115 subtests fail | pugscode.org <<Overview Journal>> | pugs.kwiki.org | logged: xrl.us/e98m
Set by iblech on 23 March 2005.
jabbot pugs - 1071 - Port of Mail::Address 1.66. 00:03
pugs - 1072 - Ported Mail::Address tests, too. 00:23
stevan iblech++ # this time for Mail::Address 00:46
iblech the mad porter :P 00:47
autrijus theorbtwo: still there? :) 01:07
01:09 stef_ is now known as cognominal
theorbtwo Mm? 01:38
I'm still here, autrijus -- are you? 01:39
autrijus theorbtwo: sure. want to be an admin?
theorbtwo ?
autrijus someone who hands out committer bits. 01:40
theorbtwo Yeah, I know... but me?
My patches are all trivial and peripherial.
autrijus has nothing to do with that... just needs to be 1)around 2)willing :) 01:41
theorbtwo I'd be honored... um, does the interface require me to know Chinese?
autrijus nope.
theorbtwo OK, sounds good to me.
autrijus ok. the criteria of handing out committer bits is 01:42
1) someone who shows willingness and capability at producing patches
2) exhibits no serious antisocial behaviour :)
theorbtwo Sounds like a plan.
A very strange one, but it's your project. 01:43
autrijus it seems to be working just fine :)
we'll revise if it does not work
rt.openfoundry.org/Foundry/Project/...?Queue=270
theorbtwo Indeed it does; I was beginning to wonder about importing bits of it into PM.
nothingmuch morning
autrijus is the page that you do invitations etc. 01:44
theorbtwo G'morning, nothingmuch.
autrijus hey nothingmuch.
nothingmuch well, actually, middle of thereof
theorbtwo Isn't it a bit early to be calling it morning?
nothingmuch 3:45
woke up from sleep (thirsty)
and i just thought i'd see if autrijus is alive
autrijus and instead of fetching a drink you hop on IRC?
nothingmuch to tell him that I don't think i can do unPair before apply
laptop is on the way to water =) 01:45
autrijus nothingmuch: after seeing your reasons and larry's post
I agree too :)
so just make the binding interface monadic then.
so much for purity
nothingmuch but instead i did pugs -I ;-)
beh, purity is teh suxx0r
autrijus was that a Corion suggestion?
I saw "Corion++"
nothingmuch basically same thing
except we didn't get around to it
and Main.hs was not nice to him
(or me, but no matter ;-) 01:46
autrijus but it's fixed no?
or what did you commit?
nothingmuch yes, it is
s/fixed/hacked/
getLibs simply getArgs again, on it's own
since by the time libs can be set, args have all been fudged
to the point where it's not @*ARGS 01:47
s/not/now/
the problem is that pugs foo -Iblah is passing a param to foo
but it will still cause blah to be in the include path
so either getLibs should be pushed upwards 01:48
or we need to find a way to propagate downwards
which is a bit tricky
i thought of appending PERL6LIB
but... eh... that is just too p5 ;-)
anyway, *yawn*
work soon
theorbtwo Why not make a function in Run that adds a library path, and call it on seeing the -I ? 01:49
(And, for that matter, allowing use lib 'foo' or the @*INC handler to call it.)
nothingmuch theorbtwo: purity and stuff, err, doesn't go well with that ;-)
and afaict, @*INC is created later
if we create @*INC earlier, that would work
but we'd have to break prepareEnv to bits
or make it do the parameter processing 01:50
theorbtwo Oh, yeah.
nothingmuch basically, there's parameters for pugs
and parameters for the program as it is now
and they're cleanly separated
except that it hard for params to pugs to have an effect on the program itself ;-)
autrijus chip: oh. pugs does this ++$named thing in S06.
chip: does your Perl6::Sub do it?
(+$named is optional; ++$named is required) 01:51
nothingmuch oh, and autrijus: we couldn't grok RRegex very well
for starters
things made sense in very tiny bits
so no :g
i think we need to abstract away the PCRE int into types 01:52
hoowa hi tang, junctive what's means in chinese?
nothingmuch and wrap it in an interface that is more p5 compat
(list context with :g, without, etc, should all be done gracefully, into a [Val])
autrijus nothingmuch: yes. yes. yes. 01:53
nothingmuch i don't know how to do the iterative yet, except hide it in a lazy list
(which is actually an elegant solution ;-)
autrijus that's the way to go.
nothingmuch so anyway, i'll larryize bindparams tomorrow, i hope
it's obviously not the way he wants it now 01:54
and i'll see what sense i can make of PCRE, someone (read: stevan) beats me to it ;-)
nothingmuch thinks he's emerging from sleep too far
*poof*
autrijus :) 01:57
kungfuftr autrijus: muwump got to tapei yet at all? 01:59
autrijus in a few hrs
jabbot pugs - 1074 - real Makefile.PL currently breaking main 02:13
pugs - 1073 - External Language build support
chip autrijus: Today, at Larry's advice, I'm settling on C<+$named is required> 02:30
autrijus: However, I personally think that C<+$named> should be required by default, and C<+?$named> should be optional
anybody remember whether C< sub foo is bar > is synonymous with any value of C< sub foo is bar(ANYTHING) >? 02:58
autrijus chip: sure! 03:28
chip: how about you convince larry and I'll implement it? :) 03:29
ingy++
# We just ported Inline.pm to Pugs!
jabbot pugs - 1076 - * backport to 6.2 03:33
pugs - 1075 - * pugs --extern
stevan morning autrijus 03:35
autrijus: I am guessing you are going to be pretty busy with YAPC::Tapaei but if you have a little free time I would really appreciate :g support :) 03:38
It would mean that I could finish File::Spec :) 03:39
autrijus hehe. 03:44
I thought you'd ask :)
stevan :) 03:45
I am soooooo close
autrijus sure sure
stevan thank you :)
after this I promise I will learn haskell :)
autrijus :D
stevan either that or just bug nothingmuch :P
I was actually thinking of trying to do the -X operators myself 03:46
i am looking at the System.IO docs, is that the right place to start?
autrijus sure. 03:47
or just ask on #haskell :)
stevan should I just make each on its own 'op1' line? 03:49
autrijus sure.
stevan :)
you may see a Haskell commit from me very soon then
autrijus wow. 03:50
you just wone s///g right?
s/wone/want/
not m//g
stevan yeah s is all I really need (at least now)
autrijus kay. 03:51
stevan I should put -X pretty low down in the precidence list right?
autrijus I think so. 03:53
stevan k, I am reading pugs.kwiki.org/?CreatingANewOp now
autrijus s:perl5:g/// ok?
stevan perfect 03:54
autrijus s:perl5:g/// landed. 04:05
stevan excellent
have a sec for a question about -X?
I tried this: op1 "-d" = boolIO doesDirectoryExist
but it always returns true :)
autrijus because it only returns false when it's an error 04:06
stevan ah
ok
so do I need to make a new boolIO function?
autrijus sure...
stevan ok I will figure this out then 04:07
thanks much :)
autrijus :)
stevan autrijus++
autrijus =)
I'll brb.
jabbot pugs - 1077 - * s:perl5:g/// landed 04:13
stevan autrijus: when you get back, I am having issues with -d and -f 04:34
I can get them working, but then when I do !-d it doenst work 04:35
parsefail basically
nevermind the parse fail stuff,.. I'm an idiot :P 04:49
it parses now, just working on the precedence
jabbot pugs - 1078 - * preserve inlined code too 05:13
pugs - 1080 - Delete unneeded sample files 05:23
pugs - 1079 - * $?EXECUTABLE_NAME and $PROGRAM_NAME
pugs - 1081 - my first Haskell hack of -d and -f, they 05:33
stevan whooo hoooo, r1081, thats my first Haskell hack :P 05:34
however the precedence on !-d is not right (see the tests if anyone wants to fix it) 05:35
t/operators/file_test_ops.t
autrijus stevan++ 05:56
stevan thanks :)
hey, any tricks for using GHC 6.4 on OS X? 05:57
autrijus hm? just install the .dmg no?
stevan not wokring
autrijus how not?
stevan but I think i didnt remove all the 6.2.2 stuff
autrijus oh. set path?
stevan Chasing modules from: src/Main.hs
Compiling Config ( src/Config.hs, src/Config.o )
/tmp/ghc20837.s:1612:Unknown pseudo-op: .subsections_via_symbols
make: *** [pugs] Error 1
thats the error it dies on
autrijus weird. you on 10.3? 05:58
stevan yup
autrijus have the latest XCode tools?
(10.3 ones should work)
stevan yup
ok let me see if its something I did
I think it is
(its getting late here) 05:59
autrijus get XCode 1.5
it seems that you don't have XCode 1.5.
stevan nope, I do
autrijus "Note also that the GHC 6.4 binary distribution requires at least version 1.5 of Apple's "XCode" developer tools to be installed. Otherwise you'll get an error related to ".subsections_via_symbols"."
could it be an old assembly in PATH?
stevan odd
autrijus s/assembly/assembler/ 06:00
like, GHC 6.2 before GHC 6.4?
not sure at all.
GHC 6.2 should still be happy with pugs tho
stevan I am going to try and clean out all of 6.2.2 and try again
autrijus k.
gugod I have no such issue
crysflame gcc -v returns 3.3? 06:08
stevan no 3.1
crysflame gcc_select 3.3 06:09
then it'll work
must do this as root
so sudo gcc_select 3.3
stevan invalid argument 3.3 it says
crysflame uh
gcc_select --help
i'm 2 miles away from my mac. that utility lets you change to gcc 3.3. 06:10
stevan ok
cool
crysflame if you cannot change to gcc 3.3, your xcode installation is severely damaged or misconfigured
stevan thanks crysflame
crysflame ya
for what it's worth XCode 1.5 doesn't imply gcc 3.3 as DEFAULT compiler
just that it's available. someone should report whatever line you come up with that works back to ghc so they can correct their instructions (well, state them more clearly) 06:11
stevan ok worked now 06:32
I had to reinstall XCode
i think I had messed it up myself
I wont bore anyone with the details :P 06:33
autrijus :D 06:34
stevan ok, gotta restart for a system update brb 06:35
crysflame dool
3r,cool
ARGH
viirya autrijus: any suggestion about "return" with no arguments? 07:57
elmex morning 08:04
jabbot pugs - 1082 - fixing grep.t. 08:07
08:13 cnhackTNT is now known as cnhackTNT|away
nothingmuch morning 08:18
morning 10:09
top of thereof, to be exact
lumi Hi nothing 10:13
Not quite the top
nothingmuch isn't top of the morning eq noon? 10:14
ingy must...
hack...
pugs...
lumi I don't think so..
Apparently it's not topological but qualitative 10:15
nothingmuch WOW!
a google server error!
lumi Been having a lot of those lately :/ 10:16
Alias_ Maybe the first generation of hard drives are starting to reach their warranty 10:18
It's been almost 3 years hasn't it?
nothingmuch ugg
rdc box isn't responding
i hate this stuff 10:19
lumi That shouldn't cause an error with their architecture, afaicc
Alias_ lumi: That sounds right... 10:21
nothingmuch www.neurolux.com/thread.cfm?threadi...essages=19 10:33
" The top of the morning is just before noon"
autrijus greetings /\camels. 10:59
Alias_ has decided to implement "The Autrijus Protocol" for the PPI SF repository 11:00
As many committers as possible and feature requests required to be in the form of unit tests
autrijus :D 11:01
Alias_ www.devsource.com/article2/0,1759,1778142,00.asp
11:01 ingy is now known as whiteg2
Alias_ Funny to see Thomas spruiking Rails as much as possible 11:01
nothingmuch wonders how HAT could help him with haskell 11:02
should have to time to try it in about a half hour
11:07 whiteg2 is now known as ingy
jabbot pugs - 1083 - * use ".imc", not ".pir" as the extensio 11:07
autrijus need to run for dinner.
:) &
nothingmuch ciao!
hola gaal 11:34
gaal hey 11:50
oops, that was kinda slow of me
nonthingmuch: did you take a look at the yaml harness?
it's kinda basic and all that but it works; i wanna see if there's still interest in it. 11:51
use it like prove
only get yaml output
nothingmuch is back 11:59
how about i work on that now
gaal is stevan awake? 12:01
he mentioned he wasn't sure the diag info was necessary
which is fine by me, because i don't have it right now 12:02
also:
could you add perhaps a command to pugs to print build information? so that each yaml upload can self-ident 12:03
nothingmuch i'd like to make it do caller position
diag does that
i will make it suck that bit out
gaal hmmm
nothingmuch i can do this stuff
gaal i don't know wyh it doesn't right now
but i didn't dig into that yet 12:04
nothingmuch i don't mind
gaal also i'm at woik right now
nothingmuch i'm babysitting a build/test run
gaal: then i'll do it 12:05
i think we want to have a percentage measure for each test case
from caller pos
green is 100% fail
red is 100% fail 12:06
gaal i don't understand how you use the "term caller" pos
er
term "caller pos"
nothingmuch $?CALLER::POSITION
stringifies into "blah blah.t at line blah"
gaal ah, ah
k
"coordinates" 12:07
nothingmuch so anyway, i want to find out where each test case is
and colorize the test cases
in the output
i'll start without the distributed status thing
gaal k. note though that yaml doesn't stream output (yet).
nothingmuch just a smoke box
gaal so this is fine for batch processing
nothingmuch yah
gaal but sucks for hte user
nothingmuch nevermind, it'll all be backgrounded i think
we will make some sort of smoke kit, maybe 12:08
we could implement pastebot RCP
nothingmuch pasted test results of rxxx on linux
etc
gaal riht now btw i accumulate everything to one big doc. we can turn it into one stream+many docs
only if they fail :)
nothingmuch "Details, details", as they say 12:09
they are run once per r, at most
well, never mind
gaal well, otherwise you'll have so many test repotys after each commit, you;ll need to add a level of nopaste indirection to get to them :-p
nothingmuch we'll see what we can do 12:10
i think an http post should be good enough
i wasn't reallly serious ;-)
gaal what's your wallclock make + make test? mine is around five minutes on the slow machine, maybe half on the fast one
nothingmuch i dunno 12:11
on pasta it's ok
on safta it's fast
on syeeda it's slow
and on eris it's pretty good
syeeda takes about 6-7 mins to compile
i never make test entirely
pasta is about 3 mins for a typical incr build + test
gaal can you host the smoke server?
nothingmuch sure
2 of them
osx + linux
the linux being the result display thingy
gaal so people can watch the commit rss, svn up && make smoke automatically, and the smoke server can nopaste before jabbot picks up :) 12:13
i have some time tonight to work on this, so leave me with something to do :)
nothingmuch ok 12:14
i'll try ;-)
gaal has been pretty busy with real world stuff lately 12:16
jabbot pugs - 1084 - Begin of a URI 1.35 port. 12:17
nothingmuch what real world stuff? 12:18
(sorry, product crash, test fails... distractions)
gaal $work :)
jabbot pugs - 1085 - Port of URI::_generic 13:07
stevan gaal 13:08
morning
hey nothingmuch
nothingmuch busy 13:09
stevan GHC 6.4 is sooooo much faster
gaal oh, hello 13:11
stevan gaal, I wanted to ask you some YAML questions before I did anything 13:12
so if you have a second
gaal sure
stevan I had some ideas for the output
gaal though isa_ok($gaal, "yaml_expert") fails. 13:13
stevan is there any kind of XSLT like formatter for YAML?
gaal don't think so - but deserializing into a perl object is real easy 13:14
stevan but is($steve.yaml_knowledge(), 0) passes
have you ever used or seen the HTML output of Devel::Cover?
gaal if ram is not a concern, and you don't need any kind of streaming, write perl code to process -- probably easier on your sanity than xslt. 13:15
no, got a url for me to peek at?
stevan likes XSLT
pjcj.sytes.net/cpancover/ 13:16
gaal all rihgt, that's immediately communicative
stevan exactly
so you can look and see right away
gaal yup 13:17
though is # of passing tests as intersting as a continuum as coverage data?
stevan so you think it best to just load the YAML into perl and go from there?
gaal basically you want to see if it's ALL PASS, ALL PASS+TODO, or HAS FAILS.
theorbtwo The problem with creating something like the devel-cover output is that we don't have anything like the devel-cover input. 13:18
stevan I am thinking a bar for each test, which is seperated into the number of tests, red for fail, green for pass
gaal (note to self: tag output of passing todos)
stevan theorbtwo: it would be much less than the Devel::COver output
mostly I wanted to have that color based output 13:19
so its is easily readable
gaal, yes todos are very important
theorbtwo: like a 10,000 ft view
gaal well, we do know if something is todo or not. so we do have interesting input 13:20
stevan it probably will only make sense to see it, so give me a few minutes and I will whip someting up
gaal (also need the yaml to include plan info)
theorbtwo: what don't we have, coverage? not objectively, true 13:21
but i think stevan is aiming just for the status of those tests that are available
stevan exactly 13:22
gaal if someone wants *coverage* analysis, they can take it offline; this tool is meant to aggregate results across different platforms and build envs, to get immediate feedback of how the latest r affected the tests. 13:23
theorbtwo wonders if util/yaml_harness.pl needing an installed pugs is a bug.
gaal it does? 13:24
it requires yaml, which is sortof a bug :)
theorbtwo perl util/yaml_harness.pl t/builtins/defined.t
Can't exec /usr/bin/pugs at t/builtins/defined.t line 1.
gaal bleh 13:25
yes, that's a bug.
though
i *don't* havean installed pugs
oh oh
theorbtwo Also, it gives me a bunch of "Setuid script not plain file".
gaal yu need stuff in your env, one minute
!? 13:26
pjcj I don't know if it will be helpful or not, but Devel::Cover allows annotation plugins. If the input/output match is close enough that might save a bit of work.
theorbtwo Oh, it doesn't like being passed directories.
gaal to2, do you have the env vars mentioned here set? pugs.kwiki.org/?FrequentlyAskedQuestions
yes, and it doesnt glob eithre.
which sucks :)
theorbtwo (All my dirs under /usr/src/ are setuid, so things end up being owned by the src group.) 13:27
Ah, no, that'd explain it.
gaal i think its command line should mimic prove's, so it shouldn't glob but should recurse 13:28
and have some syntax for skipping rules/disabled etc.
theorbtwo I suspect we should really roll our own test-runner -- it'd be nice if we had failure reasons as well, for example.
gaal we do 13:29
pass -v
theorbtwo Oh.
Ah, so we do. 13:30
gaal once again mimicing prove
theorbtwo Hmm, is it just me, or are the various length-like things in t/builtins/strings/length.t not interesting because they don't have any combining chars? 13:31
gaal to2: feel free to add 'em! 13:32
no combining chars in hebrew
oh wait no that's not true
how did i forget vocalization marks
theorbtwo Or as people who speak languages that use them regularly call them, "vowels". 13:33
gaal not exactly:
the vowel is the phonetic entity 13:34
which, in hebrew, are not represented by letters
but by optional "decorators" on letters
nothingmuch well 13:35
are you guys hacking the YAML harness? 13:36
gaal stevan was about to show us something i think.
nothingmuch ah
i butchered Test.pm a bit
theorbtwo I'm hacking something to output a HTML test summary; I think other people are too... I'm just giving suggestions to other people on the harness and tests.
nothingmuch added an env variable which will cause it to always appenned CALLER::POSITIOn
gaal nothingmuch applies his famous kung-fu skillz 13:37
nothingmuch i'm going to try to make it colorize the cross index
someone ruined the aio subsystem
and, err, nothing is working on sunos and linux
so.... this afternoon, perhaps, when I get home, i can work on it a bit further
stevan gaal almost there 13:38
nothingmuch stevan: what are you doing? 13:40
gaal to2, i hope you're using the yaml data, yes?
theorbtwo Yes.
gaal the format of which is subject to change...
theorbtwo Right now, I'm wondering why I'm getting a YAML_PARSE_ERR_NO_FINAL_NEWLINE
nothingmuch probably because there is NO_FINAL_NEWLINE? 13:41
thus yielding a PARSE_ERR?
gaal maybe i sohuld tell the yamlobj to close itself?
nothingmuch in YAML_?
theorbtwo Looks like there's a final newline to me...
gaal i don't see a close method. to2, try .=ing "..." (three dots) to the data before parsing it. that's the doument terminator there 13:44
stevan nothingmuch: I am making a example of the UI I am thinking of 13:46
theorbtwo Waiddasec... it's reporting that the stream does not /end/ with newline character... at line 0. 13:47
This thing is completly borked.
jabbot pugs - 1086 - Ported some subclasses of URI.
gaal that sounds like a pretty short stream to me :)
oh 13:48
i have an idea
try going through a file
because i saw that something was polluting my stdout probably from a leftover print in Test.pm
so maybe that screws up the yaml stream
stevan where is a good place to put up HTML 13:49
gaal yaml_harness -v -o temp.yaml
stevan pastebot wont work right?
perlbot nopaste
perlbot Paste your code here and #<channel> will be able to view it: sial.org/pbot/<channel>
pasteling "stevan" at 67.186.136.119 pasted "example output idea" (1 line, 4.7K) at sial.org/pbot/8564 13:50
stevan yeah that didnt work,, ignore that
theorbtwo Your scratchpad on perlmonks.
stevan theorbtwo: good idea :) 13:51
www.perlmonks.org/index.pl?node_id=359670 13:52
nothingmuch stevan: that's a good idea
stevan really high level you know
nothingmuch i'd like to make todo not a different color
stevan nothingmuch: colors are subject to change :)
gaal stevan, each bar represents one test?
nothingmuch but rather, yellow for unexpected pass
stevan gaal: each bar is a file
gaal each square, sorry 13:53
stevan and each box within the bar is one test
nothingmuch and a sort of green for expected fail
i'd like to see that
theorbtwo That's more or less what I was thinking.
nothingmuch in the mega cross index thingamabob
and make each square clickable
stevan nothingmuch: yes
nothingmuch to jump to the line where it happens
gaal clickabel with diag info
stevan it would be really really really cool if this would integrate with the test catalog as well
theorbtwo I was thinking use titles on the cells to get mouseover diag info.
gaal alternatively if the data ain''t huge, tooltip
jinx 13:54
stevan theorbtwo: thats a cool idea :)
theorbtwo Also, another column with % pass/fail of the file.
nothingmuch gaal: the data ain't huge
it's per test case
gaal with diag? that's beginning to be big.
stevan and it can do your taxes, wash your dishes and feed the cat as well
gaal of course you only have that for failing tests
theorbtwo And make the colors different -- red=fail, pink=unsuspected success, lt-green=todo fail, green=pass.
gaal but the html is $average_test_size * 3k, which can get biggish. 13:55
stevan gaal: we could break this up by folder
each folder gets its own HTML page
nothingmuch it's probably going to be a) mod_gzippable
b) generated locally, most of the time
stevan dammit, client bug reports coming though 13:56
gotta run and attend to them, be back in a bit
nothingmuch you should tell them: "well, in theory, if i work on pugs, we can have p6, which is bound to have less bugs"
gaal more pugs though. 13:57
stevan actually not bugs, but we use the bug tracking system for things like copy changes and data changes
:P
nothingmuch ok, so don't work on pugs, see if i care. 13:58
humph!
>:(
stevan and since nothingmuch has not finished his top secret DBI port yet ....
gaal ooh
nothingmuch bwahahaha
stevan *cough* slacker *cough*
nothingmuch doesn't do DBI
i barely do CDBI anymore
i tend to wrap around *that*
gaal !bwa
nothingmuch is so cool that way 13:59
gaal !!
to2, are you getting yaml to parse now? 14:00
theorbtwo Yep. 14:01
gaal was it the file thing?
theorbtwo I'm an idiot; I was passing a filename where it wanted a string of YAML. 14:02
theorbtwo has never used YAML before.
gaal nothingmuch, you mentioned you were looking into Test.pm, if you see a spare print somewhere ("looks liky oyu failed...."), please kill it
nothingmuch ok,
theorbtwo That's on STDERR, not STDOUT, though. 14:03
gaal s/pare/tray/
oh, it is? then it's less of a problem
nothingmuch i will not buy this record 14:27
jabbot pugs - 1087 - * Added (some ported to Perl 6) tests.
stevan nothingmuch: what record?
nothingmuch it is scratched 14:28
my hovercraft is full of eels
gaal will you hold it against me if i said you have beautiful tests?
nothingmuch 's bass teach is hungarian
i would hold them against you, gaal 14:29
(most people miss the double meaning, i am surprised at you, gaal!)
(you are not most people)
gaal is not most people
see, me, i'm an *individual*. 14:30
theorbtwo , apparently, is most people, as he didn't get it.
stevan didn't either, must be a desert dweller thing 14:31
crazy from the heat and all
nothingmuch would you hold it against me, as in will you think it's not nice to say you have a beautiful body
or would you hold the body against me
gaal ha!
nothingmuch++; # literary observations regarding monty python skits
nothingmuch the reason i got it, is that thhe first time i thought 'i am no longer infected' was a bit out of context 14:32
anyway, so as I was saying 14:33
John Cleese has a wonderful hungarian accent
the nicest bit about Zoltan, my teacher:
when he is talking about the ring finger
the ama in hebrew, with the second syllable being the emphasized one 14:34
he always calls it the aaama
stevan autrijus: if you are around ... 14:36
this pugs -e 'my $path = "/path//to///a//////file"; $path ~~ s:perl5:g{/+}{/}; say $path'
does bad things :)
gaal is stealing code from prove for the yaml harness 14:38
never really understood the basic difference between prove and Test::Harness
theorbtwo Hmmpf, YAML does not appear to be unicode-clean per default.
gaal ew! 14:42
theorbtwo Checked in testgraph.css and util/testgraph.pl -- run it with a yaml file as only argument, expects it's output file to be in same dir as testgraph.css. 14:47
jabbot pugs - 1088 - testgraph -- initial checkin -- creates
gaal to2, i'm ci'ing an updated harness soon
that does recurse
it will change the yaml structure a bit 14:48
theorbtwo wonders if there's a decent way to get it to run two test scripts at once. 14:52
gaal you mean concurrently? 14:53
because it does do sequential now
theorbtwo I meant concurrently.
gaal what's your wallclock for make test?
theorbtwo I have a HT CPU, I'd rather it use all of it and take half as long. 14:54
theorbtwo is impatient.
gaal sure
nothingmuch half as long?
gaal to2, rm -rf t/rules
14:55 metaperl is now known as _metaperl
theorbtwo Eh, it doesn't actually take as long as I expected. 14:56
nothingmuch make optimized usually pays off if you've got a strong machine
much of the time is latency starting pugs
and optimized /really/ helps there
theorbtwo Hm, it didn't for me last I checked, but we have a lot more tests since then. 14:57
gaal i'll look into concurrent prove later
theorbtwo Don't worry too much; it's not actually that slow. 14:59
nothingmuch gaal: you are getting into the testing habit, aren't you?
gaal++
theorbtwo Hmm, I should ignore the t/rules dir, though; a single test file with lots and lots of tests makes the entire graph warp -- because it tries to be wide enough to fit a single NBSP for each test. 15:00
nothingmuch maybe it should be multiline then
up to 20 tests per line, then <br>
making a slightly thicker table row isn't so bad 15:01
gaal maybe split it to several files? how come there are so many tests there anyway? were they carried over from some other distrib?
nothingmuch or maybe N / x tests perl line
gaal: p5 rules test
nothingmuch would like to see someone try to graph MPEG::Audio::Frame
theorbtwo They were taken from the p5 tests automagically.
nothingmuch ranging from 10-2000 tests per file
gaal nothingmuch: well, i'm writing a test harness now, but i'm evilly not testing it before writing it. 15:02
nothingmuch heh
it's supposed to save time
gaal i know
nothingmuch you don't /have/ to do it if it doesn't make sense
gaal i know that, too :)
nothingmuch =) 15:03
gaal -rw-r--r-- 1 roo roo 432678 2005-03-24 17:02 all.yaml
r1089 - yaml harness improvements 15:06
run it as: util/yaml_harness -v -r t/ -o something.yaml
jabbot pugs - 1089 - Stole some nice features from prove 15:07
nothingmuch is going to buy a costume 15:45
and eat, and what not 15:46
ciao!
theorbtwo
.oO(Costume?)
autrijus yo. 15:48
gaal it's the local variant of haloween here.
autrijus stevan: got it. 15:49
theorbtwo Ah. I suppose Easter isn't a big holiday there...
gaal well, it happens more or less near passover, which is 15:50
nothingmuch theorbtwo: depends, there's also some christian arabs
gaal unfortunately
nothingmuch easter is always fun at grandmas
theorbtwo True enough.
nothingmuch except all my memories are from when i was 10 or so
i doubt i'll enjoy searching for stuff this much anymore
gaal because you can't get bread in stores during passover.
nothingmuch anywho, *poof*
theorbtwo Well, that would make keeping pesa dich (or however you spell it...) easier. 15:51
OTOH, it would make not keeping pesa dich much harder.
gaal exactly 15:52
stevan autrijus: thanks 16:00
theorbtwo OK, testgraph should work with the newer yaml_harness. 16:05
hcchien do we still use grep {...} @foo in perl6?
or it's changed to "grep {...}, @foo"
theorbtwo IIRC: Yes, but the grep EXPR, LIST form no longer exists. 16:06
Odin-LAP Isn't there something about no need for comma after block?
autrijus find a reference?
I think grep {...} @foo should work 16:07
but I no longer recall the reason.
jabbot pugs - 1090 - Upgrade testgraph to work with new yaml_
gaal theorb2, cool. 16:10
theorbtwo pugs -V should probably actually mention the version.
gaal and the ghc version
and platform 16:11
i tried talking nothingmuch into adding it but that guy, he only has parties on his mind.
theorbtwo Well, S06 has some examples of grep CODE LIST, but I don't see anything specific.
hcchien and something like "grep { $_ % 2 } <== @data;" :) 16:15
autrijus try not to think about it :) 16:16
tests welcome
theorbtwo Yeah, has lots of those too, but some of them are just grep { $_ % 2} @ data.
gaal ==> $pressure <==
stevan hcchien: I think that grep BLOCK LIST is still supported 16:17
I cannot imagine they would break backwards compatibility that much 16:18
theorbtwo Esp in a way that makes common things harder.
hcchien stevan: yes, I guess so.
stevan hcchien: this sounds like a question for perl6lang 16:19
gaal to2 add \n to html plz 16:20
hcchien anyway, let me add it to grep.t
autrijus hcchien++ 16:21
lets just assume grep BLOCK LIST works.
stevan: * s:perl5:g/// now really works.
thanks for the report. 16:22
stevan hcchien: actually it used to be there, but it looks like viirya changed it to grep BLOCK, LIST in r1082 16:26
autrijus++
autrijus so that needs to be undone
or rather, additionally.
grep block, list also needs to work.
hcchien stevan: yes, but I think we should test both of them? 16:27
stevan autrijus: yes
hcchien: yes I agree
jabbot pugs - 1091 - * s:perl5:g/// now really works.
stevan autrijus: have you seen all of iblech's commits? 16:28
he is the mad perl6 porter
autrijus yeah.
I'm, like, wow.
stevan exactly
autrijus not sure how to summarize it.
stevan he's porting like a muthafu*.... shut your mouth (for all the Issac hayes fans out there) 16:29
tomyan i think i can see a bug with s:perl5:g, although not sure where to put test for it
autrijus tomyan: t/rules/newthing
stevan tomyan: what is the bug?
autrijus .t
theorbtwo I'm just talkin about iblech!
tomyan compare: ./pugs -e 'my $test = "aad"; $test ~~ s:perl5:g/ad/d/; say $test;'
with
perl -e 'my $test = "aad"; $test =~ s/ad/d/g; print $test, "\n";' 16:30
autrijus both prints ad.
you need the trunk! :)
hcchien so I pushed. :)
tomyan sorry
autrijus no need to be sorry
theorbtwo gaal, you've got your \ns.
autrijus I just fixed it 5 mins ago
tomyan switched to main repository on other machine
oops
autrijus 5 mins ago is like, stone age.
stevan autrijus: GHC 6.4 is sooooo much nicer :)
theorbtwo wonders if anybody has tested this with non-firefox browsers.
autrijus stevan: in what regards? :)
better error messages?
stevan fast
autrijus right! 16:31
C--++
C-- is totally da bomb.
stevan perlbot karma C
perlbot Karma for C: 28
theorbtwo wonders who owns perlbot.
stevan perlbot karma C-- 16:32
perlbot C-- doesn't have any karma
stevan perlbot highest karma
perlbot The top 5 karma entries: autrijus: 44, nothingmuch: 42, C: 28, ~brad2901: 26, stevan: 25
gaal theorb2++ 16:33
autrijus purr. 16:34
autrijus needs to work on slides.
which is essentially Pugs Apocryphon 2.
"Pugs Hacking 101"
stevan: we may very soon require GHC 6.4 for all nontrivial stuff. 16:36
like inline haskell and TH-based parrot copiler
compiler
maintaining two different implementations is getting out of control.
6.2 can still build basic pugs evaluator 16:37
just not the fun things.
stevan if 6.4 is stable, then probably best to force the change earlier rather than later
less people to complain :)
jabbot pugs - 1094 - r4336@hcchien: hcchien | 2005-03-25 00
pugs - 1093 - r4328@hcchien: hcchien | 2005-03-25 00
pugs - 1092 - testgraph -- more newlines in the HTML (
autrijus ok :)
so you think it may be better to just drop 6.2 support 16:38
theorbtwo I tend to agree.
autrijus instead of doing it halfway?
stevan autrijus: yes
autrijus that's that then.
stevan because the more cool stuff you put in, the more people will want to try it
autrijus true.
stevan the transistion is much less painful now
gaal ghc 6.4 is alpha?
autrijus gaal: no it's stable. 16:39
it's like perl 5.8
pjcj 6.4 isn't even in debian unstable yet, but it's not hard to build yourself - just takes a while
autrijus I'll drop 6.2 support before 6.2 then :)
gaal well, perl6 is like perl6! :-p
autrijus I need to also work on GHC 6.4 porting for AIX :-/
gaal so i "me to"
autrijus the sexy stuff in External.Haskell is so nice. 16:40
a complete haskell parser, transformer and emitter
stevan autrijus: maybe by the time pugs is ready for AIX, so will 6.4
gaal thouhg: with the speed of things, p6 will be out before ghc6.4 becomes prevalent. :p
autrijus in 30 or so lines.
stevan: sure :)
stevan gaal: pugs might push the adoption of 6.4
theorbtwo Oh: There's 6.4 i386 debs in experimental.
I suspect it will. 16:41
autrijus s/might/does/ :)
stevan i mean after all, all the /cooool/ kids use 6.4,... dont you want to be /coooool/
gaal that's the second language it pushes the development of.
is the 6.4 build faster, too?
autrijus yes.
much.
stevan gaal it is for me
so are the tests
gaal apt-get -t experimentals.
autrijus and its test message makes sense :) 16:42
s/test/fail/
s/fail/error/
gaal s///++ :)
autrijus stevan: ah, right. Data.Map is the reason.
obra has pugs caused any new haskell features yet?
stevan autrijus: shouldnt it be s:perl5/test/fail/
theorbtwo stevan, in this case, the p5 re and p6 rule are the same... I think. 16:43
stevan Data.Map++ then
theorbtwo: not in pugs :)
autrijus stevan: and I need to survey the new .*Map stuff
that will likely to speed up things another 20% or so
theorbtwo G'point.
autrijus and is cumulative with the compiler backend
obra: the Unicode stuff, for one
stevan you know,.. i think I will miss the days of everyone on perlmonks whining about how long perl6 is going to take 16:44
autrijus nothing much else, though.
stevan damn you autrijus
autrijus I think those days are already gone :)
gaal theorbtwo, what's the package name in experimental?
stevan autrijus++ # for stopping the whining
autrijus #perl6++ :)
theorbtwo ghc6. You need to explicitly install ghc6/experimental, though. 16:45
stevan ghc6??
theorbtwo (Hmmpf, this is removing ghc6-prof and hat for me...)
Yeah, it only refers to the major version, for some reason.
That's pretty normal for debian.
gaal yes i know. just weirdly the error message about the virtual package mentiones something else.
autrijus I was very delighted reading savige's perlmonk post today
brings back the olden memories
theorbtwo Odd, I didn't get a message about a virtual package. 16:46
autrijus # perlmonks.thepen.com/437487.html
gaal odd, i don't get the 6.4 package :)
(the virt package error is when you try "ghc" 16:47
)
clkao /win25
gaal savige++; # Lighter Side of Perl Culture series 16:48
theorbtwo wonders about the recent rise of the word "scintillating". 16:49
autrijus I don't even know what it is supposed to mean. 16:50
gaal ah, silly me. i had experimental commented out in my sources.list
autrijus (other than the obvious define: in google) 16:51
gaal: very prudent of you :)
gaal :)
theorbtwo bugs.debian.org/cgi-bin/bugreport.cgi?bug=300343 ++ 16:52
gaal i notice it depends on perl5 16:53
so does pugs :) 16:54
autrijus it's this roundabout boostrapping thing. 16:56
(ghc was bootstrapped by perl5 long ago.)
the part done in perl5 was named "Evil"
and the part bootstrapped with haskell was named "Glorious." 16:57
gaal heh
autrijus now only the Evil Mangler remains.
the other parts are now all Glorious.
shapr grins
gaal that's what the G in GHC stands for, no? Glorious
Odin-LAP Glasgow, wasn't it?
shapr The full name is the Glorious Glasgow Haskell Compiler.
gaal aaah, right. 16:58
theorbtwo No, that stands for Glasgow... but ghc -V gives Glorious Glasgow Haskell Compiler.
shapr There were more pieces written in perl before.
oh, I see autrijus was just saying that, oops.
17:08 Lynx_ is now known as Lynx_awy
gaal beh, yaml_harness doesn't catch pugs diagnostics from parsefails 17:09
gaal--
theorbtwo gaal, while you're at it -- put in the time of the test run? 17:10
Er, when it ran, not how long it took.
gaal good idea. (prolly should use GMT?)
more metadata == good, if anyone has any ideas 17:11
theorbtwo just give a time(). 17:12
gaal righto.
lhooq and dump %?OS (or whatever the name is) 17:16
17:16 lhooq is now known as dada
theorbtwo Oh: Skip m/[A-Z]/ (or is that m/Disabled/?) 17:24
gaal data: i'd rather that go into pugs -V, which i dump anyway 17:27
jabbot pugs - 1095 - testgraph: Output sane HTML, %success co
gaal theorbtwo, should i hardcode that in? 17:30
oops, dada :)
theorbtwo boggles -- it actually works on IE, Firefox, and Opera! 17:35
Odin-LAP What does? 17:36
theorbtwo testgraph.html
Odin-LAP Is such a thing even possible?
theorbtwo It presently doesn't try anything /that/ complicated, but I thought the title attribute on non-a tags didn't work on IE. 17:37
gaal [off topic] www.cs.huji.ac.il/~yweiss/Colorization/ be sure to scroll down
theorbtwo desert-island.dynodns.net/misc/testgraph.html if you want an example without generating your own. 17:38
Ah, yeah, saw that before -- it's pretty nifty. 17:39
gaal apart from the obvious insane compression, there's also the neato recoloration app 17:40
theorbtwo++; # testgraph looks really nice now!
need to write an aggregator of yaml inputs
hmph, it's not so trivial to catch deaths. 17:42
theorbtwo Can't you just catch both STDOUT and STDERR?
gaal it isn't me :) 17:43
theorbtwo Oh.
malaire does that testgraph-script generate one html-page for each revision? 17:48
Is there any way to quickly see test-result-diffs between revisions?
theorbtwo No easy way at present. 17:50
gaal hmm, is there a way to get the svn r number? 17:52
(without peeking :)
rgs not really. try svn info 17:53
theorbtwo src/Version.h
malaire I was thinking of a format quite similar to what you currently have. Each test-file has single line in html-output with colored blocks, but each block represents that test-files result for e.g. last 20 revisions.
i.e. one colored block per test-file per revision 17:54
jabbot pugs - 1098 - Added t/examples/golf.t 17:57
pugs - 1097 - * Ported some other subclasses of URI.
pugs - 1096 - Ported Commands::Guarded 0.01 (complete
theorbtwo I don't know if a file-level resolution is useful. 18:00
malaire It would of course be better to get some format which shows test-level diffs between revisions... 18:01
e.g. Could the colored block in current version contain more information that just the color. (e.g. add a single 'x' (or something else) for each block which has changed from last rev 18:02
s/block/blocks/ && s/that/than/ 18:03
theorbtwo Hmmm... if we keep around enough data, I could make each block show "r..." for the last time it changed -- and possibly put a border on those that changed recently.
malaire border might be good. It works also for small blocks which are too small to contain any letters.. 18:04
jabbot pugs - 1099 - * pugscc --runparrot examples/mandel.p6 18:07
elmex PUGS 18:08
autrijus mandel runs.
elmex thats a major feature ;) 18:12
autrijus yes. 18:13
mugwump please, nobody tell /. :)
autrijus lol.
elmex not?
autrijus probably wait till 6.0.13.
mugwump nods
autrijus # autrijus.org/mandel.imc 18:14
nothingmuch evening 18:15
theorbtwo Find yourself a nice costume?
autrijus nothingmuch: hey. --runparrot mandel.imc works.
nothingmuch got a pink wand 18:16
with sparkly stuff
you mean parrot in pugs?
mugwump pugs spits out imcc and passes it to parrot I think
not embedded yet
malaire it really works :) and it's quite fast... 18:17
mugwump which will be needed for eval $string
jabbot pugs - 1100 - * simple optimizer, suggested by Coke, t 18:27
autrijus is there a debian package for parrot or something? 18:32
there's no parrot in fbsd ports :-/
cls_bsd_: *hint* *hint*
jabbot pugs - 1101 - * Coke pointed out we need no intermedia 18:37
gaal has been abducted by $work 18:39
elmex woho 18:48
PugS 18:53
? 18:54
i-blech-bot?
gaal where's a good place to put a small data file that's useful only for people running smoke tests? 18:58
in the tree that is
util/?
docs/?
top level?
theorbtwo util/, probably.
autrijus util/
gaal k thanks
theorbtwo I should move the testgraph.css...
ninereasons are we going to be able to say things like " $i .. -$i " ? 18:59
or is it a good thing to have to reverse a range, to count down? 19:00
theorbtwo Well, there's always 10..-10:-1.
malaire hmm.. will that interfere with proposed radix-syntax :16<abcd> ? 19:01
ninereasons I get "-1", did I do it wrong? 19:02
or is it a TODO?
theorbtwo This is a good question, mal. 19:03
autrijus eh. 19:04
10..-10:by(-1)
is the syntax.
that's a TODO.
theorbtwo Ah. 19:05
ninereasons oh. I like that syntax.
Khisanth 10..-10:by(1) will emit a warning? :) 19:06
autrijus maybe :)
Khisanth I don't know ... seems like it would better if the "steps" were an absolute value 19:07
ninereasons bleh. coworker bumped my power cord 19:15
never happened before
Khisanth badly place power cords? 19:16
ninereasons I didn't think so. I've changed my mind 19:17
jabbot pugs - 1102 - Ported most of URI. 19:27
iblech Sorry for iblechbot rejoining that often -- my gateway has extreme power supply problems, reboots every ten minutes or so :/
ninereasons boy, it takes so long for pugs to be compied with "make optimized" 19:31
I suppose it's not really worth it, at this point.
autrijus iblech: I'm in awe. 19:35
five modules in a day. 19:36
stevan iblech: me too 19:40
very impressive
nothingmuch iblech++? 19:43
stevan too late nothingmuch
nothingmuch perlbot karma ibelch
perlbot ibelch doesn't have any karma
nothingmuch eh
stevan his gateway sucks
nothingmuch perlbot karma iblech
stevan huh
perlbot Karma for iblech: 4
nothingmuch see, not enough. 19:44
iblech++
stevan iblech++
jabbot pugs - 1103 - * benchmarking time 19:47
elmex i want more carma and women 19:51
jabbot pugs - 1104 - more detailed oo/attributes tests; some 19:57
autrijus rgs: you'll like this. 20:15
gugod and ingy ported Kwid to Spork/Kwiki.
and they wrote a Spork to S5 generator.
so you can just write Kwid
and generate S5.
nice eh?
it's going to power YAPC::Taipei speakers.
journal up. zzz & 20:16
pjcj autrijus_journal =~ s/ibelch/iblech/ 20:20
or whatever the P6 syntax is ;-)
theorbtwo iblech++
rgs autrijus: ooh nice indeed. 20:22
gaal hi, i updated the yaml harness a bit again
autrijus fixed.
gaal 1105 already! wow 20:23
jabbot pugs - 1105 - cleanups; optionally identify tester 20:27
Limbic_Region wonders if there is enough p6 support in Pugs to write his Rotational Prime Number finder code 20:34
p5 regexes are supported right?
pjcj via pcre 20:35
Limbic_Region right, so /e isn't supported - will need to modify that piece 20:37
jabbot pugs - 1106 - Added Config::Tiny 2.01 (complete with t
gaal whee!
i like that module.
Limbic_Region out of curiosity, will these modules be revisited/refined as Pugs becomes more feature complete? 20:39
gaal depends on the individual authors+porters, presumably 20:40
ninereasons I certainly hope so. They'll be museum pieces within a year, otherwise, I suspect.
Limbic_Region I am not afraid of code being untouched that is solid - I just think there might be an awful lot of "work-arounds" ATM as the features aren't available 20:45
ninereasons well, and things aren't settled down, entirely, right? 20:46
the differences between "Perl 6 Now" and Pugs are noticeable.
that pace of change will continue, at a slowing rate, won't it? 20:47
stevan Limbic_Region: none of the code in modules/ is expected to work for a while 20:55
The idea of the modules/ directory is to start porting your favorite module to perl 6 and as things get completed they can serve as pseudo-integration tests for Pugs 20:56
Corion A good evening! Mad props to nothingmuch++ who implements stuff I dream up
stevan if you wanna see hacks and work arounds,.. check etc/FileSpec/ :)
howdy Corion
Corion I dreamt up the -p , -n and multiple -e switches, which is what I'll start implementing now (as soon as "svn up" does its deed) 20:57
-e print -e "q(" -e hello -e " " -e world -e ")"' should work IMO 20:58
(and AFAIK it works in Perl5
qmole wow, so it does 20:59
stevan Corion: but why would you want it too?
Corion stevan: I find -p and -n very convenient (in P5), and the multiple -e is just a side effect of implementing -n and -p. Also multiple -e switches get me around ugly quoting issues from time to time 21:00
stevan does mostly web stuff
Corion but feel free to discuss this more before I implement it, so I don't waste my time (well, it's not wasted, as I learn Haskell from that)
gaal multiple -e works in sed too :)
Corion stevan: Heh - I relatively seldom do web stuff (except for zaphod.datenzoo.de ) 21:01
( SubEthaEdit for those who don't have OSX but just a browser)
For -p and -n, I will need to keep a buffer of the "current script", where -p and -n fill the head and tail of the buffer, while the -e's accumulate the body of the script 21:02
What does the "G" flag of "svn up" mean? 21:03
gaal conflict iirc 21:04
Corion gaal: Ah - that's plausible ... I should delete my hacks to make way for nothingmuch's implementation :)
theorbtwo svn revert filename 21:05
Corion theorbtwo: Ah - that wipes all local modifications?
stevan actually G is not conflict 21:06
G is merge
(i think)
yeah G is merged
svnbook.red-bean.com/en/1.0/re28.html
Corion stevan: That would be ugly, because nothingmuch reimplemented what I hacked on, but differently, I think ... Or svn is far more magic than I think :)
stevan corion, you might want to rm that file and svn up it agin 21:07
Corion stevan: Yep
jabbot pugs - 1109 - Ported Getopt::Std 1.05 (complete with t 21:17
pugs - 1108 - Work on build process for inline Haskell
pugs - 1107 - add --exclude (-X) so that t/rules/Disab
Corion Yay. I just fixed 10 "skip" tests on Win32 to "pass", at least for Win2k, but they should pass under Win9x too - should I paste the change onto pastebot for incorporation? 21:20
gaal you don't have commit privs, Corion?
please do 21:21
Corion gaal: Naah - I was here yesterday evening for one hour - commit privs don't come /that/ quickly (and I don't want them yet either)
gaal you'd be surprised :)
Corion Gah - no paste url in the scrollback :)
gaal perlbot nopaste 21:22
perlbot Paste your code here and #<channel> will be able to view it: sial.org/pbot/<channel>
theorbtwo Let me see the patch?
Corion gaal: Thanks! I have to get accustomed to having an infobot on the channel :) I ran one in a different chanel+life :)
pasteling "Corion" at 217.86.57.96 pasted "Make t/examples/examples.t Pass instead of Skip on Win32" (46 lines, 964B) at sial.org/pbot/8588 21:23
gaal sure
Corion I pasted the whole new code instead of just a diff
I also added more noise via some diag() calls, but I think that more noise is good while we still hunt for small stuff like $?OS being "MSWin32" instead of "win32" :) 21:24
theorbtwo Corion, sure you don't want commiter bits? 21:26
Corion theorbtwo: Currently my tree is a mess, but it would be convenient to have them, sure. The thing is, I don't know how long/often I'll commit stuff. On the other side, the Win32 part needs a strong avatar :) 21:27
theorbtwo Yes, it does.
And avatar is not the correct term.
Corion theorbtwo: I see it as "with big privileges, there comes big responsibility" - and I don't really want the responsibility :)
jabbot pugs - 1110 - include svn revision info in report
gaal Corion++, r1111
theorbtwo Not really, no. 21:28
gaal snv has the responsibility.
svn
theorbtwo Don't write code that rm -rf's anything important, that's about as far as it goes.
If you write something that's crap, it'll get backed out, everybody's good. 21:29
Corion theorbtwo: According to Wikipedia ( en.wikipedia.org/wiki/Avatar ) , Avatar is not too far. But Knight In White Armor is close enough :)
theorbtwo: Ah, if I were malicious, I'd simply release stuff on CPAN.
s!close enough!closer!
gaal champion is a suitable word.
theorbtwo Champion works well.
Corion Yeah - Champion would be the term I was looking for :) 21:30
So, I'll commit three new files as failing tests first, that's the general spirit of development, right? Even if I won't be around to fix the failures within the next few days ?
theorbtwo [email@hidden.address] ? 21:31
Corion ... or should I note in the tests how I envision the implementation ?
theorbtwo Yes, it is.
Corion theorbtwo: True
gaal yes. the failing test is your bug report.
Corion gaal: It's not a "bug" per se, it's more missing features, like -n and -p
theorbtwo In that case, they should be todo tests.
gaal your todo_is is your feature request.
Corion BTW, has anybody applied my "win32 fix" already ? 21:32
gaal yes,
Corion Ah - todo_is it shall be then
gaal r1111
Corion, when someone says something that ~~ /r\d+/, they usually mean they just committed :)
Corion gaal: Aaah :) So I should quickly svn up again :) 21:33
gaal: Yeah - I'm still learning the lingo :)
gaal is the dependency on Spiffy new?
just noticed my build fails on msys 21:34
which has a cranky ol' 5.6.1
Corion Ugh ... The build should not depend on Spiffy ... But that's to be expected if we have Ingy on board ...
gaal i guess.
at least it doesn't depend on perl6. yet. ;) 21:35
Corion Spiffy doesn't even work on all Perl 5.8 - it wants at least 5.8.4 I think 21:36
chip Why doesn't Spiffy like early 5.8?
jabbot pugs - 1111 - apply Corion's patch to allow tests on w 21:37
Corion chip: Don't ask me. I only know it doesn't work with 5.8.0
gaal theorbtwo, there's some new metadata for the htmlizer if you want it.
Corion BTW, is there an inverse operation to slurp() ? I think it's called spew(), but it doesn't seem to be used/tested/implemented ? 21:38
gaal also something to skip t/rules/Disabled
`cat $var > $file` 21:39
i've seen that done :((
chip ha. spew (FILE|HANDLE), STRING
Corion gaal: Ha ha ha ...
chip what a great idea
Corion chip: Ah, so it should exist ... I'll write a test with it and see how badly it blows up :)
chip still wishes that the chip and chimp operators had been accepted into Scalar::Util 21:40
gaal say, with =<file> syntax, is file the literal file name?
theorbtwo Wanted an epinonymous operator in the core, chip? 21:41
Corion chip: True :)
gaal because if the syntax is =<$filename>, i don't understand how that's going to work in the sigil-free dialect
theorbtwo Oh: Corion, on your first commit, add yourself to AUTHORS please. 21:42
Corion theorbtwo: Ah - good self-maintenance :)
gaal Corion: you may find t/README useful if you haven't read it yet.
Corion It's sorted alphabetically by first name?
gaal no, it we keep randomly shuffling it but it always comes out looking that way. 21:43
theorbtwo Yes. 21:45
To whit, it's sorted unicodeographically, but ignoring all chars that do not match [A-Za-z].
See utils/fixauthors.pl
Juerd hopes Perl 6 will have a built in way to sort image3 < image20 easily. 21:47
Corion I pretend to understand the Unicode that FireFox renders as ??? in the agreement :) 21:48
theorbtwo Hmm, autrijus said he'd fix that.
He pointed at an English translation, which I long ago lost, but it's nothing bad.
Juerd theorbtwo: Yes, but you are a participant in a plain text chat network, and thus not trustworthy :) 21:49
It's a legal, binding agreement. I always wonder why people accept those things so lightly. 21:50
Corion Juerd: Well, Autrijus will have a hard time to enforce it in Germany :)
theorbtwo Juerd, there are two reasons: 1) Trust in other people who say that they are OK.
2) A blind assumption that they will not come into play.
Juerd Corion: You'd be surprised. 21:51
theorbtwo: But how can you know those people can be trusted?
chip =($file) is great for the zsh command line
Juerd theorbtwo: You could be some h4x0r who's inserting packets into your irc tcp stream...
After having changed the chinese agreement 21:52
theorbtwo That is true. 21:53
ninereasons someone said "carelessness is an autrijus mantra"
Juerd carelessness is great when hacking
theorbtwo However, people who have a long relationship with me will notice certian patterns in my typing and modes of speach.
Juerd Not when virtually signing a legal document.
Corion Hmmm. svn doesn't give me no love when I try svn commit -m "message" --user corion --password secret 21:54
Juerd theorbtwo: So does the h4x0r who has sniffed your plain text irc connection for the past four months.
theorbtwo That seems like a lot of effort to go to.
gaal corion, try w/o a user at all
press enter at the passwd
then enter your username and password
Corion gaal: For commit, not for check out :)
gaal i know
Juerd They could easily use a markov chaining algo to verify that what they type agrees to your patterns :) 21:55
theorbtwo: Certainly.
gaal it is clever :)
Juerd theorbtwo: But if we build things on trust, then why the written agreement?
Corion gaal: Nope, doesn't work.
Juerd These things are not in synch.
gaal corion, what shell are you using, mingw?
Corion gaal: Nope - plain cmd.exe 21:56
gaal and the svn, is it a pure svn client?
Corion and svn tries to connect to svn.openfoundry.org:80
gaal: Yep - some Win32 svn client
(and checkout worked well, apparently)
Anyway - I should hack a bit on making -e print -e 1 work instead of worrying about this :) 21:57
gaal what error message are you getting?
Corion gaal: Authorization failed 21:58
(resp. the German translation of that)
gaal try --username instead of --user, though i doubt that's it :)
Corion gaal: I already tried --username (typo up there) 21:59
gaal beh, i'd try a clean checkout and with no auth at all on the command line
Schwern $ perl Makefile.PL
Ambiguous use of -XXX resolved as -&XXX() at ../../inc/Module/Install/Pugs.pm line 29.
...
at ../../inc/Module/Install/Pugs.pm line 29
Corion gaal: Yep
Schwern: I got the same (and ignored it for the time being)
gaal Schwern, it depends on Spiffy now, it seems. 22:00
Schwern Corion: Hard to ignore. I can't build anything!
Corion Banish Spiffy :(
Schwern I have Spiffy
Corion Schwern: Yeah - I'm running on yesterdays build
Schwern pugs really shouldn't be using Spiffy 22:01
Corion Schwern++
Schwern Not until it has some bloody documentation
And the line in question is a comment. :)
Corion Schwern: Spiffy is a source filter, so it's supposed to mess up your script in undecipherable ways 22:02
gaal added in 1108 it seems.
Schwern $ make test
chmod: too few arguments
Try `chmod --help' for more information.
make: *** [/.exists] Error 1
Why are they building pugs with MakeMaker? 22:03
theorbtwo svn ci --username corion filename
Schwern INST_LIBDIR and INST_ARCHLIBDIR aren't getting set. :( 22:04
gaal Schwern, it's a comment?
Schwern gaal: Yes. The line with XXX in it is a comment.
gaal not in my tree
Corion theorbtwo: I don't want to checkin the AUTHORS file, I just want to commit the changes :)
Schwern # $mm->{'INST_AUTODIR'} = 'XXX$(INST_LIB)/$(FULLEXT)'; 22:05
gaal that's not line 29
use Spiffy -XXX;
Schwern But that's the only lien which uses -XXX so I delted the use Spiffy line, too
gaal "Spiffy can export a few debugging functions [...] XXX dies with the output" 22:07
Schwern *thud*
THANKS FOR BREAKING THE BUILD 22:08
stevan somebody wake up ingy 22:09
Schwern Somebody with committ access just remove the code
gaal no, eeverybody svn up
r1112
Schwern Thank you
gaal my pleasure.
Schwern Still won't build, but hey 22:10
gaal pesky
buildingforme
what are you getting?
theorbtwo gaal, do you have write permission for /?
Cannot write /.exists:Permission denied at /usr/share/perl/5.8/ExtUtils/Command.pm line 127.
gaal what's /? 22:11
Schwern $ make
chmod: too few arguments
Try `chmod --help' for more information.
make: *** [/.exists] Error 1
I see exactly where the problem is
He's rewriting init_INST
gaal how come it works for me then?
Schwern In Module::Install::Pugs::set_makefile_macros()
gaal ah
*that* /
Corion Oh great. And it now fails on Win32 due to badly built path names (forward slashes instead of backslashes :(((( )
gaal wait wait way
let's take this one at a time :)
Corion ... in nmake test. But I'm away hacking on different stuff anyway :) 22:12
gaal i have msys, so very possibly my / is writable.
forward slashes ought not break on windows!
since c system calls honor them
unless there's some stupid string comparison?
Corion: spaces in pathnames do break the build on win32, at least used to. 22:13
Corion gaal: That's expected, because spaces in pathnames break the build everywhere. But forward slashes as the path separator don't work in the Win32 shell, and that is what I'm seeing here.
gaal: The Win32 shell (cmd.exe in my case) does care for backslash vs. forward slash, as the forward slash is the switch initiator for cmd.exe 22:14
gaal all right, lets fix that then. do you see where it happens?
Corion gaal: (and as the Makefile goes through the shell, stuff breaks)
gaal: I haven't looked into the Makefile 22:15
Schwern The problem is he's not setting INST_LIBDIR
gaal is_extension_build i see assumes /
jabbot pugs - 1112 - unbreak the build 22:17
Schwern Fixed it 22:18
gaal can you ci?
perlbot nopaste
Schwern Don't have access. nopasting
perlbot Paste your code here and #<channel> will be able to view it: sial.org/pbot/<channel>
gaal i'll ci
Schwern sial.org/pbot/8590 22:20
theorbtwo Looks good to me, Schwern. 22:22
Do you want a comitter bit?
Waitasec, you have one.
Nevermind.
Schwern I do?
gaal checking in
theorbtwo Um, you're listed in AUTHORS, which I suppose is not actually the same thing. 22:23
gaal r1113, please test (it worked for me before Schwern's fix too)
Schwern One of these days I have to figure out how to get rxvt to display unicode
pjcj I think you have to specifically ask not to be a committer, don't you? 22:24
Schwern Well the tests don't work
gaal at all?
Schwern pjcj: Specifically ask *not* to be a committer?
pjcj seems that way ;-)
gaal oh crap, they dont' for me either.
Schwern t/builtins/defined.........................sh: line 1: /Users/schwern/devel/pugs
can't run t/builtins/defined.t. No such file or directory
theorbtwo Works for me now. 22:25
Schwern err, blib/script/pugs is missing
gaal bonks himslef on the head for ci'ing before make test 22:26
wolverian Schwern: you use urxvt. :)
Schwern gaal: Well, its "less broken" now
gaal i used to have this borkage, i think i fixed it by upgrading makemaker.
Schwern What sort of maniac uses q*foo* quoting in a one-liner?
gaal: I assure you I am using the latest MakeMaker 22:27
theorbtwo Ah, now fails after make clean.
gaal heh
theorbtwo Schwern, one who wants to be shell-independent, and thus can't use ' or ".
Schwern But *?!
jabbot pugs - 1113 - unbreak the build, mk2
theorbtwo Yeah, * is kind of strange. 22:28
Schwern * just screams "I'm a shell metacharacter"
q{} is safe 22:29
gaal blib or no, why is pugs even installed into script? 22:30
Schwern q[] surprisingly is not
Corion Weirdly enough, it still creates a forward slashed path to blib/script/pugs.exe on Win32, where it shouldn't.
Schwern Its what Module::Install::Pugs wants
Corion: That's ok.
Corion Schwern: Nope, it's not, as the shell sees it :)
Schwern The shell can handle it 22:31
As long as its not mixed / and \
gaal my eyes
Corion Eh - weird indeed. The shell can maybe handle it, but it is right about blib/script/pugs.exe not being found, as it's not there.
theorbtwo script at that level should probably be named bin. 22:32
Corion ... because pugs.exe now lives in blib6 :)
theorbtwo pugs.exe doesn't belong in blib6, IMnsHO.
Schwern nopaste.snit.ch:8001/2168 22:33
That fixes the q** problem 22:34
gaal Schwern, is it a problem, or a weirdness?
Schwern It has a high probabability of being a problem
gaal does it fix anything for you now? i'd rather not commit w/o asking the original author (autrijus? ingy?) it unless it does 22:35
Schwern nopaste.snit.ch:8001/2169
This gets the tests working.
Nothing is broken for me but it'll probably break on non-Unix platforms. 22:36
Its a MakeMaker and portability issue and a minor edit. 22:37
theorbtwo Hmm, I thought the problem was that it was in blib6 instead of blib?
Schwern That change appears to have been done by iblech
theo: This is a different issue.
jabbot pugs - 1114 - ~~50% of MIME::Lite 3.01 port done.
Schwern theo: nopaste 2169 solves the blib vs blib6 problem. 2168 is a probable portability issue. 22:38
gaal Schwern, I'm surprised, because q** has been there for a while i think and autrijus used to use activeperl
Schwern I so want a "show me all the changes to line X in file Y with context" command. 22:39
gaal: Yes, it might work out ok but its playing with fire. q{} is known portable.
gaal svn blame?
Schwern svn blame only goes back to the last change
Then I hav eto manually work backwards through each change
gaal wait, how is it a portability issue? 22:41
Schwern Looks like its been there since the file was added.
gaal: * is very likely to be a metacharacter to somebody
gaal: Or some make
gaal but not to perl, surely?
Schwern gaal: Its not Perl code, its make
gaal well, in that case, better change it in Pugs::MakeMaker too. hold on. 22:44
Schwern Yep
Well, the tests work now. 22:45
103 failures
gaal yes, i'll vommit both changes in a minute.
awwaiid Is anyone working on a port of the CGI module yet?
theorbtwo 103 sounds about right, Schwern.
pjcj I hope not ;-)
theorbtwo Sure you don't want comitter bits? 22:46
gaal oh, can't perl6 abolish the web, while it's at it?
Schwern theo: Can't hurt
theorbtwo Email?
Schwern All this just to try out the manbelbrot optimization
[email@hidden.address] 22:47
Which pugs loses out on but still much, much faster.
jabbot pugs - 1115 - File::Spec::Unix is DONE and tested; Fil
Schwern Ugg. I hope File::Spec does not live on into Perl 6 22:48
stevan Schwern: it does for now
Corion (it could do with a syntax change :) )
stevan we will replace it soon though
Corion: I couldnt agree more 22:49
I just spent the past week porting the internals of it
Juerd File::Spec is a horrible productivity killer
I'd much prefer a generic path notation to be converted to whatever is needed
stevan BTW that is a working version of File::Spec::Unix,.. not just a port which is waiting for features
theorbtwo Odd, you already had an of.org account, but were a member of no projects?
Anyway, you should be able to commit now. 22:50
pjcj 103 subtests fail | pugscode.org <<Overview Journal>> | pugs.kwiki.org | logged: xrl.us/e98m 22:50
theorbtwo Thanks for your work on this! 22:50
Schwern Something more like Path-Class
gaal r1117, Schwern++
Schwern But a bit more transparent.
stevan Schwern: once objects are in place, I would like nothing better :) 22:51
Schwern stevan: I thought they were.
stevan nope
t/oo/*
all todo tests
Schwern theo: Except I don't know my password
stevan: Ahh
stevan modules don't even work right yet
theorbtwo There's a "forgot my password" thing.
(We have require, but not use.)
stevan and subroutine args are still immutable 22:52
Schwern String overloading?
stevan return values are foobared in many cases as well
gaal there'd have to be traits for that to be otherwise, no stevan?
stevan Schwern: nope
(well in some cases, but not controlable from perl)
not to brag (becuase it is nothing to brag about), but the ext/FileSpec stuff is state of the art Pugs code 22:53
it pushes it to its current limits 22:54
and god it is ugly :P
(and you thought perl5 File::Spec was ugly)
gaal i like the bit where you do sub cwd returns Str { system("pwd") } :-) 22:55
stevan gaal, ... if you wanna see HACK,.. look at File/Spec.pm
how I do the cross platform stuff 22:56
it exploits unfinished/broken features
my goal is to make this work, and to keep it working until I can make something better
gaal heh, i made a moose smiley not on purpose.
jabbot pugs - 1117 - Schwern: "q** is probably unsafe. q{} is 22:57
pugs - 1116 - Ported more of MIME::Lite, but the rest
Schwern My work here is done. If you need me again just admit that you're screwed and die. 22:58
theorbtwo Bedtime for me -- night, guys!
pjcj Does anyone know of anything like Pretty I can use to dump an entire AST, or how to get Pretty do dump subs and blocks? 22:59
stevan night theorbtwo
gaal bedtime is a good idea actually.
night all&
stevan nite gaal
pjcj Ha - you all leave when I have a question ;-)
stevan pjcj: I think that the pugcc stuff is built around the idea of dumping ASTs
look in src/Compile/* 23:00
pjcj Yes, but I don't see subs implemented there.
Are they?
stevan I have no idea
sorry
gaal i thought there was a command line option that did that
must have been removed at some point
stevan dump ASTs?
maybe 23:01
or something in the pugs shell
. <exp> = show the syntax tree of an expression
in the shell
gaal ahh
stevan ok dinner time,.. see ya'll later 23:02
gaal yeah bye
pjcj I don't think that will dump a sub though.
chip Good news: Perl6::Subs can be CPAN'd. Bad news: $dayjob doesn't like Same Terms As Perl. They want LGPL.
pjcj Never mind - I'll try learning some more Haskell ...
chip Not that it'll affect anybody, LGPL is fine, it's just a pain in the @$$..
pjcj as is licensing in general 23:04
ninereasons "real 192m6.770s" 23:31
theorbtwo returns. 23:33
LGPL is pretty easy in a perl environment. 23:34
Keeps you from freely cut-and-pasting, but you can use with wild abandon, pretty much.
pjcj Half an hour's sleep should be enough for anyone!
theorbtwo I didn't go to sleep. Just bed. 23:36
Corion Hah. I changed the command line option parser around so it now will accept multiple -e switches. Now I "just" have to reintegrate the stuff with Pugs... 23:37
Odin-LAP theorbtwo: Is that intentionally ambiguous? :> 23:38
theorbtwo Yes, it is.
My sleep cycle is pretty independent of wall clock time. My girlfriend's is not.
Odin-LAP doesn't ask for clarification, then. ;)
pasteling "Corion" at 217.86.57.96 pasted "prototype of collecting multiple -e switches" (84 lines, 2.8K) at sial.org/pbot/8598 23:39
Corion (but I really think that the command line parser should be a real (Haskell) parser instead of this hackery :) ) 23:40
... unfortunately, all I know is this hackery, and no Haskell parsing techniques, so that's all I'll be able to do :) 23:41
theorbtwo Hm, I don't like how different -e and files seem to be. 23:46
For example, why does -c work on files but not on -es?
Corion theorbtwo: I don't like that either, but I haven't factored that handling out - it's very ad-hoc at the moment, with the different calls to doRun and doCheck ... The function to call should be another parameter, but I have to find out how to stuff all these parameters into a hash 23:48
... that gets passed around 23:49
(see my patch on the first approach with an accumulator variable in _run)
... and I still can't seem to commit - I must be doing something wrong :)
anyway - sleepy time now... I have to be fit tomorrow :) 23:50

Powered by Cro and the Raku® Programming Language.

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

Thank you!