Chip needs help!!! www.geeksunite.net | pugscode.org/ <Overview Journal Logs> | pugs.kwiki.org | paste: sial.org/pbot/perl6
Set by theorbtwo on 10 July 2005.
svnbot6 r5529, nothingmuch++ | Move around metamodel business for stevan 00:00
nothingmuch done
sorry again 00:01
nothingmuch--
nothingmuch goes to bed
stevan nite nothingmuch++
wolverian is there a syntax to specify that a method inherited from a superclass should instead be handled by a role that our child class does? some sort of delegation, that is, but using a superclass's method to handle it instead of our own. 00:04
stevan wolverian: that is how Roles work 00:08
the role method do not override the locally defined methods, but they do override those defined in the superclass(es)
wolverian stevan, the role's method doesn't have the same name as the superclass's 00:09
stevan, the role also shouldn't know that this particular subclass wants to use that role to override its superclass's method
s/shouldn't/doesn't/
stevan wolverian: I am not sure what you mean then 00:11
you want a role's method to override a superclass's method of a different name?
and you want the local class to decide this? 00:12
if that is so, then you cannot do this automagically
make the local class's method override the superclass's and delegate to the role instead
wolverian stevan, right, that's what I want to do, but I'm not sure how to delegate it. I thought I could use 'handles' somehow. I can of course write the body of the method myself, but I was just wondering if there is a way to do it with something else perl6 provides. 00:14
stevan handles is not implemented in pugs IIRC 00:15
but handles only works on attributes
has $.foo handles 'bar';
wolverian stevan, that doesn't matter, right now, this prototype is fully perl6 as in AES
stevan something like that
ok
wolverian stevan, has &.foo, right? okay. just wondering if I could just do 'method foo handles <bar>;'
stevan can you not do a role? can you do a proper delegation relationship
wolverian but I guess I can't without writing the body.
stevan wolverian: not that I know 00:16
wolverian stevan, can you rephrase that bit? I mean, proper delegation relationship?
stevan a has-a relationship
so instead of using a role and superclass 00:17
you local class will just have an instance of the superclass
but I am not sure if that is appropriate
wolverian no, that's not appropriate here, I'm afraid. an isa relationship is appropriate.
stevan ok 00:18
:)
wolverian hmm. if a superclass provides method foo, and I have has &.foo handles <bar>; in the child class... what happens when I do ChildClass.new.[foo|bar]?
stevan no idea
wolverian I'll just punt this particular class then, and move to the next. :) thanks a lot for your help. 00:19
oh. another question, actually.
can you .wrap methods?
stevan sure
I assume so, but I am not really sure
wolverian a child class wants to do some extra checking before calling an inherited method.
stevan I would probably just use SUPER:: there
rather than try to wrap it 00:20
wolverian that feels dirty, for some reason. I'm not sure why..
stevan its standard OO practice
its just not very "functional"
wolverian right
stevan being a functional fan, I understand what you mean
but the OO in me says,.. use SUPER::
wolverian oh well. I'll do that later too, then :) 00:21
stevan :)
wolverian the nice thing about this is that this doesn't need to run, so I can just modify it to whatever perl6 looks like this week.
that didn't make much sense, I think. I need to sleep. 00:22
stevan well if you make it run in pugs too 00:23
then you can also have to change it based on what is working right in pugs and what is not :)
wolverian yeah. I'd like this to run this year, at least.
I haven't looked a lot at other people's perl6 code. do others name their roles so that the 'does' lines make sense in english? 00:24
svnbot6 r5531, Stevan++ | Perl6::MetaModel - renaming the container tests (nothingmuch++); and making the rest of the test suite run again (no container types though);
stevan wolverian: I think that is a good way to name roles 00:25
however, it does not always work
wolverian I have a role called Explode here. :)
stevan Explodeable would be my choice
wolverian yeah, it seems to be I've just hit a case where it doesn't _quite_ work.
stevan, 'does Explodeable' doesn't work at all 00:26
stevan but thats because I look at roles as just using methods
wolverian: its american english, all bets are off :)
wolverian I was taught british english, so. :)
stevan :P
I be a 'merican, and I ain't talk very good 00:27
wolverian I also have a role called Belong, which is a bit more awkward
but it still works. :)
stevan but seriously, I tend to look at Roles as being a collection of behaviors which works well with the *able names
wolverian yes, I do that too, coming from Ruby 00:28
but
look at it from a theatre perspective
"Your role is to Sing Funny songs" class Me does Sing[Funny] { ... } 00:29
or so.
Odin- Whaddya mean, "does Explode" works just fine.
It just implies the wrong thing. ;)
wolverian Odin-, wrong, how? :) 00:30
Khisanth is Explosive? 00:31
Odin- wolverian: Oh, the code *does* blow up in your face when you try to use it? :p 00:32
wolverian Khisanth, ech. no. this kind of a thing is exactly what roles are for. 00:33
Odin-, no, the object does. :)
Odin- wolverian: I ... see. 00:34
wolverian this is getting obnoxious. I have twelve classes in one file
wolverian refactors
perl6's OO system makes ruby seem pitiful. so much for poignancy!!! (re: why's poignant guide to ruby) 00:45
okay. traits are magic spells. 00:53
putter nothingmuch: ping? 01:02
dudley_ putter: any luck on the exploding build? 01:14
putter Not on fixing it. So I snarfed r5509, and am svn merge'ing, but as work continued on moved files, that's turning out to be a pain. 01:17
mugwump Install paths are an issue that need to be resolved, really. There is also the issue that we need to have modules installed with the same short name but different long names 01:20
putter The simplest approach would be to revert head to r5509, and disable make install. Folks can svn merge the last day of changes out of the repository and back in. And install fixed at lesure. 01:25
Where folks is mostly nothingmuch and stevan. 01:27
stevan: ping? 01:28
mugwump there are two pure_site_install rules in the generated Makefile 01:36
the first one shouldn't be there 01:38
autrijus greetings. I woke up at Ovid's journal. 01:43
wolverian you were sleeping in Ovid's journal? :)
autrijus no, I was sleeping in use.perl homepage.
brentdax Unexpected Win #1: ~gather { ... } simplifies writing functions that generate [HT|X]ML.
autrijus greetings brentdax the summer coder. 01:44
mugwump hi autrijus. I'm just picking through the inc/Module/Install/* to try to find out why this rule is ending up being inserted twice...
brentdax Greetings Autrijus the genius language implementor.
autrijus brentdax: I used to use [~]gather{}
brentdax: ~ gives you extra whitespace you may or may not care
brentdax Mmm, good point. 01:45
autrijus url complained during my yapcna speech that ~= is made harder to type
and he uses .= all over his program
er, I mean uri.
so I suggested [~]gather{...}, but I'm not sure he agreed
mugwump [~]gather{ ... } will be faster by design 01:46
brentdax I love how it's not necessary to declare a variable and use that name all over the place--gather/take makes it all implicit in a very handy way.
autrijus indeed
brentdax It's almost like working directly with print or say statements, but you get a string instead. 01:47
autrijus mugwump: so you think you can fix this in the next few hours? I was about to give up and rely on moving lib/* into runtimes/perl5/*
mugwump using lots of ~= gives you O(N^2) performance, assuming that appending to a string is O($string.length)
putter mugwump: while you're at it, keep an eye out for why blib6/lib isn't being created....
autrijus and then adding them as explicit LIB 01:48
mugwump heh. well, I'll give it a crack - but you know the history with this makefile stuff ;)
most people end up running away screaming
autrijus indeed I do. so let me try this very very simple fix. sec
putter I tried that. Forgot to open the door. Sitting back down subsequently seemed indicated.
:-) :o) :x) 01:50
mugwump ok, so in Module::Install::Makefile::fix_up_makefile, the postamble specified in Makefile.PL (where the more correct-looking pure_site_install rule lives) is appended, duplicating the section 01:52
nothing seems to be written to remove the duplication 01:53
putter Courtesy of my "I really should delete those sometime" pile of old pugs, I note the duplicate pure_site_install entries date back to at least June 7.
autrijus I think I've fixed it. 01:55
testing.
putter which one, install, blib6, or both? 01:56
autrijus all
a sec
putter gathers little flags at hand to wave in celebration.
brentdax Interesting... 02:21
Is qq{ foo { bar } baz } specced to ignore the inner closure?
jp-autark I think it is supposed to be interpolated. currently a bit buggy. 02:22
brentdax M'kay, just making sure.
I'm working around it by just using a different delimiter. 02:23
autrijus please grep for a test and if not there, write a :todo<bug> test :) 02:30
svk++ # local branch makes this possible 02:44
svnbot6 r5532, autrijus++ | r23238@not: autrijus | 2005-07-13 10:32:25 +0800
r5532, autrijus++ | r23239@not: autrijus | 2005-07-13 10:36:17 +0800
r5532, autrijus++ | * revert #1
r5532, autrijus++ | r23240@not: autrijus | 2005-07-13 10:37:31 +0800
r5532, autrijus++ | * move Perl6-Config back
r5532, autrijus++ | r23241@not: autrijus | 2005-07-13 10:38:37 +0800
r5532, autrijus++ | * revert #3 - makemaker
r5532, autrijus++ | r23242@not: autrijus | 2005-07-13 10:40:01 +0800
r5532, autrijus++ | * #4 - move things to perl5/
r5532, autrijus++ | r23243@not: autrijus | 2005-07-13 10:40:21 +0800
r5532, autrijus++ | * remove lib/Inline-Pugs
putter ooo, rX@not! 02:47
jp-autark *miff* make stopped working 02:49
autrijus did you rerun perl Makefile.PL?
putter Actually, configure stopped working. 02:50
Could not open 'lib/Perl6-Pugs/lib/Perl6/Pugs.pm': No such file or directory at /usr/lib/perl5/5.8.5/ExtUtils/MM_Unix.pm line 3079.
autrijus oh. right.
fixed
wolverian autrijus, heh, r20k? that's nice :)
putter clean build of 5533. same error. 02:53
or not quote, 02:54
Could not open 'lib/Perl6/Pugs.pm': No such file or directory at /usr/lib/perl5/5.8.5/ExtUtils/MM_Unix.pm line 3079.
s/quote/quite/
dudley_ yep. Except mine looks for 'lib/Perl6-Pugs/lib/Perl6/Pugs.pm' 02:55
svnbot6 r5533, autrijus++ | * more path realignment
r5534, autrijus++ | * rename P5::PIL::Run to PIL::Run
autrijus r5535.
jp-autark *wee* 02:56
Darren_Duncan I got the same error, but an update plus tossing a few dirs plus an update fixed it 03:00
svnbot6 r5535, autrijus++ | * add back missing files; finish reverting 03:01
r5536, autrijus++ | * add perl5/ to layout document.
autrijus ok, that was drastic, but hopefully sane
Darren_Duncan currently making 5535
autrijus 5536 is just docs
Darren_Duncan so is it preflight time now, or are there still known issues? 03:02
autrijus I don't know, I'm just starting to run smoke
I need to fix quicksort.p6
and look at module failures 03:03
that's all
Darren_Duncan I'm going to do the whole "make test" thing in the background while I do other work
svnbot6 r5537, brentdax++ | Test for a (currently present) closure-interpolation bug. 03:06
autrijus rafl: hey. sorry for effectively reverting all your hard work 03:07
rafl: but it does not look like it can be resolved for 6.2.8 03:08
putter r5535 resolves make test and make install problems. 03:09
putter wearily waves flag of celebration.
autrijus :) 03:10
mugwump a white flag?
putter lol
... for a long time.
note to self: an you wanted _another_ reason to switch to svk? here you go - - - 03:12
svnbot6 r5538, autrijus++ | * brent forgot to update the plan
putter brentdax: autrijus fixed it. 03:14
autrijus heh 03:15
putter and make pirtest is sane. 03:17
mugwump Schroedinger's bugs: that's when you don't know whether or not the bug is fixed until you commit it to svn
s/commit it/commit the test for it/
Odin-LAP Schrƶdinbug.
putter So, while building optimized, I note 03:20
t/builtins/arrays/isa..............................Name '&ok' not found
under pirtest. 03:21
autrijus putter: exports may not be visible inside closures? not sure
svnbot6 r5539, autark++ | * Fix for (it seems) qq{ ... { ... } ... } 03:35
r5539, autark++ | * un-bug relevant tests
putter Did .namespace start working at some point? Replacing ok(isa( with &main::ok(isa( moves the error from &ok nf to &isa nf. 03:40
But there is a seemingly appropriate .namespace["main"] in -CPIR.
Once upon a time (ie, a few weeks ago), .namespace was unhelpful. Did that change at some point? 03:41
autrijus there's always .namespace["main"] after each .namespace["other"]
I don't think the unhelpfulness changed in general 03:42
putter so "we switched to using namespaces pugs-side, and parrot still cant deal, so we loose"?. 03:43
autrijus yup.
putter ok.
I note one thing which seemed odd in the generated code. it seemed like it sets Test::foo, then sets main::foo from it, then it sets main::foo from itself. or something like that. the last step left me puzzled. 03:46
nopaste 03:47
pastebot nopaste
channel title++
pasteling "putter" at 66.30.119.55 pasted "odd code" (25 lines, 655B) at sial.org/pbot/11728 03:49
putter random observation. 03:50
oh. I didn't follow the control flow. so the oddity is the inits look like dead code. 03:52
stevan putter: pong (from a while ago) 03:57
putter ? oh. I was wondering if you would mind a 12 hour rollback, and merging your work back in, since I was having trouble svn merge'ing automagically. but autrijus invoked svk magic and fixed the build. 04:00
is -B/-C Parrot still real? 04:01
stevan putter: its in a VCS, so rolling back and forth is easy :)
time travel :)
putter back and fourth, yes. but svn merge wasnt too happy with merging changes from files which had moved in trunk but not in the local copy. 04:07
autrijus putter: no, it's alias for PIR
Alias_ seen autrijus?
autrijus Alias_: yes?
Alias_ oh good, the autrijus bot is in here
:)
autrijus I need to run for lunch and some errand rsn
Alias_ me too 04:08
gaal_ mornin'.
autrijus so speak quick
Alias_ meeting, just a quickie
Is it possible to get the pugs svn converted to cvs
putter alias? ./pugs -BParrot -Iblib6/lib t/builtins/arrays/isa.t yields
Alias_ if not ongoing, then just a one-off
svnbot6 r5540, putter++ | enum.t - increased number of tests expected to match number run.
autrijus Alias_: yes, svn2cvs.tigris.org/
Alias_ ok, goodo
putter Null PMC access in get_string()
current instr.: 'EVAL_1' pc 10 (EVAL_1:7)
Alias_ I'll have a cvsmonitor of pugs up shortly :)
autrijus cool. 04:09
Alias_ Looking for things to talk about in my cvsmon oscon talk
Mono, Enlightenment, Pugs
Things with strong public interest that nobody ever really sees the inside of
anyways, gotta dash
putter oh, so does -BPir. something else is wrong. never mind. 04:10
re svn merge, or perhaps it worked fine, and I just became confused. perhaps. 04:16
svnbot6 r5541, putter++ | p5/method.t - untodoed working a working test; wrapped eval around a dieing test. 04:24
04:25 brentdax_ is now known as brentdax
putter end of day. good night all. 04:34
rafl_: sorry the installation improvement didnt work out. we can try again post-release.
&
svnbot6 r5542, putter++ | 10-dash-uppercase-m.t - :todo<bug> one test 04:35
pasteling "putter" at 66.30.119.55 pasted "r5538 make test output. fyi." (591 lines, 26.9K) at sial.org/pbot/11729 04:48
masak i can't get the url in docs/notes/plan to work 04:50
rt.openfoundry.org/Foundry/Project/...n?rev=5465
is that temporary? 04:51
sanxiyn masak: works for me 04:56
masak now for me too :/
must have been temporary then :) 04:57
Darren_Duncan those preludes must be doing something good ... the tests do seem perceptibly faster 05:24
... than a couple weeks ago
Khisanth maybe someone secretly upgraded your system! :) 05:25
Darren_Duncan Fyi, my 'make test' of 5542 shows: Failed 6/357 test scripts, 98.32% okay. 5/6373 subtests failed, 99.92% okay. 06:04
Actually its 5/356; the extra one is a file I had sitting in there that does nothing 06:05
putting that aside, there appears to be a bug in the test harness ... it says 5 subtests failed, but the count of failed tests is 8 when counting the individual files 06:07
brentdax How do you push a pair onto an array? I think they're being interpreted as options. 06:16
Khisanth wow, that is much faster startup 06:20
svnbot6 r5543, autrijus++ | * correct plan number of 10-dash-uppercase-m.t 07:01
r5544, autrijus++ | * make ext/DateTime/t/Date/last-day.t runnable from basepath.
autrijus brentdax: this is a S29 problem; 07:02
multi sub Array::push (@array is rw : *@values) returns Int {
*@values defaults to Item, i.e. does not accept either junctions nor pairs 07:03
so you can't push pair or junctions into an array
this is of course weird. can you please bring this to p6l?
(and worse, there's currently no way to work it around.)
I blame a certain larry to rule that pairs are junctivated :) 07:04
theorbtw1 We couldn't say (@array is rw : Any *@values) ?
07:04 theorbtw1 is now known as theorbtwo
autrijus sure, then it would also allow you to push junctions into @array 07:04
sure we could, the problem is which builtins get that treatment
it's not only push.
so S29 needs updating anyway
theorbtwo OK. 07:05
autrijus fwiw, I'm not entirely happy with this surprising treatment of pairs in function calls, but at least it's consistent and easy to explain
Aankhen`` Morning. 07:06
autrijus brb 07:07
Aankhen`` Nooo, don't run away! :-P
Alias_ autrijus: BTW, so I can study up to OSCON, is there "WTF?" page somewhere that lists what all the various bits and pieces of the Pugs/Perl6 stuff is, what all the cronyms are and mean, and how they all fit toghter?
coral yti, bidkwia. 07:09
Khisanth hmm why is 1=>2 not the same as the 1=>2 in 1=>2,3=>4?
Aankhen`` What are "characters classes in globs"? 07:15
# groups-beta.google.com/group/perl.p...8c243fa88c 07:16
coral *.[gb]z
?
Aankhen`` Uhh... how does that apply to PGE?
coral hmm, misinterpreted globs 07:17
theorbtwo Aankhen``, I think PGE supports both glob-regexes and perl6-regexes. 07:21
(The former is really just an alternate syntax for a subset of the latter.) 07:22
Aankhen`` Ah. 07:23
Kewl.
rafl autrijus, putter: OK. 07:32
autrijus, putter: Soory I didn't had time yesterday to fix that.
Alias_ search.cpan.org/~davidnico/?D=D 07:34
erk
(ignore that)
autrijus smokers, please start a new round. 07:35
I think as of r5546 the main test suite is all pass.
I'll now move to modules.
lightstep wheree does this test belong? 07:37
for ("a\n", "b\n") -> {my $x = chomp $_; say $x}
rafl autrijus: Module::Install looks into lib for its stuff to install per default, right? Where's the code for that? I can't find it.. :-/
autrijus rafl: it's just EU::MM code.
EU::MM does that by default, not us
M::I is just preprocessor for M::I
rafl The last sentence seems a bit weird to me. 07:38
Aankhen`` For EU::MM, maybe?
rafl Yes, M::I for EU:MM or the other way round? I assume the former.. 07:39
svnbot6 r5545, autrijus++ | * eliminate the quicksort bug: An explicit empty signature should
r5545, autrijus++ | not be taken as an implicit *@_. This is a regression that happened
r5545, autrijus++ | after 6.2.7, probably during hackathon.
r5545, autrijus++ | sub foo () { ... } # should mean ()
r5545, autrijus++ | sub foo { ... } # should mean (*@_)
r5546, autrijus++ | * TODOify inheritance.t tests. `make test` should all pass now.
autrijus lightstep: t/subroutines
er, for EUMM.
lightstep: I'll fix rightaway
lightstep how can i test what domething prints? 07:40
Alias_ hijcak STDOUT
jack
autrijus lightstep: you can just push it to sometihng
lightstep ok
autrijus btw. I think that is bogus 07:41
for ("a\n", "b\n") -> {my $x = chomp $_; say $x}
this should not bind $_ anyway
because -> has no implicit signature
it means for ("a\n", "b\n") -> () {my $x = chomp $_; say $x}
which discards the arguments
wolverian I hope that'll warn :) 07:42
lightstep lots of code depends on it working
wolverian lightstep, just s/ -> // and it'll work again :)
autrijus wolverian: it's Eval.hs line 376
feel free to throw in a warning there.
wolverian hmm.
wolverian pulls
lightstep btw, for ("a\n", "b\n") -> {print $_; my $x = chomp $_; say $x} 07:43
does bind $_
autrijus lightstep: and _that_ is a bug.
I'll fix it now. 07:45
lightstep line 376 makes for ... -> {} work like for ... {}, doesn't it?
i figured i have no idea how to test this stuff
autrijus no, it discards $_
er
it discards the iterator 07:46
there will be no topic
lightstep how do i test if a string is a substring of another? 07:50
QtPlatypus $string ~~ /substring/ 07:51
lightstep if the second is in a variable?
Khisanth "PackFile_unpack: Bytecode not valid for this interpreter:" < do I need to get a newer revision of pug|parrot or is there something else going on? 07:52
QtPlatypus $string ~~ /$substring/
autrijus lightstep: the fix is in
QtPlatypus I think.
autrijus Khisanth: what are you doing?
lightstep: index()
Khisanth well I was just trying s:3x:3rd/a/b/
that wasn't the full error though 07:54
autrijus the x and rd modifiers won't work
and you need parrot 0.2.2
Khisanth ah, need older parrot :) 07:56
svnbot6 r5547, autrijus++ | * Implicit variables -- $^a $^b etc -- now _only_ works in bare blocks,
r5547, autrijus++ | in accordance with S06. Existing code may or may not break.
autrijus anyone running smoke for r5549? 07:59
svnbot6 r5548, autrijus++ | * unbreak fp's `map` and `grep` 08:05
r5549, autrijus++ | * move modules that has no chance of pass test suite to modules/.
gaal|work i just started a smoke, autrijus, but am busy at work so i won't be able to be very active in the release process 08:06
autrijus that's fine
I'll deal :)
gaal|work kinda suspected that :) 08:07
autrijus (and if you're busy, then ^C the smokE)
since it's known to be bogus -- r5550 fixed it
gaal|work ah
autrijus I'll start a smoke myself then
it only takes 15 minutes now.
:)
much better than ~40 08:08
gaal|work heh.
svnbot6 r5550, autrijus++ | * oops, extraction was disabled for blocks too. reactivated 08:12
autrijus Still burning: DateTime, Net-IRC, Perl-Compiler, Set, Span, Test-Builder, Test 08:18
gaal|work NameGen here too
autrijus namegen is part of P-C
gaal|work walla. 08:19
autrijus all of them looks curable anyway. I'll start from Test and work backwards.
I wonder what's wrong with Net::IRC's export failure
autrijus is multitasking between pugs and $work -- bbiab 08:20
Aankhen`` No subs are being exported. 08:32
svnbot6 r5551, autrijus++ | * fix Test's 3-output.t by disabled TEST_ALWAYS_CALLER
r5551, autrijus++ | for the subprocess.
gaal|work still t\pugsrun\10-dash-uppercase-m.t: *** No compatible subroutine found: "&load_test"
svnbot6 r5552, autrijus++ | * fix Span.pm's failures.
Aankhen`` (from Net::IRC) 08:33
Lunch, BBIAF. &
gaal|work statements/for.t crashes 08:34
autrijus gaal|work: fixed 08:35
will commit shortly
gaal|work dollar_underscore fails one
implicit_params has an unexpected pass 08:36
subroutine\subroutine fails 2
smoke up in a jiffy
autrijus down to 6 failures in modules 08:38
svnbot6 r5553, autrijus++ | * more test cleanups
gaal|work perlcabal.org/~gaal/smoke_xp.html
autrijus danke
gaal|work eh, i don't know why it says 5549 in the build info
it's 5550
brentdax ...I didn't notice that before... 08:52
./pugs -e 'my $x=1; $x.foo().bar()' complains about the lack of bar() first; only if I remove bar() does it complain about foo(). This might help explain the weird method-chaining bugs I've been seeing. 08:54
svnbot6 r5554, Aankhen++ | * ext/libwww-perl/t/api.t: use `use_ok` instead of hand-rolled variant. :-)
gaal|work methods/instance has a yellow. 09:09
svnbot6 r5555, autrijus++ | * relax implicit $_'s existence in subroutines without formal parameters.
Aankhen`` paints gaal|work pink. 09:11
gaal|work examples.t and general/basic fail..... smoke in a bit.
gaal|work is multicolored 09:12
like a parrot?
Aankhen`` Nah, you're just pink. 09:15
brentdax Can the Time::Local oject localtime returns actually do anything? 09:17
*object
Aankhen`` brentdax >> It can look pretty... :-D 09:19
brentdax Oh, it's in Prelude... 09:20
Hmm, seems the method bug is really a bug when attaching a method to any subroutine--localtime.sec isn't happy, but $localtime_obj.sec is. 09:21
autrijus yeah, known issue
brentdax Oh, alright. I'll keep working around it, then. 09:24
That one fix will probably cut about a quarter of the line count off... 09:25
svnbot6 r5556, autrijus++ | * eliminate cruft: $*PACKAGE, envStash 10:20
r5557, autrijus++ | * nested `use` no longer cripple the middle module's environment.
r5558, autrijus++ | * It's "generating" prelude, not "installing" prelude
gaal|work perlcabal.org/~gaal/smoke_xp.html # r5558 # 7591 ok, 389 failed, 1014 todo, 107 skipped and 0 unexpectedly succeeded 10:45
autrijus gaal|work: msys? 10:50
gaal|work native win32 10:51
autrijus can you look into the failure of 01-string-with-spaces
and see if we should just excise that test?
gaal|work sure
autrijus 323 out of 389 is that single tet.
gaal|work it's a fs permission issue 10:53
i say we junk it for now :)
it's also a very slow test, it creates a process that many times
autrijus mv into t_disabled or rm away, either is fine with me, please excise 10:54
Elly is there a version of perl6 that I can play with right now? 10:55
autrijus Elly: yes, we are about to release 6.2.8 10:56
but you can download the previous version -- see pugscode.org/ and Perl6::Pugs on CPAN
Elly How's Perl6 on Parrot going? 10:58
gaal|work autrijus, done as r5559
meeting &
autrijus Elly: we can run the test suite with parrot as of 6.2.8; only 3% or so of tests pass currently.
Elly coolness 10:59
I'm waiting on Perl6 ;D
autrijus but basic subroutine calls and flow control, arithmetic, string manipulation etc all works
Elly: but Perl6 on Perl5 may be even more fun to you
please see svn.openfoundry.org/pugs/docs/notes/plan . 11:00
Elly do you guys think parrot is likely to be the sort of thing that will be installed on a large proportion of computers? (like Java) 11:01
svnbot6 r5559, gaal++ | move away problematic test until its fs permission problems are resolved
autrijus Elly: I have no idea. it may or may not. 11:02
also do you think firefox to be on a large proportion of computers?
Elly well...allegedly around 10%
autrijus yeah
so maybe one day firefox will bundle parrot. 11:03
Elly that would be cool
autrijus for me that would be good enough :)
but for now... I'm more focused on getting Perl6 on a large proportion of computers, instead of parrot 11:04
that may also have something to do with this being #perl6 not #parrot :) (#parrot is on irc.perl.org)
Elly well, a computer with perl6 is a computer with parrot isn't it? 11:05
autrijus not neccessarily; for example, currently you can install GHC instead of Parrot to run Perl6.
my hope is that eventually you can install perl5 instead of Parrot to run Perl6.
or even javascript.
I mean, run perl6 in a browser's javascript engine
Elly Now that would be awesome. 11:06
ToniEisner that could change a lot of things...
Elly I shudder to think of how hairy that would be though
autrijus :)
Elly how would that work, you'd have a javascript wrapper with Perl6 code inside it?
autrijus no, it works by compiling perl6 into javascript, much like how C is compiled to assembly 11:07
Elly oh
hey, that's cool ;D
except 11:10
wouldn't that still be really hairy? =\
autrijus yes it would, except it will be invisible to the user, I think
Elly well 11:12
yeah, I guess
You'd do perl6 --compile-to="javascript" myfile.pl > myfile.js?
or something? 11:13
autrijus currently it's
pugs -C Parrot myfile.pl > myfile.pir
pugs -C Haskell myfile.pl > myfile.hs
etc
Elly right
autrijus we don't yet have a javascript backend written
but iff we do, then yes, that's how it will look like.
Elly so will eval() work? 11:14
ToniEisner and that curently work on windows and linux the same way?
autrijus ToniEisner: yes
ToniEisner thkx
Elly I suppose not?
autrijus Elly: eval() will work once the compiler itself is ported from Haskell to Perl6.
(which is part of the plan)
Elly I meant in JavaScript
autrijus if that happens, then the compiler itself can be compiled to javascript to make eval() work.
Elly I wouldn't think it'd be possible to do a lot of the cooler perl stuff in JavaScript 11:15
autrijus because you will then have the entire compiler running on js.
not saying it will be fast :)
but should be possible.
Elly like threads and sockets and `` and qx()
Wait, we are talking about the same language right?
JavaScript?
but there's whole masses of Perl6 that won't work on JS and can't be emulated
autrijus yes... javascript. if it's running inside a browser, then you won't have `` primitives
but if you're running inside SeaMonkey in priviledged mode
then you can do pretty much everything. 11:16
(remember much of the firefox extensions is written in javascript)
Elly ah, yeah
autrijus see for example JSSh, a shell server written in javascript.
Elly I thought you meant webpages would want to execute perl client-side via JS
autrijus they will, but then qx() will be forbidden 11:17
just like java applets in sandbox
Elly qx() and threads and sockets and open() and all the stuff that makes perl hacking worth doing ;( 11:18
autrijus well, look at it this way... you are writing AJAX
Elly but I see what you mean with the priveleged thing
AJAX?
autrijus instead of using two languages, one for server side and one for client side
wouldn't it be fun to use a single lang for both 11:19
en.wikipedia.org/wiki/AJAX
broquaint pugs -c HTML leetcode.pl > leetcode.html
autrijus maps.google.com type hackery
theorbtwo (maps.google.com hacking)++ 11:20
theorbtwo gets an idea.
Elly heh, wow 11:21
so perl6 will compile into a lot of different languages? 11:22
some of which are themselves interpreted?
autrijus that is the current plan.
broquaint Perl6 doesn't really exist it's purely an Aristotelean concept.
autrijus I will at least see that it targets three: Parrot, Perl5, GHC
Elly will Perl6 compile into Perl6? :D 11:23
autrijus that is a good question :)
I'm interested in seeing it ultimately do
but that depends on other people who want to work toward that as well.
theorbtwo Hm, JSRef's parser isn't yacc-based. 11:25
autrijus (we need emitter not parser)
Elly I meant that as a joke ;\
can you make it compile to a standalone binary?
autrijus Elly: pugscc foo.pl
already does that
and the resulting exe has no external deps.
theorbtwo Yeah, I know. I need a parser, though, to make sense of maps.11.js.
autrijus oh :) 11:26
Elly sweet
so Perl6 currently exists in a form in which I can code?
autrijus Elly: sure!
theorbtwo Yes is the short answer.
autrijus see the presentation slides in the left top corner of pugscode.org
Elly so perl6 is really going to kind of do everything
theorbtwo The slightly longer answer is svn co svn.openfoundry.org/pugs/
Elly compile to binaries, be a client-side language for web browsers, and be an all-around awesome language for everyone
theorbtwo Make easy things easy, and hard things possible, yes.
autrijus and also make the impossible just happen! 11:27
Odin- The longer answer is "depends on what you want to do"...
autrijus run infinite loop 10x faster!
broquaint Just stick the impossible in a junction and let the multiverse do the rest.
autrijus Elly: par.perl.org, my other project, let you pack perl5 into exe as well. 11:28
but enough advertisement... dinner, bbiab &
Elly I've seen par 11:29
it's cool
theorbtwo (pugscc and par take quite different ways to do it, though.)
JaffaCake Can anyone tell me how to find out the exact pugs command that is being run by the test harness for a particular test? 11:31
svnbot6 r5560, iblech++ | * Usual svn properties (including svn:ignore for the module dirs). 12:11
r5560, iblech++ | * Added some more info to pugs::hack, section about the svn props.
Elly why does svnbot6 append '++' to all nicks? 12:14
dudley karma iblech 12:15
Elly oh, perlbot keeps track of karma? 12:16
dudley jabbot did, I thought
perlbot, karma dudley 12:17
perlbot dudley doesn't have any karma
dudley well, there you go 12:18
Elly perlbot, karma iblech 12:19
perlbot iblech doesn't have any karma 12:20
nothingmuch perlbot: highest karma 12:25
perlbot The top 5 karma entries: tybalt89: 3, $|: 2, Yaakov: 2, hobbs: 2, ': 1
nothingmuch i think it was reset
autrijus had around 800 a month ago 12:26
svnbot6 r5561, iblech++ | * Moved ext/POE to modules/POE, as it's 1. far too small for a ext/ and 2. 12:36
r5561, iblech++ | doesn't work.
r5561, iblech++ | * Noted that change in pugs::hack.
r5562, autrijus++ | * For methods, an extra *%_ is appended to the signature 12:52
r5562, autrijus++ | at compilation time instead of binding time.
r5562, autrijus++ | * Also, methods does not implicitly extract $_ anymore.
autrijus preflight: please start another round of smoke 12:53
I'm working on ChangeLog
will push a new tarball
nothingmuch seen stevan 12:54
perlbot: seen stevan? 12:55
okay, screw that 12:56
castaway_
.oO( lastlog stevan)
nothingmuch i need a volunteer to make a decision for me
gaal|work the answer's no!
castaway_ last seem 06:01 CEST
theorbtwo Try a coin, nothingmuch.
nothingmuch it's not yes/no
castaway_ s/seem/seen/
theorbtwo 23?
gaal|work the answer's mu!
nothingmuch Test::TAP::Model::Consolidated isa Test::TAP::Model 12:57
both need to have Test::TAP::Model::Colorful mixed in
optionally
how would I make it not very ugly?
12:57 brentdax_ is now known as brentdax
gaal|work why optionally? 12:57
nothingmuch because ::Colorful is in another package 12:58
I'm thinking ::Colorful just adds methods to Test::TAP::Model
but another way is to start using the wierd perl5 hacks like spiffy, mixin, or Class::Traits 12:59
gaal|work datetime has fails. is that okay?
latest r
nothingmuch gaal|work: that's the dispatch problem
autrijus er, sorry, I didn't check in my round of test fixes
so the smoke will be bogus
r5563 13:00
gaal|work resmoke?
-p?
autrijus resmoke on r5563
yeah
gaal|work ack
autrijus should be all pass
nothingmuch is also starting a smoke 13:01
so what do you guys think WRT ::Colorful?
gaal|work how often are colorful methods called? is calling can impractical? 13:02
how right is the coupling conceptualy? maybe offer a dummy plugin for when you don't have colorful?
svnbot6 r5563, autrijus++ | * huge set of test cleanup 13:03
nothingmuch gaal|work: they are called from the template
->css_class
castaway_ if TTMConsolidated isa TTModel then load it in TTModel, on demand?
nothingmuch or the percentage color thing
castaway_: i'm inclined to do that, but I'm worried about pollution
gaal|work so mix in a dummy when you don't have color?
nothingmuch a dummy?
gaal|work noop plugin
castaway_ polution of methodnames in space? 13:04
gaal|work which has the same methods as colorful
but doesn't do anything
castaway_ the question is a bit vague, IMO.. we've no idea how these things are normally used, or now colorful changes them
s/now/how/ 13:06
theorbtwo There's no reason not to have the colorful bit always in there.
What colors they show up as should be in the CSS anyway, not set on each element.
So really it's just a matter of getting the class name, which isn't a specific enough operation that it needs to be segregated.
gaal|work ext\Test-Builder\t\010_Test_Builder.t : pugs: ../../pugs.exe: runProcess: does not exist (No such file or directory)
ext\Test\t\3-output.t : '.' is not recognized as an internal or external command, operable program or batch file. # Looks like you failed 1 tests of 1 13:07
the first one is a platform oversight, need {_exe} i gues
ext\WTemplate\t\basic.t fails 2, too 13:08
autrijus gaal|work: it's a slash / backslash problem 13:09
please fix away
WTemplate worksforme
what's the failure?
gaal|work oh: no parrot in path 13:11
autrijus !
oh, please apply the same skip_rest fix as t/rules
then
and/or t_disabled/rules
autrijus is still changelogging
pasteling "gaal" at 212.179.43.225 pasted "wemplate" (15 lines, 632B) at sial.org/pbot/11733 13:12
gaal|work but wait no, different shells
huh, no. same error in w32 shell. 13:13
autrijus, which fix? 13:14
autrijus if(eval('!("a" ~~ /a/)')) { skip_rest "skipped tests - rules support appears to be missing";
}
oh and exit too
gaal|work fix alek 13:15
ok
nothingmuch www.lscheffer.com/bf2malbolge.html 13:18
gaal|work no way 13:19
gaal|work is reminded of the worlds hardest programming challenge
autrijus mm? 13:20
integral is it ok to commit a fix for `make profiled` to Makefile.PL at this stage? (to make it work with the precomp prelude)
gaal|work www.livejournal.com/users/gaal/2003/09/10/
autrijus integral: sure!
autrijus stresses again there is no freeze in pugs development
integral is a release close?
autrijus very 13:21
a preflight tarball will be up as soon as I'm done with changelogging and gaal done with the two fixes
(trivial skip_rest)
gaal|work autrijus, does 010_test_builder workforyou? i have the path ..\..\pugs.exe but it stll fails. 13:22
autrijus gaal|work: it worksforme but it skips that 13:23
I'm tempted to just skip that test.
gaal|work oh! hmmm.
autrijus skip_rest it?
for now
gaal|work k, sec
but line 58 isn't skipped on your version 13:24
my $res = system ....
autrijus just skip_rest and exit in line 57. 13:25
I think.
svnbot6 r5564, bsmith++ | Fixed `make profiled` to also build the precompiled prelude.
gaal|work done as r5565. 13:26
starting new smoke?
or waiting for tarball?
autrijus wait for tarball :) 13:27
gaal|work considers a ZT interpreter
nothingmuch oof! 13:28
autrijus wow, this release took a full month
nothingmuch inheritence scheme is more complicated, i forgot
autrijus last one was june 13rd
er, 13th.
nothingmuch june tenthird 13:29
dudley autrijus: slowing down in your old age? :-)
nothingmuch if anyone is willing to help me discuss this problem, please /msg 13:30
i hate inheritence
svnbot6 r5565, gaal++ | skip problem tests
nothingmuch i think delegation support is the thing I'm most happy about in p6 13:31
gaal|work "delegation support"?
nothingmuch has $.attr handles <method>>
Limbic_Region no preflight announcement in the topic???
nothingmuch etc etc
Limbic_Region salutations all btw 13:32
autrijus dudley: no, mostly because huge architecure changes :)
nothingmuch pugs has more inertia than any project I've ever seen 13:33
gaal|work i have to go to another meeting soon - i'll try to get another smoke going if the tarball lands by then - does anyone volunteer to test 'make install' for safety? 13:36
autrijus nothingmuch: dead project has infinite inertia
gaal|work inert, ja.
dudley I can try out 'make install'
gaal|work cool, thanks. dudley++ 13:37
nothingmuch autrijus: i don't understand that claim
_Jonathan_ nothingmuch: Inertia is a measure of how hard it is to get something to move...and dead projects don't move. :) 13:38
integral or a measure of how hard it is to slow down... 13:39
QtPlatypus _Jonathan_: Inertia is a mesure of how hard it is to change an obects velosity.
_Jonathan_ OK, this is why I only just scraped through physics. :)
13:40 iwlx is now known as wilx
nothingmuch _Jonathan_: ah, ok 13:40
gaal|work meeting & 13:41
autrijus nothingmuch: you can't move a dead project, hence it has the most intertia
QtPlatypus pugs has alot of momentem. 13:42
autrijus uploading 13:44
Limbic_Region some may say that Pugs has reached critical mass 13:45
which IMO it has
autrijus I'd rather that it had reached critical energy
but wait, they are interchangeable
Limbic_Region heh
autrijus eta 1min for upload 13:46
Limbic_Region well - I wasn't talking about the connatation for nuclear reactions or halting the expanding universe
_Jonathan_ wishes he could convert some of his mass into energy...
autrijus that's what exercise is for
svnbot6 r5566, autrijus++ | * Preflight #1!
_Jonathan_ Hey, I'm trying. Gym, walking holiday... 13:47
Seems my smokes are broken due to me having an out of date copy of some module...
autrijus perlcabal.org/~autrijus/Perl6-Pugs-...re1.tar.gz 13:48
1308325 sha1(9b13ea64b60ed1a5e4de14356e77c3f573c4315e)
somebody /topic it
clkao autrijus: did you see my error yesterday?
autrijus clkao: p5 error?
clkao can't run any test, it says can't find test.pm
no, pure pugs
autrijus oh. if it's during my sleep 13:49
then it's likely due to rafl's changes
it should be fine now
clkao k
Limbic_Region We are in Preflight #1perlcabal.org/~autrijus/Perl6-Pugs-...re1.tar.gz | Chip needs help!!! www.geeksunite.net | pugscode.org/ <Overview Journal Logs> | pugs.kwiki.org | paste: sial.org/pbot/perl6 13:50
kolibrie help? my 'svk up' says: Syncing //mirror/pugs(/mirror/pugs) in /var/projects/pugs to 5138. 13:53
what does that mean?
but it does nothing
Limbic_Region depending on your version of svk - you may have to sync and then up 13:54
Limbic_Region found this out the hard way
autrijus kolibrie: you may want "svk pull".
actually, you certainly want that :)
"svk up" is an offline operation.
kolibrie aha
doing something now
autrijus because the repository is at your local hard drive
kolibrie autrijus: thanks 13:55
autrijus np :)
Limbic_Region heh - so svk pull does a svk sync and a svk up in 1 operation?
autrijus Preflight #1: perlcabal.org/~autrijus/Perl6-Pugs-...re1.tar.gz | pugscode.org | pugs.kwiki.org | Paste: sial.org/pbot/perl6 | www.geeksunite.net 13:55
Limbic_Region hmmm - I guess I should update my update script then 13:55
autrijus Limbic_Region: yes, and if you are working with local branch, then it also does a merge. 13:56
dudley how do you commit w/svk, svk push?
autrijus (from vendor trunk to local branch)
dudley: for mirror path, "svk ci" will commit back
if you have local branch, then svk ci will ci to hard disk
then svk push will push it back.
Limbic_Region svk ci -m "your message"
dudley cool. 13:57
autrijus # perlcabal.org/~autrijus/svk-overview.png
dudley eagerly anticipates the exploding hard drive from 'make install' 13:58
autrijus clkao: set perlcabal.org to svk.elixus.org whitelist? 13:59
greetings iblech
iblech Hi :) 14:00
autrijus: linide.sourceforge.net/pugs-svngraph-6.2.8.png
dudley 'make install' works just fine 14:01
autrijus cool!
I checked in some minor ChangeLog fix
people with english-fu please proofread 14:02
(r5567)
iblech: cool, still nicely linear
castaway_ english-fu, at last, one I can actually claim :9 14:03
svnbot6 r5567, autrijus++ | * META.yml and ChangeLog cleanup
nothingmuch given a few smoke reports 14:04
and the conslidated view
you typically want to distinguish between them
along axes like revision, platform, build options, etc
does anybody think they can take in N yaml reoprts, and programmatically create a distinguishing string 14:05
it's simple for simple cases: 14:07
take N reports, generate a list of attrs, for revision, platform, etc
and take away all the fields that are the same for them all 14:08
jp-autark smoke for pre1-tarball: home.ifi.uio.no/~davidra/smoke.html 14:09
svnbot6 r5568, iblech++ | ChangeLog -- Fixed one grammatico.
QtPlatypus triggers a new error "No entries on IntReg_Stack!" 14:10
theorbtwo Whee, fun. 14:11
Parrot, I take it?
QtPlatypus pugs has gained a majour speed boost.
Limbic_Region runtime anyway 14:12
Limbic_Region is still waiting for the compile to finish
QtPlatypus nods to theorbtwo "I think so, most likely in PGE"
nothingmuch also reload my smoke test occasionally
does anybody think they could do the YAML thing? 14:13
Limbic_Region starts nmake test on preflight #1 (Win32 with no bells and whistles i.e. p5, haskell, parrot) 14:14
nothingmuch - what is it you want again
Limbic_Region is up for a p5 task if that's what is on the table
nothingmuch it is
the yaml_harness produces a big structure
given N such structures, i'd like a small snippet of code that goes through them 14:15
and creates a distinguishing string for each
it does this by comparing platform, revision, arch, settings, etc
and then strips off everything that's the same
Limbic_Region ok
nothingmuch ideally you have a 10 char string that is humanly distinct per report 14:16
Limbic_Region where everything the same means everything across all N structures?
nothingmuch yes
if you have an idea on add more DWIMmery, that's good
Limbic_Region let me explain what I think you are after in my own words 14:17
have you confirm/deny
and then ask a question or two
nothingmuch sire 14:18
sure
Limbic_Region you have a base set of qualities that are going to be the same across all N structures
you want a 10 char string that can be generated that says how any given structure differes from those base qualities
nothingmuch so far so good
(WRT to other structures in the same set) 14:19
Limbic_Region is the base set of qualities ever expected to change?
kolibrie autrijus: ChangeLog looks good to me (english-foo-wise)
autrijus cool
jp-autark: can you sync up and rerun t/unspecced/p5/array.t t/unspecced/p5/method.t by hand? 14:20
nothingmuch is amazed by his cat
a week ago I gave him a bath
ofcourse, he didn't really like it
but started having some fleas
appearantly more than we thought, becuase all the past week, despite the trauma, he has been so happy and calm 14:21
Aankhen`` " Implicit variables like `$^a` now only work in bare blocks"
svnbot6 r5569, autrijus++ | * TODOify failing perl 5 tests, courtesy jp-autark
Aankhen`` Shouldn't that be "placeholder variables"?
nothingmuch why is that?
jp-autark autrijus: how do you run that by hand?
nothingmuch sub { $^a + $^b }; # why is this bad?
autrijus jp-autark: ./pugs t/... 14:22
iblech jp-autark: ./pugs t/.../
autrijus nothingmuch: it's against the spec.
Limbic_Region jp-autark - if you haven't done make install then you will need to copy Test.pm to the same location as pugs executable
Aankhen`` That aside, aren't they "placeholder" variables?
nothingmuch i mean, why does the spec forbid it?
jp-autark ah...
gaal|work rehi
Limbic_Region nothingmuch - any response to my YAML question project?
autrijus nothingmuch: because implicit vars are meant only for places where there cannot be formal params.
nothingmuch: it's one or the other
gaal|work preflight status?
nothingmuch Limbic_Region: the base set of qualities is pugs dependant
autrijus gaal|work: see topic
nothingmuch sorry, i missed it 14:23
autrijus: fair enough
gaal|work worth starting that one?
autrijus nothingmuch: if you want to pushback to p6l, please go ahead. I'm fine either way
nothingmuch nonono
autrijus gaal|work: sure, known 5 failures with p5embed
nothingmuch enough pointless discussions
Limbic_Region nothingmuch - if the qualities in common is expected to change I don't see how it is possible to describe the differences in a 10 char string
gaal|work right. d/ling
autrijus I thought p6l stands for point-6-less
jp-autark autrijus: array: 2/11 TODO, method: 1/13 TODO
autrijus jp-autark: good, that's as expected
so it's otherwise clean for you 14:24
autrijus waits for Limbic_Region and gaal|work's report to trickle in
Limbic_Region is on builtins/strings/length ATM
but it is moving right along now 14:25
nothingmuch - you get my point?
if I change the YAML structures to automobiles
nothingmuch Limbic_Region: one second, mom calling from england
Limbic_Region and say they all have this type of engine, this model, etc
k
will continue to type confusion though
Aankhen`` If anyone's editing the ChangeLog, please sync up and merge...
gaal|work compilation started. will GHC get -j 3 sometime?
Limbic_Region so I build a string that says what "extras" each individual car has and encode that in a 10 char string 14:26
svnbot6 r5570, Aankhen++ | * ChangeLog: two small edits.
r5571, iblech++ | util/livecd/mklivecd.pl -- Adjusted default initrd size again, as pugs gains size :)
iblech autrijus: linide.sf.net/pugs-livecd-6.2.8.iso
autrijus thanks!
Limbic_Region my program breaks if I start handling cars with a new add on I previously didn't know about 14:27
now in oo/methods btw 14:29
14:29 brentdax_ is now known as brentdax
dudley Limbic_Region: optimized build? 14:31
Limbic_Region yes 14:32
unfortunately, as a $work machine - Pugs has to compete for resources with - well $work stuff
nothingmuch bac 14:33
Limbic_Region: i think the best solution is to have an algorithm, and a list of fields 14:34
the list of fields can be maintained
and the algorithm compares the fields in N yaml reports, and returns a hash of (string => structure) 14:35
gaal|work first pass build completed - about 9 minutes.
i think "generating precomiled prelude" can be further golfed to "precompiling prelude", no? :)
Limbic_Region nothingmuch - I don't have any issues with the project if "what is in common" is static because then I only have to worry about describing the differences - which will have a decoder 14:36
autrijus gaal|work: no, just keep it that way :)
there are two "generating" messages anyway
Limbic_Region if the "what is in common" changes - not only does the encoded string change but so does the legend
I don't have a problem with that either (programmatically speaking)
nothingmuch the string should be independant, and is throwaway
let me show you an example
gaal|work yay, second pass isn't too bad, just ~2 minutes 14:37
Limbic_Region but I was under the impression that this was supposed to be human readable
nothingmuch yes
only humans read it, not machines
nothingmuch.woobling.org/example.html
under dummy.t you should have two strings
the user provides them
since our yaml_harness has more data than just the test results, we can create a string
like have the first line be 'PUGS_EMBED=""' and the second have 'PUGS_EMBED="perl5"' 14:38
if that was the only difference
Limbic_Region I understand that just fine
nothingmuch so i don't see the problem
Limbic_Region ok - let me try and explain my perspective 14:39
nothingmuch okay
Limbic_Region let's say each position in the 10 char string stands for a different attribute
i.e. what type of imbedding is enabled
0 = none, 1 = haskell, 2 = parrot, 3 = p5, 4 = p5 + haskell (all the way up to 7 using the unix file permissions mode) 14:40
that would be a single position in the string
you have a decoder so you can go look up what that information represents
nothingmuch 10 chars is not the absolute limit
Limbic_Region right - not the issue I am concerned about 14:41
adding more types of embedding is easy - since the single position can be any readable ascii character
QtPlatypus Limbic_Region: Wouldn't it be better to have 4=p5 and 3=haskall+parrot ?
nothingmuch i think if it's too much info it should have '+p5, win' and if there's only one diff, like platform the string should be 'windows'
Limbic_Region QtPlatypus - was just an example
QtPlatypus nods. 14:42
nothingmuch i don't want the users to twiddle bits in their heads
i want a simple readable, freeform bit of text that says how this report differs from the others
Limbic_Region you mean how this individual report in side this larger individual report 14:43
not - this individual report from any other individual report inside any other larger report
nothingmuch yes
no, not the former
Limbic_Region see - that's what I was confused about
nothingmuch okay
err, i created more confusion
Limbic_Region if one overal report needs to stand against another and stuff can change inbetween I was at a loss 14:44
nothingmuch the string needs to be distinct only for the bunch of reports in this consolidated report
gaal|work ext\Test\t\3-output.t: '.' is not recognized as an internal or external command,
Limbic_Region k
gaal|work didn't i fix that?
Limbic_Region that seems doable nothingmuch
nothingmuch if one overall report needs to stand against another the strings are regenerated
the world-distinct data is already in the YAML
we don't need to reformat it
Limbic_Region so in theory - every single attribute needs to have a way of being described simply 14:45
even though in most cases there will be a large overlap
the algorithm needs to be able to handle unexpected changes 14:46
nothingmuch what are unexpected changes?
Limbic_Region make test almost complete btw
things like a brand new feature
ok - so where can I get my hands on the YAML 14:47
autrijus gaal|work: I can't duplicate that -- please hunt it down or disable the test
nothingmuch Limbic_Region: yaml_harness produces them 14:48
gaal|work on it
Limbic_Region autrijus - sifting through the results now
gaal|work may be lineendings
Limbic_Region All tests successful, 89 subtests skipped. 14:49
Files=354, Tests=6029, 1319 wallclock secs ( 0.00 cusr + 0.00 csys = 0.00 CPU)
the module test results didn't fair quite so well
nothingmuch the modules tests are all passing except for WTemplate
nothingmuch.woobling.org/pugs_test_status/ <-- 20 mins ago
a new one should finish shortly 14:50
gaal|work oh, no, the test is expected to pass unexpectedly, but it only TODOs.
heh.
ahhh, i think stderr isn't being captured. 14:52
pasteling "Limbic_Region" at 129.33.119.12 pasted "make test results for preflight" (27 lines, 1.3K) at sial.org/pbot/11734
gaal|work 2>&1 isn't portable.
btw, anyone else getting this (just a warning)? t\pugsrun\09-dash-uppercase-c.t *** Useless use of a constant in void context: VInt 1 14:53
autrijus, shall i skip the Test test?
autrijus gaal|work: do so, then 14:57
gaal|work done as r5572. all other tests OK. 14:59
svnbot6 r5572, gaal++ | skipping unportable test
gaal|work i see a minor issue: pugs -V gives all paths with the correct separator ( \ ) 15:00
except for "sourcedir", which uses / 15:01
hey! interactive mode works!!!
someone++
putter Should we say something about the
/usr/lib64/perl5/5.8.5/x86_64-linux-thread-multi/CORE/pp.h:130:1: warning: "RETURN" redefined
in INSTALL?
autrijus putter: it's harmless no?
putter yes 15:02
autrijus well, if you feel like saying something in INSTALL, go ahead, but I doubt people will look at it
putter ok
I note the worrisome
/tmp/ghc16620.hc: In function `PugsziAST_zdwcharInc_entry':
/tmp/ghc16620.hc:39: warning: cast to pointer from integer of different size
this with a recent 6.4.1 stable x86_64 snapshot. 15:03
autrijus src/Pugs/AST.hs:charInc x = chr $ 1 + ord x
must be this.
not very worrisome if you ask me
but it does look like a GHC issue
gaal|work are we stil waiting on smokes? 15:04
putter any time ghc is getting 64 bit warnings I worry, as until recently 64 bit compatibility has been... a work in progress. 15:05
Hmm, also
/tmp/ghc16620.hc: In function `__stginit_PugsziRunziPerl5_':
/tmp/ghc16620.hc:3117: warning: implicit declaration of function `getStablePtr'
autrijus gaal|work: no, not anymore; #2 will be up in a bit
gaal|work eeek, 1058 todos :) 15:06
autrijus actually our todo rate is quite constant.
(and improving gradually)
putter Err, what is this
/usr/bin/perl -Iinc util/version_h.pl src/Pugs/pugs_version.h
Can't exec "svk": No such file or directory at util/version_h.pl line 26.
Not writing src/Pugs/pugs_version.h because 0 == 0
in make test 15:07
autrijus eek.
JaffaCake putter: I think the gcc warning is harmless, but I'll look into it at some point 15:09
autrijus suddenly notices JaffaCake in #perl6
er, hi :)
JaffaCake is wrestling with pugs today 15:10
hiya ;)
I'm trying to reproduce that GHC bug, and having no luck
autrijus which one? EVACUATED or fromJust?
JaffaCake EVACUATED.. I gave up on the fromJust one 15:11
I get test failures for t/oo/destruction, but not the EVACUATED crash
autrijus right because as I said we already ditched performGC 15:12
putter JaffaCake: I may have unrecoverably lost it when I did a recent ghc install, but until then I had a tar file of a pugs build which repeatably did fromJust.
JaffaCake I've checked out the exact revision from SVN as per the bug report
autrijus oh. hrm
JaffaCake r4916
now trying with r4733, which is another report of a crash 15:13
putter JaffaCake: and fyi, the ghc warnings were from 6.4.1.20050711 x86_64 binary stable snapshot. (src didnt build)
JaffaCake it's difficult because our x86 boxes are running RH9, which has a too old version of perl
putter: thanks
autrijus (preflight #2 is uploading) 15:14
putter if you would ever like me to dustoff the fromJust generater and try something on it, just let me know.
autrijus Preflight #2: perlcabal.org/~autrijus/Perl6-Pugs-...re2.tar.gz | pugscode.org | pugs.kwiki.org | Paste: sial.org/pbot/perl6 | www.geeksunite.net 15:15
autrijus preflight #2: perlcabal.org/~autrijus/Perl6-Pugs-...re2.tar.gz 15:15
1308436 sha1(c84083d43fac3ac8f92114b8606669b30b90cdd6) 15:16
putter on #1 I have a dubious on t/unspecced/p5/method.t
autrijus putter: yes, that's cured
Khisanth so I guess the "make install" issue has been fixed?
autrijus I intend to push #2 out as the release
Khisanth: yes
but I'd still like a round of `make test install` to make sure it works as intended
so please test away 15:17
gaal|work (in)
putter on #1 I have a dubious on t/unspecced/p5/array.t
autrijus that's cured as well. 15:18
pasteling "gaal" at 212.179.43.225 pasted "build warning, no doubt platform dependant" (6 lines, 380B) at sial.org/pbot/11736 15:19
autrijus nod. just like putter's x8664 warning, not going to worry about it now
gaal|work ack 15:20
putter JaffaCake: oh, re ghc warnings, make (optimized). looks like make unoptimized doesnt do it. 15:21
JaffaCake putter: your snapshot probably has the x86_64 native code generator, which is used for unoptimised, hence no gcc warnings 15:22
nothingmuch nothingmuch.woobling.org/example.html
JaffaCake I merged the x86_64 NCG a couple of days ago
autrijus: is it likely that using hsplugins will make any difference? or PUGS_EMBED? 15:24
nothingmuch btw, does anyone have any suggestions for improving the summary view? 15:25
(reload the link)
it's jjust not as pretty
autrijus JaffaCake: both unlikely. weird 15:26
JaffaCake: around that revision I could reliably reproduce the bug in freebsd 5 x86
JaffaCake no joy with r4733 either
autrijus :(
anyone else than gaal smoking? 15:27
nothingmuch autrijus: always 15:28
but not the tarball
jp-autark yes, smoking. 15:29
dudley I will be shortly
putter we should nudge the syck folks to fix this:
gram.c:1115: warning: implicit declaration of function `sycklex'
src/syck/syck.c:492: warning: implicit declaration of function `syckparse' 15:30
autrijus I wonder if you get that from compiling syck's src tarball too
putter a previous rev had some actual errors from failing to declare things. it looked like sloppiness. 15:31
nothingmuch starts working on the smokeserver catalyst app 15:32
i will need lots of help
mostly for the HTML
nothingmuch has to stop typing /quit when he means /leave
Limbic_Region nothingmuch - sorry I have been trying to get yaml_harness running on my work machine 15:34
nothingmuch Limbic_Region: it's not urgent 15:35
Limbic_Region ok - now that I have YAML and Test::TAP::Model installed - how do I go about making yaml_harness.pl work? 15:38
svnbot6 r5573, autrijus++ | * move DateTime tests to t/* from t/Date/* to unbreak `make test`
r5574, autrijus++ | * do not run svk for version_h.pl in a release
r5575, autrijus++ | * remove the now-empty t/Date/ from ext/DateTime
nothingmuch Limbic_Region: read make smoke
i mean util/run-smoke.pl
Limbic_Region k
jp-autark smoke for pre2: home.ifi.uio.no/~davidra/smoke.html
nothingmuch it's basically like prove 15:39
Limbic_Region well - I was a bit confused by
Can't exec /usr/bin/pugs at
since this is a Win32 box
nothingmuch pugs? odd...
putter for the future, it might be nice to add a "happy" first line to Makefile.PL output, so the first thing folks see isnt
ghc-pkg: cannot find package plugins
nothingmuch set HARNESS_PERL ?
putter ;)
Limbic_Region nothingmuch - learning curve - will overcome - I have never ran a smoke before is all 15:40
putter It might also be worth mentioning the "test something, fail, *** message" idiom, to save first time users the "ok, something went wrong! oh, hmm, the *** means it's (all?) ok, I think" experience. 15:42
JaffaCake, btw, thanks for your work on x86_64 ghc. it's appreciated. 15:48
JaffaCake no problem, it's fun :) 15:49
I have GHCi working now
gaal|work all green!
autrijus good. anyone else still smoking? 15:50
putter ! :) # ghci
Aankhen`` putter++ # "happy" first line ;-)
Alias_ ew
oh, _that_ smoking 15:51
Alias_ hides head in shame and leaves channel
autrijus JaffaCake: ooh, and TH too, then
JaffaCake yup, though I haven't tried it yet
there's some slight flakiness: I can't dynamically load an object with a data reference to a shared library 15:52
putter What does 15:53
Not writing src/Pugs/pugs_version.h because 0 == 0
mean?
autrijus it means something that should be squashed. a sec 15:54
Limbic_Region autrijus - I had to rebuild to squash some more Config.pm errors - but as the entire main test suite passed last time I don't see any reason to suspect there would be new errors this time 15:56
svnbot6 r5576, autrijus++ | * add "happy" first line to Makefile.PL
r5576, autrijus++ | * version_h.pl should no longer print "0 == 0" message
r5576, autrijus++ | (both reported by putter)
autrijus okay then.
let me do a round of sanity check before flying... 15:57
putter i still have two make tests runniing, but one of them is unoptimized, so it wont be done for bit. 15:59
autrijus ok. let me know as soon as the fast one is finished :) 16:00
putter :)
has anyone tried make install since yesterday's changes?
dudley yep, it worked fine
putter k
dudley preflight 1, that is...
Limbic_Region putter - that's the reason I had to rebuild from scratch (Pugs that is) 16:01
me applauds AS for providing a zip file that doesn't require installation
but they really should have done something more sane with the Config.pm and Config_heavy.pl
nothingmuch kungfuftr: ping 16:02
kungfuftr pong
putter has anyone run with an external parrot? (that's my unoptimized test which isnt going to happen)
autrijus I had.
putter k 16:03
Limbic_Region C:\TEMP\perl----------------------------please-run-the-install-script----------------------------------\bin is not very helpful to have all your config paths set to
Limbic_Region wanders off for food fully expecting pugs to be in flight soon
autrijus I'm only waiting for putter now :) 16:04
putter: where are you currently? 16:05
(which test, i.e.)
putter #1 vs #2 delta - make test on #1 only ran t/. make test on #2 is now off running ext/ tests. 16:08
#1 make test ended in
make: *** [test_dynamic] Error 255
autrijus #2 should fail in ext/Date/.
if that is the case, then we're all gree. 16:09
green.
I mean, ext/DateTime
(because I fixed and verified ext/DateTime/ as well as the other ext/*)
putter All tests successful, 49 subtests skipped.
Files=354, Tests=6029, 1689 wallclock secs (942.18 cusr + 16.86 csys = 959.04 CPU)
autrijus oh, even better!
putter that's just t/, ext/ is still running.
autrijus after that, try "make install"?
with sudo if needed 16:10
putter yep,
FAILED--no tests were run for some reason.
make[1]: *** [test_dynamic] Error 9
make[1]: Leaving directory `/home/net1/perl6/pugspl/ext/DateTime'
make: *** [test] Error 2
autrijus good.
try make install?
putter eep. 16:11
autrijus it won't hose you I promise :)
autrijus waits patiently 16:12
dudley 'make install'ed just fine...
autrijus that should be fine, then 16:13
putter err, make -n install shows it installing into /usr/lib... I'd really rather not... sorry... :(
autrijus that's fine :)
okay. fly! 16:14
Pugs 6.2.8 released! | pugscode.org | pugs.kwiki.org | paste: sial.org/pbot/perl6 | www.geeksunite.net 16:14
autrijus pugscode.org/dist/Perl6-Pugs-6.2.8.tar.gz 16:15
SIZE = 1309147
SHA1 = efd32419dcddba596044a42564936888a28b3c69
on PAUSE now.
release notes published on p6l, p6a, use.perl.
autrijus cheers
Aankhen`` Yey!
Pugs++
autrijus #perl6++
dudley woohoo! 16:16
Aankhen`` autrijus++
putter whoot! whoot!
dudley I'll second that, autrijus++
nothingmuch so I'm thinking the smoke server is like this:
there's an upload API for test.yml
Aankhen`` nothingmuch >> No, wrong, sorry. :-(
nothingmuch there's a report browsing interface
?
kolibrie my 'make install' worked correctly too
Aankhen`` :-P
nothingmuch Aankhen``: context? 16:17
Aankhen`` [21:46:25] <nothingmuch> so I'm thinking the smoke server is like this:
autrijus 00:18 * CPAN upload: Perl6-Pugs-6.2.8 by AUTRIJUS
Aankhen`` It was a joke, actually. :-)
I do these things occasionally.
nothingmuch ah, ok... i thought you were speaking about something else
autrijus thanks everybody for getting this release out.
nothingmuch okay, shut up now
autrijus I really need to sleep now. :-)
Aankhen`` SIR YES SIR
nothingmuch have a good one, autrijus!
Aankhen`` autrijus >> Sweet dreams!
kolibrie good night!
autrijus have fun!
autrijus waves &
nothingmuch Aankhen``: what did I say? 16:18
putter minor note, make install rebuilds the Prelude on a fully make;make test'ed build.
Aankhen`` nothingmuch >> Er... "okay, shut up now"?
nothingmuch yes
Aankhen`` Good, we're both clear on that. :-D
nothingmuch it seems to me you're not listening
where's an op when you need one
acme autrijus++
dudley putter: mine didn't
Aankhen`` $Aankhen.lip.zip();
nothingmuch #perl6 doesn' thave enough power struggles
so anyway, there's this typical crud interface to reports 16:19
you can search and so forth
and here's the catch
it's like a shopping cart
you say 'add to list' or whatever
and various reports accumilate in a list
putter putter runs tests on release...
nothingmuch and from your little list you get a consolidated graph
putter good night autrijus 16:20
Aankhen`` nothingmuch >> Sounds good. 16:21
BBIAF, dinner
.
nothingmuch ciao Aankhen`` 16:22
kungfuftr nothingmuch: ullo? 16:24
nothingmuch kungfuftr: you helped make htmlmatrix pretty
now do it again: nothingmuch.woobling.org/example.html
kungfuftr nothingmuch: how so? 16:25
nothingmuch it's ugly
make it pretty
you have html+css fu
also have a look at the detail view: nothingmuch.woobling.org/detail.html 16:26
kungfuftr heh, i'm afraid my fiancee is visiting at the moment, so i'm quite busy at the minute
drop me an e-mail and i'll put it in my action list
nothingmuch damnit, is sex the only thing on your mind? what are you, some kind of animal?
you geeks ought to give up your lives, you know
putter Hmm, INSTALL doesnt actually say how to config the install to put it in place X.
kungfuftr nothingmuch: nope... getting the wedding organised is what's currently on my mind... =0) 16:27
Limbic_Region (wedding planning)-- # especially if you only have 6 weeks to do it in 16:28
kungfuftr Limbic_Region: i'm not far off that 16:29
putter Could someone with a windows box check jwcs.net/~jonathan/perl6/ jwcs.net/~jonathan/perl6/pugs-win32.zip and just make sure that isnt a snapshot of a p5 smashing rev?
It doesnt actually say when/what version it is. :(
dudley putter: How do you configure the install prefix? 16:31
Aankhen`` dudley >> perl Makefile.PL PREFIX=/foo/ 16:33
Limbic_Region putter - I finally got pugs compiling on my off-limits work machine - not going to look back now
seriously - if no one has volunteered by the time this build finishes I will check it out
putter tries to find a version number in strings pugs.exe. But the zip file dates are 07-13-05 15:59, so that looks ok. 16:34
dudley Otherwise it defaults to your perl5 prefix?
Arathorn congrats on the release, folks :) 16:35
putter as long as fetching is happening too...
Aankhen`` dudley >> I dunno. 16:36
dudley That's where it goes for me...
putter ok, re ~jonathan/perl6/, you dont actually do an install, yes? so even if it's outdated, its not a threat. does that sound reasonable? 16:41
Anyone know a way to ask a Makefile.PL where it plans on putting things by default? 16:43
Aankhen`` nmake -n install? 16:45
putter Limbic_Region: does this sound reasonable? : re ~jonathan/perl6/, you dont actually do an install, yes? so even if it's outdated, its not a threat?
16:46 brentdax_ is now known as brentdax
kolibrie putter: I just look in the Makefile itself, after that has been generated 16:46
putter Aankhen``: :) I can just see the INSTALL paragraph now, to install someplace other than the default, do perl Makefile.PL PREFIX=/foo/ . To find out what the default is, read through make -n install. lol :)
Aankhen`` Oh, for the user... :-P 16:47
putter :)
Limbic_Region putter - sorry, was AFK - you want to know if it is a dangerous version so you can warn ppl right? 16:48
rafl Where to get the release from? It' not yet on CPAN.. 16:49
putter Limbic_Region: actually, I'm wondering if a prebuilt win32 thing is even capable of being dangerous. if one doesnt run make install, then... no? 16:50
rafl: pugscode.org/dist/Perl6-Pugs-6.2.8.tar.gz 16:51
rafl Thanks. 16:52
Limbic_Region putter - duh 16:54
it is just a tar ball really that you unzip 16:55
putter Oh, that's... unfortunate. PREFIX=mumble only affects _some_ of the installed files. not others. :(
Limbic_Region IIRC it doesn't have the test suite or a lot of of things that comes with a source checkout
putter - IOW - it is only as dangerous as the guy unzipping it 16:56
putter Limbic_Region: ok. thanks.
:)
The Heisenbug corollary to Murphy's Law, "if it wasn't tested, it doesn't work". PREFIX wasnt tested, so... 16:58
dudley: just to make sure you saw, PREFIX doesnt actually work. 16:59
dudley putter: thanks. 17:00
I won't commit INSTALL claiming that it does, then :-)
putter good call :) 17:01
nothingmuch: re smoke, dont forget we'll likely have -BPerl5 rsn, and will need to smoke it. Perhaps others too. 17:05
so at least three smokes near term, "vanilla", -Bpir, and -Bp5. 17:06
(something about calling haskell "vanilla" just seems very... odd)
Aankhen`` LOL. 17:07
putter dudley: or maybe commit, but add a big visible XXXXXXXXXXXXXXX - PREFIX doesnt work yet note? 17:10
dudley alright
putter needs to find a small ascii art picture of a pug for some future release celebration 17:16
nothingmuch putter: since I'm partly responsible for -BPerl5 i doubt it'll be rsn, but ;-)
this is Limbic_Region's responsibility now
putter :)
nothingmuch loves delegation
Limbic_Region nothingmuch - which responsibility is that? The yaml_harness project? 17:17
sili hmmm
i reported a yaml bug and it was thoroughly ignored 17:18
putter though actually, if parrot development ends up being a bottleneck on -Bpir, -Bp5 may get more attention.
nothingmuch Limbic_Region: yep
sili: YAML.pm? or YAML the standard?
sili YAML.pm
Limbic_Region ok - still trying to get it running
putter -Bpir to keep us honest, -Bp5 to be stable and visible 17:19
actually, what _is_ the state of -Bperl5?
(curious minds want to know(tm)) 17:20
btw, dudley++ # assorted stuff
:)
dudley :-)
Does 'make haddock' work? 17:23
Aankhen`` make haddock 17:24
Apparently not...
Limbic_Region is there any particular reason every single make results in pugs.run being compiled, linking, the precompiled prelude, then the pugs run being compiled again, and then final linking? 17:26
putter nothingmuch: oh, btw, re having a one-line per rev "rXXXX - N tests, ....", to have a cumulative file on a server somewhere, so one could eyeball the state of tests across revisions.
nothingmuch okayh 17:27
should be trivial
putter :)
nothingmuch want to write it as a Test::TAP::Model excercise?
putter looks over his sholder; thinks "who is nm talking too..."
;)
Limbic_Region putter - duck 17:28
putter where!?!
Aankhen`` And we have a winner! *points at Limbic_Region*
putter is it typed?
putter should really be doing a job search. but "job search? projects? job search? projects? hmm. maybe job search tomorrow." :/ 17:30
Aankhen`` pugs> ceil(1.5) 17:31
*** Undeclared variable: "&do_ceil"
at src\perl6\Prelude.pm line 290, column 9 - line 291, column 5
Whaa?
do_ceil does exist...
dudley may need to do a job search soon if he doesn't quit playing with pugs during the day...
putter make haddock worked for me... at least until it realized I didnt have `haddock` installed. 17:32
Aankhen`` Hmm, `round` and `ceil` both don't work...
dudley haddock doesn't build for me, so I can't test it
putter ah. 17:33
Aankhen``: they do unqualified sub calls inside of a class... which isnt the most stable thing at the moment. 17:34
svnbot6 r5577, dudley++ | Document PREFIX and more make targets in INSTALL, plus a little English-fu.
Aankhen`` Okay, I'll make them qualified calls.
Bleh, not happening. >_> 17:37
putter Here is a project for someone - a read-eval-print loop in p6. the current one is a bit painful. 17:40
Aankhen`` Where is the current one? 17:42
putter re ceil, sigh, my little tests all work. trying Num... 17:45
re repl, checking...
./src/Pugs/Shell.hs 17:46
Aankhen`` Ah, I have no Haskell-fu. 17:47
putter We'll need a p6 version at some point... might as well start now... :) (modulo the usual "unless the current state of pugs is getting in the way") 17:48
Limbic_Region nothingmuch - I think I have yaml_harness working now via make smoke 17:49
nothingmuch goody
Limbic_Region the yaml files will get outputted where?
or rather - instead of running through the whole smoke - is it safe to break out and verify it is working or does it not write out files until the end?
putter wont be able to do pugs internal stuff, which makes it less useful, but could do nifty multi-line input, and persistant history, and ... 17:50
googling for perl read eval print loop turns up some things, including an old p6c question... I thought there was a p5 faq on it, but I dont immediately see it... 17:53
could find and port a p5 one... 17:54
Aankhen`` starts hacking on examples/eval.p6 17:57
putter Aankhen``: for ceiling to work, one needs to make $^n an explicit arg, and qualify both do_ceil and round_gen. 17:59
or could leave them as a test of implicit args and unqualified calls working... :)
re examples/eval.p6, :-) 18:00
kungfuftr q 18:02
putter q? 18:03
autrijus: fyi, make pirtest is sane in an unoptimized external parrot, but of dubious sanity in an optimized embedded parrot. Errors of: 18:06
Null PMC access in get_string()
current instr.: 'EVAL_1' pc 10 (EVAL_1:7)
Aankhen`` Er... how do I get fallthrough behaviour in a `given`? 18:07
putter ? 18:08
Aankhen`` I want to group a bunch of `when` clauses... 18:09
Ah, `continue`.
putter nothingmuch: ping?
Aankhen`` Hmm... apparently I have to use &goto... how the heck? 18:10
"You can explicitly break out of a when block (and its surrounding switch) early using the break verb. You can explicitly break out of a when block and go to the next statement by using continue. (Note that, unlike with C's idea of falling through, subsequent when conditions are evaluated. To jump into the next when block you must use a goto.)"
vcv its all goto in the end anyway ;)
Aankhen`` Yes, but I want an example of how to use it... do I give a specific line number?
vcv oh. hm.
putter labels arent implemented yet, i believe 18:11
nothingmuch putter: pongv 18:12
i can implement break, i know how
nothingmuch spent all yesterday evening trying to fix next/redo/last
Aankhen`` What is the ord() value for ^Z? 18:16
putter 26 18:21
perl -e 'print ord("\cZ");'
Aankhen`` Thanks.
putter np 18:22
nothingmuch: so, what is the state of -Bp5? what's done, what needs to be? 18:23
Aankhen`` pugs> chr(26) 18:27
?
Erm, that is a "right arrow" there.
It doesn't seem to recognise it as "^Z".
kolibrie anyone seen this error building pugs with emb parrot and perl5? 18:28
/tmp/ghc9183.hc: In function `__stginit_PugsziRunziPerl5_':
/tmp/ghc9183.hc:3223: warning: implicit declaration of function `getStablePtr'
dudley That looks familiar...
kolibrie then a few lines later I get:
Linking ...
/usr/bin/ld: cannot find -lperl
collect2: ld returned 1 exit status
make: *** [pugs] Error 1
builds just fine with emb parrot alone 18:29
so I think it is a perl5 thing
dudley hmmm, this last one built just fine. 18:30
Aankhen`` gaal, you around?
kolibrie I'm on r5576 18:31
dudley oh, wait, there it is.
no linking problems, though
kolibrie so the linking problem is something else 18:32
nothingmuch putter: p6 on p5 requires an implementation of the runtime 18:33
that is, containers, the object model, etc
(that has good progress)
(but we still need a lot of things)
kolibrie I also have this warning, which may or may not be related:
/usr/lib/perl/5.8/CORE/pp.h:134:1: warning: "RETURN" redefined
nothingmuch and the PIL to p5 compiler, that just generates p5 code that uses the runtime, and does the actions described in the PIL 18:34
this is stalling
dudley kolibrie: it looks that way. I'm on r5577 BTW
putter perl -e 'print chr(26);'|less shows ^Z
dudley kolibrie: But 5577 only updated INSTALL
putter ./pugs -e 'say chr(26);' |less too
kolibrie: the /tmp/ghc9183.hc warning should be unrelated to the link failure. 18:35
kolibrie I'm mostly just concerned about the link failure
Aankhen`` Hmm, it does here too...
But the terminal itself only shows a right arrow. 18:36
putter do you have a libperl around? (some os/versions dont have one)
kolibrie putter: where do I look for libperl?
/usr/lib/libperl.so.5.8 18:37
/usr/lib/libperl.so.5.8.7
nothingmuch gaal et al, please try the darcs repo versions of TTH
i want to make a release
putter /usr/lib/perl5/5.?.?/mumble/CORE/libperl.so
maybe? 18:38
nothingmuch nothingmuch.woobling.org/Test-TAP-HTMLMatrix/ and nothingmuch.woobling.org/Test-TAP-Model/ contain head revisions
kolibrie putter: don't have any under perl5, but do in plain /usr/lib 18:39
see above
Juerd just read two weeks of autrijus' journal
putter hmm, do you need a link to /usr/lib/libperl.so.5.8 from /usr/lib/libperl.so for it to be found (don't remember)
kolibrie hmm, could try that
Juerd The speed at which things go here is even more impressive if you're away for a while.
Aankhen`` Hi Juerd. 18:40
nothingmuch hola Juerd
dudley Alright, enough pretending to work for me. Adios!
putter no, it doesnt look like you do.
Juerd Don't consider me back yet, by the way
Aankhen`` Hrm... I need to store a value in a variable and print out the `.perl` version of it, but since it's in a variable, I get a '\' in front of it; is there any way to avoid that? 18:41
putter so the next step is to figure out what your load path looks like.
kolibrie listens
Juerd # multi foo { ... } is now a shorthand for multi sub foo { ... } 18:42
good! I was expecting that :)
nothingmuch patiently waits for 'sub foo is now a shorthand for multi sub foo'
Juerd Well, I wouldn't mind full mmd 18:43
putter putter looks at make file trace to figure out who's doing what...
nothingmuch my semantics proposal (docs/notes/mmd_match_order.txt) was sort of warnocked
a thread followed
but I think that still no one really knows what perl6's mmd will look like 18:44
were you here at the time, Juerd?
Juerd I don't know 18:45
My memory is extremely bad
nothingmuch in any case, i'd be grateful for some feedback
Juerd Something is happening to me and all I know is that it's not good.
putter kolibrie: take a look at the make output... there should be lines with both -lperl and various -L/someplace. btw, what os? 18:47
kolibrie putter: debian
putter call - bbiab
kolibrie perlbot nopaste 18:48
perlbot Paste your code here and #<channel> will be able to view it: sial.org/pbot/<channel>
pasteling "kolibrie" at 66.239.158.2 pasted "ghc ... -lperl ... -L... line" (1 line, 1.2K) at sial.org/pbot/11740 18:51
kolibrie putter: do I need to put a link to my libperl in one of the -L paths? 18:53
putter: adding link to /usr/lib/perl/5.8/CORE did not help :( 18:59
gaal Aankhen``, pong? 19:10
putter kolibrie: ok, sorry, where were we... 19:25
kolibrie putter: pasted the ghc line, trying to figure out where libperl needs to be 19:27
no luck yet
putter ah, so it should be
/usr/lib/perl/5.8/CORE/libperl.so yes? 19:28
kolibrie I put a link there
but still gives same linking error
well, I left the 5.8 on the end
putter try without? 19:29
gaal nothingmuch, ping
kolibrie tries without 19:30
no error yet... good sign 19:31
putter 's fingers are crossed... 19:32
nothingmuch: how much is needed to just get as far as make test sanity? in other words, as far as pir?
kolibrie woo hoo! 19:33
putter: awesome!
nothingmuch uh, i dunno
putter :)
nothingmuch PIL -> perl 5 needs to be done
i think it's best done in haskell or perl5, but not in perl5
i think it's best done in haskell or perl5, but not in perl6
putter right. is the plan... ah. (preanswered questions:) 19:34
nothingmuch i think I want to give haskell perl5 generation a try
Code::Perl is not complete enough
and I think not flexible enough
gaal nothingmuch: i'm d/ling darcs for win now. what do you thing needs testing with THH? 19:35
nothingmuch PIL is not well suited enough for the job
gaal TTH
kolibrie so, if libperl.so is not in any of the -L paths, can 'make' detect that and work around, or prompt user?
nothingmuch err - try to run example.pl in TTH on stuff you care about
by ripping out it's logic
like two smoke results
putter "PIL is not well suited enough for the job"? huh? I thought that was the core of the plan...?
nothingmuch try to go over the docs
err
PPI, not PIL
PIL has nothing to do with perl5 code generation 19:36
it just happens to be the thing being generated
i think another plausible approach is to serialize PIL into perl5 object representing PIL
and just walk that
so there's no compilation and code emittion
putter kolibrie: one would probably have to do a test compile, notice if it failed, and complain...
kolibrie sort of like what I did? 19:37
putter lol
gaal one thing to put in the next release of TTH is the darcs repo url ;-)
kolibrie putter++ # for being so patient and helping me embed perl5 19:38
putter nothingmuch: my understanding is the plan is to create a PIL->p5 backend, supported by a native p5 runtime. no?
tnx. de nada.
nothingmuch putter: yes 19:39
by any means possible
putter it would be nice if there was an easily parsable pil dump format... 19:43
gaal nothingmuch ping 19:44
many fails in latest TTH on win32
fs path errors i think, from glancing.
nothingmuch pong 19:48
use lib "t/lib" ?
gaal perl Build test
shouldn't that do that?
nothingmuch no, it's in the tests
gaal 55/112 subtests failed, 50.89% okay. 19:49
nothingmuch should i be saying use lib File::Spec->catdir("t", "lib")?
gaal ahhh
prolly yes
nothingmuch oh man
wait, doesn't windows allow / in paths?
gaal my brain is tired
oughta 19:50
but win32 sucks!
sucks! i tell you
nothingmuch eep =(
i'll try it out at work tomorrow
gaal yeah, i should probably leave the computer myself at this point :) 19:51
in fact, that's exactly what i'll do
good night :)
nothingmuch ciao 19:53
svnbot6 r5578, Aankhen++ | * added naive, basic, simple, whatever-you-call-it-just-don't-laugh read-eval-print under examples/eval.p6. 19:54
r5579, Aankhen++ | examples/eval.p6:
r5579, Aankhen++ | * added POD.
r5579, Aankhen++ | * added todo notes.
r5579, Aankhen++ | * minor refactoring (gasp).
r5580, Aankhen++ | * examples/eval.p6: more TODO.
putter tries to figure out where the -Cpir output comes from... 19:55
Limbic_Region nothinmuch - tests.yml the file I want?
nothingmuch yep
Limbic_Region ok - gonna have to wait til tomorrow at this point
Chewie[] Aankhen``: Ping. 19:56
Limbic_Region (locked down work machine)--
putter ah, src/Main.hs nothingmuch.woobling.org/pugs_test_...s/haddock/ ++ 20:00
revdiablo is 'pugs -BParrot ...' still the way to run pugs through parrot backend, or has this changed?
putter -BParrot and -BPir are equivalent at this point. oh, and yes. 20:01
kolibrie using my new embedded perl5, I notice that using perl5 modules works in a script, but not in the pugs shell 20:02
putter I was earlier trying to talk someone into beginning a p6-based shell... I think Aankhen`` is beginning an examples/eval.p6... 20:03
Aankhen`` Chewie[] >> Pong. 20:04
putter >> I already began it. :-)
Chewie[] Aankhen``: Where have you been?
putter :) !
Aankhen`` Chewie[] >> When?
putter >> Feel free to hack on it.
Chewie[] Past few weeks?
Aankhen`` "XXX" marks the spot. ;-)
Chewie[] >> Uhh... here?
I was out of town from the 17th of June to the 3rd of July; been here ever since. :-) 20:05
Chewie[] Heh.
Aankhen`` (and as `svn log` will show, been hacking away on libwww-perl :-)
Chewie[] Aankhen``: I've got a very good base64_encode working in Parrot, but I haven't been working on the decode for a while.
It's also very parrotish. It'll have to be munged into the new calling conventions. 20:06
Aankhen`` That's neat.
Chewie[]++
Or perhaps that should be: Chewie[] ōæ½xBB++
Chewie[] Considering I've had well over a few months to do it.. :-D
Chewie[] has been working on various other things.
Aankhen`` Think you could up-port it to Perl 6? :-)
Chewie[] Heh.. I don't even have Pugs building locally yet. 20:07
If I did, I'd be doing a lot more with Perl 6.
Aankhen`` The time is now!
putter :) 20:08
Aankhen`` perlbot karma ōæ½xBB
perlbot Karma for ōæ½xBB: 1
Aankhen`` Bah.
Chewie[] Hmm. Debian still hasn't upgraded from ghc6.2. 20:09
Aankhen`` Anyway.
integral Chewie[]: 6.4's in unstable at least
Chewie[] integral: Heh. I'm not going to switch to unstable just yet.
Aankhen`` G'night. 20:10
putter G'night
ok, there's a scary concept... with just a little bit of preprocessing, it looks like PIL, a haskell code dump, would parse as p6... 20:14
Limbic_Region multi foo { ... } is now shorthand for multi sub foo { ... } so I have to still spell out multi method foo { ... } ? it takes more keystrokes to type method than it does sub - some might argue that sub is used more but p6 may change that 20:22
svnbot6 r5581, fglock++ | Span.pm - added internal module "Span::Code"
PerlJam That's one short hand I don't think we should have "multi foo" it should always be "multi sub foo" 20:26
Limbic_Region I was reading from the 6.2.8 release notes 20:28
I agree - sub/method should always be explicit
putter Is there a cannonical test for perl code to determine if it's running under p5 or p6? 20:37
svnbot6 r5582, fglock++ | set svn properties
PerlJam putter: in the absense of "use v6;", "module" or "class" it's perl5 :) 20:38
Darren_Duncan I downloaded and tested the pristine 6.2.8, and its a-ok 20:39
now back to the Svn head
putter Trying to do a runtime test. even eval('class Foo{}; 'p6''') or 'p5' wont work, because there are p5 modules which emulate p6... 20:41
I suppose one could duck type languages... eval it, and if it works, great... but it would be nice to have a predicate...
Darren_Duncan shouldn't you be using 'perl5' now? 20:42
putter ?
huh?
Darren_Duncan some parts of Perl 6 / Pugs used to have 'p5' to specify using Perl 5, but now that was removed in favor of 'perl5' 20:43
or at least the syntax for using Perl 5 modules in Perl 6 is that way
putter Ah, right. No, I'm after something different... 20:44
Sometimes one wants to write code which runs under either p5 or p6.
Darren_Duncan good luck with that
putter One could imagine a use PBoth; module which in p6, made things more p5 like, and in p5, sucked some of the "make p5 like p6" modules. 20:45
But even without that, a lot can be done in the overlap.
After all, as long as one limits oneself to function calls and a few data types, one can write code which works in both perl and ruby. 20:46
Now normally, you can write a predicate sub is_this_p6() which just eval's some p6 code, and if it works, you know where you are. 20:47
But it looks like p5 is getting a lot of "make me look like p6" modules, which confuses things. 20:48
So I was wondering if there is a "flag" yet, to determine where one really is.
PerlJam putter: use v5; ... use v6; ... use v5; ... (ad nauseum) 20:49
putter: there isn't (and won't) be something automated I don't think
putter & 21:25
svnbot6 r5583, fglock++ | Span::Code works !!! 22:17
23:13 brentdax_ is now known as brentdax 23:19 brentdax_ is now known as brentdax
Alias_ What's the CVS commit count up to now 23:24
wolverian CVS? :) 23:28
brentdax Is that a trick question? 23:29
Alias_ err, sorry.
mistype
The VCS count
:)
Alias_ is looking at a svn2cvs run 23:30
wolverian 5583, says svk pull
Alias_ wonder how hard that will effect the server 23:31
unfortunately, it looks like it doesn't handle multiple users 23:32
So I lose most of the best information 23:33

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!