Think twice before running "make install" for Pugs | moritz.faui2k3.org/irclog/ | pugscode.org | sial.org/pbot/perl6 | ?eval [~] <m oo se> | We do Haskell, too | > reverse (show (scanl (*) 1 [1..] !! 4)) | "Perl 6 Today" video from YAPC::Asia: xrl.us/v6op
Set by agentzh on 6 May 2007.
00:05 DarkWolf84 joined 00:09 pbuetow joined 00:47 nipotaway is now known as nipotan 00:49 B0HR1 joined
jrockway B0HR1: try to be civil, please 01:00
B0HR1 fuck you jrockway
you got me banned in #perl
i'm not saying a thing in #perl6 due to my own consideration
Juerd This is not #perl.
If you're not here for Perl 6 stuff, please do leave.
B0HR1 this is not #perl. i'm not saying anything here.
i'm a fan of anything perl 01:01
buu Juerd: He was accusing people of not using perl6 because of audrey's sexuality in #perl. He was then banned.
Juerd buu: Could you add some parens so I know how to read that? 01:02
wolverian meh. trolls these days. no class at all.
Juerd Big precedence issue there.
buu Juerd: Which part is confusing?
jrockway Juerd: in #perl == location of accusing, sexuality == reason for not using perl6
Juerd buu: because of--relates to accusing people, or to not using perl 6?
jrockway anyway, i think it would be best to drop the subject 01:03
Juerd Probably
B0HR1 i'm not saying a thing while son of a bitch jrockway frothing his mouth
jrockway the people in question don't need their time wasted with this shit
B0HR1 then demand civility
Juerd B0HR1: You are saying things here, and frankly what you say makes me a bit angry.
The way you say it isn't great either 01:04
B0HR1 i have to defend myself out of neccesity
jrockway B0HR1: please either sit back and listen or contribute intelligently
perl6 will be released faster if *you* help
01:20 B0HR1 left
meppl gute nacht 01:20
jrockway guess he got bored :) 01:29
01:30 navigator joined
rhr Juerd: does feather swap much? swapstorms could cause what you're seeing 01:58
processes in R or D state contribute to loadavg, you can find them with ps uaxw|grep ' [RD][ slLN<+]' 01:59
also, 18/sec timer tick hasn't been around for years, it's either 100, 250, or 1000Hz 02:00
actually they're in the process of eliminating periodic ticks altogether 02:02
02:06 audreyt_ joined 02:13 khisanth_ joined 02:15 jisom joined 02:16 araujo joined 02:17 khisanth_ is now known as Khisanth
rhr Juerd: anyway, next time the load spikes just do: while :; do ps uaxw|grep ' [RD][ slLN<+]'; done (from a reniced root shell) it gives a much clearer picture than top of what's chewing cpu or getting stuck in io 02:27
jrockway why in a while loop? 02:28
rhr to catch things that only run for a little while, top misses a *lot* 02:29
jrockway ah, i see
i might have to run that on our work dev server, which routinely has load averages of 3000-5000 for no apparent reason 02:30
rhr wow 02:31
jrockway (other than 60 developers with their own apache + huge mod_perl app)
02:38 ggoebel joined 02:44 araujo joined
Khisanth jrockway: lots to short lived processes? 02:45
jrockway no idea
fortunately i am not a sysadmin, so i get to complain without trying to fix the problem ;)
the sysadmins are uber-incompetent, that's probably the problem 02:46
we have a huge app that uses no CPAN modules because apparently the sysadmins didn't know how to setup CPAN (and admittedly the early developers weren't that bright either)
now we just copy the CPAN modules into our app because the sysadmins won't install them
(etc.)
it's painful
Khisanth things seem to crawl even at 10, I don't think I would have the patience for 3000
jrockway it sucks at 3000 02:47
even at 1 it's terribl
it takes about 15 seconds to load a static page
Khisanth I can tolerate 2 :)
jrockway imagine how fun the debugging cycle is (with no unit tests of course)
Khisanth I think we are living in the same world :)
jrockway i'm trying to move everything to my workstation; much easier
Khisanth well mine is slightly better!
jrockway unfortunately that has to be windows because we have a dumbass VPN at work 02:48
they go out of their way to make it impossible to write good software, sadly
jrockway <-- slightly bitter
Khisanth s/slightly/much/
Khisanth is playing around with Catalyst :)
I guess you aren't using that for work? Catalyst and no modules seem like a match made in Hell 02:50
jrockway yeah, we use mason 02:51
it's basiclly PHP with perl syntax
very depressing
Khisanth it almost sounds like we work for the same company :)
jrockway i unfortunately started programming professionally right when frameworks were becoming popular, so i never really experienced the horrors of "web 1.0" websites
jrockway is performics / doubleclick (now google) 02:52
Khisanth I don't think Mason is that terrible, I mean you don't HAVE to mix stuff like that ...
jrockway right
we're trying to factor things out into modules
but everyone is very junior and just plain can't program
(not everyone, but the bulk of people that write code everyday) 02:53
buu jrockway: Come play with us
We're way cooler.
Khisanth impossible, you're there! 02:54
jrockway Khisanth: ah, but people that know how to program have meetings all day
plus i have to work against the coworkers that don't know how to program
it sort of sucks
Khisanth that makes no sense at all do they really enjoy the meetings?
jrockway i don't understand meetings 02:55
imagine spending 2 hours talking about what day to have a "test driven development" training session
uh oh that paves the way for a new rant
so a few weeks ago we paid some consultant to come in and talk about TDD 02:56
Khisanth I thought you said there were not unit tests?
jrockway i'm getting there :)
Khisanth s/not/no/
jrockway i capitalized on the "testing excitement" and said "we can't do TDD if nobody knows that Test::More is"
so for the next 3 weeks, we're having training for all perl devs to learn how to test
which is great
unfortunately 02:57
this week the guy is coming back to teach us how to write tests
except he only knows java
and none of the devs here even know what a test is
can you say, poor planning?
*sigh*
Khisanth this sounds very confusing
Aankhen`` Ouch.
jrockway mangement thinks that paying some consultant will magically write tests for us 02:58
Khisanth only if you are paying the consultant to write tests!
jrockway i say we should learn the syntax first, and then tweak our skills after we've gotten some experience with Test::* and friends
Khisanth: exactly, we're not :)
he is providing small-group training sessions
training = "here's how i would test your code" 02:59
it's really going to be a nightmare since all of our logic is in mason
and we have a homegrown ORM
Aankhen`` Double ouch.
jrockway which ... uh ... i'm not even going to comment on
it's really weird
most of the developers have never heard of "refactoring"
they don't know that programmers NEVER cut-n-paste
i guess... nobody ever told them that? 03:00
it's scary.
i'm mostly sticking around to accumulate capital for a consulting company or something :)
google just bought us, which could be good... but C++/Java/python suck pretty hard... *sigh* 03:02
Khisanth C++ and Java? I guess it's not so bad that I have no chance of getting hired by Google :) 03:04
jrockway google is not that great 03:05
they don't pay very well
you're stuck in a cube all day
the code is a mess
it's pretty depressing, i'm told
Khisanth that does sound inferior to current situation
jrockway (i won't know for another few months)
cubes suck for developers... worst idea ever
Khisanth which isn't really bad :)
just slightly messy code 03:06
jrockway i think the key to good code is working someplace where all the software is haskell or lisp :) 03:11
people that write bad code don't tend to be interested in those langauges :)
03:15 perlDreamer joined 03:24 thoughtpolice joined
rhr ?eval sub baz(Code $x, *@y) { $x.(@y) } baz { say @^x } , 1, 2, 3 03:25
03:25 evalbot_r16281 is now known as evalbot_r16282
evalbot_r16282 OUTPUT[123ā¤] Bool::True 03:25
rhr TimToady: it looks to me like Perl-6.0.0-STD.pm doesn't allow the space before the , there, is that right? 03:26
i.e. it has to be unspace?
03:27 weinig joined
rhr ?eval sub baz(Code $x, *@y) { $x.(@y) } baz { say @^x }\ , 1, 2, 3 03:27
evalbot_r16282 Error: ā¤Unexpected ","ā¤expecting "."
rhr in fact, shouldn't I need a ; before I call the sub? 03:32
or \n 03:33
spinclad heh. that does look like { $x.(@y) } is taken as the last arg of &sub; i expect that's how it's parsed; le'me see if that's what the synopses say. 03:38
rhr how is S04:167 implemented in Perl-6.0.0-STD.pm? agentzh++ 03:41
the endlist part of token block allows anything, no? 03:43
spinclad yes, S04:167 applies, in that C< sub ... {...} baz ... > is two terms in a row. 03:55
i think you're right, there.
rhr yep, I'll trust S04 here 03:56
03:58 araujo joined 04:08 araujo joined 04:11 penk joined, dduncan joined, stevan_ joined 04:17 offby1` joined
svnbot6 r16283 | rhr++ | Add tests for space/unspace after blocks 04:26
04:47 audreyt_ is now known as audreyt 04:58 offby1` is now known as offby1 05:05 BooK joined
svnbot6 r16284 | agentz++ | [docs/feather/script] 05:14
r16284 | agentz++ | - niced all the shell commands demanded by Juerd++
05:17 amnesiac joined 05:23 lisppaste3 joined
rhr ?eval sub named_array2(:@x, @y) { (+Ā«@x, 42, +Ā«@y) }; named_array2(:x, :x) 05:50
05:50 evalbot_r16282 is now known as evalbot_r16284
evalbot_r16284 Error: Missing required parameters: @y 05:50
rhr should the second :x bind @y to (:x), or is it an error?
I'll write the test assuming it should succeed, let me know if that's wrong... 05:52
05:57 rfordinal joined
rhr ?eval sub named_array2(:@x, @y) { (+Ā«@x, 42, +Ā«@y) }; named_array2(1, 2) 05:59
evalbot_r16284 (42, 1)
rhr is that what it's supposed to do? 06:00
svnbot6 r16285 | agentz++ | [docs/feather/script] 06:15
r16285 | agentz++ | - more nice :)
r16286 | rhr++ | Add tests for foo(:x, :x) and foo(:x, (:x)) type stuff 06:27
rhr @tell TimToady could you check the tests marked XXX in t/blocks/sub_named_params.t for correctness? thanks! 06:29
lambdabot Consider it noted.
Juerd rhr: Reniced root shell isn't even needed; the box stays responsive! 06:31
rhr: But thanks a lot for the hint! 06:32
rhr: And nope, it doesn't swap at all. Once it starts swapping, it's just half an hour away from dieing
rhr good to hear it's doing better :) 06:34
Juerd Killing idle pugs processes helped 06:36
I don't know if the constant sigalrms they had, or something else, kept the system boiling
But it wasn't visible on any of my radars (top, vmstat, iostat, dstat) 06:37
Yet killing them did help instantly
rhr yeah, I'm not surprised to hear that was the problem 06:38
Juerd I'm partially surprised 06:39
rhr things can get synchronized so they always run when top's alseep, so you never see them
Juerd You'd think a performance should show up somewhere, as cpu load, or as io load
But the box could be 85% idle and still be slow. 06:40
s/performance/performance problem/
rhr cpu load is very hard to calculate accurately...
Juerd And smart people like audreyt and some people over in #haskell said it wasn't the problem, and that it was benchmarked before
06:41 agentzh joined
Juerd Hi agentzh 06:41
agentzh: Thanks for nicing your processes
agentzh Juerd: hi, now only cp and cd are not nice'd.
lambdabot agentzh: You have 1 new message. '/msg lambdabot @messages' to read it.
agentzh :)
in case you think there is a need... 06:42
Juerd agentzh: You could just have used a single wrapper :P
I don't think there is
agentzh right, from cron table. 06:43
Juerd e.g.
06:46 iblechbot joined
Juerd audreyt: Please nice evalbot, svnbot, and svk (if only at "1", that's better than "0") 06:48
audreyt Juerd: please nice them for me 06:49
svnbot6 r16287 | agentz++ | update-syn - no longer nice individual commands, we now nice the whole script from cron.
Juerd audreyt: o
audreyt: ok
06:50 franck__ joined
audreyt I had niced them at 20 before but apparently missed that from the prev start 06:51
so 20 is just fine
06:51 gugod joined
agentzh uses 5. 06:52
Juerd 5, 10, 20, as long as non-interactive processes don't get 06:53
<= 0, it's fine :)
20 may be a bit overkill
agentzh grins.
dduncan audreyt, haven't read you here in awhile ... how are things in general with you?
audreyt dduncan: family health issues has been dominoeing 06:54
so did not get much hack cycle
dduncan that's unfortunate ... I hope the health issues improve
audreyt they shall... I'm again @hospital atm 06:55
so will have to run in a bit
dduncan is it just other family members, or are you also unwell?
Juerd Does anyone have tuits to implement manuals.textdrive.com/read/book/8 on feather?
lambdabot Title: Running Trac on TxD | 06:56
audreyt aside from sleep deprivation and mild depression I think I'm doing rather well
also, hacking on 6.28.0 local branch helps
audreyt is atm fixing the long-standing "$_ is not truly lexical" problem
dduncan that part's good, at least
audreyt yeah :) 06:57
agentzh audreyt: take care :)
audreyt will do :)
dduncan please keep us appraised regularly as you can or is appropriate 06:59
Tene Juerd: can you give a name/description to the task instead of that url?
Juerd Tene: Implement tracd and proxy apache2 to it 07:00
audreyt dduncan: will try to do better at this regard :)
Juerd trac currently uses cgi scripts, which are slow
Which is very noticeable in dev.pugscode.org/
lambdabot Title: Pugs - Trac
Tene Hmm. Doesn't look too bad. 07:02
I've never worked with trac before, though. 07:06
Juerd: I could manage an hour or so of tuits right now. 07:07
Juerd Tene: Great 07:08
Tene: What's your feather login?
Tene I don't have an account on feather. 07:09
Juerd Let's get you one 07:13
Tene 'tene' is preferred. 07:14
Juerd Could you email your realname and preferred login to [email@hidden.address]
Tene I will.
sent 07:15
Juerd Tene: You've got mail 07:20
agentzh wow, load average on feather drops to 1.02 :)
Tene wtf you're right! are you spying on my inbox?! 07:21
rhr t/operators/precedence.t is saying "Test header seen more than once!" does anyone know what that's about?
Juerd Tene: Oops. Pretend I said nothing.
agentzh rhr: TAP headers like "1..10"? 07:22
rhr agentzh: I don't understand
agentzh @google TAP 07:23
lambdabot www.flytap.com/
Title: TAP Portugal
agentzh sigh
rhr: you're running prove6? 07:24
rhr yup
agentzh try prove6 -v t/.../.t ?
rhr that's what I'm running
agentzh it supports the -v option? 07:25
rhr yes
Juerd @google test anything
lambdabot search.cpan.org/dist/Test-Harness/l...ss/TAP.pod
Title: Test::Harness::TAP - Documentation for the TAP format - search.cpan.org
agentzh rhr: there's an extra 1..10 at the end of the TAP output. 07:27
sorry, i mean "1..42"
rhr yeah, where is that coming from?
I see it when I run it with pugs
agentzh rhr: have you planned the test number?
rhr yeah, the plan is correct
I don't see anything in the .t that should print that 07:28
agentzh looking
rhr let me commit some fixes first...
look at r16289, it doesn't parsefail 07:30
agentzh k
rhr kicks svnbot 07:32
07:34 rfordinal joined
agentzh it's interesting 07:37
Juerd ?eval 1+1 07:38
07:38 evalbot_r16284 is now known as evalbot_r16288
evalbot_r16288 2 07:38
Tene Juerd: what user should tracd run as? 07:39
agentzh rhr: debug Test.pm? 07:40
agentzh currently doesn't have a good machine.
Juerd Tene: The user that used to execute the cgi, so you don't have to change permissions, or a new user of your choice, if you don't mind digging up which permissions need to be changed
rhr it must be Test.pm:306
agentzh rhr: nod
$Test::num_of_tests_planned should have been set by sub plan, but in case it has been destroyed by the tests or... 07:41
rhr yeah, but I don't see anything that could clobber it.. 07:42
Tene Juerd: any preference on the port it should run on?
agentzh rhr: add a warn statement to Test.pm:306 ?
not sure if it helps. 07:43
Juerd Tene: < 7000
Tene kk
rhr agentzh: what kind of warning? not sure what you mean 07:45
is sub proclaim what gets run for each test?
agentzh yes 07:47
rhr k
strange, adding 07:49
Tene Juerd: do you know what domain name trac wants for its realm?
rhr warn "XXX "~$Test::num_of_tests_run~"\n" to proclaim shows it stays defined 07:50
Juerd Tene: dev.pugscode.org
agentzh rhr: adding debugging code to proclaim may help tracing $Test::num_of_tests_planned's value.
rhr what else happens after all the tests are run?
agentzh: that's what I just did :)
agentzh rhr: the END block triggers. 07:51
rhr: that's what END is for. :)
07:51 DarkWolf84 joined, DarkWolf84 is now known as OmegaKO
rhr it's still 49 in END 07:52
very strange
agentzh rhr: what if at the beginning of sub test_ends? 07:53
i mean the sub body
and it could be a bug in the defined builtin ;) 07:54
rhr yup, it's still 49 inside test_ends
agentzh then write a test for defined?
rhr how come this doesn't affect other tests!?!?
agentzh maybe defined dislikes 49? 07:55
agentzh shrugs
or ! or if..
you can trace further :)
rhr I'm stumped 07:56
agentzh anyway 07:57
rhr ok, 16290 commits a fix that I don't understand at all 07:59
agentzh svnbot6 is not here, hence r16290 08:00
heh, it looks funny :) 08:02
rhr: write specific test for it?
(as long as you can reproduce it in a new context)
rhr I'll try, I haven't been able to reproduce yet 08:03
08:03 lichtkind joined
agentzh rhr++ # locating a subtle bug 08:03
rhr ?eval module XXX; my $x=2; if(!defined($XXX::x)) { say "wtf" } 08:06
08:06 evalbot_r16288 is now known as evalbot_r16289
evalbot_r16289 OUTPUT[wtfā¤] Bool::True 08:06
rhr that's it
agentzh my? 08:07
Juerd $Foo::bar can refer to lexicals?
agentzh doesn't think so.
Juerd Then why would it be defined? 08:08
rhr no, that's not it
08:08 buetow joined
agentzh it's "$Test::num_of_tests_planned;" in Test.pm:18 08:09
rhr I honestly have no idea
agentzh ?eval module Foo; $Foo::x; $Foo::x = 49; if ! defined($Foo::x) { say "oops" } 08:10
evalbot_r16289 undef
agentzh it's much more complicated to reproduce than i thought. :( 08:11
rhr agrees 08:12
agentzh ?eval module Foo; $Foo::x; sub bar { $Foo::x = 49; } bar(); if(!defined($Foo::x)) { say "oops" }
evalbot_r16289 undef
lichtkind le
agentzh gives up. 08:13
maybe we should call bar() from another namespace. who knows?
or it may be a bug related to yaml precompilation. 08:14
rhr I'll add some more comments to Test.pm describing what's wrong
agentzh audreyt once said the yaml compilation stuff had some peculiarities.
Tene Juerd: potential problem. you still around/
?
agentzh rhr: you have precompiled Test.pm to yaml, right?
the last bug i found in Test.pm in last summer was that END {} did not trigger at the exit time. 08:16
and it was related to yaml precompilation.
Tene @tell Juerd tracd uses a literal 'md5crypt' to check passwords in the htpasswd file. Patching tracd to recognize the '{SHA}' sequence in the htpasswd file is an option? 08:18
lambdabot Consider it noted.
rhr agentzh: yes, I'm precompiling. let me try without... 08:19
agentzh rhr: cool
Tene @tell Juerd python-support/trac/trac/web/auth.py:268 08:20
lambdabot Consider it noted.
Juerd Tene: Certainly an option, please document it 08:21
lambdabot Juerd: You have 2 new messages. '/msg lambdabot @messages' to read them.
rhr I still get a 1..49 not precompiled
agentzh rhr: okay
Juerd Tene: Consider sending the patch to the trac people, there are probably more who would benefit from this hack
08:21 BooK joined
rhr I changed config.yml and did make, that's all I needed to do, right? 08:21
agentzh rhr: no 08:22
rhr what else?
agentzh rhr: remove .yml in blib6/lib/
rhr k
agentzh or move them to other places.
in case there are more...a find -name *.yml may help.
or *.yaml
another way is use -I or set the env. 08:23
*to use
the latter is easier :)
Tene Okay, it looks like I get to learn python tomorrow. 08:24
rhr ok, I did find blib6/lib -name '*.yml' -exec rm {} \; make and I still get a 1..49
agentzh it runs much slower, is it? 08:25
rhr can't tell, it's a very fast machine
agentzh k
Tene Juerd: looks like there are three instances of trac libraries installed on feather (locate auth.py), which should I modify, or should I make a new copy elsewhere?
rhr should I do a make realclean to be sure? 08:26
agentzh rhr: one tip: bundle the old buggy Test.pm to t/ and write a test for it.
Juerd Tene: I have no idea which one(s) is/are in use
rhr good idea
Juerd Tene: Perhaps audreyt knows
08:30 elmex joined
Tene dev.pugscode.org:6543/trac/ -- runs fine, just can't log in 08:30
lambdabot Title: Pugs - Trac
Tene I'll look into it tomorrow. Sleep now. 08:31
agentzh rhr: morning or afternoon in your place? 08:37
rhr agentzh: err... 4:40am, but I woke up "today" at 9pm 08:38
my sleep schedule is busted lately...
agentzh rhr: oh
rhr: lambdabot told me it was 4:39 and i was shocked. 08:39
rhr :)
after a realclean and fresh make I still get a 1..49 08:40
agentzh rhr: okay, time to write a test for the buggy Test.pm :)
rhr what' the easiest way to run a different Test.pm? 08:41
agentzh rename it?
then use it?
that way you can't get confused yourself :)
08:41 dduncan left
rhr do I need to make a new ext/Test-bug and s/Test/Test-bug/ everywhere within? 08:42
agentzh rhr: ext/ is not the right place for Test-bug
rhr: t/ is.
rhr ok. does anything special need to be done to load a module from t/? 08:43
agentzh t/xx-uncategorized, maybe.
use t::blah::Foo? 08:44
not sure.
in perl 5, we can write use lib 'blah/blah'; or use blah::blah::Foo
08:44 IllvilJa joined
agentzh or write a .t which does a system call. 08:45
rhr I'll give it a try and see what foes wrong :) 08:46
agentzh system "$*EXCUTABLE_NAME -It/path/to/buggy/test-pm t/some/test.pl" or something like that.
rhr *goes
agentzh rhr++
08:47 the_dormant joined
rhr hold on, I re-fixed Test.pm but I still got a 1..49 so it must still be using a precompiled one somehow... 08:49
agentzh ah
rhr: use the following command: 08:50
./pugs -Iext/Test/lib t/operators/precedence.t
rhr: it works on feather and i didn't see the "1..14" stuff at the bottom. 08:52
rhr ok, hold on, I just started another fresh make... I had removed Test.pm from precompile_modules:, now I'm trying precompile_prelude: false 08:53
agentzh rhr: well, i believe you needn't do that...it's a bit overkill ;) 08:54
rhr but it must have been getting precompiled... or my fix no longer works?
agentzh it works on my side. 08:55
(my side == feather)
rhr yeah, I noticed the last smoke before mine at r16273 didn't show the problem 08:56
mine was r16286 08:57
agentzh rhr: there is a trick to prevent using the wrong version of Test.pm 08:58
cd t
../pugs -I../ext/Test/lib operators/precedence.t
rhr [106 of 112] almost there :) 08:59
agentzh rhr: how long does your smoke take?
6 min?
rhr ~6.5 min 09:00
agentzh wow, a dual-core machine?
rhr 2xdual core xeon 3GHz
agentzh aha, mine is core2duo 2.4GHz, but it's at home. 09:01
rhr ok, ../pugs -I../ext/Test/lib operators/precedence.t shows no 1..49 now
agentzh rhr: nice
rhr and the unfixed gets no 1..49 either! 09:02
agentzh we need a tool that can updates Test.pm.yml. 09:03
rhr so it must be precompile
agentzh rhr: are you using the proper -I ?
or you're updating the same Test.pm? 09:04
rhr same command as before 09:05
agentzh okay, another yaml precompilation nit. :)
maybe gaal or audreyt or lumi or other haskell people can fix it soon
rhr yep, I haven't made a copy of Test.pm yet
so a test won't show the bug :(
unless it's precompiled somehow 09:06
agentzh rhr: save a copy of Test.pm and leave a note there.
rhr I'll leave the old code in the comments 09:07
agentzh when someone has the tuits to do some haskell hacking, direct him/her there. :)
rhr k
agentzh rhr: preferrably a whole copy since others may update Test.pm and change the context 09:08
well, it's just a suggestion 09:09
agentzh goes to supper &
09:09 agentzh left
rhr wait a minute...when it was precompiled, I was able to insert debugging code in Test.pm and it was run, but the bug was still there 09:13
and my change fixed the bug...so..WTF?! 09:14
rhr cries in his soup
09:16 zamolxes joined 09:39 bloonix joined
rhr ok, I re-enabled precompiling for Test.pm and did a realclean in the hope that the bug was a one-off cosmic ray-type fuckup, and we could all just forget this ever happened 09:43
but the bug is back
this time editing Test.pm to re-enable my fix doesn't fix it, which is expected with precompiling 09:46
so WTF was happening at the beginning of this fiasco? 09:47
but ./pugs -Iext/Test/lib t/operators/precedence.t gets the fix as expected 09:49
./util/prove6 -v t/operators/precedence.t sees the bug again
and that's what I was using at the beginning 09:50
rhr gives up for the 3rd or 4th time 09:53
10:00 ddwagnz joined
audreyt prove6 uses the pugs in path 10:04
and hence will use the installed Test.pm.yml
usually in sitelib
rhr makes sense 10:06
what's confusing is that when I was first debugging I wasn't using -I or anything, but I was able to insert debug statements that worked 10:08
later testing seemed to show that the bug only occured when precompiled, but that doesn't jive with the above in any way 10:09
audreyt gotcha. 10:11
rhr it's too weird for my little brain
audreyt is still backlogging 10:12
10:12 svnbot6 joined
audreyt ok, I think I get it 10:13
can you do a pugs -V and rm -rf the sitelibs and see if it occurs again? 10:14
rhr I never did make install, so there's no /usr/local/share/perl6, if that's what you mean 10:17
audreyt ok. do this:
find . | grep Test.pm.yml
in the pugs path
10:17 araujo joined
audreyt and see if there's more than one 10:17
rhr ./blib6/lib/Test.pm.yml
that's it
rm it? 10:18
audreyt yeah. "make blib6/lib/Test.pm.yml" can recreate it
but rm it for now and see of prove6 agrees with command line
*see if
rhr prove6 doesn't show the bug now 10:20
audreyt good. now make that .yml
and then run prove6 again
and see if it reoccurs
rhr k
bug's back 10:21
audreyt ok, so it's precomp bug. you've commited all local changes?
it's known that precompilation can't deal with closures, nor END blocks
it's also fixed here ;)
rhr yes, no local changes
did you ever see the bug? 10:22
audreyt ./pugs -Iblib6/lib t/operators/precedence.t
try that?
(with .yml intact)
rhr bug is present
audreyt ok, so it's not mysterious
rhr ok, I must have screwed something up 10:23
audreyt my local branch does not exhibit the bug with precomped Test.pm.yml
10:23 ddwagnz left
audreyt so I think the best course of action for me is simply to get some food, continue hacking tonight 10:23
rhr nobody has seen it but me...
audreyt and pushback ASAP :)
rhr audreyt++
audreyt afk, bbl 10:25
rhr ohhh, I'm an idiot! I was (of course) doing makes after my edits with the debugging, that's why it worked 10:27
no mystery indeed 10:28
10:30 ruoso joined
rhr thought of something. let me try a build in my 32bit chroot 10:36
10:42 xern joined 10:52 the_dormant joined 10:57 nferraz joined 11:00 penk joined
rhr hmm, nope. bug is still present in 32bit build 11:05
11:08 chris2 joined
svnbot6 r16293 | rhr++ | Update comments in Test.pm; bug wasn't quite as mysterious as thought 11:09
11:20 bonesss joined 11:35 ggoebel joined 11:40 nipotan is now known as nipotaway 11:47 rindolf joined 12:02 ggoebel joined, fglock joined 12:03 polettix joined 12:13 kanru joined 12:14 penk joined 12:26 the_dormant joined 12:53 nferraz joined 13:29 agentzh joined
agentzh heh, glad to see audreyt speaking here :) 13:29
@tell Aankhen`` could you take a look at that load event handler when you have the tuits? it's indeed running 10+ or even 20+ times during a single loading. (location.hash = '...' is not the cause) 13:31
lambdabot Consider it noted.
13:32 fglock joined, weinig joined 13:37 araujo joined 13:38 stevan_ joined 13:41 DarkWolf84 joined, agentzh left 13:55 [particle] joined
moritz re 13:59
14:21 kanru joined 14:30 shafire joined
shafire hi 14:30
when will be perl6 avialable?
masak shafire: by christmas. 14:31
moritz shafire: we don't know which christmas, though ;)
masak a better question would perhaps be: what would it take for it to be this christmas? 14:32
moritz if you prefer a more serious answer: hopefully next year ;)
masak: a miracle? *SCNR*
shafire Haskell?
moritz well, that's one step 14:33
but we want an offical perl6 implementation in perl6
masak shafire: I like that :)
moritz so one possibilty is to finish pugs, and then reimplement it in perl6
shafire: do you want a commit bit?
14:43 REPLeffect joined 15:04 polettix joined 15:06 rindolf joined 15:16 ifnotwhynot joined
ifnotwhynot hello 15:17
moritz hi 15:18
ifnotwhynot hi moritz 15:24
are you the perl of the ocean? 15:25
moritz ifnotwhynot: no, I'm just a grain of sand at its shores ;)
ifnotwhynot oneday you will be a perl, its only time 15:26
is this the perl elite on mirc or just general chat?
[particle] yes 15:27
moritz ifnotwhynot: "this" is just a channel for perl 6 development - partly perl "elite", partly not 15:28
PerlJam Anyone who uses perl is elite if you ask me :)
[particle] mostly not :)
moritz certainly isn't elite, just an interested user ;)
15:28 chris2 joined
moritz PerlJam: only if you compare it to PHP ;) 15:29
[particle] ifnotwhynot: wanna help implement perl 6? be a member of the perl elite?
GET A COMMIT BIT NOW!!!
free, for a limited time only*
*until the heat death of the universe 15:30
ifnotwhynot anyone want to help a perldummy(description- thought perl is pruduced by mother nature) 15:31
PerlJam ifnotwhynot: We'll help with perl 6 problems but your best bet for perl 5 problems is to try #perl or #perlhelp etc.
moritz #perlhelp is better than #perl for that matter, propably ;) 15:32
ifnotwhynot thx let me go bug them rather 15:33
15:34 rio_41de joined 15:35 shafire left 15:50 jjore-w joined, the_dormant joined 15:53 rhandom joined 15:54 perlDreamer joined
masak moritz: re larry quote "everyone wants the semicolon" -- you do indeed have it wrong: it was not the semicolon but the colon, and it was about everyone wanting to use it as part of new syntax for some proposed feature for perl6 during the RFC phase 16:00
16:00 marmic joined 16:01 buetow joined
moritz masak: ok, thanks 16:01
rhandom I do like the semi-colon/period analogy though. When I say semi-colon - you don't need to try and parse if the things that come next were in someway part of the current sentences. It imposes finality. 16:09
I think with a well written recursive grammar they could be made optional - but at the loss of syntax errors for other common errors. 16:10
[particle] foo($a) if $b == 3 bar($c) 16:11
moritz rhandom: how on earth could you decide if 'foo bar' means 'foo() bar()' or 'foo(bar())'? 16:12
[particle] er... foo($a) if $b == 3 { bar($c) }
rhandom you can't disambiguate all cases
[particle] right
i'm agreeing, with code :)
perl 6 tries to disambiguate everything 16:13
moritz rhandom: but then intuitivity suffers
rhandom A year ago I wrote a TT3 (Template Toolkit) parser - in TT3 semi-colons on continued statements are optional.
However,
they are required if you need to disambiguate a post-operative directive which is the same thing as a statement modifier
So I do have a working, recursive TT3 grammar, that works without semi-colons - but you can use them as well. 16:14
moritz but is it worth the effort to do that for perl6? 16:15
rhandom Some cases you can't disambiguate without a semi-colon - but in those cases it would fall to the longest token rule.
Which is what the TT3 grammar does.
The foo($a) if $b == 3 { bar($c) } would be a parse error 16:16
because foo($a) if $b==3 was the longest token
moritz but { bar($c) } is allowed in "normal" code
rhandom you then found a bare block - which I guess isn't an error - but would instead run straight away
so like I said - you "could" optionally not require them - but at the loss of valuable syntax errors 16:17
that catch errors when you actually want them too.
By saying could, I don't mean that Perl 6 should make them optional
I'm more a fan of having them there
moritz rhandom: but then 'if' would nearlly always be parsed in the form "$statement if $cond" 16:18
even when it's followed by a block
rhandom right - and in the TT grammar - you MUST put a semi-colon in that case
moritz ok
rhandom the case that you don't want it to be a post-operative directive
I personally prefer having the lines end with semicolons, or periods, or braces - it is a simple way to disambiguate that I don't really mean for anything else to follow. 16:20
moritz same for me 16:21
16:23 veritos joined
perlDreamer How do I get my SVN password reset? 16:25
I'm coming up on some spare time during $dayJob, and I'd like to work on the test suite. 16:26
16:26 fglock joined
moritz perlDreamer: I hope I can do it via commitbit 16:27
if it would only answer... 16:28
damned, I though feather's performance problems were mostly solved
veritos it's been taking me ~15s to open a shell 16:29
16:30 franck__ joined
moritz perlDreamer: I just sent the POST request to reset/resend your password 16:31
perlDreamer ++moritz. Thank you very much.
It will be nice to hack for a project that appreciates tests again 16:32
moritz hopefully it will arrive before it times out
oh yes, we do ;)
and parrot does as well ;)
perlDreamer will it email me with instructions?
moritz I think so
if not, you can ask again 16:33
perlDreamer that's cool, I heard the test suite is supposed to be broken out into its own distribution for cross-project use.
moritz well, the perl6 test suite currently is a bit pugs specific, and there is work going on to amend that
[particle] perlDreamer: that's a bit off
moritz is right
moritz and perl6 on parrot ("not onion") and perhaps kp6 might use parts of it 16:34
perlDreamer what is kp6? 16:35
16:35 rodi joined
moritz perlDreamer: fglock++'s perl6 implementation in perl6 16:35
uhm, in miniperl6 that is
[particle] "kinda-perl 6"
16:35 azawawi joined
[particle] kp6 is a subset of perl 6, built on mp6 (a smaller subset of perl 6) 16:36
perlDreamer so it bootstraps like lisp then 16:37
moritz its in v6/v6-KindaPerl6/ in the pugs repository
s/its/it's/ # english--
[particle] perlDreamer: yes 16:38
moritz can somebody else please try to reinvite perlDreamer? my http proxy has fairly strict timeouts :( 16:44
[particle] trying now 16:45
moritz now it worked
[particle] okie
TimToady @massages 16:46
lambdabot rhr said 10h 17m 27s ago: could you check the tests marked XXX in t/blocks/sub_named_params.t for correctness? thanks!
16:51 justatheory joined, ashleyb joined 16:55 thepler joined 16:58 ifnotwhynot left
TimToady @tell rhr You're running into trouble with those tests because they're based on an illegal declaration. Arguments are allowed in any order, but parameters are not. Positionals must always be declared before named parameters. See S06:581 and S06:619. 16:59
lambdabot Consider it noted.
ingy hola 17:02
TimToady: sorry for telespamming you the other day
that was really weird 17:03
I was so confused for a minute
17:03 kanru joined
rhr oops, I had thought I remembered reading that it was legal but ill-advised. obviously not, I'll fix 17:03
lambdabot rhr: You have 1 new message. '/msg lambdabot @messages' to read it.
rhr @moosages
lambdabot TimToady said 3m 56s ago: You're running into trouble with those tests because they're based on an illegal declaration. Arguments are allowed in any order, but parameters are not. Positionals must
always be declared before named parameters. See S06:581 and S06:619.
azawawi ?eval say "hello!"; 17:07
17:07 evalbot_r16289 is now known as evalbot_r16293
evalbot_r16293 OUTPUT[hello!ā¤] Bool::True 17:07
17:09 rhandom left, rhandom joined
rhr TimToady: still a question. I've changed the sub to named_array2(@x, :@y), what should named_array2(:y, :y) do? 17:13
svnbot6 r16294 | lwall++ | allow \h* between block and comma/colon
TimToady fail for lack of a positional
rhr k
TimToady also fixed your block vs comma question 17:14
rhr TimToady++ 17:17
svnbot6 r16295 | rhr++ | Fixes to t/blocks/sub_named_params.t
moritz perlDreamer: did the mail arrive? 17:24
svnbot6 r16296 | rhr++ | Fix comma/colon after block in arglist tests
perlDreamer Yes, all 4 of them :). Thanks, moritz.
17:24 the_dormant joined
moritz perlDreamer: ok, then olny the response timed out, not the request. I had no way of telling which one did ;) 17:24
17:35 ozo joined 17:41 kunwon1 joined 17:45 b_jonas joined 17:48 Psyche^ joined 17:53 veritos joined 17:55 kunwon1 joined 17:58 sreeram joined 18:05 Psyche^ is now known as Patterner 18:07 IllvilJa joined, sreeram_ joined 18:08 sreeram_ is now known as sreeram, bernhard joined 18:11 edenc joined 18:35 oceanlab joined, justatheory joined 18:50 jisom joined 18:51 penk joined 18:55 japhb joined 19:01 drupek12 joined 19:08 ludan joined 19:28 explorer joined 19:42 dduncan joined, edenc joined, kanru joined 19:46 polettix left 19:50 Aankhen`` joined 20:05 kanru joined 20:08 bonesss joined 20:09 REPLeffect joined 20:13 veritos joined 20:18 as joined 20:37 rhandom left 20:38 bonesss is now known as bones`cochilo 20:59 edenc joined 21:01 ruoso joined 21:03 justatheory joined 21:04 justatheory_ joined 21:05 qmole joined, the_dormant joined, SubStack joined 21:08 justatheory_ joined
Tene Hmm. Are there still microgrants available? 21:28
moritz I didn't hear of any more being used up 21:29
so you'd have a good chance
Tene Looks like I'll be suddenly unemployed soon.
perlDreamer Are all of the tasks in t/TASKS up to date and still valid? 21:40
[particle] tene: yes, there's still a pile of money waiting 21:42
21:44 mr_ank joined 21:50 buetow joined, catch23_ is now known as catch23
Aankhen`` ?eval "\n".say 22:02
22:02 evalbot_r16293 is now known as evalbot_r16296
evalbot_r16296 OUTPUT[ā¤ā¤] Bool::True 22:02
22:03 larsen_ joined 23:09 justatheory joined 23:11 Limbic_Region joined 23:13 stef__ joined
Limbic_Region anyone know where audreyt's presentation(s) are available for direct download - perlmonks.org/?node_id=615432 23:17
lambdabot Title: Re^2: What's wrong with Perl 6?
wolverian Limbic_Region, doesn't google video let you download? 23:22
moritz just use ``clive -y $youtube_url'' 23:23
damned, wrong one
which presentation exactly? 23:25
Limbic_Region wolverian - if you read the note, there was a reason google video was suboptimal 23:28
moritz vp.video.google.com/videodownload?v...iStrings.x 23:29
lambdabot vp.video.google.com/videodownload?v...vnN6XskFbQ
NI985HEC49rMTUGjOxPM9dHKfm-voSCoxAZQTpVeXy9CHDwUR1-y7kAfXuZ9UUxwl0rcL3Ym9Hi0H9lGBUTKM5chvC0xSKQDr1MbiSZmf3IlJpcWo4kPDDZfKIg&sigh=nhqOt7vXWUYWESUT3Fr9ZwLzN-8&begin=0&len=3360900&docid=-3876155376103839
772&messagesUrl\u003dvideo.google.com/FlashUiStrings.x
Limbic_Region that worked out well
moritz sorry about that 23:30
Limbic_Region besides, this isn't for me but for a monk on perlmonks
replying there will help him and many more
moritz the link is "perl 6 today"
and too damned long :(
Limbic_Region wanders off 23:31
offby1 whatever happened to lambdabot's magical tiny-url-ification? 23:32
lumi It gave up in disgust? 23:33
moritz perhapos tinyurl doesn't handle URLs that long ;) 23:34
lumi That's what I said! :)
moritz tinyurl.com/ypuwv5 23:35
23:37 mr_ank joined 23:51 IllvilJa joined 23:56 edenc joined

Powered by Cro and the RakuĀ® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!