncan now smoking r8875
03:26 * putter puzzles over what the Var.hs change means...
03:27 dduncan speed seems normal
03:27 dduncan I should know in about 30 min whether all the ext/ pass or not
03:27 dduncan er, don't die that is
03:28 audreyt putter: the Eval.Var findSub dispatch system was written before S12's dispatch order was specced clearly
03:28 audreyt back then there were no clear distinction between single and multiple dispatch
03:29 audreyt and so both were lumped in a single findSub loop
03:29 audreyt so "$x.foo.bar" is no different from bar(foo($x))
03:29 audreyt (rather, bar(foo($x:):))
03:30 audreyt but MMD dictates that you have to find which of the candidate &bar to be dispatched to
03:30 audreyt before you fully evaluate its argument foo($x:) under its specified context
03:30 obra audreyt: what's your favorite win32 perl these days?
03:30 obra (perl5)
03:30 audreyt obra: I'm not on win32 anymore, but I hear camelpack is quite nice
03:31 obra I'll have a look
03:31 audreyt putter: but you can't find out which candidates of &bar to dispatch to, unless you know the type of foo($x)
03:31 * obra is poking at a new tabletpc
03:31 audreyt putter: that's a chicken-egg problem
03:31 audreyt putter: so I wrote a tiny type inferencer to guess a type of an Exp without actually evaluating its side effects
03:32 audreyt putter: the idea is that foo($x) is first inferred, then uses that inferred type to decide which bar() to call, and finally evaluate foo($x) in full
03:32 putter (call bbiab
03:32 audreyt using the argument type expected by bar()
03:32 audreyt but the infer engine didn't handle indexed expressions well
03:32 audreyt $x[0] etc
03:33 audreyt in any case, that tiny inferencer is obsolete under the new (November 2005) S12 and docs/notes/multimethods.pod
03:33 audreyt which would be implemented at PIL layer
03:34 audreyt the inferencer fix I just committed for indexed expressions
03:35 audreyt just checks if both the indice and indexee are simple expressions (i.e. things that can be evaluated without side effects)
03:35 putter back
03:35 audreyt and if so, it just evaluates them to find out their actual type
03:35 audreyt end of explanation
03:35 audreyt (feel free to haddockify this)
03:35 putter got it.
03:35 putter ::)
03:36 putter ok, will do
03:37 audreyt mixing return type polymorphism ("want") and MMD naturally leads to bizzare ambiguities
03:37 audreyt and I'm happy that S12-Nov05 addressedthis
03:41 grayson` joined perl6
03:41 grayson` left perl6
03:46 dduncan okay, so both ext/Perl6-[Container-Array|Value-List] died during the smokes
03:47 dduncan of 8875
03:47 evalbot_8875 joined perl6
03:47 dduncan as did Set-Infinite
03:49 dduncan same reasons putter mentioned
03:51 putter in, r8876
03:51 svnbot6 r8876 | putter++ | Eval/Var.hs: added audreyt's description of the old findSub dispatch system.
03:52 dduncan what did that fix for you?
03:52 putter We do that a couple of times and pugs will be wizzily documented.
03:52 dduncan that was just documentation?
03:52 putter re what did that fix for you?, my brow is no longer furrowed in puzzlement
03:52 putter :) yes
03:53 dduncan when I first saw you did an update, I thought it was for the ext/ failures
03:53 SamB joined perl6
03:53 * putter takes a moment to look at S12...
03:54 audreyt lunch, bbiab
03:54 putter audreyt: current state?
03:54 putter summary of?
03:56 putter dduncan: nope.  I think I'm done with serious work for the evening.
03:56 dduncan well ext/Span did not die
03:56 putter you mean span.t ?
03:56 audreyt putter: still can't quite narrow down the failure modes of the const VArray bug
03:57 dduncan both span.t and unicode.t did not die, as they did in 8864
03:57 audreyt will continue after lunch; once it's there I'll review/resume chglogging
03:57 dduncan I think they died before
03:57 audreyt and then push out preflight
03:57 audreyt also a worrisome heisenbug
03:57 audreyt t/rules/from_perl6_rules/capture.t
03:57 dduncan at least, a sudden switch from green to red often means a death
03:58 putter audreyt: ok, thanks.
03:58 audreyt fails tests 1..17 20..22 24..29 ~half of the time here
03:58 audreyt PGE sefaulting it seems
03:58 dduncan and hovering over one of the red squares in my browser shows 'stub' rather than a fail message, which suggests death
03:58 audreyt but half of thetime it's just fine
03:58 putter hmm.
03:58 audreyt suspecting external parrot's CGP switch
03:58 audreyt trying to disable it now
04:00 audreyt (i.e. change parrot -C to just parrot)
04:01 wolv joined perl6
04:01 audreyt doesn't seem to help.
04:02 audreyt anyway, lunch first :) bbiab
04:03 putter &
04:04 wolv is now known as wolverian
04:04 putter dduncan: yeah, all died except for the infinite sets, which, well, infinitely looped.  those tests are now commented out.  so if spam, err, span isn't dying (remember when spam was simply an obscure canned meat product?), progress.
04:05 dduncan okay, besides the 3 ext/ I mentioned, all others did not die
04:06 dduncan the 3 being P-C-A, P-V-L, and S-I
04:09 feng joined perl6
04:10 beppu_ joined perl6
04:11 putter ok.  i svn up'ed to 8876 and am poking at capture.t.  one thing to do might be to walk the tests which were failing, with ./pugs -Iblib6/lib t/... , and make sure the "fixed" ones are entirely passing.  random thought.
04:13 vel joined perl6
04:13 wolv joined perl6
04:14 wolverian oops.
04:14 buu evalbot_8875: ?join #buubot
04:18 putter dduncan: so, setCatchSigwinch isn't used in Shell.hs any more, but still is in the Makefile.hs test snippet.  want to see if we can get rid of it there?
04:18 dduncan no
04:18 putter lol
04:18 putter ok
04:19 dduncan we left the sigwinch in the makefile on purpose, as a test to whether any existing readline is Tiger's broken version or not
04:19 * putter suspects it was not setCatchSigwinch which was the missing ingredient of the shorter test, but rather the readline call.
04:19 justatheory joined perl6
04:20 dduncan apparently, when the broken readline is linked into pugs, there were issues with the shell, such as incessant beeping
04:20 dduncan changing just the makefile so readline isn't linked in fixed this
04:21 putter re we left the sigwinch in the makefile on purpose, err, no.  the test snippet was intended to mirror the Shell.hs usage.  the shorter test ceased to do that in several respects, and ceased to work (ie, fail).  the makefile should, eventually if not tonight, again reflect the current Shell.hs usage.  which means setCatchSignals, _and readline_ but no setCatchSigwinch.
04:22 vel joined perl6
04:22 dduncan whatever is done, the shell has to actually work on Tiger with its default pseudo-readline
04:23 dduncan that is, pugs should just work by default on a user's Tiger install even when they didn't upgrade readline
04:23 dduncan it works right now
04:24 putter right.  I hypothesize the current snippet's behavior will be unchanged by the removal of the setCatchSigwinch.  and that this is the right thing because Shell.hs no longer uses setCatchSigwinch, and the test snippet should mirror Shell.hs needs.
04:25 dduncan well, don't commit any such changes until after I've tested them
04:25 audreyt dduncan: you can take out line 182 in Makefile.PL
04:25 audreyt ."  setCatchSigwinch False\n"
04:25 audreyt and run Makefile.PL again
04:25 putter ;) thus the: dduncan: so, setCatchSigwinch isn't used in Shell.hs any more, but still is in the Makefile.hs test snippet.  want to see if we can get rid of it there?
04:25 dduncan yes I can
04:25 audreyt and inspect Makefile to see whether PUGS_HAVE_READLINE is still no there
04:26 dduncan however, I"m going to let my current smoke complete first
04:26 audreyt if it's not, then commit Makefile.PL sans that line :)
04:26 audreyt sure
04:26 audreyt (though running Makefile.PL does not affect smoke)
04:26 audreyt so you can do both in parallel
04:26 audreyt brb
04:26 dduncan I should get to this in about 1 hour
04:27 * putter has found updating Makefile.PL not only mangles smoke, but an require a complete rebuild. :(
04:27 dduncan I prefer to play it safe myself
04:27 putter s/mangles/can mangle/
04:28 putter but not a running smoke I guess.
04:29 dduncan mine is currently on t/oo
04:29 putter could always make a second copy of pugs....
04:30 dduncan I"m not that impatient
04:30 putter but there's nothing really pressing about this.  prerelease is likely still an hour or few away.
04:30 dduncan then we're okay
04:30 putter oh, and there's still lots-o changelog'ing to be done.
04:31 dduncan I've done what I can on that front
04:31 dduncan while the smoke is happening, I'm changelogging one of my own projects that will be released simultaneously with pugs
04:33 putter audreyt: is that with external or embedded parrot?
04:33 putter k
04:41 putter audreyt: with external parrot, I am unable to generate a capture.t failure.  15+ runs.  half on pure 8868, half with the new Var.hs and Shell.hs.  embedded parrot stopped working entirely for me a month or few back.  but local system changes occurred (new libc, etc) during the same time, so that may just be me.
04:42 justatheory joined perl6
04:45 justatheory joined perl6
04:47 putter smoking pil2js is not a quick thing...
04:48 clkao zzzzz
04:54 audreyt okay, another ext/ failures down (P::V::L), two more (P::C::A and S::I) to go
04:54 svnbot6 r8877 | audreyt++ | * Perl6::Value::List - add "is export" to multis is sufficient to make all its tests pass.
04:59 * audreyt meanwhile releases M::Sig 0.53
05:04 * putter notes two failures aren't _quite_ identical:
05:04 putter *** Cannot cast into Hash: VRef <Array>  in current ext/Perl6-Value-List/t/array-lazy.t
05:04 Daveman Hey everyone :)
05:04 putter *** Cannot cast into Hash: VRef <Array::Const>  in current ext/Set-Infinite/t/set-infinite-functional.t
05:05 putter hello Daveman
05:14 Cryptic_K joined perl6
05:14 putter past end of day for me.  how useful would people find a r8869 smoke-js?  I've one running, sort of almost done.  Wondering whether to stay up for it.
05:16 joepurl joined perl6
05:34 BeoTW joined perl6
05:39 meppl guten morgen
05:40 putter r8869 smoke-js up.
05:40 putter guten nacht &
05:45 BeoTW finger meppl
06:00 dduncan my r8875 smoke for haskell/darwin is now up
06:01 dduncan showing only 172 failures of 10,936
06:03 dduncan also, there was no magicals/env.t failure reported this time ... which I couldn't reproduce before anyway
06:04 dduncan the only failures seen now are those 3 ext/ distros
06:04 dduncan now that that's done, I will look into that Makefile.PL simplification thing ...
06:07 dduncan I tried removing line 182 of Makefile.PL ...
06:07 dduncan the one containing:                         ."  setCatchSigwinch False\n"
06:07 dduncan it is making now
06:07 dduncan the Makefile.PL output does *not* say that readline is disabled
06:09 dduncan when I put that line back, it *does* say that readline is disabled
06:10 dduncan so, audreyt and putter, removing the sigwinch line does break things
06:23 dduncan now checking r8877
06:25 dduncan it appears that, under ext/Perl6-Value-List/,  t/array-lazy-overloaded has gone infinite-loopy, rather than just dying
06:25 dduncan (it died 2 revs back)
06:28 kanru joined perl6
06:35 evalbot_8875 is now known as evalbot_8877
06:39 gaal $morning!
06:40 tewk gaal: morning, looks like audrey, dducan, and putter just about have a preview ready
06:41 gaal cool! i just svn upped, am making a clean build and popping into the shower.
06:45 nothingmuch joined perl6
07:10 gaal embedded parrot is still broken on colinux.
07:14 azuroth joined perl6
07:16 dduncan morning
07:16 gaal hey dduncan
07:17 dduncan so, afaik, all tests pass on normal haskell but 3 ext/ distros
07:17 dduncan of those, 2 die and one hangs
07:17 dduncan as of r 8877
07:17 dduncan as of r8875, the hang was a die
07:18 gaal on colinux, extrernal parrot doesn't seem to work at all!
07:18 dduncan I've personally never been testing with anything but the pure normal haskell ... no options linked in at all
07:18 BakedB joined perl6
07:19 dduncan is 'colinux' short for something?
07:20 gaal dduncan: cooperative linux. it's a win32 process with a special driver that lets it run linux at very close to native speeds.
07:20 dduncan I C
07:20 dduncan that's fairly new, isn't it?
07:20 gaal a year maybe?
07:21 gaal from 2004 actually
07:21 marmic joined perl6
07:21 gaal twice as old as pugs :)
07:22 hachi and no longer maintained sadly, last I saw
07:24 gaal hachi: I just got a note from the original author saying that it is, but slowly because he has other projects.
07:25 gaal 2.6.10 isn't *that* bad.
07:26 hachi no, it's not... I use it at my day job
07:28 gaal oh! external parrot isn't even looked up in PARROT_PATH, no wonder it doesn't work
07:28 gaal (I don't have parrot installed, just built in a homedir tree)
07:28 gaal shouldn't it also look there, though?
07:29 gaal in fact it should probably look there first...
07:31 audreyt .rehi
07:31 gaal yo!
07:31 * audreyt resumes releng after lunch
07:37 kcwu joined perl6
07:38 gaal oh, darn, I had an old parrot in my PATH :-(
07:38 gaal audreyt: should PARROT_PATH take precedence?
07:39 gaal removing the old parrot made it work :)
07:39 audreyt gaal: I think so, yeah
07:40 gaal ok, trying parrot emb first
07:41 gaal whee, they both work now.
07:41 gaal fixing the search order.
07:41 audreyt whee
07:44 iblechbot joined perl6
07:50 kattana_ weeeeeeeeeee
08:01 nothingmuch joined perl6
08:04 svnbot6 r8878 | gaal++ | * change search order of external parrot, giving PARROT_PATH
08:04 svnbot6 r8878 | gaal++ |   precedence over PATH so that an old executable installed in the
08:04 svnbot6 r8878 | gaal++ |   system doesn't get in the way of a new one.
08:04 gaal now, to do the same for embedded parrot. would that be in the makefile?
08:05 audreyt Makefile.PL you mean?
08:05 gaal audreyt: my imperative style in haskell is clunky, do you have a suggestion for how to improve it?
08:05 gaal audreyt: yeah :)
08:06 audreyt let me look... (just done with P::C::A triaging, only S::Infinite to go)
08:06 gaal oh, style can wait for after the release :-)
08:07 gaal hmm. Makefile.PL *looks* like it's doing the right thing for embedded parrot, but experiment showed it did not.
08:07 svnbot6 r8879 | audreyt++ | * triaging of Perl6::Container::Array finishes.
08:11 webmind joined perl6
08:20 svnbot6 r8880 | audreyt++ | * ditto for Perl6::Value::List - I gave up fixing from_single and force_todo'ed
08:20 svnbot6 r8880 | audreyt++ |   related tests.
08:24 r0nny joined perl6
08:26 dduncan pulled r8880, and now making
08:30 dduncan with P-C-A, nothing dies now, but one test fails
08:30 dduncan t/array-operations.....FAILED test 26
08:32 dduncan with P-V-L, all tests pass/todo
08:33 audreyt ok
08:33 gaal I'll have a smoke of 8880 up sometime... at t/builtins/control_flow now.
08:33 audreyt retestnig array-operations
08:33 gaal (that's pretty much in the beginning)
08:34 dduncan it sure is
08:34 gaal how long does an ext/-less smoke take nowadays?
08:34 dduncan I think about 2/3 of the time of an ext-with smoke
08:34 dduncan for me, both together take 2 hours
08:34 audreyt ~30min here
08:35 dduncan normal haskell runcore
08:35 audreyt ditto
08:35 dduncan audreyt, what kind of hardware do you have?
08:35 dduncan mind you, mine's 3.5 years old
08:35 audreyt mine's pentium-M laptop
08:35 audreyt strangely, test 26 worksforme
08:35 dduncan its merely dual-1.0 ghz G4
08:36 dduncan I'll try running that test by itself
08:36 gaal dduncan: you do know about test concurrency, yes? if you have a dual cpu box, that can speed things up.
08:36 audreyt gaal: dduncan reported the other day that concurrency doesn't quite work for him
08:36 gaal oh :(
08:36 dduncan well, I set my process count to 2 in config.yml long ago, but it doesn't seem to do anything
08:36 audreyt oh, it requires SMP?
08:37 dduncan according to top, only 1 pugs process ran at a time
08:37 gaal it doesn't require it, but the gains are very low w/o it
08:37 gaal must have bitrotted: it's been a while since I had an smp machine to play with
08:38 audreyt we don't have a MacBook-powered committer yet
08:39 audreyt ok, I've finished this round of triaging here
08:39 lypanov ummmmacbooks
08:39 audreyt I'm making smoke now
08:39 svnbot6 r8881 | audreyt++ | * Finish triaging for Set::Infinite.
08:41 gaal we should add a progress bar to the harness. :-)
08:41 audreyt good idea :)
08:41 audreyt gaal: have some cycles to catch up the remaining chglogging?
08:41 gaal sure
08:41 audreyt woot
08:41 audreyt I'll move back to Taipei from Tamsui
08:41 audreyt the MRT takes ~2hr
08:42 audreyt so maybe a preflight tarball once I hit taipei
08:42 dduncan not ok 26 - kv #   Failed test (../t/array-operations.t line 98, column 3-27)#   Expected: '1'#     Actual: 'undef'
08:42 audreyt will work on other polishing / sanity check on MRT
08:42 dduncan that is the one that failed for me ... it did not say todo
08:42 gaal audreyt: if nobody had edited the chlog after me, you and stevan have to add your piln/mm stuff
08:42 audreyt gaal: sure, will do
08:42 gaal cool
08:44 audreyt dduncan: r8882
08:44 svnbot6 r8882 | audreyt++ | * force_todo an erratic feailure in P::C::A as reported by dduncan.
08:44 dduncan I have pulled r8881, currently retesting Set-
08:46 dduncan now pulled r8882
08:46 dduncan Set-Infinite passes all its tests
08:47 dduncan now retrying container
08:48 dduncan that all passe
08:48 dduncan so I'm not aware of any more pugs test failures
08:49 audreyt excellent
08:49 audreyt that means we probably have a green tree now
08:49 dduncan since its almost 1am, I don't plan to stay up to smoke/test any release candidate now
08:50 audreyt nod
08:50 dduncan if, for whatever reason, that happens in around 10 hours, I can do it then
08:50 audreyt I'll see you tomorrow :)
08:50 dduncan otherwise I'll just smoke the actual cpan release
08:51 audreyt k
08:51 audreyt sleep well
08:51 svnbot6 r8883 | audreyt++ | * capture.t: disable the erratically-failing tests.
09:17 svnbot6 r8884 | gaal++ | ChangeLog to 8884
09:20 gaal smoke up, has errors. looking
09:20 gaal # http://m19s28.vlinux.de/iblech/stuff/pugs-​smokes/pugs-smoke-6.2.10-r8880-linux-norma​l--1138699085-2604--9075-8486-589-1255-900​-0--7434cbdf7ecb18134078ed503cd8b922.html
09:22 gaal they're (almost) all parrot related crashes it seems
09:29 gaal it may be difficult to make the first test in t/unspecced/times.t fit all sizes
09:30 gaal it tends to fail on my machine
09:32 gaal gotta go for ~2hr
09:32 kane_ joined perl6
09:44 svnbot6 r8885 | audreyt++ | * stdrules.t: skip_rest the erratically-failing tests.
09:44 svnbot6 r8885 | audreyt++ |   (/me praises city-wide (albeither intermittent) wireless network)
09:45 audreyt hm, my brain-autocompletion sees "eit" in "albeit" and completetd "either"
09:47 wolverian heh.
10:07 nnunley joined perl6
10:46 gaal albethernet?
10:55 nothingmuch joined perl6
11:10 G2 joined perl6
11:24 leo_ joined perl6
11:29 comet^ joined perl6
11:38 EdLin joined perl6
11:51 Alias_ joined perl6
11:56 Alias_ left perl6
11:58 kcwu joined perl6
12:16 nothingmuch FYI: http://www.osdc.org.il/schedule.html
12:24 drbean joined perl6
12:25 pjmm joined perl6
12:55 RNoise joined perl6
12:58 iblechbot joined perl6
13:14 kolibrie joined perl6
13:16 Baked joined perl6
13:32 G2 joined perl6
13:54 gaal so, how's the takeoff taking off? I'm off to grab some lunch but will be back soon for preflight help
14:06 Qiang joined perl6
14:13 rantanplan_ joined perl6
14:25 sysfault joined perl6
14:25 sysfault left perl6
14:26 Baked left perl6
14:26 hugues23 joined perl6
14:27 cdpruden joined perl6
14:27 Baked joined perl6
14:28 Baked left perl6
14:41 Juerd audreyt: Could you make darcs start niced please?
14:41 Juerd I'll renice the running process
14:42 chris2 joined perl6
14:43 rafl Juerd: Do we still have performance problems?
14:43 Juerd rafl: No
14:43 Juerd rafl: But darcs runs continuously, and the CPU still is a simple pentium 4
14:44 Juerd darcs uses close to 100% cpu
14:44 Juerd (darcs tag, that is)
14:45 rafl Ah..
14:45 stevan_ audreyt: ping
14:50 * Juerd wonders if darcs is used for anything
14:52 webmind what is it ?
14:52 Juerd A revision system
14:53 webmind k
14:54 stennie joined perl6
14:54 lypanov i think its just a backup system no?
14:55 Juerd I suspect it's nostalgia
14:55 stevan_ lypanov: no, its a VCS written in Haskell
14:56 stevan_ Juerd: I think you are correct, I dont think we have many Darcs users left
14:56 lypanov stevan_: yeah... i know
14:56 lypanov stevan_: i mean. its a back vcs only. not the main one
14:56 * lypanov uses darcs principally :P
14:56 stevan_ lypanov: oh,.,. I see
14:56 stevan_ :)
14:56 * lypanov grins
14:56 stevan_ lypanov: yeah I am not sure anyone but you uses it then :)
14:56 * lypanov giggles
14:56 lypanov i know plenty :)
14:57 lypanov i don't use the pugs darcs tho
14:57 lypanov i learnt svk recently anyways
14:57 stevan_ lypanov: no, I mean for pugs
14:57 lypanov not even i use it for pugs ;)
14:57 * stevan_ knows a few non-pugs-hacking darcs users :)
14:57 lypanov ;)
14:57 * stevan_ adds lypanov to the list ;)
14:58 lypanov ;)
14:58 * lypanov is using svk for his new projects tho
14:59 stevan_ svk is very nice :)
14:59 stevan_ clkao++
14:59 * lypanov agrees
14:59 lypanov i chatted to him 6 months back and was more than impressed, but couldn't get the darn thing installed :P
15:00 lypanov now i turn around to find .dmg's that Just Work
15:00 stevan_ yeah, I had the same issues,.. the .dmg made my post-Tiger-upgrade-svn-install job much easier too :)
15:01 * lypanov grins
15:03 rafl pugs tells me: Could not find module `plugins'. But ghc-pkg list shows plugins. Any pointers?
15:05 * robkinyon wishes he could use a .dmg
15:06 vel joined perl6
15:15 cdpruden joined perl6
15:22 SamB rafl: pugs says that, or the configure/build stuff says that?
15:25 rafl SamB: ghc called by Makefile.PL.
15:25 rafl SamB: It looks to me like try_compile is used wrong.
15:26 audreyt rafl: that might be the case, yes
15:26 rafl SamB: You normally call it try_compile($code, $ghc, $run) where $run is optional. But for the plugins package it's called like try_compile($code, $ghc, '-package', 'plugins')
15:26 audreyt I'll prod dons again
15:26 svnbot6 r8886 | audreyt++ | * some light editing of ChangeLog
15:27 SamB what is $run?
15:27 rodi joined perl6
15:27 audreyt $run is a misfeature.
15:27 audreyt removed.
15:28 audreyt rafl++
15:28 Juerd audreyt: Have you read my lines?
15:29 audreyt Juerd: yes, will do
15:29 svnbot6 r8887 | audreyt++ | * remove the $run misfeature in Makefile.PL's try_compile;
15:29 svnbot6 r8887 | audreyt++ |   it was added to probe libreadline, but is not used anymore,
15:29 svnbot6 r8887 | audreyt++ |   and furthermore interferes with hsplugins probing.
15:29 Juerd Thanks
15:29 Eimi joined perl6
15:29 audreyt done
15:30 Juerd audreyt++
15:32 rafl Can I build pugs against parrot without having a parrot executable installed?
15:32 rafl (I have headers and libs, though)
15:32 rafl audreyt: Thanks!
15:33 putter joined perl6
15:34 audreyt rafl: maybe. not sure. probably yes
15:35 * rafl tries
15:35 svnbot6 r8888 | audreyt++ | * bump version to 6.2.11.
15:35 rafl It should result in linking ./pugs to libparrot.so.$(SOVERSION) if it worked, right?
15:35 audreyt yup
15:36 * rafl wonders how it knows which libparrot to use.
15:36 audreyt it asks parrot-config iirc
15:36 rafl I made parrot install both shared and static libraries.
15:36 audreyt nod
15:37 rafl Ah, parrot-config libparrot gives $(LIBPARROT_SHARED).
15:37 audreyt I'm dozing off already... will check back in 9hrs
15:37 audreyt I think I'll call for a round of general test for pre0
15:37 rafl The Makefile needs interpolate that somehow.
15:37 nothingmuch joined perl6
15:39 wolverian hm, reading the "Haskell is not not ML" paper at lectures is a great way to speed up your learning. you really see how much more interesting the paper is than the lecture. :)
15:39 audreyt rofl
15:39 audreyt okay, pre0 up.
15:39 putter anyone going to osdc, please nag Beni Cherniavsky to get his "S-expression syntax for Python with 1st-class macros" stuff online.  tnx ;)
15:40 audreyt changes topic to: 6.2.11-pre0 calls for test: http://xrl.us/jsza | pugs.blogs.com | pugscode.org | pugs.kwiki.org | paste: http://paste.lisp.org/new/perl6 or http://sial.org/pbot/perl6
15:40 putter yay
15:40 audreyt and with that... goodnight :)
15:40 * audreyt waves &
15:40 putter g'night audreyt
15:41 eric256 joined perl6
15:41 eric256 left perl6
15:43 nothingmuch joined perl6
15:44 vel joined perl6
15:45 _bernhard joined perl6
15:45 nothingmuch_ joined perl6
15:47 hexmode joined perl6
15:50 stevan_ is now known as stevan
15:52 putter I've never been clear on whether these represent a problem or not (nor on whether they are fixable):
15:52 putter Compiling Pugs.AST         ( src/Pugs/AST.hs, dist/build/src/Pugs/AST.o )
15:52 putter /tmp/ghc21601.hc: In function `PugsziAST_zdwcharInc_entry':
15:52 putter /tmp/ghc21601.hc:39: warning: cast to pointer from integer of different size
15:53 putter Two more in Pugs.Lexer
15:54 justatheory joined perl6
15:57 putter boom.  build failure with r8888 (HEAD).
15:57 putter Building Pugs-6.2.10...
15:57 putter Build failed: 0 at util/build_pugs.pl line 86.
15:57 putter make: *** [pugs] Error 2
15:57 putter Command exited with non-zero status 2
15:58 elmex joined perl6
15:58 putter anyone successfully make'ed pre0?
15:58 gaal putter: just starting a build now.
15:59 gaal could be the default heap size needs to be increased
15:59 vel joined perl6
16:00 putter sometimes the preN's also get out of sync (local changes) vis svn HEAD
16:00 sub_chick joined perl6
16:00 gaal svn was not very busy when audrey created the image, so probably not this time
16:00 gaal but when you build a pre, you get a fresh config.yml
16:01 putter ok, pre0 has a slightly different makefile than 8888
16:02 vel joined perl6
16:02 eric256_ joined perl6
16:02 eric256_ left perl6
16:02 putter hmm, as a separate issue, pre0 has a somewhat pruned file tree.  was the exclusion of ./debian and examples/cookbook, and the other stuff, intentional?
16:03 putter of simply MANIFEST oversights?
16:05 putter gaal: btw, I hope I didnt leave ChangeLog too messy.  I was separating the "grovel over the log and snag stuff for log" task from the "make log readable" task.  and never did the latter.
16:06 putter rafl: ./debian should be in the release, yes?
16:06 rafl putter: No
16:06 gaal putter: me either :) audreyt++ seems to've done that :-)
16:07 putter rafl: ok. isn't.
16:09 putter gaal: do I have to do anything to have changed config.yml take effect?  uncommented the heap line.  simply make fails with the same error.  hmm, 8885 worked.  svn log'ing...
16:09 gaal putter: yes, perl Makefile.PL again.
16:09 gaal putter: fwiw, I'm now in Pugs.Parser and no errors yet :)
16:10 putter it did the first compile pass, created the .a, and then boom.
16:10 putter ok, will try Makefile.PL
16:10 gaal that does sound like a heap exhaustion problem
16:11 gaal though those usually give evacuate or other errors
16:12 putter I also note r8888 is still saying its 6.2.10, despite r8888 ("change to 6.2.11")
16:13 gaal whoops, I got the same error as you.
16:14 gaal stracing.
16:14 putter 500m heap has no effect
16:16 eric256_ joined perl6
16:16 eric256_ left perl6
16:18 gaal I think I fixed it, testing.
16:19 putter ?
16:19 gaal it created the .a with 6.2.10, but make looked for 6.2.11.a and exited after not finding it.
16:21 putter ah, Pugs.cabal.in ?
16:21 gaal fixed.
16:21 svnbot6 r8889 | gaal++ | bump version number to 6.2.11 in Pugs.cabal.in; this was what was
16:21 svnbot6 r8889 | gaal++ | causing preflight 0 to fail.
16:23 gaal strace++ :)
16:23 putter gaal++ :)
16:24 gaal that's kind of you to say, after I sent you on a wild moose chase :)
16:24 putter Though it's kind of sad that our system is so opaque that the easiest thing to do is to go all the way around it and look at it's interaction with the os.
16:24 Bit-Man joined perl6
16:26 gaal we rely on a lot of existing stuff... to achieve cleanliness with our level of variety would mean implementing a lot of that ourselves
16:28 putter chasing wild goose/moose - caveat - safe distance from a mad goose is order 10 m.   and they will generally flee.  but low order 100 m may not keep you safe from a pissed moose.  and a moose chased is likely to be pissed.
16:29 gaal Attention Preflight team: if you want to test out pre0, please s/6\.2\.10/6.2.11/ in your Pugs.cabal.il *before* running make. If you forgot and get a build error, please rm src/Pugs/PreludePC.hs and pugs && perl Makefile.PL && make
16:29 * gaal never meta moose he didn't like
16:29 putter "it is now safe to move around the Cabal"
16:29 putter ;)
16:30 putter oh, *groan*  # meta moose
16:30 putter that's reaaally bad
16:31 gaal There was once a Metamoose Gaalery
16:31 putter ?
16:32 gaal a gaalery of metamoose. what? :)
16:32 gaal if we meet in YAPC::NA this year, I'll show you :)
16:32 putter oh, oh, oh, ouch
16:34 gaal ouch is what I just said. hitting the power button with my knee wasn't so smart. :(
16:34 putter :/
16:35 awwaiid joined perl6
16:36 putter stevan: is the exclusion of misc/Perl-MetaModel/lib/Perl from the release intentional?
16:37 putter oh, its an empty directory.
16:38 elmex joined perl6
16:38 putter which leaves the disappearance of examples/cookbook as the big oddity
16:39 gaal smoke entered.
16:42 putter 'Edit your settings in config.yml." still seems excessively imperative.  there is really no need to edit config.yml.
16:45 gaal go ahead and soften the phrasing
16:46 putter ok
16:46 Juerd Someone just changed all /does something/ to "do something" in POD I wrote.
16:47 putter pil2js has a lot of regression.  but I guess not something to worry about for this release.
16:47 Juerd No longer "Creates a new object and returns it.", but "Create a new object and return it."
16:47 Juerd Now *that* is excessively imperative.
16:49 Yoti joined perl6
16:50 lypanov it doesn't read well at all for documentation...
16:51 Juerd It's just plain wrong.
16:51 lypanov yeah
16:52 Juerd Documentation isn't comments that translates what the code does.
16:52 Juerd [this method] Creates a new object and returns it.
16:52 Juerd The guy who changed that apparently wasn't aware of that linguistic implication.
16:52 putter so config.yml only contains _some_ settings.  ie, no parrot or embedded info.  ok...
16:52 Juerd What I don't get is that he *changed* my documentation.
16:53 putter welcome to anarchy ;)
16:53 Juerd I can imagine that one writes new documentation in that way, but not that one would actively go and screw existing documentation that is written according to widely used standards
16:54 SamB clearly, this guy is not lazy enough
16:54 Juerd I would change his documentation if he wrote it imperatively :)
16:54 Juerd I'm also not lazy in that respect.
16:54 SamB well, people with bad taste should be lazy
16:54 Juerd SamB++
16:55 putter gaal: how about something like
16:55 putter *** Default build config file created.
16:55 putter    You can edit some settings in $filename (and then rerun Makefile.PL).
16:55 putter thoughts?
16:56 gaal putter: sounds good to me!
16:56 stevan putter: misc/Perl-MetaModel is a very old attempt at a p6 mm.. it is pretty much disposable
16:58 putter now, now.  there are some schools of documentation writing in which the change is clearly the Right Thing(tm).  I don't agree with them, but that isn't to say they don't have principled reasons for believing it, or that the editor didn't think they were correcting obvious problems.
16:59 putter stevan: ok.  so that leaves only examples/cookbook as a possibly odd omission from pre0.
16:59 stevan putter: I guess so,.. I am without enough tuits to participate in releng right now :(
16:59 putter after this release is out, before the next, perhaps we should do some "spring cleaning".
17:00 stevan no doubt, a very very good idea
17:00 putter stevan: 's ok.  $job++
17:01 putter (anyone needs a p5. py, rb, lisp, prolog, etc, hacker in the Boston area, email putter...;)
17:01 putter (dduncan also sounds like he could use some remote p5 work:)
17:02 putter s/use/also use/
17:07 svnbot6 r8890 | putter++ | inc/PugsBuild/Config.pm: Changed
17:07 svnbot6 r8890 | putter++ |   *** Default build config file created. Edit your settings in $filename.
17:07 svnbot6 r8890 | putter++ | to
17:07 svnbot6 r8890 | putter++ |   *** Default build config file created.
17:07 svnbot6 r8890 | putter++ |       You can edit some settings in $filename (then rerun Makefile.PL).
17:07 svnbot6 r8890 | putter++ | to make clear that editing is not required.
17:07 putter 30/30 skipped: BACKEND_PUGS does not support lazy lists yet depends     lazy lists
17:07 putter err, what?
17:08 gaal that's the output of :depends, isn't it?
17:08 putter ahhhh
17:11 stennie joined perl6
17:13 svnbot6 r8891 | putter++ | lazy_lists.t: added a period to the skip message so the test log reads clearer.
17:14 putter is anyone currently actually using embedded parrot?
17:15 stennie joined perl6
17:15 putter if not, it could be worth adding a note to INSTALL "using external parrot is recommended for this release"
17:16 putter anyone at all using embedded parrot?  or recently?
17:17 gaal putter: this build I'm smoking is embparrot
17:17 gaal (...I hope :-)
17:17 gaal ah, and there are crashes, I just noticed :(
17:17 gaal t/rules/from_perl6_rules/prior.t
17:18 gaal t/rules/from_perl6_rules/litvar.t
17:18 gaal pugs: internal error: stg_ap_p_ret
17:19 putter gaal: you have a, err, "normal" platform?
17:19 gaal is colinux normal? :)
17:20 gaal most of the time emparrot does work, these tests must be ram-hungry.
17:22 putter ah.  my x86_64 has been segfaulting for a while, but a good default on such things is "it's just me".
17:22 putter anyone running vanilla linux/bsd (or even windows:)  with embedded parrot?
17:23 putter re "is colinux normal?", err, no.
17:24 * cdpruden builds with parrot embedded, but doesn't really run anything with that backend
17:25 gaal cdpruden: if you do "a"~~/a/, you're using PGE on parroy
17:25 gaal *parrot
17:25 Amnesiac joined perl6
17:25 putter cdpruden: could you do a  ./pugs -Iblib6/lib t/rules/from_perl6_rules/prior.t  and see if it works?
17:25 gaal putter: nope, parrot emb doesn't work on win (tried it)
17:25 cdpruden sure
17:25 putter tnx
17:26 putter if cdpruden's also goes boom, and perhaps even if it doesnt, it would seem an INSTALL commented is warranted.
17:26 cdpruden segfault on 8864; sorry, not on release yet (so maybe that doesn't help) debian
17:27 putter cdpruden++  thanks
17:27 cdpruden updating to see if release is any different
17:30 cdpruden putter, still segfaults with the latest svn
17:30 gaal cdpruden: can you try all of t/rules to see if some pass?
17:31 cdpruden sure
17:31 gaal prove t/rules
17:31 gaal (assuming you have HARNESS_PERL=./pugs and PERL6LIB=blib6/lib )
17:34 svnbot6 r8892 | putter++ | INSTALL: "Using embedded parrot is NOT recommended with this release (6.2.11)."
17:35 gaal http://m19s28.vlinux.de/iblech/stuff/pugs-​smokes/pugs-smoke-6.2.11-r0-linux-normal--​1138728858-3210--11109-10519-590-1457-1109​-0--855915cdd2fdf2c558a609c214440c83.html
17:35 gaal 613 files  11109 test cases: 10519 ok, 590 failed, 1457 todo, 1109 skipped
17:35 clkao hm,, changelog doesn't have new backend jsperl5 ?
17:35 gaal the failed are mostly parrot related.
17:35 gaal clkao: please add anything you thing should be there
17:35 clkao ok.. but i am going to sleep
17:36 putter gaal: audreyt was seeing parrot instability even with external parrot.  so I think parrot is just flakey with this release, and minimizing the flakiness by recommending against embedded is the right thing, regardless of additional cdpruden results.
17:36 gaal that request also goes out to putter, stevan, iblech, kane, rafl, fdglock, etc :-)
17:37 gaal okay. the only other problem on my machine is times, for which it is apparently too fast.
17:38 gaal other than that, I'm good for release.
17:38 svnbot6 r8893 | putter++ | Makefile.PL: "Linking against Parrot is NOT recommended with this release 6.2.11."
17:39 putter INSTALL should probably say test-pir instead of pirtest, but let's punt, as "why expose an api if it may change in the next release".
17:40 putter gaal: your undef.t failures were rules related?
17:41 gaal putter: yes.
17:41 putter is the times.t problem easily fixable?
17:41 gaal it's a rule that doesn't match. I think it's one of my first commits to pugs :-)
17:42 putter :)
17:42 gaal putter: the times.t problem is that there's a busy loop used as a subsecond sleep. of course it's system dependent.
17:42 gaal I don't know of a good way to fix it
17:43 theorbtwo use Time::HiRes.
17:44 putter gaal: if you bump the loop up an order of magnitude, to 10_000, does it pass for you then?
17:45 putter theorbtwo: it's a  <look at time> for 1..1000 { 1+1 } <see if time changed>
17:45 putter err, see if times() changed.
17:47 pasteling "cdpruden" at 69.212.29.10 pasted "putter, from_perl6_rules -- all failed w/ embedded parrot" (111 lines, 5.2K) at http://sial.org/pbot/15682
17:47 putter thanks cdpruden.
17:47 gaal putter: it passes, yes, but that would be gratituously slow on slow machines.
17:47 cdpruden np
17:48 gaal hey wait! eairlier today embparrot didn't always segfault for me. why is it doing that now?
17:51 putter ok, a 5-10 sec difference at 1GHz.  could you try 5000?   that seems perfectly reasonable.  even at 233 MHz :)
17:51 * putter looks at gaal, bemused ;)
17:53 gaal 5k worked too.
17:54 gaal shall I commit?
17:54 gaal seems silly.
17:54 gaal I mean, what is it testing? what cosmic significance does 5k have? :-)
17:55 gaal but ok, ok :-)
17:55 putter lol
17:57 putter actually, maybe that's not such a good idea
17:57 svnbot6 r8894 | gaal++ | times.t: update fudge value so test passes on fast machines
17:59 putter oops.  ah well.  fine.  putter turns out to be running at 3GHz.   so that's perhaps a 15-30 hit on a 1GHz, twice that on 500MHz.  eh, still fine.
17:59 putter but embedded was working for you earlier today?
17:59 putter There was a change to Parrot.hsc at one point.
18:00 putter s/15-30/$1 sec/
18:00 gaal remember when?
18:00 putter checking...
18:01 putter r8848
18:01 gaal no, it worked after that
18:01 putter ah.  curious.
18:02 putter the only other change to Parrot.hsc was the search path change of r8878
18:02 gaal that was my own fix :)
18:02 putter yeah :)
18:03 gaal well, rebuilding HEAD now to see what happens there.
18:03 putter so it worked then... hmm...
18:04 putter and it doesnt look like anything significant has changed since.  Makefile.PL tweaking which shouldnt have an impact.  ???
18:05 gaal no :(
18:05 gaal well, this build will be ready in a while and if it works there, i'll just diff the trees.
18:05 gaal bbiab &
18:06 vel joined perl6
18:07 putter r8889  make test  passes.
18:09 putter so the only issues I have are - what happened to examples/cookbook?; - pil2js is buggy (but I'm fine for release); - embedded parrot seems to have issues (but I'm fine for release); - ChangeLog is still marked as "needs editing".
18:11 feng123 joined perl6
18:11 f0rth joined perl6
18:14 putter bbiab
18:15 Limbic_Region joined perl6
18:26 gaal eep. r8894 embparrot segfaults for me. I'll try with external parrott...
18:28 putter given that this is the "anniversary release", and the first in a while, and thus likely to draw attention, let's be careful with expectation management.
18:28 gaal the first anniversary release in a while :)
18:29 putter ie, there are regressions, but we're releasing anyway because it's been a while, and we're intentionally not fixing some things because as soon as this release is out the door, we're doing a major core change / overhaul.
18:29 putter :)
18:35 feng joined perl6
18:36 putter audreyt: btw, re "triage", my understanding is that's "focused on assessment, quick look at everything".  tag the dead, tell the mobile to move, briefly touch the critical, and the key "and move on".  so when it comes down do debugging individual files with known issues, I don't think that's "triage" any more.  random thought. :)
18:38 Yoti left perl6
18:38 gaal external parrot does work for me.
18:39 Debolaz joined perl6
18:42 sub_chick left perl6
18:45 putter med: http://en.wikipedia.org/wiki/Triage  http://nielsenhayden.com/maki​nglight/archives/006708.html soft: http://software.ericsink.com/​articles/Four_Questions.html   http://mail.python.org/pipermail/py​thon-dev/2000-September/009414.html
18:46 putter http://geekswithblogs.net/srkpra​sad/archive/2004/08/20/9961.aspx http://www.stickyminds.com/sitewide.asp?ObjectId=​6323&amp;Function=DETAILBROWSE&amp;ObjectType=COL
18:48 putter http://channel9.msdn.com/S​howPost.aspx?PostID=26641
18:48 putter ok, that's it ;)
18:51 putter one more  http://wiki.kdenews.org/tik​i-index.php?page=Bug+Triage
18:51 xinming joined perl6
18:58 vel joined perl6
19:04 Juerd Will Perl 6 have goto?
19:04 gaal טקד
19:04 gaal םםפד
19:04 gaal oops
19:04 gaal yes.
19:05 Juerd Even goto LABEL?
19:05 gaal Juerd: Test.pm even uses goto &sub
19:05 gaal I think it probably will have goto LABEL, yes.
19:06 gaal but then again, I'm basing this on recollection of a pretty old post of Larry's.
19:07 Juerd goto &sub is different, and not what I meant. I was not clear enough.
19:07 Amnesiac longjmp!
19:10 Limbic_Region Juerd I asked this question myself on the list
19:11 Limbic_Region I am searching for the reply so that I don't tell you something incorrectly
19:12 Amnesiac L~R
19:13 Limbic_Region salutations Amnesiac
19:14 putter audreyt: ok, so maybe the ongoing use of "triage" is fine.  never mind. ;)
19:14 Limbic_Region Juerd - If you want to search for "Formal Parameters To While Blook", the first response from luqui implies the answer is yes but not explicitly
19:15 Limbic_Region and did the cpu issue on feather get resolved?
19:16 Juerd Limbic_Region: Thanks.
19:16 Juerd Limbic_Region: It was a memory issue. The memory was somehow very slow.
19:17 Juerd Limbic_Region: It was resolved by using a combination of three entirely different memory modules AND hard coding the timings in the BIOS. Typically, a very bad setup. But at least it works.
19:17 Limbic_Region odd - I have never heard of memory being slow.  I have heard of having the wrong kind of memory and fast memory but not slow memory
19:17 beppu joined perl6
19:17 xinming audreyt: in 8840 revision, Pugs will segfault, But Now, the segfault appear again. :-/
19:17 putter goto with in a block is straightforward.  so even if the language doesnt provide it, an ast macro can.  goto's across blocks are much more problematic, and thus much more difficult to do a macro for, and perhaps less likely to be in the language.
19:17 Juerd This was very fast memory, but not in this board.
19:17 Juerd It's the most expensive non-ECC memory available to me.
19:17 Juerd non-ECC non-registered
19:17 Juerd It's a mobo/ram incompatibility
19:18 Juerd That even only occurs with dual sided memory. The 512 MB dimm I had from exactly the same brand and series, did work well.
19:20 putter what happens if the memory's actual latency > what the mobo thinks it is?
19:23 putter http://dev.perl.org/perl6/doc/design/syn/S04.html says "To jump into the next when block you must use a goto."
19:28 xinming         for @{$<pairs>} => $key, $val {
19:28 xinming             say "Key: $key";
19:28 xinming             say "Val: $val";
19:28 xinming         }
19:28 xinming by the way, Is this example above valid?
19:29 putter Apr '05 larry has goto LABLE: http://groups.google.com/group/perl.perl6.la​nguage/browse_thread/thread/c06c8f07c23cab9/​4c6c4e39f58b2088?lnk=st&amp;q=goto+perl6.lan​guage&amp;rnum=2&amp;hl=en#4c6c4e39f58b2088
19:29 putter LABEL even
19:33 xinming anyone here would confirm my question?  for @{$aRef} => $odd, $even { ... } would work?
19:33 Juerd I remember that post, and my reply :)
19:33 Juerd Thanks, putter
19:34 xinming It's from S05.
19:34 putter np.  there have been other mentions since, but perhaps not in a thread which included @larry[0].
19:35 putter xinming: shouldnt that be "->" rather than "=>"?  but I could be wrong.
19:36 xinming putter: That's what I would ask, as => in S05 appeared many times.
19:36 xinming which doesn't use '->'
19:41 putter xinming: yeah, that looks like a bug.  note that S03 is using ->.  good catch.  I'm not sure how one reports Sxx bugs.  anyone?
19:41 xinming putter: hmm, I might post it to p6l
19:42 xinming bbl
19:42 putter anyone?  how does one get Sxx bugs fixed?
19:43 gaal putter: p6-l, I guess.
19:43 putter ie, typos, rather than referals for reflection
19:43 gaal bbl &
19:44 * putter has very fuzzy recollection of someone saying "email me patches"...
19:44 putter ok, tnx gaal.
19:48 putter http://aspn.activestate.com/ASPN/M​ail/Message/perl6-language/2620898   the design documents are all available from http://svn.perl.org/perl6/doc/trunk .  And I've already volunteered to review/apply patches to the design documents or forward them to the appropriate people for review -- there just haven't been a lot of patches submitted.  (p6l and/or p6c are probably appropriate forums for this.)
19:48 putter - Patrick R. Michaud
19:49 putter I'm not sure whether that means emailing Pm directly, or p6l.  :/
19:57 putter
20:01 putter audreyt: Since it's already Feb 1 where you are...  My congratulations and thanks for a year of Pugs.  It's a beautiful beast.  Thanks for letting us come along for the ride.
20:01 lypanov :D
20:05 putter xinming: so yes, emailing p6l sounds like a plan. xinming++
20:11 vel joined perl6
20:13 Juerd Hm, I even already explicitly asked if goto will be in Perl 6
20:14 Juerd And Larry's answer was: Hmm, well we do still have bare labels for loop control, so I'd
20:14 Juerd probably still put goto LABEL into the same category.
20:14 Juerd Which I interpret as "yes"
20:14 Juerd This will need a hell of a lot of tests.
20:15 Juerd And definitions. What does LABEL: my $foo; do, the second time it sees "my $foo", after being sent back to LABEL with goto LABEL?
20:15 Juerd In Perl 5 it does nothing.
20:15 Juerd (The value of $foo is not reset to undef)
20:15 Juerd This is intuitive or counterintuitive depending on how much you know about Perl's guts.
20:17 Juerd Message-ID: <20050420171714.GV19716@c4.convolution.nl>
20:28 Limbic_Region seen robkinyon
20:28 jabbot Limbic_Region: robkinyon was seen 1 days 1 hours 36 minutes 49 seconds ago
20:29 dduncan joined perl6
20:31 hexmode joined perl6
20:34 rodi left perl6
20:41 robkinyon limbic_region: pong
20:43 putter Juerd: "L: my $foo;" seems clearish, given that declarations are currently described as relocated to the top of the enclosing block.  but what about  L: my $foo = bar();  ?  Does the initializer run again?  So yes, goto will need some limits.  Perhaps a heuristic might be "goto can only be used in cases which don't drive us nuts". ;)
20:44 Juerd putter: That means: goto cannot be used.
20:45 * putter is actually comfortable with the initializer being rerun in that case.  it's just a variant of  my $foo; loopy { $foo = bar(); }.
20:46 Limbic_Region robkinyon - just wanted to let you know that I finished my quest to get arbitrarily nested loops that were dynamically constructed finished (in C)
20:47 putter and actually, on reflection, working out a semantics for general goto might force the creation of a clean semantics for everything else :)
20:47 SamB general goto?
20:47 SamB that sounds... quite evil
20:49 robkinyon limbic_region: How many lines?
20:49 Limbic_Region 158
20:49 putter oh no no, evil would be being able to goto into the middle of a sub in a continuation in another thread.  hmm...! ;)
20:49 Limbic_Region but really the meat is only about 100
20:50 robkinyon huh ... that's pretty impressive
20:50 robkinyon i'd like to see it, if you don't mind
20:50 Limbic_Region I am looking to make it general purpose now which will stretch my C knowledge even further
20:50 putter Limbic_Region: strictly nested?  or allowed to overlap?  (eg, Duff-ish games)
20:51 Limbic_Region robkinyon - well, I will share it with you most definately but I would prefer to wait until I have a generic solution as the project I wrote it for is uber top secret
20:52 putter reminder that #perl6 is logged...
20:52 Limbic_Region putter - not sure, can you explain the difference
20:52 DaGo joined perl6
20:54 putter http://en.wikipedia.org/wiki/Duff's_device  http://www.lysator.liu.se/c/duffs-device.html ,  re strictly nested, whether
20:55 putter the loops are like { { {} } {} } or can be like {a {b {c a} b} c}  ;)
20:56 putter where abc are subscripts identifying the the urm, size, of the particular loop.
20:56 frederico joined perl6
20:56 Limbic_Region putter - strictly nested though I can see how to support overlap
20:56 dduncan I am now smoking pre0
20:56 putter neat :)
20:56 putter dduncan: dont forget to edit Cabal
20:57 putter first
20:57 Limbic_Region the problem with supporting overlap is one of bookkeeping since each iteration the values (or even the number of loops) below the current point are subject to change
20:57 dduncan I edited config.yml ... why would I edit something else?
20:57 putter because pre0 has a bug which prevents successful compilation? ;)
20:59 putter In Pugs.cabal.in, change 6.2.10 to 6.2.11
20:59 dduncan something like that would prevent compilation?
21:00 dduncan nevermind
21:00 dduncan we want correct smokes
21:00 putter it makes a 6.2.11.a and looks for .10, or the reverse, or some such.
21:01 dduncan I also notice that that file says copyright 2005 only, not 2006
21:01 putter oo, where
21:01 dduncan line 5
21:01 dduncan Pugs.cabal.in
21:02 dduncan in the pre0 version
21:02 * putter greps for all 2005...
21:03 xinming joined perl6
21:03 dduncan I always do my copyrights as a range; in this case, 2005-2006 ...
21:04 dduncan some people do comma lists, but I find the range works better
21:06 putter ok, that looks like the only one.  good catch dduncan
21:07 putter dduncan: would you like to fix it, or shall I?
21:07 dduncan I'll do it
21:07 putter k
21:08 putter dduncan++
21:10 dduncan I just did a search myself of my pugs checkout dir, and there are a lot more statements just saying copyright 2005
21:11 dduncan eg, in debian/copyright
21:11 dduncan what was your search pattern that you only found one occurance?
21:13 putter debian isn't in the release ;)
21:13 dduncan so the release excluded some of the repository?
21:13 putter re cabal, it looks like the example format is Copyright: Copyright (c) 2004, Angela Author   http://www.haskell.org/libraryInfras​tructure/proposal/index.html#EXAMPLE
21:13 putter yes
21:15 putter the exclusion I remain unclear on is examples/cookbook.  maybe they are out of date?
21:15 putter rafl says the debian exclusion is the RightThing.
21:16 dduncan reason?
21:16 putter hmm, but the cabal example format from the proposal soo doesnt match the file.  so ignore me.
21:17 putter debian?  no idea.  that's rafl's bailiwick
21:17 BakedB left perl6
21:18 dduncan even if it's excluded, is there any disadvantage to adding 2006 to it?
21:20 putter that's the thing.  I have no idea what constraints apply to ralf's stuff.  I could certainly imagine it mattering.  We're in picky debian legal land there.
21:20 dduncan fine, I'll leave debian alone
21:20 dduncan now, I was also going to change a few core "2005, 2006" to "2005-2006"; do you see any problem with that?
21:21 putter I have no opinion.
21:21 putter though I note that it's second guessing whoever edited the files in the first place.
21:22 putter not to be negative or anything ;)
21:22 dduncan I suppose, for minimal impact, I can then just format my change with the comma, to be consistant
21:23 putter the ChangeLog still says "needs editing"... ;)
21:25 vytautas joined perl6
21:25 justatheory joined perl6
21:25 xinming joined perl6
21:31 dduncan okay, I did just the one file
21:35 svnbot6 r8895 | Darren_Duncan++ |  r2279@darren-duncans-power-mac-g4:  darrenduncan | 2006-01-31 13:27:41 -0800
21:35 svnbot6 r8895 | Darren_Duncan++ |  Pugs.cabal.in : added 2006 to the copyright
21:38 eedthf joined perl6
21:39 dduncan I also notice that whether or not the prelude is generated seems somewhat random
21:40 dduncan like it tends not to happen on the first make, but does later, for example
21:41 dduncan rather than smoke the pre0, I'll smoke the repository instead
21:42 dduncan and Larry already posted that pre0 has some problems anyway
21:42 dduncan on p6c
21:44 putter the INSTALL file and Makefile.PL messages both now say embedded is not recommended with this release
21:44 putter dduncan: re prelude, the regular pugs prelude, or the pil2js prelude?
21:45 dduncan regular pugs
21:45 putter oh.  that would be a new one.  what happened?
21:45 dduncan this has happened for weeks, if not months ... where the pugs prelude is sometimes not generated
21:46 dduncan besides the config.yml being set to do it, I'm not sure how to kick this
21:46 dduncan the first run of pre0 did not build it, but my repository with r8895 does build it
21:46 dduncan s/repository/checkout/
21:47 dduncan anyway, I'm currently smoking r8895
21:49 putter curious.
21:49 putter platform?  ghc rev?  gcc rev?
21:50 dduncan platform is darwin
21:50 dduncan the behaviour occurred on both X 10.3.9 and 10.4.4
21:50 dduncan gcc 3.3 and 4.0 respectively
21:50 dduncan ghc 6.4.1 on both
21:51 putter hmm.  I wonder who else is using darwin.  whether they are seeing something similar?
21:52 dduncan sometimes it seems that a clean and a rebuild kicks it into being made
21:52 dduncan as if some other generated file's presence was used as a clude that the prelude should be done
21:52 dduncan s/d//
21:54 xinming1983 joined perl6
21:59 Eimi joined perl6
22:01 putter Ok, that's just depressing.  Google search hit an Apr 2002 ruby-talk post I did looking forward to computational algebra support in p6.  Almost 4 years ago.  Sigh.
22:05 comet^ Don't worry Marvin^Wputter, things will get better. :P
22:06 Odin- Ohhh. Here I am, brain the size of a planet...
22:07 Juerd I've got this terrible pain in all the diodes down my left side.
22:07 Odin- Guess which part they never replaced?
22:07 Juerd His personality :)
22:07 Odin- That, too.
22:08 Odin- How old did he get to be, again? Twenty-seven times the universe?
22:08 putter comet^: Oh my.  Until I saw Odin-'s comment, I thought you meant Marvin Minsky.  And got _very_ depressed.  Hey parking cars is fun!  >:)
22:09 * Odin- shrugs. Maybe he did.
22:10 putter (Minsky, ai researcher since forever, got to grow old while watching people _not_ explore his ideas.)
22:10 comet^ Nah, Odin- was correct.
22:11 comet^ That is depressing indeed.
22:13 xinming joined perl6
22:19 * putter listens to Marvin's 1980's singles... ;)
22:22 Juerd In 2002, people wondered if learning Perl 5 was worth the trouble, with Perl 6 in development.
22:22 Juerd In 2006, people google for Perl 6, wonder why their distribution hasn't packaged it yet, and start learning Perl 5 anyway.
22:23 Odin- Heeh.
22:23 Odin- They can't be using Debian (unstable, at least), then.
22:23 comet^ Stephen Fry was a fitting choice for the voice of Marvin in the HHGTTG movie.
22:24 Odin- He was?
22:24 comet^ Imo.
22:24 rhesa Odin-: testing has a prehistoric version of pugs as well ;)
22:24 * comet^ runs sid as always. :P
22:24 Odin- rhesa: Neat.
22:26 rhesa how risky would it be to switch my workstation to unstable (from stable)? note that i'm running amd64 (with a i386 chroot for testing stuff)
22:26 putter Ooo, maybe we can do an April 1st pugs release! :)
22:28 comet^ Depends what you have running really and what do you mean on stable, rhesa.
22:31 rhesa i don't think i'm running anything out of the ordinary. gnome, firefox, svn clients, perl, xmms, that's about it. oh, and OO in the chroot. I mostly work remotely through ssh, so local stuff is just email and browsing, and toying around with perl[56] :)
22:31 comet^ I think it would be relatively painless then.
22:31 comet^ You might have to edit a few configs here and there but that is all.
22:32 comet^ (Afaik)
22:33 comet^ I don't know if there is any official way to switch between versions, but I've always edited /etc/apt/sources.conf and just dist-upgraded.
22:33 rhesa i wanted to see which pugs version is in unstable, but packages.debian.orgg is down
22:33 rhesa yes, that's what i would do too.
22:33 comet^ again?
22:33 comet^ :\
22:34 rhesa since jan 6th apparently
22:34 comet^ Version: 6.2.10-4
22:34 comet^ If that means anything thats what apt-cache show pugs gives.
22:34 rhesa testing has 6.2.9
22:34 Juerd audreyt: Please have a look at darcs; it still runs with nice 0
22:35 Juerd audreyt: Darcs is resource hungry, and this affects build speed for people who are waiting for it.
22:35 rhesa i suppose the upcoming release will be in unstable pretty soon, right?
22:35 comet^ Yeah.
22:35 comet^ (I guess so, unstable is usually pretty quick)
22:36 Juerd rhesa: Ask rafl
22:36 putter You would think there would be a webpage devoted to Marvin quotes.  But no.  Or if there is, I can't find it.  Unless it's the link where the host was down.  And even if I did find one, the link would be broken soon anyway.  And the web only gets worse from here on.
22:36 comet^ oh he's the maintainer of that deb. package
22:36 comet^ wikiquote?
22:37 comet^ http://en.wikiquote.org/wiki/The_Hitch​hiker%27s_Guide_to_the_Galaxy#Marvin_.28The_Paranoid_Android.29
22:38 rhesa thanks guys. i'll look at upgrading my box this weekend.
22:41 putter Second billing to a dilettante.  And a human.  With human selected by committee.  That sounds like fun.
22:41 putter drat.  s/human/humor/
22:42 putter thanks comet^ :)
22:42 comet^ yw.
22:43 comet^ doh.
22:44 comet^ I'm really slow getting humor today.
22:45 putter lol :)
22:47 comet^ Sorry.
22:47 comet^ :)
22:49 putter Oh, so np.  I love delayed reaction "gotchas". ;)
22:53 putter If the timing works out, one could have sooo much fun with a Apr 1 release.  "This weeks release of pugs adds support for ruby, python, R5RS scheme, Common Lisp and APL, all passing their respective language test suites.  No, we're not kidding.  Really.  This is not an April Fools joke!".  And maybe, perhaps, it just might not be...
22:53 integral if not, we could stretch out development to next year...
22:54 putter laughing.. too... hard... to... type...
22:55 drbean joined perl6
22:55 Juerd It'd be fun if the first real release of perl 6 would be accompanied by no more than a mere notice saying "Okay, here it is. Have fun with it."
22:55 Juerd That would be the ultimate anticlimax after years and years of development.
22:58 putter :)
22:59 putter does Haskell have parallel (ie, declarative) array/matrix ops?
22:59 putter s/Haskell/ghc/
23:03 stevan comet^: Marvin's voice in the movie was Alan Rickman, not Stephen Fry actually
23:03 integral putter: there's some maps in Data.Array.IArray, but nothing like matrix multiplication
23:03 stevan Alan Rickman was also Professor Snape in Harry Potter movies too :)
23:03 comet^ hm, really?
23:04 comet^ I distintly remember Steven Fry's name.
23:04 stevan comet^: just double checked on the website
23:04 stevan Stephen Fry was the narrator
23:05 stevan http://www.imdb.com/title/tt0371724/
23:05 comet^ ah
23:05 comet^ you're right.
23:05 comet^ Guess I was wrong.
23:05 putter http://haskell.org/hawiki/HaskellOhSix has some fun numeric prelude discussion
23:05 * stevan just watched it the other night, otherwise I would not have remembered or cared :)
23:06 putter There's also http://haskell.org/hawiki/LinearAlgebra and http://haskell.org/hawiki/FunDeps
23:07 putter integral: tnx
23:08 * integral did some matrices coursework over finite fields (one version used type-level numbers!) with it; it's usable just
23:09 putter oops, late.  bb(much)l. good night &  (still wondering about examples/cookbook)
23:09 putter hmm, type level numbers... :)
23:14 cdpruden joined perl6
23:16 Juerd Does Larry read this channel, did anyone contact him directly, or can he read minds? :)
23:16 comet^ His perl script reads minds.
23:18 Juerd comet^++
23:18 Debolaz2 joined perl6
23:20 Khisanth joined perl6
23:44 xinming1983 joined perl6
23:45 xinming1983 ?eval my $a = "ABC"; $a ~~ s/A//;
23:46 xinming1983 ?eval "hello"
23:46 evalbot_8877 is now known as evalbot_8895
23:46 evalbot_8895 bool::true
23:46 xinming1983 died... :-P
23:46 evalbot_8895 "hello"
23:46 xinming1983 how to compare 2 strings in perl 6?  is "ABC" == "abc"  correct?
23:51 Juerd xinming1983: Are you serious?
23:51 Juerd No, you're not.
23:59 xinming1983 ok, I think I know, using ===. :-/
23:59 bluefeet joined perl6

| Channels | #perl6 index | Today | | Search | Google Search | Plain-Text | summary

Perl 6 | Reference Documentation | Rakudo