pugscode.org «Overview Journal Logs» | r1773/winxp: 587/4256 smoke/win2k (occasional): xrl.us/fqum (158/4302, r2032) | pugs.kwiki.org
Set by autrijus on 18 April 2005.
mugwump Junctions do seem to be an arse to deal with, in terms of writing visitor patterns like serialising or stringifying 00:03
It's the return of (if (ref $foo eq "ARRAY") { ... } elsif (ref $foo eq ...
machack666 you almnost have to store the method bny which they were created 00:05
I i'm getting sporatic readline bugs 00:06
visual artifacts, doubled letters (of which there is only one)
crysflame perhaps someone put a junction in your prompt
machack666 snorts 00:07
also got two segmentation faults and several illegal instruction crashes that shot me out of pugs 00:08
all when dealing with junctions
mugwump ouch 00:17
any reproducible ones?
machack666 nothing consistent 00:18
a lot when I was dealing with the junction comparisons
ok checked in an associativity test
jabbot pugs - 2105 - Make $<delim> not interpolate in quoting 00:21
pugs - 2106 - Tests for the associativity of junctions
machack666 what does the double-U mean in svn update?
(as a status) 00:22
never mind: `svn help status` 00:23
jabbot pugs - 2107 - Fixed wrong plan count for t/builtins/ha 01:21
obra where's the paper on implementing make in ocaml? 01:40
www.venge.net/graydon/talks/mkc/html/ 01:48
mugwump So, going to implement a make in haskell, obra? :) 02:00
crysflame www.cse.unsw.edu.au/~pls/thesis-top...nmake.html 02:03
In this project you will design a clean, simple, portable replacement for `make', as a Haskell EDSL.
mugwump maybe I should go move to NSW and complete that for a masters 02:04
obra mugwump: no. a firend wanted it. 02:14
jabbot pugs - 2108 - updated the ChangeLog with all of tadays 02:21
lucs Hola bradb 03:43
bradb hey lucs 03:53
so, i'll be a bit late tomorrow evening. hitting up the plateau with my real estate agent.
lucs Good luck with that. 03:57
I'm beat.
lucs & # ZZ
jabbot pugs - 2109 - quote generator: add a few phrases 05:11
pugs - 2110 - quote generator: up the samples to incre 05:21
pugs - 2111 - quote generator: put a cap on annoying p 06:11
nothingmuch morning 06:17
06:40 castaway_ is now known as castaway
mj morning 06:42
nothingmuch morning, mj 06:43
castaway morning
jabbot pugs - 2112 - quote generator: a little more fiddling 06:51
pugs - 2113 - delete obs file
pugs - 2114 - * restore `%h<str> = want()` to impose s 09:21
Jonathan_ The Win32 daily builds are back for good. www.jwcs.net/~jonathan/perl6/ Sorry for the gap in putting them out. 09:29
jabbot pugs - 2115 - * oops, repair readline() 09:31
kungfuftr bah! 09:39
autrijus: don't suppose you know of a tool for converting PDF to text that'll cope with east asian languages? 09:40
09:45 chady_ is now known as chady
nothingmuch hola 10:10
Jonathan_: want to arrange for smoke testing too? 11:07
akin to: nothingmuch.woobling.org/pugs_test_status/ 11:08
Corion only puts them out once in a while, i think
shapr waves the lambdacamel pom-poms 11:12
nothingmuch hola shapr
shapr y0 wassup? 11:13
nothingmuch not much
theorbtwo Allo, shapr.
And all, since I haven't said anything today.
nothingmuch tests covered executable %
26 12071 56404 21
shapr greetingz Mistah Palantir.
nothingmuch 1 38462 171598 22
beh
theorbtwo 's adventures in macland are very much not working.
nothingmuch tests being 1 on windows is nonsensicle, it's all the tests
the rest is line counts 11:14
theorbtwo What's that from, nothingmuch?
nothingmuch how the fsck are there 3x as many executable lines of code?
i hate complilers! i hate c++!
our project, and code coverage with a few tests
first line is linux (g++, gcov)
second line is window
s
dunno which tools
shapr obra: there's at least one nifty make tool in Haskell. Possibly up to three. 11:15
Jonathan_ nothingmuch: Sure, got details of how I can set that up? 11:16
nothingmuch yup
there are some scripts in util
Jonathan_ Also, is anyone seeing the makefile problem I posted on the mailing list yesterday?
Ah, OK.
nothingmuch yaml_harness runs the test suite, and creates tests.yml
then testgraph.pl reads that yaml (very slow, you can hack it to storable if that bothers you) 11:17
shapr Why not a 'make smoke' command that sticks all the steps together?
nothingmuch and uses Test::TAP::HTMLMatrix to generate an HTML
uh
for starters everyone has a slightly different webserver setup
i don't know really
then you run catalog_tests.pl to create backlinked tests and docs 11:18
and then you shove all that on a webserver somewhere
with the css
shapr I'd like to have everything but shove in a single command.
nothingmuch nothingmuch.woobling.org/pugs_test_status
i have a shell script
there's run_smoke.pl
which i haven't gotten along with yet
and since mine worked for me i keep using it
Jonathan_ OK, I'll see if I have time to get that set up later on today. 11:19
nothingmuch perlbot nopaste
perlbot Paste your code here and #<channel> will be able to view it: sial.org/pbot/<channel>
pasteling "nothingmuch" at 212.143.92.226 pasted "test_report.sh" (39 lines, 1K) at sial.org/pbot/9482
Jonathan_ Thanks. I may be able to get smoke tests on a Linux box going too. 11:21
nothingmuch goody
shapr cheers
Jonathan_ And this time I promise not to run off to Spain and leave it all broken, like I did with my Win32 Pugs build. :) 11:22
nothingmuch =)
Jonathan_ Shouldn't be much risk of that, it's back to university on Saturday. 11:23
Ugh...exam term.
theorbtwo Grrr. I keep writing #elsif instead of #else for some stupid reason.
11:32 chady is now known as [s]AINT 11:33 [s]AINT is now known as {s}AINT
mj WinXP, s/todo_//, r2115 - 1000/4459 subtests failed, 77.57% okay. 54 subtests skipped. 11:39
theorbtwo pings nothingmuch.
nothingmuch chimes 11:55
castaway wonders if other people chiime differently, and whether pinging can play tunes..
shapr pings nothingmuch, castaway, and theorbtwo in that order 11:56
It's like a chrismas bells concert! 11:57
nothingmuch ding!
theorbtwo My mac still isn't playing nice, after getting Panther CDs.
castaway dong! 11:58
nothingmuch is it cursing at you?
theorbtwo WooOOOooong.
castaway hmm, of course the lag makes the tune come out all wrong :)
nothingmuch rings, reverbing theorbtwo's pong
shapr grins
nothingmuch wmmmmmmmmmmmmmmmmmmmmmm
theorbtwo It doesn't seem to be doing anything in purticular, just sitting at the "preboot" screen mostly. 11:59
nothingmuch can you boot debian?
if you can get dd on the ramdisk, then you can fdisk the disk
put a cd sized partition at the end, and an HFS one elsewhere 12:00
and then eject the debian CD
put in panther #1
dd if=/dev/cdrom of=/dev/hda2
castaway beige doesnt boot from anything but an apple-cd apparently
nothingmuch or actually 3 on macs
oi
you need bootx?
i could mail you an OS9 CD
castaway (so sayeth the linuc-ppc channel)
theorbtwo #debianppc, actually. 12:01
nothingmuch if i can find one
shapr I can probably scrape up a copy of System 7 around here somewhere.
nothingmuch wishes castorbway would have bought an imac or something
theorbtwo We tried.
nothingmuch system 7.5.3 can be downloaded free, IIRC
it is also bootx capable
castaway on floppy images, but we have no floppy drive ,)
theorbtwo Right, but only on floppy, which won't help me without a floppy drive.
arcady_ yeah, installing linux on the beige macs is a real pain
and booting it too 12:02
castaway arcady_: actually we want osx on it
nothingmuch arcady_: ever tried the nubus macs? yuck!
arcady_ panther won't work
theorbtwo Right, have Jaguar.
castaway it should with XFactoWhatsit
nothingmuch what does xpostdefacto do, btw?
theorbtwo Sorry, meant Jaguar when I said Panther above.
arcady_ ah
nothingmuch does it modify the binaries?
theorbtwo It seems to be mostly a set of kernel extensions.
arcady_ why do you want OS X on it? 12:03
nothingmuch or does it fudge the puter so it thinks it's a newworld?
castaway to run SEE
arcady_ yeah, basically it installs a few kexts
you used to have to do it yourself
it just does it for you nicely, I think
nothingmuch theorbtwo: another path could be trying to install darwin itself
and bootstrap OSX from an installation of that
theorbtwo That's an idea. 12:04
nothingmuch or netbsd
arcady_ darwin probably won't work
theorbtwo I was wondering also if I can put the hdd in one of my pc boxes, and copy what I need onto there.
castaway umm.. ebay: mac floppy drive new: buy now 5 euro
nothingmuch theorbtwo: sure
if it's ide, which at that era i'm willing to bet it is
castaway (assuming thats the right sort of drive)
it has IDE
nothingmuch so it's no problem 12:05
but you should have a linux with support for macos style partition tables
arcady_ oh, and OS X doesn't support internal floppies... or didn't anyway
since no mac has one
castaway eh? 12:06
theorbtwo Hmm? There's a connector for one. Even came with a cable.
castaway what do they have then?
nothingmuch and from there bootx
arcady_: but that could be used to get 7.5.3
arcady_ those macs that have one
they don't exist anymore
nothingmuch to get dd
or whatever tools
castaway cgi.ebay.de/ws/eBayISAPI.dll?ViewIt...eName=WDVW (anyway)
nothingmuch or at least run xpostdefacto
theorbtwo OpenDarwin 7.2.1 does not support the g3 beige.
nothingmuch darwin 7 == osx 10.3 12:07
arcady_ apple doesn't include drivers for apple hardware in darwin
nothingmuch darwin 6 == osx 10.2
etc
castaway 7.5.3 does tho, arcady_
nothingmuch arcady_: but rudimentary support should be possible, no?
i mean, if they don't have graphics accel or networking, it doesn't really matter, right?
arcady_ potential problem with 7.5.3 though, is that it doesn't support HFS+
nothingmuch, I don't think they have support for ADB either 12:08
nothingmuch oh my
arcady_ you could netboot it headless, and that's about it, I think
nothingmuch yuck kyuck yuck
castaway eep
arcady_ else the linux folks would have taken the drivers long ago
theorbtwo www.opendarwin.org/downloads/6.6.2/...LL.ppc.txt says that opendarwin/ppc 6.6.2 will insall and boot on all PowerPC-based Macintoshes supported by Mac OS X 10.2.6. 12:09
Which AFAIK should include this one. 12:10
nothingmuch if the osx boot cd won't work, i doubt opendarwin would either
castaway wonders if install/boot inclues "run" :)
arcady_ by the way, have you tried Open Firmware?
theorbtwo sighs.
castaway the boot cd may just be broke
theorbtwo Yes. I still get the preboot screen. 12:11
Which seems quite odd.
castaway yes, arcady_ .. people are claiming that beige's don't boot from anything but an apple-cd
theorbtwo Hold down alt-apple-o-f while booting, right? 12:12
...and that should work independent of any sort of boot media, even on oldworld macs, yes?
arcady_ yes 12:13
BUT
oldworld macs will show the OF screen on the default device
which, for oldworld, is the serial port
nothingmuch oi
arcady_ doesn't hurt to try anyway
nothingmuch is soooo glad that his booting-funny-OSs-on macs started with an ibook 12:14
and then progressed into history
and not the other way around
shapr yay! iRiver > Sony
nothingmuch ?
shapr I bought a Sony 1GB Network Walkman, it refused to work with out Windows. iRiver 1GB costs less, has more features, works fine with Linux. 12:15
arcady_ but the sony one looks cooler, or something 12:16
shapr Sony has lotsa copy protection, digital rights management, etc
arcady_ and sort of almost supports mp3!
castaway 20GB++
shapr Truly, the Sony does look cooler. Too bad you can't initialize the Sony without a copy of windows.
The iRiver worked out of the box.
kungfuftr nothingmuch: that HTML seems to output very oddly (lots of random horizontal lines) 12:17
nothingmuch kungfuftr: where?
between table rows?
theorbtwo Is the serial port the bottom one or the top one?
kungfuftr nothingmuch: yar 12:18
nothingmuch xshot?
kungfuftr *blink* wtf... i switch back to em and they're gone... bloody odd
nothingmuch theorbtwo: either, but i think the one marked with a telephone is the first one
<em>?
theorbtwo I have no case, and thus no markings.
nothingmuch for the links?
oh,
uh, i think the top might work
and if not the top then probably the bottom 12:19
castaway laughs
nothingmuch do you have a mac serial to rs232 converter?
theorbtwo I think so.
kungfuftr nothingmuch: thought about marking todo tests with opacity? (yes, it's possible on all browsers)
nothingmuch conceptually how will that work?
i don't mind, i just don't know how
but IMHO todo tests should be darker, like they are 12:20
because everyone already knows what it means
theorbtwo Eh? TODO tests are /lighter/.
nothingmuch there's a bug right now 12:21
they are yellow
but in my darcs repo they are dark green like they used to
and unexpectedly succeeded todo tests are yellow
kungfuftr todo = light red... todo unexectedly pass... light green? 12:22
nothingmuch red implies failure, IMHO
todo+fail is OK
that's expected
red catches the eye 12:23
kungfuftr well yeah, but it's still a failure overall
nothingmuch but then we'll be swamped in failures which aren't real
castaway orange! no, blue!
nothingmuch and green/red mixed together looks like poop, not green or red
kungfuftr hhhmmm... orange is good 12:24
as acme would agree
nothingmuch disagrees more
q[acme] agrees with orange
kungfuftr =0)
nothingmuch when i look at test results i would normally look for things in the following order:
how many tests failed/passed etc (bottomline)
if i see fail, scroll up/down looking for red things 12:25
i want nothing else to stand out in this sweep
and i don't want to get confused
then i can go and look for a test to untodo, or a feature to implement
kungfuftr nothingmuch: ever heard of semiotics? 12:26
castaway so why are they in there at all? :)
shapr What about red for failed test, yellow for fatal, and green for passing? Maybe unexpected successes can be blue?
nothingmuch what's so bad about the current scheme?
shapr The dark green for fatal confuses me, green means success to my unit test training. 12:27
kungfuftr nothingmuch: different colors mean different things to different cultures
semiotics++
shapr The dark green fatal color is the only one that really gets in my way.
theorbtwo Huh? Dark green doesn't mean fatal.
nothingmuch shapr: green *is* success
kungfuftr: everywhere but PRC, if i recall correctly green is YES and red is NO
theorbtwo Green is "what we expected to happen did". 12:28
castaway (just give them classes, and let whomever has problems, customize in the CSS.. ,)
shapr yeah, good idea
theorbtwo Dark green is "expected failure".
nothingmuch basically the colors are a measure of how much the implementation and tests are in synch
sync
if the implementation moves a little
creating bugs, or fixing things
theorbtwo We do, castaway, except in the right hand column.
nothingmuch then you notice things
castaway well why are there complaints then?
nothingmuch because you see red and yellow popping up 12:29
nothingmuch doesn't know
castaway (of course, this makes for nice confusion when comparing outputs :)
shapr Ok, never mind. I like what I just saw on datenzoo.de/pugs/win2k.html
nothingmuch shapr: that has a buggy Test::TAP::HTMLMatrix
kungfuftr castaway: todo fail is green, which seems like this test is ok... no need to worry about implementing it to actually pass
nothingmuch compare with nothingmuch.woobling.org/pugs_test_status
kungfuftr: the test is ok, it's not a bug
it's the lowest priority action after: fix bugs, untodo tests 12:30
theorbtwo OK, I now have an open firmware prompt on the screen of my apple.
castaway kungfuftr: then change your css, and stop bothering people :)
nothingmuch theorbtwo: yay!
castaway wow!
kungfuftr castaway: =0P
castaway how did you do that?
nothingmuch t/operators/quoting.t is an interesting example
roie_m's fixes changed things
for the better, and for the worse 12:31
(i think it was him)
the changes caused two tests to fail
and 6 tests to start becomming ok
the rest of the green stuff is like it was before
shapr nothingmuch: Ah, I don't like the dark green there. I don't know what it should mean. But I'll take castaway's suggestion.
nothingmuch find a better meaning for it
to me it's like a diff
where dark green needs work
but you better take care of red and yellow first
btw, i don't know how to represent tests with no outpput 12:32
kungfuftr black?
nothingmuch Test::Harness thinks they are errors
what meaning do you want to give to those blocks?
black? why?
kungfuftr nothingness... lack of something 12:33
nothingmuch think of the tests as a sort of system where you mark points of sanity in the implementation
testing lets you change code with confidence
the stable situation is one where you start from
everything is as expected 12:34
some things are not yet ready
you change xyz to make some dark green tests become green
kungfuftr reminds me... gotta set up a tindering system soon
nothingmuch you can see which ones started working when they turn yellow
and which ones break when they turn red
if your changes make only yellows, then you can just go to the test
s
and say "ok, this is the new stable situation, you are no longer todo"
you confirm them 12:35
todo tests that are failing should not attract attention, because they are normal
and because we have so many of them
kungfuftr ran TAP::Matrix stuff with PDF output, wasn't bad
nothingmuch html to pdf?
kungfuftr nope... PDF::API2
nothingmuch ah, instead of Petal? 12:36
kungfuftr yar
nothingmuch could you put that up on CPAN?
kungfuftr++
kungfuftr nope, fraid not
nothingmuch licensing issues?
kungfuftr commercial... =0(
nothingmuch pdf2::api, or your visualizer?
kungfuftr me own visualizer thingy
nothingmuch oh well 12:37
kungfuftr will have a look at doing it at home though
nothingmuch /me is glad it's being useful though
kungfuftr and it got ported to another framework too... so don't even have the code no more
nothingmuch you lost me, 12:38
kungfuftr we have our own harness framework... i did it in Test::TAP initially as a test to see if it was possible
nothingmuch ah
kungfuftr management like colorful PDFs 12:39
q[acme]: defenitely nutty
stevan morning all 12:49
theorbtwo: ping! 12:50
nothingmuch hola stevan
i'm home in 40 mins
i'd like to know what's new
cool?
stevan nothingmuch: yes
nothingmuch leaves now...
ciao!
stevan adios
kungfuftr: are you trying to make the test output in PDF? 12:51
stevan only half read the backlog today
kungfuftr stevan: i did do 12:55
stevan kungfuftr: even better :)
which PDF lib did you use?
kungfuftr PDF::API2
stevan yuk :P
stevan prefers pdflib-lite
kungfuftr yar, but the most complete 12:56
stevan kungfuftr: have you tried pdflib-lite? It
is pretty complete
kungfuftr stevan: nope, mainly because it's all i have available here currently
stevan :)
I have actually been working on a DBI like module for PDFs 12:57
kungfuftr stevan: got a url?
stevan yeah
this is the commercial site
www.pdflib.com/
for the more robust lib
kungfuftr k 12:58
castaway wonders where theorbtwo went :)
stevan but if you are on BSD and have access to the ports tree, try /ports/print/pdflib-perl/
I have used that several times on FreeBSD servers with no issue
kungfuftr: its a straight port of the C-api,... so its nothing fancy, but I think that in some cases the approach is simpler than PDF::API2 12:59
PDFs are very procedural creatures anyway
kungfuftr stevan: mainly looking for a way to extract text from PDF (including non-latin characters) in perl
stevan because all they really are is a sequence of PostScript commands
kungfuftr: the pdflib has pretty sophisticated PDF-parsing capabilities, but you need to pay for it 13:00
I think the license is like $400 USD
kungfuftr stevan: hhhmmm... bah, don't think boss would got for the licence
stevan yeah thats the hard one :)
jabbot pugs - 2116 - Added script which reads output of "svn 13:01
stevan PDF parsing is pretty difficult since different programs and different libs will create different PDFs
we tried to create 2 pdfs, one with PDF::API2 and one with pdflib and tried to compare/diff them 13:02
totally different files
kungfuftr stevan: we'd need the $10,000 licence it'd reckon
stevan :)
13:02 webmind_ is now known as webmind
kungfuftr though should check with activestate to see if they have anything 13:02
stevan I think you will find most PDF parsing libs cost $$ 13:03
but if you do find a free one,.. please let me know :)
kungfuftr stevan: well, if activestate have something, it'll not cost us anything 13:04
(owning opensource language and tools company)++ 13:05
shapr waves pom-poms
castaway deduces that kungfuftr works for Sophos
kungfuftr castaway: filthy lies!
*blink* um... yes
anyways, back to work with me 13:08
castaway :)
stevan later kungfuftr 13:09
howdy shapr, I aint seen you 'round these parts lately
shapr Hiya stevan, I've been busy.
stevan me too (I finally had to do some $work) 13:10
shapr I should be busy now too, but I got my first wearable digital music player so I'm being geeky.
stevan very nice 13:12
what did you get?
shapr iRiver 1GB, after taking back the win32-only drm-only Sony Network Walkman. 13:13
It's quite nice, fm radio, easy voice record/compress/playback on the player itself, and standard mp3/ogg/something else playing. 13:14
theorbtwo has been trying to mess with the mac.
stevan I have been condsidering getting one,.. but I find I am rarely without my laptop anyway :) 13:15
shapr I haven't yet tested it in-flight on my unicycle, but that's about to happen once I have some music loaded.
theorbtwo (And largely failing, but I at least have auto-boot mode off, so it now boots into open firmware.)
stevan theorbtwo: our fearless Mac-Warrior has returned
theorbtwo: just FYI, i have OS 9, 10.0, 10.1, 10.2, 10.3 Cds
all you need to do is ask and I will post images for you
shapr stevan: what sort of $work do you do? 13:20
stevan shapr: mostly web-based database reporting tools
although we do plently of PDF and Excel exporting too
so it is not always web based 13:21
shapr: and yourself?
pugscode.org «Overview Journal Logs» | r1773/winxp: 587/4256 smoke/win2k (occasional): xrl.us/fqum (158/4302, r2032) Mac OS X (62/4459, r2116) | pugs.kwiki.org 13:22
theorbtwo Hm, these may come in handy. 13:22
shapr Sort of similar stuff. Lately I make websites for EU projects.
It mostly consists of webmonkey creation of input/edit/view for whatever type of info the EU project wants. 13:23
stevan ah
stevan has to get some more coffee, bbiab 13:24
shapr We use Zope/Plone because we need control flow abstractions, i18n support, document handling, mailing lists, message boards, membership, etc etc
stevan I have been meaning to check out Zope, but I live so much in a perl world I know I will only get jealous :) 13:25
shapr Zope has good and bad points.
It's the best thing I've seen yet, but I still very much want something better. 13:26
Computing in general needs a better i18n abstraction. ResourceBundles/gettext/etc just don't cut it. 13:28
nothingmuch afternoon 13:33
theorbtwo Have you read through the i18n modules at search.cpan.org/~autrijus ?
Nothing mindblowing, but looks quite nice.
Allo, nothingmuch!
castaway didnt notice nothin' go..
theorbtwo has the machine booting to OF (without silly stuck keys stuff).
nothingmuch wee! 13:34
now what can you do? 13:35
nothingmuch neever dabbled with OF
except for 'reset all' or whatever
and 'boot'
stevan: ping 13:40
hola Corion 14:00
did you get around to sending that test result?
Corion hola nothingmuch
nothingmuch: No - when run manually, it succeeds :(
nothingmuch oi, 14:01
oh well
Jonathan_ Gotta run...looks like it's tomorrow morning for getting the smoke tests running, providing nothing gets in my way. Bye 14:02
nothingmuch ciao, Jonathan_
good luck!
what's everyone's favourite slide show tool? 14:03
Corion s5 14:04
shapr I like whatever metaperl was using.
Corion I even wrote a stupid converter from pod to S5, but there are hundreds of those out there already ;)
autrijus I use Spork+Kwid+S5
or rather, Kwid=>Spork=>S5
nothingmuch doesn't want to learn latex every time he wants to use prosper... good calls 14:05
autrijus: got some scripts for that?
autrijus nothingmuch: it's on CPAN
just install Spork and Spork::Formatter::Kwid 14:06
then
spork -new
edit config.yaml to say
formatter_class: Spork::Formatter::Kwid
that's all
if you want to generate S5, use
plugin_classes:
- Spork::S5
gugod - Spork::S5ThemeFlower 14:07
nothingmuch goody
search.cpan.org/search?query=spork&mode=all
yielded good results
can't find Spork::Formatter::Kwid 14:10
autrijus ingy: you forgot to release it? 14:11
autrijus.org/tmp/Spork-Formatter-Kw...def.tar.gz
nothingmuch autrijus++ 14:12
it's in SVN either way
google finds only that
Corion Hmmm. At least my DSL still works. They were supposed to upgrade it today, but that didn't happen. 14:14
Oooo. nmake optimized now makes ghc.exe (6.4) panic. "ghc.EXE: panic! (the `impossible' happened, GHC version 6.4): mkWWcpr: not a product PugsziAST.Param{tc r1f9}" 14:16
Should we report these somewhere?
nothingmuch i had that exact error too 14:17
it went away after a while
Corion nothingmuch: I still have it - since yesterda
nothingmuch i had that 3 days ago, i think
Corion in r2116
nothingmuch and it went away starting yesterday morning, when i started trying again
Corion nothingmuch: BTW, is there a way to get at your fixed version of Test::TAP::HTMLMatrix without installing darcs ? Like, a new CPAN release? :) 14:18
nothingmuch wget -rp 14:19
i could release it, but i think it's a bit silly 14:21
i'd like to be sure there are no other issues yet, that are as stupid
i thought of waiting 2 days or so
autrijus Corion: weird. does "optimized" after "clean" produce that error? 14:22
nothingmuch autrijus: for me it did
Corion autrijus: Aaah - maybe I should make clean before that ... Will retry after the smoke run went through.
autrijus nothingmuch: it did produce the error? weird. 14:24
nothingmuch yep 14:25
nevermind, it went away
Corion nothingmuch: Ah, but to fix all the stupid bugs, that's what version 1.0 is for. We are at 0.01, so you're still free to do incremental fixes :) 14:28
nothingmuch okay
nothingmuch releases
Corion nothingmuch++ # quick releases 14:29
nothingmuch it's been two days, so I think it's OK 14:30
Corion My releases often have a quick (three hour or so) fixup release after them ;) 14:31
nothingmuch: Also, I just want to see if the cpansmoke test produces another failure now for the module (and this time I won't upgrade other modules in the meantime ! I promise!) 14:33
nothingmuch okies =)
PerlJam does a "svn up" in his pugs dir after several days away from pugs ... let's see what happens ;) 14:35
PerlJam fully expects the world to have shifted.
Corion And there is a new mail worm mailing *.de addresses. Unfortunately, I have such a mail address :(
Oh. And Pugs versioning is broken again (stopped at r2097) 14:36
r2116 smoke - datenzoo.de/pugs/win2k.html 14:37
nothingmuch Corion: 2116 should be at the end of the 'extra' yaml 14:38
Corion nothingmuch: It's there, but it's not up in the version message. And Pugs reports itself as being r2097, while it's actually r2116
I wonder how I could write a good test for that ... It would need to ask svn ... Not really feasible... 14:40
Ah. I see now. Whoever "fixed" it, made it circumvent my fix. Oh well. 14:41
Or maybe it's just me being stupid. Much more likely. 14:42
nothingmuch hola lightstep 14:48
lightstep good afternoon 14:49
autrijus yo lightstep-san 14:50
:r no longer segfaults!
lightstep happy bday 14:51
what was the bug?
autrijus thanks
lightstep: seems getEnvironment is not reentrant for some reason
all works when I changed it to just getEnv
lightstep: are you interested in modelling the Larry List?
Corion Larry List ? 14:52
autrijus defined in www.nntp.perl.org/group/perl.perl6.compiler/622
Corion reads
autrijus our Array currently are simple one-part lists
Larry Lists (LL in short) will allow pushing to infinite lists. 14:53
and even popping of.
stevan pop-ing infinity :)
autrijus updating our IArray to LLs will rock.
nothing else needs changing -- just the IArray data structure itself. 14:54
lightstep push and pop work at the end of the list?
autrijus aye
theorbtwo push and pop from the end, unshift and shift from the beginning.
lightstep and infinite list are different from finite lists? or is it element-lists vs. generator-lists?
stevan shift and unshift dont work on Inf arrays yet either
autrijus: did you see the Changelog? I tried to add all of yesterdays commits to it 14:55
theorbtwo Apparently, all lists have an element section, followed by a generator section.
Of course, either of them can be empty.
lightstep eek. 0... has a last element
autrijus lightstep: it seems like LLs has an FixedArray part and a Gen part
theorbtwo Yep, and that last element is Inf. 14:56
We don't do full Cantor transfinites, just boring old infinity.
autrijus stevan: right. thanks for the monumental task :)
Corion The LL sounds sensible to me, at least, it handles push/pop on infinite lists in a sensible way
stevan autrijus: it was easy actually,.. most of the commmits were tests :)
autrijus it is sensible. it's just a matter of implementation :)
stevan BTW - there are commented out tests for these Inf lists in t/builtins/array/push|pop|shift|unshift.t 14:57
autrijus yeah, I've seen'em
stevan not very extensive though,.. since @larry hadnt decided on it yet 14:58
autrijus *nod*
stevan is off to the dirt world to do some errands
autrijus have fun :)
stevan I can try to write better ones when I come back
Corion nothingmuch: BTW, I put some fixes for Win32 paths into util/testgraph.pl - do you want them for the main class? Currently it doesn't translate from local filenames to http filenames ... 15:00
(otherwise, I'll keep them local instead of committing them. They consist of the kludgey $link =~ tr[\\][/]; # for My::File and My::Subtest 15:01
nothingmuch Corion: yes 15:02
theorbtwo D'oh, I should have thought of that. 15:03
I suspect that tests_catalog has a similar bug.
Corion nothingmuch: I suggest (as a general solution) $link =~ tr[\\][/] if $^O =~ /MSWin32/
nothingmuch (reverse?) 15:04
Corion theorbtwo: I think alias wrote a module for that, but it always trips me up as well :)
nothingmuch why not just use File::Spec?
i'll put the fix where it ought to be
Corion nothingmuch: Yes - you could use File::Spec - that's the saner approach ;) 15:05
nothingmuch what is the diff?
i
am not sure where the problem is yet
Corion nothingmuch: ? Ah - when I run the unmodified testgraph.pl, I get links like t\builtins\foo.t#line25 15:06
nothingmuch oh
Corion ... which is not really good if I want to use it via HTTP instead of the file system
nothingmuch i get it
it's too much File::Spec
;-)
Corion nothingmuch: You need to join "/". File::Spec is good for local files but not for HTTP :)
nothingmuch thought URI::file would take care of that
it's taking a $path (opaque, comes from File::Find, $subtest->test_file, which is parsed from <pos:...> 15:07
File::Spec is never in the way actually 15:08
then I do return URI::file->new($path);
I see: $u4 = URI::file->new("c:\\windows\\", "win32"); 15:09
Corion nothingmuch: Hmmm. Won't that create an absolute path?
nothingmuch hmm, from the docs it sounds like it should work
lightstep what is the arrangement of src/ ?
(ie. which modules are in the Pugs namespace, and which are not?)
nothingmuch • [syeeda:~] nothingmuch % perl -MURI::file -e 'warn URI::file->new("foo\\bar.txt", "win32")' 15:10
foo/bar.txt at -e line 1.
nothingmuch is stumped
Corion lightstep: All internal stuff went below Pugs/, but non-#perl6 libraries stay up there
theorbtwo ...is, I think, the theory.
Corion I guess that URI::file is the thing then. That looks good.
nothingmuch • [syeeda:~] nothingmuch % perl -MURI::file -e 'warn URI::file->new("c:\\foo\\bar.txt", "win32")'
file:///c:/foo/bar.txt at -e line 1.
makes sense
Corion For example the unicode is not internal, as is the regex lib 15:11
nothingmuch i'm already using it, though, that's what's odd about the whole mess
theorbtwo The reality is that I moved everything except the things that autrijus told me not to move.
Corion theorbtwo: :))
nothingmuch Corion: could you test how URI::file behaves without "win32" sent in, by default? 15:12
lightstep and what is Pugs.Types.Array ?
nothingmuch lightstep: Itypes implementations
lightstep oh
theorbtwo It's the implementation of the Array IType.
lightstep but all it contains is a class definition 15:13
theorbtwo Yeah.
Corion C:\>perl -MURI::file -e "warn URI::file->new('foo\\bar.txt')"
lightstep oh, i get it
Corion # foo/bar.txt at -e line 1.
nothingmuch oh crap 15:14
Corion Ooo. After a "make clean", "make optimized" compiles.
nothingmuch I forgot to do it for the links, it's only done for the CSS 15:15
Corion++; # really a bug
autrijus lightstep: the actualy IArray definition is in AST.hs 15:17
line 1026 15:18
obra 'morning
autrijus hi obra.
obra how goes?
autrijus brain hurts.
nothingmuch Corion: want to try now?
autrijus 4th full day of banging-against-the-bamboo-wall
Corion nothingmuch: Try now what?
autrijus in a murderous mood :)
shapr autrijus: you need a machete!
Ok, never mind
Corion Or many many pandas 15:19
nothingmuch Devel::ebug is my new machete
Test::TAP::HTMLMatrix
shapr autrijus + murderous mood + machete = bad
Corion nothingmuch: From where?
nothingmuch wget -r nothingmuch.woobling.org/Test-TAP-HTMLMatrix
autrijus frags shapr mercilessly
nothingmuch ... or you could also get a darcs binary
shapr respawns in another part of the level. 15:20
Corion nothingmuch: Ah, I've got 2 vcs installed already :)
nothingmuch so?
autrijus obra: struts.apache.org/flow/remote-example.html
nothingmuch ;-)
theorbtwo shapr++
autrijus obra: I'm converging there.
jabbot pugs - 2117 - add a couple of phrases 15:21
nothingmuch Corion: darcs is very unintrusive, i think you'd like it if you don't want more bloat 15:22
obra huh.
Which parts of that? Are continuations now a part of this?
shapr has at least four VCSs installed
nothingmuch woo! you can set $^O 15:23
nothingmuch has cvs, svn, svk (unused), darcs
i should get around to learning svk finally
but darcs is so fun
shapr Yeah, darcs is addictively easy.
autrijus obra: uh, more like the ajax help part
obra ah :) 15:25
Corion nothingmuch: Tested and installed OK here.
obra pk. makes more sense
nothingmuch how about the links?
(i did not yet check in tests for URI 15:26
Corion nothingmuch: I'm generating the graph at the moment :)
nothingmuch with or without your tr///?
Corion Ah, and btw, there is a potential division by zero error in the ::Model
nothing currently testing without the model
nothingmuch i know, fixed 15:27
0.03 coming together
Corion Hmm - no, still broken # datenzoo.de/pugs/win2k.html . But maybe that's a stale file. 15:30
No, the local copy is the same. 15:31
nothingmuch okay, i wrote the test, now i'm working with that 15:32
well, i've violated the darcs mantra 15:40
unupull the URI::file in links patch
and then darcs pull
i've amend recorded a published patch because I shouldn't have published it
ninereasons hi 15:52
shapr niner niner, we read ya good buddy. 15:53
What's your 10-20?
theorbtwo The svn --version should be working again. 15:54
ninereasons puddletown usa, goodbuddy; backatcha
jabbot pugs - 2118 - Make --version's r\d+ work right again ( 16:01
pugs - 2119 - * serializable subcontinuation monad imp 16:11
pugs - 2120 - * err, typo.
shapr That's nearly quotable.
"serializable subcontinuation monad implementation" "whoops, just a typo, really"
integral heh 16:12
autrijus heh.
qmole lol 16:14
nothingmuch hah
16:14 nothingmuch is now known as someone
someone tee hee 16:14
16:14 someone is now known as nothingmuch
ingy morning 16:16
shapr Fnord 16:17
nothingmuch shapr++
hola ingy 16:18
how many pounds in the last 24 hrs?
ingy hi nothingmuch
um 1-2
hopefully I'll be in maintenance mode this time next week
getting tired of not eating 16:19
nothingmuch i wonder why that is
ingy although I must say that not eating is far less ditracting
nothingmuch Devel::Cover is soooo sloooow 16:20
pjcj I optimise via Moore's law! 16:21
nothingmuch pjcj: explain
pjcj in 18 months it'll seem twice as fast
theorbtwo ...assuming the machines it's being run on are new.
lightstep chuck moore sells 60-GIPS chips for $1 US
nothingmuch heh 16:22
nothingmuch 's computer is 5 years old
lightstep: ref?
lightstep forthfreak.net/misc/25x.html 16:23
ninereasons thanks for fixing the ":r" segfault, autrijus. 16:24
autrijus ninereasons: no prob :) 16:25
lightstep those chips are a little silly, though: 0.5 W of power, and no muptiplication opcode 16:26
nothingmuch this is not really going to be the next big thing, is it? 16:29
shapr Isn't it similar to the Cell's tradeoff?
lightstep he's making these designs for the last decade or so 16:30
ingy perlbot: nopaste? 16:42
perlbot Paste your code here and #<channel> will be able to view it: sial.org/pbot/<channel>
nothingmuch welll, does anybody else have Test::TAP::Foo bugs? 16:49
nothingmuch considers finally releasing 16:50
hammerred out div by 0 cases in ratio
now it DWIMS
and have actual_cases method
and $subtest->planned;
jabbot pugs - 2121 - * `:l filename` no longer needs double q 16:51
ninereasons neato, typing savings 16:52
nothingmuch yay! drinking age in toronto is 199
i was afraid it was 21
s/9$//;
Juerd 199; grin. 16:54
theorbtwo (In the revision 2121, if Pugs is still alive...) 16:56
autrijus hm?
theorbtwo Old song.
Not a good one either.
frodisman.com/2525.html
Oh, also it doesn't have a 2121 verse, just 2525. 16:57
Limbic_Region chip about? 17:03
seen chip?
jabbot Limbic_Region: chip was seen on Fri Apr 8 19:06:05 2005 GMT
nothingmuch beh, releng never ever ends 17:09
nothingmuch releases 17:17
jabbot pugs - 2122 - * assigning a List into a Scalar vivifie 17:21
nothingmuch tarballs on nothingmuch.woobling.org/Test-TAP-Model and nothingmuch.woobling.org/Test-TAP-HTMLMatrix 17:22
wow, i love darcs SOOOOO much 17:35
example.pl snuck into Test::TAP::Model
you can massage it out of history
nothingmuch feels like Richard Nixon
ninereasons revisionist revision control 17:36
nothingmuch uhuh 17:37
and all with branching =)
no 'p4 obliterate'
autrijus darcs is supernice when you only have a couple developers :) 17:38
<- developed Pugs up to 6.0.2 with darcs
nothingmuch autrijus: what happens when more enter the picture? 17:39
nothingmuch doesn't have project which are that busy
i rather like SVN now that I know it more
it's copying metaphore is very elegant
autrijus nothingmuch: the slowness kicks in. 17:40
nothingmuch but centralization bites me
i guess svk is prolly best for larger groups
while darcs is fun for your own kind of thing
autrijus: ah....
they're working on it though
lots of improvements, lately
and David is taking a sabbatical, to work on the conflict resolution stuff
wow... /me sold $12 worth of photos at shutterstock 17:42
autrijus cool. but svk gained the darcs dependency algorithm :)
nothingmuch is impressed with himself
but it can't undo
autrijus nothingmuch: actually we figured out how to do that too :)
nothingmuch that's actually my #1 reason for liking darcs now
really?
autrijus `svk obliterate`
only works on fsfs though.
nothingmuch in that case, make an interface as nice as darcs, and =)
fsfs?
autrijus the new svn 1.1 backend 17:43
nothingmuch is clkao coming to the hackathon?
autrijus it works by zeroing out the delta.
don't think so... he's going to euroscon and yapceu last I heard
nothingmuch bummer
nothingmuch promised he'd help
but then time suddenly diminished
maybe you'd like to hack a bit of svk at the hackathon 17:44
well, lets not make plans
we'll see what we want to do when we get there
clkao bwahaha 17:53
nothingmuch bwahaha?
crysflame hi, clkao 17:58
nothingmuch URI::Title <-- interesting
Corion Hmm - the build is currently broken (r2122) - src/Pugs/Help.hs: file name does not match module name `Main' 18:02
autrijus huh? 18:03
worksforme
can other people confirm?
Corion "nmake clean && perl Makefile.PL && nmake" gives that problem. 18:05
ninereasons no problem seen here
Corion Weird ... I don't see any problem either, but ghc doesn't want it ... 18:06
src/Main.hs is there, it finds it, but then complains about src/Pugs/Help.hs ...
autrijus Corion: "svn di" for local differences 18:07
and do a "make clean"?
failling that, try fresh checkout?
Corion Ah. My usual problem. No newline at end of file - which doesn't get caught by a fresh checkout, but which makes ghc cry. 18:08
autrijus: Checkouts don't catch that...
Yep - now builds properly.
autrijus you really should switch to a newline-happy editor :) 18:09
theorbtwo wonders if there's a way to get openfoundry to reject files with missing newlines at the end.
Corion Why is ghc broken that way? (and why is my editor broken that way too)
I'm also looking for a fast/nice/tabbing editor that knows UTF8. I don't want the hog that is Eclipse, don't want vi* or emacs*
autrijus well.. that doesn't leave much choices 18:10
Komodo? ;)
theorbtwo I hear good things about kwrite.
Corion autrijus: Hmmm - never looked at it.
I should continue working on Zaphod.
autrijus there is Haste.
Corion An editor that is scriptable in JS has many advantages.
Corion goes looking for Haste 18:11
autrijus # haste.dyndns.org:8080/screenshots.php
it doesn't help your perl files :)
Corion Ugh. Also uses Java. Big-slow-bloat.
autrijus is it?
Corion Oh. Now the error mutated.
autrijus I thought it's totally in GHC.
Corion src/RRegex.o(.text+0xd8):ghc2480.hc: undefined reference to `RRegexziSyntax_zdwpolyzuzez126Uz126U4_closure'
... and some more. 18:12
autrijus Corion: a clean+make will almost surely fix that
Corion while Linking ...
autrijus: Did that already. Maybe I should do a fresh checkout...
Or maybe you did forget to check in a relevant file?
autrijus maybe, yeah. sounds weird
I didn't.
ninereasons Corion, I like Helios Textpad, which has some support for utf8 encoding 18:13
Corion autrijus: Oh well - I'll wipe src/ and do a fresh checkout then.
Yep. Wiping src/ and checking it out again worked. I wonder what caused that, because svn status claimed I was clean... 18:16
ninereasons finds himself typing " :t subname " in pugs. 18:18
mild case of Haskell infection 18:19
mostly, I seem to be immune
Corion I need a faster machine. Building Pugs (optimized) already takes as long or even longer than building Perl ...
theorbtwo We could make that work, possibly, though the types of subs I don't think are currently hugely interesting. 18:20
ninereasons it seems to be largely a function of how much memory you have, Corion
shapr hasn't recovered yet 18:21
clkao i/win24
Corion ninereasons: I have 1GB of real RAM, and GHC only seems to be using 200MB or so. Maybe I should look into giving ghc more RAM ...
shapr four years and counting
Corion shapr: Complaints about ghc? :) 18:22
shapr Only that it has more flags than the UN.
ninereasons hahah
autrijus Corion: splice.t all passes.
Corion autrijus: Oh whee! That means I need to uncomment some of the harder tests there! Yay! :)))
autrijus Corion: actually I uncommented them :) 18:23
see r2123.
Corion autrijus++
autrijus now you can convert the rest 5
Corion autrijus: Sure will do :)) And I really have to look at your changes to learn what you did :)
clkao autrijus: damnit. let's do svk hackathon in yapceu
autrijus clkao: I won't be @yapceu
Corion clkao: Could be interesting ;)
autrijus euroscon, yes 18:24
unless I manage to find another sponsor
Corion (but I don't know/have svk)
clkao where will you be around that time?
autrijus Corion: just download the win32 installer and start s/svn/svk/g
Corion: everything will be the same except magically faster
clkao: dunno, it's mostly funding issue
Corion autrijus: And I'll have a local repository where I can commit to. Hmmm ... 18:25
clkao autrijus: er, to #svn-dev and talk to brane. he is resolving the bad apr_uint32_t issues
autrijus: funding, i thought you earned enough money last year :p 18:26
autrijus clkao: ...and I'm going to invest all of them to this box thing :p
clkao oh.
magical box
shapr Pandora's box?
autrijus rubic's box
shapr What's rubic's box? Like Rubik's Cube? 18:27
autrijus right. misspelling
theorbtwo wants to see a Rubik's Cube case mod now. 18:28
Preferably motorized.
shapr Would be an amusing way to dump heat.
Watch the bits spin.
theorbtwo I'm not sure it'd really dump heat. 18:29
Friction. 18:30
Corion autrijus: I get No compatible subroutine found: &splice\n App "&splice" [] [Var "@a",App "&prefix:+" [Var "@a"] [],Val (VInt 0),Val (VInt 11),Val (VInt 12)] # now 18:31
... which is the first splice call I want to test.
autrijus Corion: really.
Corion: is it from splice.t?
Corion Yeah - line 53 is offending I think. 18:32
If I comment out lines 52-56, some more tests pass and then it dies after test 7
autrijus this is maximally weird. 18:33
I'll check out a fresh copy.
Corion The first test is splice(@a,+@a,0,11,12); - maybe it doesn't like that inline list at the end ? I'll check my local status with svn diff - maybe the newlines got broken again and I didn't notice.
autrijus no, it all works for me 18:34
nothingmuch nothingmuch.woobling.org/YAPC/NA/20...utline.txt
autrijus that test, I mean
Corion No, all clean according to svn diff.
and r2123 is the most current version.
nothingmuch cleaned up a bit, now that I can read better
Corion And just before I started looking closer, I saw splice.t work
I think I need to wipe the whole directory and check out completely fresh :( 18:35
autrijus it worksforme from a fresh copy :(
Corion autrijus: I think it's something broken in my local copy and I need to nuke it from space. It's the only way to be sure.
theorbtwo (Jurassic Park solution)++ 18:36
Corion Oh. "make" works. "make optimize" doesn't. 18:37
This is really weird. If we accept ghc as bugfree, that is.
Corion builds pugs as optimized again.
theorbtwo GHC is not bugfree.
Corion theorbtwo: I know.
autrijus we are exercising all sorts of cutting edges. 18:38
theorbtwo You might be interested in adding a -dcore-lint, to tell you where ghc thinks it's buggy.
Corion ... but in an ideal world, the optimized version would be functionally identical to the unoptimized version ;)
autrijus: The GHC devs should worship you, as you supply them with a possibly infinite list of bugreports :)
"potentially"
theorbtwo tries to remember what he was working on.
Corion theorbtwo: resurrecting your filesystem? 18:39
theorbtwo No, did that several days ago.
Mostly, at least.
Corion No. It was merely a bad build, however I did that. Now it just plain works, even with the optimized build.
theorbtwo /var and /home were on a different partition.
Corion doubts his sanity.
(a bit more than every other day) 18:40
autrijus theorbtwo: Evalable? :D
theorbtwo Ah, no, but close.
Writing tests for the existing eval_haskell.
castaway eval_lisp !
theorbtwo Including a bunch of TODO tests.
autrijus delete.t passes. yay.
jabbot pugs - 2123 - * splice.t all passes. 18:41
pugs - 2124 - * delete.t passes.
autrijus praises the high test coverage
machack666 pugs> my $j = 1|2 18:42
undef
$j = 5
*** Error:: Can't modify constant item
I get that the assignment is being autothreaded, but does this mean that once a variable is declared a junction, it cannot be reassigned? 18:43
i.e., is there some way of saying $j, the variable, rather than $j, the junction of the two values held in the variable 18:44
nothingmuch maybe if you take a ref to a junction, then it become a non-autothreading junction?
autrijus I don't think assignment is autothreaded.
it's just a simple bug :) 18:45
machack666 I think we're getting the equivalent of 1 = 5 | 2 = 5
Corion autrijus: Do we know how big our test coverage actually is? In a Devel::Cover sense, I mean?
nothingmuch meant in the general case
autrijus machack666: add a test for that?
nothingmuch Corion: Paul Johnson thought that maybe we could pull something off with hat
machack666 sure ... is that an eval ... ; 1 case?
autrijus machack666: eval_ok, yeah
Corion nothingmuch: Nice - so we can zero in on the white spots of the map :) 18:46
pjcj nothingmuch: no - that looks unlikely now
nothingmuch oh, you are paul johnson
pjcj ;-)
theorbtwo HAT doesn't work for a lot of the more odd features of GHC.
nothingmuch pjcj++; # Devel::Cover has been very useful for me lately
pjcj thanks - I'm glad it has 18:47
right, pugs uses libraries hat can't deal with
nothingmuch maybe we can be a bit more brutish, and just note every successful pattern match and the function it took? 18:48
the only thing left after that is branches
i think
theorbtwo And sadly it fails very uncleanly.
nothingmuch how hackable is GHC? 18:49
shapr Somewhat hackable.
autrijus GHC is written in GHC
shapr On the good side, the primary GHC developer shows up on #haskell most days, so once you get into it, you can ask him deep questions.
machack666 ok, added a test in t/junctions/misc_junctions.t
nothingmuch autrijus: that does not compute
autrijus machack666: danke 18:50
machack666 de nada
autrijus shapr: oh, wait, who is that?
nothingmuch shapr: I hope that was a hypothetical 'you'
Corion autrijus: I get two fails now in splice.t, with weird tests that I don't really understand. But I verified them in Perl5. (r2127)
jabbot pugs - 2125 - * exists.t passes. 18:51
pugs - 2126 - Test for reassignment of a junction vari
pugs - 2127 - Converted more Perl tests
machack666 off to find more corner cases...
nothingmuch Corion: any results from tapfoo yet? 18:56
Corion nothingmuch: tapfoo?
nothingmuch Test::TAP::*
Corion nothingmuch: Just installing the new versions from your tarballs 18:57
Both tested (and installed) clean
nothingmuch how about the URI mess? 18:58
machack666 ok: 18:59
pugs> my $j = 1|2 19:00
pugs> my $k= \$j
pugs> $k.ref
({ref:<Scalar::Const>})
nothingmuch wow 19:01
how do you find all this stuff?
machack666 looks like it's taking a ref to the contents, and not to the junction itself
i like to break things
:D
nothingmuch i meant how do you figure out what to try and break?
machack666 instinct, i guess 19:02
nothingmuch machack666++
machack666 my first karma point !!
:)
nothingmuch machack666++
machack666++; # deserves more
masak machack666++ 19:03
machack666 my second karma point !!
nothingmuch perlbot karma machack666
perlbot Karma for machack666: 4
masak :)
nothingmuch =)
machack666 my "third" | "fourth" karma point !!
masak lol
machack666 so what is the type of a junction ref supposed to return, so I can write a test for it?
theorbtwo perlbot, highest karma 19:04
perlbot The top 5 karma entries: C: 93, autrijus: 83, nothingmuch: 59, stevan: 49, theorbtwo: 31
machack666 C++ :P_
shapr C--
theorbtwo C--
machack666 must come up in conversatin
shapr C-- is at www.cminusminus.org/
machack666 considers changing his nick to machack666++ 19:05
;)
nothingmuch lets talk about C-- in the context of how GHC uses C-- to create C code. Does GHC use C--?
C-- is much better than the cee that shall not be autoincremented
doesn't see, to work
shapr laughs
That which shall not be autoincremented. 19:06
nothingmuch I think C-- is really cool, only I don't really know what C-- is. Anyone care to explain C--?
shapr perlbot: highest karma
perlbot The top 5 karma entries: C: 84, autrijus: 83, nothingmuch: 59, stevan: 49, theorbtwo: 31
shapr wow, that was fast.
machack666 perlbot: lowest karma 19:07
perlbot The bottom 5 karma entries: <: -417, -: -89, php: -70, <!: -36, ": -15
nothingmuch where's rindolf?
q[acme] heh <!
shapr summons schlomi
machack666 perlbot: middlingest karma
nothingmuch middest karma? 19:08
theorbtwo C-- C--
shapr midden karma?
C-- C-- !
machack666 C----!
C-=100?
machack666 contemplates breaking the perlbot... 19:09
edge cases, dammit!
Corion_ nothingmuch: datenzoo.de/pugs/win2k.html
machack666 ;)
mj C--
shapr perlbot: highest karma
perlbot The top 5 karma entries: autrijus: 83, C: 78, nothingmuch: 59, stevan: 49, theorbtwo: 31
shapr ahh, much better.
nothingmuch Corion: with your hacks? 19:10
Corion_ nothingmuch: No, without any hacks.
nothingmuch yay!
castaway orange!
nothingmuch goes to play with sister
autrijus waves goodnight
nothingmuch orange what?
ciao, autrijus!
jabbot pugs - 2128 - eval_is '...' won't find the lexical '$j 19:11
Corion_ Hmmm. Should \xa0 break in a whitespace delimited list? I think not.
castaway colours
Corion_ goes to investigate
castaway later nothingmuch
theorbtwo I think not as well.
G'night, autrijus.
nothingmuch castaway: which colours? in the percentages?
theorbtwo Er, G'night, nothingmuch.
shapr good night autrijus!
castaway yup 19:12
nothingmuch will come back =)
castaway: aren't they pretty? i like around 65%ish the best
shapr throws lambdas
castaway pretty++
shapr C--
theorbtwo ducks.
castaway catches and tosses back 19:13
theorbtwo Careful, those lambdas have sharp points!
Allo, Corion!
Are my flixes for the version thingy working for you?
Corion__ Hmm. The connection seems a bit spotty :( 19:14
theorbtwo Oh, the DSL-ISP switch finally went through?
Corion__ theorbtwo: Partially. I'm now on a 3mbit line. But I still don't have the 1und1 account data %-) 19:16
I'm using my T-Online account now, which is only good for a 1mbit line, supposedly. But I've also read that others are using their 1mbit account on 3mbit lines.
shapr Does it cost three times as much?
castaway sneaky
Corion__ shapr: With T-Online it costs three times as much (usually), 70 EUR as opposed to 25 EUR. But with the switch I will pay 7 EUR. 19:17
Oh. I got a mail from their customer support. Let's see.
castaway is that including the line?
Corion__ And some more ... mailworms :( 19:18
I need to rewire my mailscanner to delete them on the pop account so they don't clog my mailarchive :(
Grrr. Their customer "service" reply is merely a redirect to the phone support. Which is closed for today. Great. 19:19
(but I expected bad customer service)
shapr I recently got k-lined by my ISP because they had network problems so I flooded the MAC address cache on the switch to explore the local net. They laughed when I told them I was actually trying to fix the network rather than DoS it.
Corion__ shapr: Take over their Cisco routers and reconfigure the network yourself. 19:20
shapr I'm tempted.
19:20 Corion__ is now known as Corion_
shapr But I have enough stuff to do, and they're limping along by themselves for the most part. 19:20
19:21 Corion_ is now known as Corion
Corion Hmm. I should find some big file to download, to see if the connection is capped or not ;) 19:22
shapr download me! 19:23
Corion shapr: url?
gaal reads the poem for Config.hs and giggles
shapr um
Sadly, I'm not network enabled to that level yet.
gaal s/poem/riddle/ really. 19:24
theorbtwo Corion: Download yourself a new kernel source tarball. 19:26
shapr Oh, good idea. You could darcs get the linux kernel repo
Corion theorbtwo: Yeah, that's an idea.
shapr: I don't have darcs :)
shapr You could download darcs first ;-) 19:27
Oh, speaking of which I should migrate the DarcsWiki to moin 1.3 and upload it to darcs.net
Now that I finally got around to asking for a login for darcs.net. 19:28
jabbot pugs - 2129 - unTODOed unexpected succeedings 19:31
gaal perl5 $Config{sig_name} is ugly imo. think we can change it from a string to something else? 19:34
Corion Ah hah. Seems like Haskell words() regards \xA0 as whitespace. 19:37
... which is bad/unexpected.
boogie introducing the first Pugs web application 19:40
boogie says it's very slow, so I'll remove from my server, but you can check it now. :)
perl6.wish.hu/cgi-bin/index.cgi
Corion boogie: Nice :)) 19:44
theorbtwo Cool. 19:45
Hm, there's an awful lot of "some nondescript dude" tiles. 19:47
castaway camels++ 19:48
boogie thanks. :) 19:49
stevan boogie++
boogie: you should put this into CGI/examples/ 19:50
theorbtwo boogie++
theorbtwo wonders what perl6.wish.hu/pics/card10.jpg is.
shapr It's a camel.
jabbot pugs - 2130 - Fixed whitespace parsing. \xA0 is nonbre 19:51
boogie stevan: the code has a lot of ugly hacks, i don't think it's a good idea. :)
shapr theorbtwo: I'm guessing it's part of the Perl Tarot. 19:52
stevan boogie: so does CGI.pm :) 19:53
theorbtwo didn't know there was a perl tarot. 19:54
shapr didn't either.
theorbtwo Of course, I also don't recognize all the faces in this game.
Corion Bah. Looks like the latest "fix" I committed did break more than it fixed (as usual). 19:55
shapr That's how we all learn.
Corion shapr: True, but I should learn to run "make test" _before_ I commit :) 19:56
shapr ah, good point.
theorbtwo doesn't recognize guy-with-brown-stripe-in-the-background, guy-with-pale-blue-background, or guy-with-grey-background.
boogie who wonders: 3 - larry wall, 4 - leo tötsch, 5 - me, 6 - autrijus tang, 8 - damian conway, 9 - randal schwartz 19:57
ninereasons fun stuff, boogie. 19:58
boogie stevan: It's really ugly. But I'll think about it. ;)
theorbtwo Are you sure 6 is Autrijus?
shapr card7 maybe? 19:59
boogie theorbtwo: I'm not sure just me and Leo (I met him personally), the others from Google Images
the correct list: 3 - larry wall, 4 - leo tötsch, 5 - me, 7 - autrijus tang, 8 - damian conway, 9 - randal schwartz 20:00
stevan two matches so far ;)
boogie 6 - dan sugalski
stevan boogie: I think the code (regardless of the number of ugly hacks) is useful for people to see 20:01
we all promise not to judge you by it :)
theorbtwo Who is 6, then?
Ah, thank you, boogie. 20:02
stevan I got Larry, Randall and a Camel :) 20:03
doh,.. the server went away ??
ninereasons shut down now, for me
well, boogie said that its' being removed 20:04
... and now it's back 20:05
boogie oops, i've gzipped the whole directory. :)
it's seems that the server is OK now, I think I will remove it later 20:06
the load is about 6, what's not normal, but working 20:07
lightstep what is PAR? 20:10
Corion PAR is the Perl answer to JAR - a self-contained Perl package 20:11
(can be converted into a self-contained executable9
lightstep autrijus wrote about #par in his journal, but it seems rather empty
Corion There is a talk by autrijus on PAR ( he held it on the German Perl Workshop) 20:13
I think CPAN also has lots on PAR, resp. the documentation of PAR.
Corion goes looking for the talk.
wagner.elixus.org/~autrijus/par-tut...start.html 20:14
testgraph++ # makes comparing against the current test run very convenient 20:15
jabbot pugs - 2131 - Working fix for nonbreaking whitespace. 20:21
ninereasons www.letterjames.de is nifty
lightstep the we-want-cookies-and-you-don't page isn't available in english 20:23
Corion r2131 - datenzoo.de/pugs/win2k.html 20:36
pugscode.org «Overview Journal Logs» | r1773/winxp: 587/4256 smoke/win2k (occasional): xrl.us/fqum (91/4373, r2131) Mac OS X (62/4459, r2116) | pugs.kwiki.org 20:37
mj my @index = ( 20:43
hash ( a => 1, b => 2, ),
{ c => 3, d => 4, },
);
say join " ", @index.ref, @index[0].ref, @index[1].ref; # Array Hash Pair
theorb Hm, well, 2 out of 3 ain't bad. 20:44
stevan :) 20:46
theorb Hm, which is better for eval_haskell if $*PUGS_HAS_HSPLUGINS is 0: Don't plan the rest of the tests, or skip them? 20:49
Corion theorb: I think skipping is better, as skipping produces some output that will allow you to diagnose the problem. 20:51
Perl also skips tests when features are disabled. 20:52
theorb Hm, a couple problems with skip, though... 20:54
1) My test code has to keep checking with every test if $*PUGS_HAS_HSPLUGINS.
2) They constantly show as skipped on the testgraph. 20:55
boogie Memory game's source code has been added to the SVN repository...
Corion theorb: No. skip 4, "$feature not implemented"; exit. And yes, they show on the test graph, which is my intention (for diagnosing)
ninereasons yay, boogie++
theorb Mmm, OK. 20:56
Corion ... because if you think you have hs-plugins, and you see the skips, you should start to wonder
shapr skips and hops
Corion theorb: Later we should get SKIP: { ...; skip; ... }; # like we have in Perl5
theorb Hm, should we have a skip_rest that skips all of the tests that the plan says we should have left? 20:57
Corion theorb: That'd be nice, yes.
jabbot pugs - 2132 - The first web application... ;) 21:01
stevan boogie++
theorb Wow. Test.pm has a global variable named "$loop", uncommented. 21:03
stevan we should make all those $globals into $GLOBALS 21:05
Corion g'night 21:13
theorb G'night, ... oh, already missed him. 21:14
jabbot pugs - 2133 - Test::skip_rest (thus far undocumented), 22:11
mj g'night 22:12
ninereasons night, mj 22:15
jabbot pugs - 2134 - * restore IMC to a runnable state 22:21
pugs - 2135 - * restore Eclipse sanity. 22:31