Run Perl6 Now! & pugscode.org ('Overview', 'Journal') & pugs.kwiki.org & channel log: xrl.us/e98m & < autrijus> damn, my journal keeps getting obsoleted by #perl6.
Set by Juerd_ on 12 March 2005.
nothingmuch perlbot nopaste 00:00
perlbot Paste your code here and #perl will be able to view it: sial.org/pbot/perl
stevan theorbtwo: better get moving,.. or you wont make it to Pugs nirvana
nothingmuch++ 00:01
nothingmuch shit, pasted to #perl
stevan perlbot karma nothingmuch
perlbot Karma for nothingmuch: 16
stevan now we are even
pasteling "nothingmuch" at 212.143.91.217 pasted "diff of sub_named_params.t for linking" (44 lines, 1.4K) at sial.org/pbot/8350
nothingmuch in diff format, to stress what changed 00:02
theorbtwo Looks great!
stevan nothingmuch: can you paste the whole thing too
diff does not give the whole picture
nothingmuch yup
Khisanth nothingmuch: could just replace the nopaste definition :) 00:03
pasteling "nothingmuch" at 212.143.91.217 pasted "sub_named_params.t - the complete modified version" (104 lines, 3K) at sial.org/pbot/8351
theorbtwo I don't think you can replace them.
Just add a new one.
Khisanth perlbot: forget nopaste
nothingmuch Khisanth: that would annoy everyone on #perl, though, wouldn't it?
perlbot removed nopaste from the database
Khisanth nothingmuch: #perl doesn't use it much!
nothingmuch ah 00:04
Khisanth I mean that specific definition
stevan nothingmuch: this looks good to me
nothingmuch++
nothingmuch nothignmuch--; # please, don't get carried away
theorbtwo perlbot, learn nopaste as Paste your code here and #<channel> will be able to view it: sial.org/pbot/<channel>
perlbot added nopaste to the database
nothingmuch ingy: does that clarify things?
perlbot nopaste 00:05
perlbot Paste your code here and #<channel> will be able to view it: sial.org/pbot/<channel>
nothingmuch hmm.. doesn't it have sprintf like formatting for channel, user who called it, etc? that seems, err, mandatory in a bot
and then it should implement ??{ } so we can, err
theorbtwo perlbot, highest karma? 00:06
perlbot The top 5 karma entries: autrijus: 21, ingy: 18, nothingmuch: 18, stevan: 16, ~wegstar: 13
theorbtwo Wow, nm, you're kicking ass.
nothingmuch hah
stevan and its nothingmuch and ingy neck and neck
ingy nothingmuch++
nothingmuch ingy *= 2
ingy riight
theorbtwo nm, ??{} would make the bot trivial to root. 00:07
nothingmuch autrijus.=factorial
ingy Ok, color me dense but I still don't totally grok it
stevan autrijus + Inf
nothingmuch theorbtwo: uhh, err, i didn't mean it in /that/ sense, we would only make *good* use of it ;-)
ingy what is supposed to happen automagically
nothingmuch ingy: this lets us run catalog_tests.pl
and get pretty links in the output to the synopses, as were explicitly specified
perhaps with the match stripped fof 00:08
but much more importantly
theorbtwo ingy: Produce a HTML or whatever version of the test itself with links to the design docs, and a version of design docs with links to the tests.
nothingmuch let us generate copies of the synopses with backlinks to the tests themselves
so that we can know what parts of the specification is tested where
stevan sort of a Devel::Cover for the Synopsis
ingy Are you trying to suck in parts of the synopsis? 00:10
nothingmuch not suck them in, just backlink from the synopses to the tests
theorbtwo Can't do that; we didn't specify an extent.
nothingmuch and place the backlinks in the right place in the synopses
ingy ah. ok!!
got it
nothingmuch so, final thoughts before we get this started?
ingy um sure 00:11
one thing is this... 00:12
put your section name in double quotes
this is part of Pod
nothingmuch ingy: i'm scared of that, because we need to premunge the links to make them actual pod
i think this can be the output
but i wouldn't like to fudge with quoting 00:13
ingy L<Text|page/"Section Name" other stuff>
stevan ok,.. dinner time,..
nothingmuch: all sounds good to me
nothingmuch come back quick
ciao, stevan
stevan nothingmuch: not likely,.. but I will be back later tonight
ingy in fact Pod requires the double quotes for multi word sections
nothingmuch ingy: i know that, but i find it, err, annoying. In theory. You've probably dealt with this.. is it really crappy to implement? 00:14
ingy nothingmuch: is your L<...> supposed to be Pod?
nothingmuch it's easier to be strict in what you emit
i mean, if we translate this pseudo-pod or pseudo-kwid link format to pod or kwid, we'll make it valid
but much harder to be lax in what you permit 00:15
well, you know what, why don't we double quote
ingy ok, then i suggest making the psuedo link look like neither Pod nor Kwid
nothingmuch it's not really that complicated
ingy no it's not
nothingmuch ingy: i actually disagree with that.. I think it should be intuitive 00:16
and if it's quoted and stuff, it might actually look valid to ...
not that that's the issue, since we're converting
basically, slightly modifying an existing format gives us both an OK design to start with, and familiar semantics 00:17
ingy nothingmuch: if you generate html from the pod/kwid in the page, would it include your links?
nothingmuch L<> is obvious as a link to anyone who wrote a bit of pod, ever
yes
ingy so it is part of the Pod
nothingmuch but the generation of either (by catalog_tests.pl, and by the reverse for syns) will first munge that first 00:18
i think
ingy and not a psuedoback link
nothingmuch like it makes some pod from the individual test cases
ingy so why not use real Pod?
nothingmuch because it's not flexible enough
theorbtwo I think it being intuitive is much much much more important then it being POD or Kwiwi. 00:19
ingy =begin backlink
...
=end backlink
nothingmuch that makes it impossible to stick in trailing comments
and makes it 3 lines, instead of 1
ingy ?
nothingmuch and something that must be parsed wholly and translated
which can't just be done with perl -ne
unless you modify $/ to be something very clever
ingy perl6 gives you this though
for free
nothingmuch our scripts are going to be perl5 for now, we want it working tonight 00:20
and when perl6 will have rules to munge this, we may translate
ingy %=POD
well why not try a litle harder now
nothingmuch oh, in that sense,
$perl + expression(); # %=POD ... 00:21
is that it?
ingy no
%=POD is a global that contains the POD DOM
nothingmuch ah,
ingy it is not currently available
nothingmuch right
i think we're beeing too hypothetical 00:22
people come in and they ask 'what can i do?
theorbtwo /metoo.
nothingmuch and we say 'uh, we have no clue'
ingy you should really read Damian's p6l post carefully
nothingmuch that is a problem that needs to be fixed here
ingy anyway,...
nothingmuch got a message id?
ingy what to do now.
I would make a new non-L code for this 00:23
nothingmuch ingy: can you whip up a good format in 30 seconds?
ingy since it is different, or is it?
I can whip up something quickly, sure 00:24
theorbtwo It's not.
nothingmuch ingy: it's really L<> semantics, exactly
theorbtwo This is something that it'd always be good to be able to do with links, thta presently can't be.
that.
nothingmuch just with additonal 'skip until you match the following regex' thing
ingy I like the idea that you make it a comment for now
nothingmuch: ok
nothingmuch which, given your comment about double quoting sections, might actually be valid, too
if the separator is not a third slash, but whitespace 00:25
but i bet that's unspecced, and thus unimplemented by 99% of pod parsers
ingy nothingmuch: but what is even better is to not just get a starting point but also an ending point
then you can INCLUDE<...>
a section is currently includable
nothingmuch ingy: i think that complicates the output format
but here is an idea i shared:
sorry, i had, but didn't share yet: 00:26
if the pattern matches at the first part of a regex, it's placed before, if it matches after or in the middle of a section, it applies to it as a unit
errm, i don't like including
this is hypertext
and we are in the tabbed browsing era
theorbtwo Hmm... 00:27
I'm not sure what you mean by first part/middle/after...
ingy including can be a good thing, but continue
nothingmuch ingy: do you have a good argument for inclusion?
theorbtwo But we could allow REs to capture.
ingy nothingmuch: it is actually called /transclusion/
nothingmuch theorbtwo: like, the links i used in sub_named_.*.t
ingy that's the real term for this 00:28
nothingmuch they reference the middle of the sections
they apply to it all
ingy any part of the Pod DOM is currently transcludable
nothingmuch but if i were to say, match the begining of the sentance
but i don't like this idea
ingy: why would we want to do that in the test/syn mapping?
theorbtwo ingy: And by "currently" you mean...
nothingmuch to get verbatim sections of synopsis copied into the output htmls of tests? 00:29
theorbtwo++ # transclusion by re captures is very beautiful
"@+" should do it, no?
ingy yes, why not make the regexp part capturing? 00:30
nothingmuch oh, no, that's offsetts
ingy @- and @+ could do it
nothingmuch well, sure, if you want to implement that
but i really don't see how it could benefit readability or clarity 00:31
ingy $1 || substr($text, $-[0], $+[0])
nothingmuch i expect the following search path to be typical:
theorbtwo Donno... could.
nothingmuch user pops into #perl
says 'what can i do?'
$regular says 'errm, read the synopses, see what you like, and if it's already tested, there'll be a little link' 00:32
'you can see if the test looks good to you, or if you can add, or if nothing is linked at all, just add a test'
clicking on the link will give back the catatlog_tests.pl output
which has sections per test
with links to synopses, and test cases layed out
ingy seems like transclusion could be real useful there 00:33
nothingmuch with the descriptions from test cases (is(... "")
included too
just a simple, concise list
please specify exactly how... to me it would be a distraction, but maybe i just don't understand how they will be used, and wher
e
ingy nothingmuch: as long as you are using comments, just do something, and we can refactor later 00:34
nothingmuch ok
so i vote for L<(Page/"Section")\s+(.*)> being matched
ingy I would use the double quotes though =)
nothingmuch $1 remains in the link
$2 is the pattern used to place the backlink
theorbtwo Need to use a Regex::Common delimited // thing; the regexes will commonly contain >s. 00:35
nothingmuch theorbtwo: good point
ingy cool, good luck! 00:36
nothingmuch so L<(Page/Section)\s+(delimiter)(....)(closedelim)>?
lets start with /, and if we run into leaning toothpick syndrome will fix that, ok? 00:37
theorbtwo Sounds good to me.
ingy nothingmuch: did you accidentally forget the double quotes? 00:38
nothingmuch yes, i did forget them
but they are optional if it's \S+, right?
damnit.
i hate special cases!
this was supposed to stay simple =(
then again, simple to humans is not necessarily simple to implementors 00:39
ingy yes, but if you are going to use a regexp in there please use the double quotes
nothingmuch and i think the tradeoff is 1:N in favour of user simplicity
ingy and I'll support them in Kwid too
nothingmuch well L<Foo/Bar //>
that should be ok
but L<Foo/Bar Gorch //> becomes tricky
so it needs to be L<Foo/"Bar Gorch" //> 00:40
right?
just like pod
ingy right
well hmmm
so consider this
maybe the regexp can just be a part of the section 00:41
nothingmuch oh my
that doesn't sound good
ingy it is up to the processor to decide what the section maps to
the parser doesn't care
nothingmuch that was the purpose of the matches anyway
no, because we want finer granuality than sections
because tests are normally
this is a basic check for Section 00:42
this tests that special case mentioned by text 'blah.*gorch' in Section is as specified
and so on
and these things are worth mentioning from the perspective of the synopses
theorbtwo And because we don't want things to depend on processors.
nothingmuch because if someone bothered to test them, someone reading the synopses should know that the special test is tested for, and where 00:43
ingy I think that anything after the *first* slash in the link will just reported as the section
and the processor can be as smart or as dumb as it wants
so all is good
nothingmuch i disagree
it's not a section
it's where inside the section that the pattern is talking about 00:44
the section should still be intelligable to users, etc
ingy but conceptually they are all just pointers into the page Dom's content
theorbtwo No, all is bad.
When different renderers think something has different semantics, that is bad.
ingy page points to the dom 00:45
nothingmuch ingy: but not every part of the dom is linkable by heading name
ingy and everything else is a query into the dom
nothingmuch you are thinking xpath?
it's not in pod yet
ingy nothingmuch: I'm explaining it in those terms
I know
nothingmuch and we're trying to get a Q&D xpath working within 10 minutes
ingy basically we are agreeing
nothingmuch which will be ok for special cases
till pod has xpath, if it does 00:46
ingy you just don;t know it yet
I'm just saying do what you will after the first slash
nothingmuch because as i understand it it's semantically equivelent, but it's format is less backwards compatible
example please
ingy and it will not break with Pod's model conceptually
nothingmuch ok, so what advantages is your format capable of bringing over a clear distinction between "section" and /pattern/?
what about syntatically? 00:47
ingy ok I'm a bit lost...
nothingmuch then post an example
that will help us help you help us understand what you mean ;-)
(mr. ingerson) 00:48
theorbtwo I'm not concerned about breaking POD; I'm much more concerned about whatever replacing it being sane.
nothingmuch amused with himself, /me puts on simon & garfunkle
ingy but what is replacing it is really just Pod ;) 00:49
theorbtwo Yeah, that ought to get fixed.
Because POD is horribly horribly broken by years and years of being underspecified.
nothingmuch ingy! example! or i will just commit what i've already done!
ingy nothingmuch: commit it!
nothingmuch ok
ingy cool
nothingmuch++
nothingmuch r890 00:50
i think we should start a background see session on the converter now
and work hard on creating a large volume of tests for this
tests with this
as a higher priority 00:51
whomever gets bored works on the converters a bit,
theorbtwo Is there an easy way to see the diff for a given r?
nothingmuch are you gusy in?
ingy all I am saying is that with L<page/...> whatever you put in ... will probably be just fine in the end
nothingmuch darcs diff -p
theorbtwo And how do we link to ml posts?
nothingmuch ;-)
theorbtwo [message-id://] 00:52
i think
let google do the work
. o O ( parsley sage, rosemay and cheese ) 00:53
theorbtwo? 00:54
ingy nothingmuch: 890 looks groovy
theorbtwo I don't have darcs.
nothingmuch theorbtwo - i was kidding.. but i meant about the message-id stuff
theorbtwo What you have there looks good.
Um, don't like that, finding the RFC now.
news:msgid. 00:55
(With no angle brackets on the msgid.)
Straight out of RFC 1738
nothingmuch inside square brackets, perlmonks style? 00:56
theorbtwo No, as an L<>.
With a regex, if you want, but news arts are generally short enough that you don't need one.
nothingmuch L<news:foo@woobling.org>?
theorbtwo Yeah.
nothingmuch okay, works for me 00:57
see://woobling.org
theorbtwo My PearPC has been annoying lately, but will try again. 00:58
Oh, anybody know if there is a way to do remote-desktop from xwindows to OSX?
nothingmuch theobtwo: vnc
but that's it
apple has an rdesktop tool, but it: 00:59
a. sucks
b. really
c. is also propietry
theorbtwo And there's no linux binary-only version? Idiots...
nothingmuch maybe timbuktu has an xwindows client though
but you'd have to pirate^Wbuy it
ingy are you guys using SEE? 01:01
nothingmuch ingy: yes
ingy address?
nothingmuch well, i am... nobody's joined yet ;-)
see://woobling.org 01:02
nothingmuch is delighted he can finally spell synopses reliably
anyway, i'm getting myself something to serve as an ashtray
i wrote a todo in each of the perl scripts, and now will start committing links from my tests to the synopses. Then i'll take on others' tests. 01:03
hlen wonders how much longer until some kind soul gets the pipe operators working 01:06
jabbot pugs - 890 - First try at linking to synopses from te 01:07
theorbtwo ==> and <==, you mean?
hlen theorbtwo, yes 01:08
theorbtwo Jess has declared it night. I'll probably be back in a few hours, though. 01:20
Limbic_Region TTFN theorbtwo 01:23
sleep well
nothingmuch backlinking etiquette: only link what is actually tested 01:29
i wonder how nice it is to backlink to s29? ;-) 01:32
nothingmuch will avoid doing that for now
jabbot pugs - 891 - smartmatch test backlinking with synopsi 01:37
pugs - 892 - synlink of given.t 01:47
pugs - 894 - synlink subname.t 01:57
pugs - 893 - synlink of control_blocks.t
nothingmuch kula shaker! 02:05
jabbot pugs - 896 - rename, rehaul and synlink $?FILE and $? 02:07
pugs - 895 - synlink sub.t
nothingmuch wizard.. blizzard 02:11
food
was it nicholas clark who raised the \0 as quote delimiter issue? 02:14
chip anybody have the svn checkout command? 02:18
chip seens none on pugscode.org
lumimies Hi 02:19
nothingmuch svn co svn.openfoundry.org/pugs
chip doggone it
nothingmuch maybe you should add it to the wiki?
chip ah thanks
nothingmuch hola lumi
want to help catalogging tests? 02:20
err, 02:22
hola lumi
want to help cataloggin tests?
lumimies Hi
Sure
This colloquy is buggy
What can I do?
nothingmuch colloquy has issues, yes
go through tests 02:23
chip Anybody know if Str | Num where { $_ } qualifies the Num or the junction type?
nothingmuch then add a L<S04/"Section"> link wherever
see t/operators/smartmatch.t, t/*/given.t, etc
errm
that's in s12, right?
hmm 02:24
if where is a named unary
then it binds tighter than junctive or
if it's a statement modifier (that analogy doesn't fit for me), then it binds very loosely 02:25
s03 has a table at the bottom, see if it makes sense
if where is an infix named operator, then it must have more explicit precedence
in that case s03 has a bug
if it's a rightward listop, then it's less than |, in which case it applies only to the Num 02:26
. o O ( shower your love on me! )
chip what's a "rightward listop"? 02:27
nothingmuch an op that slurps a list
that is right of it
unlike leftward list ops (only ==> as far as i know) 02:28
which take the list on the left side, and moves it to the right
hmm 02:30
am i hallucinating?
chip ok... that's not what "where" is, AFAICT. I think it's a binary op. (If "of" and "is" are binary, anyway)
nothingmuch i thought i checked in t/base/subtypes.t yesterday
is and of and where and will are not listed in the s03 table
but i've seen a definition of will akin to is, in s12 02:31
talking about it being infix
but not specifying precedence
chip Asked on p6l 02:36
jabbot pugs - 897 - Add a test for subtypes (a type witha w 02:37
chip Is C<++$named> for mandatory named parameter official?
nothingmuch chip: no
it says in s06 IIRC 02:38
chip Mm, OK
nothingmuch that it will 'probably' be so
but i don't recall seeing a concrete statement
but it's tested for as such (t/subroutines/sub_named*.t)
chip Right, that's where I saw it. 02:39
which is right: "+$var of Num is required", or "+$var is required of Num" ?
nothingmuch doesn't parse 02:40
oh
chip OK, how about "$var of Num is rw" vs. "$var is rw of Num"
nothingmuch i'd say the first just because it makes sense englishly
and i'd say in that case Num $var is rw 02:41
or again, the former
damnit lumi, fix your nat!
chip yeah
ayrnieu randomly, how do you access directories in perl6? <*> and readdir() and friends all seem to not exist.
nothingmuch ayrnieu do you know if they're being tested for?
ayrnieu and I'm not clear on where I should look for perl6func =) 02:42
nothingmuch i think opendir etc are not defined enough to be implemented
ayrnieu: the recently started s29 (fetching link) project is supposed to be just that
ayrnieu nothing - nothing interesting appeared on my greps, no.
nothing - oh, cool.
nothingmuch www.rodadams.net/Perl/S29.html
okay
well, as for <*> i saw it discussed some time ago. I have no idea what it really means, that i remember it 02:43
but it might have been discussed
if you can find something sensical in s29, you might want to write a test for it
and use our new linking convention to link back to the relevant part in s29 02:44
even though it's a work in progress
oh 02:45
appearantly the apocalypse on IO stuff is not ready at all yet
errm.
bummer
anyway, i promised myself some food a while ago, bbiab
elmex woho 02:53
lumimies Actually seems it's sleeptime for me 02:57
elmex ;) 02:59
it's soooo eary ;)
early
lumimies It is, too
5am is early
Just the wrong sort of early
elmex 04:00 here ;) 03:00
nothingmuch lumimies: what, with no links done?
naughty!
lumimies Yes, sorry :/
nothingmuch anywho, your connection is acting up
hah! 03:01
lumimies It is
"I even managed to get bbedit to run and also to show me all the synopses in one window and all" 03:02
Is what I said
nothingmuch bbedit--
lumimies But still, sleep beckons
nothingmuch very expensive without a good reason
well, ciao then
lumimies What do you reccomend?
nothingmuch carbon vim
lumimies Er, 2m not 2 c
nothingmuch subetha
carbon emacs
xemacs 03:03
lumimies I'll try to findsuch
nothingmuch is very very happy with carbon vim
lumimies I don't want to run another os though
nothingmuch well, if you're not an emacs head then that's good for you
in the non cynical sense ;-)
lumimies Although it comes with its own psychiatrist 03:04
nothingmuch =)
you can also run eclipse
lumimies I think bbedit is more about milking money out of web developers nowadays
nothingmuch i find that scary though
lumimies Oh wow, overkill
nothingmuch i thought web developers like dreamweaver 03:05
lumimies Gr 03:06
jabbot pugs - 898 - synlink quoting.t 03:07
nothingmuch damn, that's just cruel
stevan hey nothingmuch 03:11
nothingmuch hola stevan
stevan how goes the linking?
nothingmuch i'll leave see://woobling.org open
i'm not going to work on it this evening 03:12
however, i linkified several test files
stevan yes I see
whats at see://woobling?
nothingmuch catalog_tests.pl, and fudge_synopses.pl (really just a todo) 03:13
stevan ok,.. I had to ditch subetha
nothingmuch maybe when ingy returns you guys can work on it
stevan it was taking over too many file types
nothingmuch ditch? why?
oh...
that can be fixed rather easily
edit it's Info.plist
anyway, the birds outside are starting to make a racket 03:14
i best be off to bed
stevan ok
what time is it there?
nothingmuch 5:15
stevan wow
ok
10:15 pm here
nothingmuch east coast?
stevan yeah 03:15
nothingmuch anyway, go back to subetha, it's fun
only make sure it doesn't get in the way (that is doable)
stevan ok I will fiddle with it
cls_bsd ok, I have stupid Q, so far pugs supports hash now?_?
nothingmuch cls_bsd: i think yes
cls_bsd nothingmuch: Thank you :) so much 03:16
nothingmuch i didn't implement it ;-)
see t/data_types/hash.t 03:17
stevan cls_bsd: hashes still have some un-implemented corner cases 03:18
cls_bsd stevan: ok, I all see tests first :)
nothingmuch cls_bsd: anything that is perl, and whose results don't end in todo_ probably works 03:19
elmex_ crap fucking internet connection hate 03:20
nothingmuch i should say, and whose results aren't checked with todo_*
anyway, good night all
elmex_ nothingmuch: gn8
nothingmuch please please please keep synlinking tests
stevan nothingmuch: I will
nothingmuch++
nothingmuch stevan++
heh
*poof* 03:21
jabbot pugs - 899 - adding in some =pod sections along with 04:27
pugs - 901 - r1153@not (orig r564): luqui | 2005-03 05:47
pugs - 900 - * regen manifest
autrijus t/operators/quoting.t is simply amazing. 05:50
(test squad)++ 05:51
actually, just woke up to say this. going back to sleep. :) &
ingy hola 06:12
chip BWAHAHA 06:48
C< sub foo ($a, $b, ?$c) { print "a=$a b=$b c=$c" } > 06:49
I've got that working in Perl 5
C< sub foo ($me: $a, $b, +$c, +$d) { ... } > also 06:52
autrijus Perl 5? 06:59
you mean search.cpan.org/~brentdax/Perl6-Parameters-0.03/ ?
so, if I added m:p5// rules support today 07:03
I don't think anyone will complain, right? 07:04
when PGE arrives we'll just hook it into the same interface, in 6.283
but p5 regex can't hurt.
masak hello
hanoi works now :)
autrijus hi masak.
masak: indeed! :)
masak and i have an openfoundry account 07:05
autrijus yay! 07:11
so commit your name into AUTHORS? 07:12
masak: also, if you'd like to write a test that exercises the hanoi bug, in t/subroutines/
ingy nothingmuch: you should look at Pod's X<> format. It is an alternative to your regexp fu.
autrijus or at the end of t/subroutines/sub_named_params.t
then I'd be ~very grateful 07:13
hihi hoowa 07:15
hoowa: chinese (utf8) variable and subroutine names completely works in pugs now... I'm adding perl5 regular expressions today
hoowa: great work on getting the translations going, btw 07:16
masak autrijus: i'd like to do both, but i am still at a loss 07:18
ingy autrijus: jesse is incoming, no? 07:19
masak i think i can handle the AUTHORS file
autrijus ingy: right
masak: ok. have you written perl5 tests before?
masak but the tests i will have to study a bit first
ingy w00t
masak autrijus: no
only looked at some
autrijus ok, read this: search.cpan.org/dist/Test-Simple/li...torial.pod
masak thx
autrijus guaranteed to change your life^Wprogramming style :)
masak lol 07:20
autrijus schwern++ # the kingly pumpking
ingy we should have a raffle to guess how many pugs patches there will be by oscon 07:23
any early guesses? 07:24
autrijus patches? 07:25
I'll just grant you admin right
and give people committer bits on the spot 07:26
ingy maybe I'll hand out freepan accounts with each oscon lanyard 07:27
jabbot pugs - 902 - New Kwid syntaxes inspired by Damian...
autrijus yup. 07:28
yapcna is earlier than oscon no?
ingy yes 07:29
autrijus so, maybe we should guess pugs patches during that :)
masak autrijus: i think you've spelled my name "musak" in your journal :) 07:35
autrijus sorry. :)
fixed.
masak thx :) 07:36
let's say i wanted to do an svn co, how would i do it? 07:37
i've only done wget so far :(
autrijus svn co svn.openfoundry.org/pugs
that's all.
masak lol
ok
autrijus to commit, "svn ci" 07:38
masak right
do i need to log in for that?
autrijus when you do that
it will ask for your username/passwd.
masak ah
is there a way for me to... run my test before ci-ing it? 07:43
autrijus sure 07:44
chip autrijus: Oh, I've gone way beyond Perl6::Parameters
autrijus make install
pugs t/foo/your_test.t
masak k, thx 07:45
autrijus if you don't like make install, you can set HARNESS_PERL env var and PERL6LIB.
PERL6LIB to ext/Test/lib
chip: wow!
chip sub foo (Int $a, Ref $b, $c of SomeOtherClass where { $_->can('foo') }) {...}
autrijus wow.
autrijus is impressed
chip can't share yet, no permission from $dayjob
autrijus and it works? or just parses?
chip works 07:46
autrijus do ask!
ask for permission that is.
chip I shall. Ironically this was driven by the need for documentation
Doxygen can't parse perl, so I made Perl easier to parse for Doxygen
autrijus fantastic. 07:47
chip My favorite useless hack: C< sub foo ($c, *@_) > initializes with a simple C< my $c = shift >
autrijus bows before chip's superior perljutsu 07:48
chip now you're just showing off. I mean, c'mon, you've got more of Perl 6 than anyone 07:49
autrijus true, I probably can't code it up in perl5 in a couple days :)
chip it's my first working source filter. now I gotta make it work with Perl6::Export or Damian will be peeved
autrijus riight. 07:50
I have this unfair technical advantage. truth be told, I tried to do pugs in perl5 first
chip well, we actually have a Plan. Perl 6 attribute plugin architecture. Probably over the hill and around the bend, ...
autrijus but it just can't be done :-/
"we" as in the Cabal we? :)
chip Yeah. Tell me about it
No, no, just Damian & me, when I discovered P6::Export and realized that P6::Subs (my working name) wouldn't play nice together 07:51
s/discovered/read/
Do you know whether "is rw" attaches to the variable name always, even if the declaration is turned around with "of"? 07:52
C< $a of Num is rw >?
autrijus aye.
I think so.
chip My example had reversed polarity from my question. Is the question yes, or is the example right?
masak :) 07:53
chip "Do not go to autrijus for counsel..."
autrijus I think both works. 07:56
yes. no.
perl is great. it contains multitudes.
autrijus grins elvishly
masak just made his first svn ci 07:59
autrijus masak++
masak :D
autrijus welcome aboard!
masak thx. i could hardly sleep last night because of pugs :) 08:00
autrijus all excited? 08:01
masak overly so
autrijus I know that feeling :)))
masak bet you do :)
autrijus so, try reading haskell.org/tutorial/ or YAHT when you have some time :)
masak will do 08:02
autrijus meanwhile, I'm sure the testing and example and p6ification squads will welcome you
masak lol
autrijus lunch, brb. 08:03
jabbot pugs - 903 - Test for correct parameter binding 08:17
autrijus poll: which one is more interesting to see it happen today? 09:14
1) Perl6 -> Haskell -> C compiler 09:15
2) m:p5// Regex
I have a beginning of both
not sure what to work on.
(and compiler = real compiler, not embedded evaluator) 09:16
jabbot pugs - 904 - use array slice rather than temp var 09:17
pjcj I don't mind, since the other will happen tomorrow, right? ;-) 09:24
autrijus no, tomorrow obra arrives 09:26
and we will hack perl5 :)
pjcj perl5 is good too :) I see regexen as more useful to people wanting to play with perl6, but the compiler as more interesting and newsworthy 09:28
autrijus Go not to pjcj for counsel... :) 09:31
pjcj yeah, my wife gets cross when she asks "a or b" and gets "yes" 09:34
autrijus she should've asked "a xor b" 09:50
but then you'll still say "yes".
so maybe "a instead of b?" is better 09:51
Schwern QUESTION! Why are many assignments in Test.pm of the := variety? 10:13
my $ok := $cond ?? "ok " :: "not ok "; for example 10:14
autrijus ANSWER! because we don't need destructive assignment.
less chance of bug that way,
if you assign $ok, that'd be an error.
Schwern So you're making it constant by aliasing it to one? 10:15
autrijus yes.
Schwern Seems a round about way of doing it.
Shouldn't it be "is const" or something?
autrijus well, := reads much more natural to me
because my mind parses it as "letrec".
Schwern := just means alias
autrijus sure.
Schwern You have to examine the RHS to know its meant to make the LHS constant 10:16
autrijus also, if you don't use $ok
then it won't be evaluated.
also there is many fun compile time tricks with :=
but well, those are not convincing reasons.
I'll just say it seemed like a good idea when I wrote it.
Schwern Wouldn't "is const" take care of that, too, in theory?
Fair nuff
autrijus no, because it still forces evaluation of RHS.
Schwern I get it 10:17
autrijus a whole class of bugs just went away if people use := instead of = :)
"side-effect bugs", that is. 10:18
Schwern Except if you never used $ok... what's the point of defining it? :)
autrijus has been brainwashed: autrijus.org/haskell_brain.html
Schwern: well, maybe I only use it in one of the many given/when branches.
it's this whole lazy evaluation thing. 10:19
Schwern So my $foo := EXPR is always going to render $foo constant if EXPR returns a constant? 10:21
autrijus yes.
Schwern Ok, that's easier to understand. 10:22
Clever
autrijus =)
Schwern Also I'm slowly gravitating towards darc 10:24
s
autrijus okay. maybe you can help on darcs/svk integration :) 10:25
Schwern Checking out pugs using it instead of SVK right now.
autrijus oh. then you'll want better log messages.
Schwern ?
autrijus currently the pathces are all named "syncup"
Schwern Oh
So the subversion repo is the real one?
autrijus currently yes.
twoway sync planned. 10:26
need clkao to be around physically
Schwern Yeah, the darcs folks were saying they had that sort of working
www.abridgegame.org/pipermail/darcs...05070.html
About the only serious complaint I have about darcs so far is its wacky diff format. 10:28
clkao *yawn* 10:29
Schwern And there's no darcs log... what the hell
clkao Schwern: btw, svk and svn/perl are both happy with new eu::mm
Schwern clkao: Excellent, thanks. 10:30
And "darcs annotate" is just uselessly gimped unless I'm doing something REALLY wrong 10:32
autrijus I have lots of darcs repo corruptions.
it feels like bdb.
i.e. when it's corrupted you can't do a damn thing about it.
and merge is slow.
but when it works it works very good. 10:33
Schwern Is the annotate format really this bad? 10:34
autrijus I don't know.
never really used it.
Schwern Have a look 10:35
nothingmuch morning
autrijus hey nothingmuch. 10:36
clkao come back use svk? :) 10:39
Schwern I'm going to yell at the darcs folks for a bit.
I'll admit the simplicity of setting up a darcs repo has grabbed me. 10:40
autrijus yes!
I've pushed co .svk/ for a while now
clkao said he is willing to do it post-1.0
Schwern I don't understand why they went with _darcs instead of .darcs though.
Yeah, per checkout .dirs seems less prone to catastrophic failure. 10:41
autrijus except _darcs is prone to rm -rf failure.
nothingmuch Schwern: the rational in a jiffy is that _darcs is important and because of that is shouldn't be hidden
Schwern nothingmuch: That's dumb, you should nevrer have to look inside of it.
autrijus I agree with Schwern. 10:42
nothingmuch not inside it, at it
Schwern And most things are already wired to ignore .files
nothingmuch i agree too
i'm jos proxying
Schwern nothingmuch: ... why would I want to look at it?
autrijus oh ok.
clkao like rm -rf *
nothingmuch s/jos/just/
autrijus I agree with nothingmuch then :)
nothingmuch size concerns
Schwern ?
Am I going to forget I'm in a darcs repository?
nothingmuch well, it grows large
so users should know
Schwern So am I going to delete it then?
nothingmuch basically that's it
i dunno 10:43
Schwern Right.
nothingmuch again, i'm only proxying ;-)
Schwern Information with no action.
I understand.
nothingmuch would like to see it be compatible. If the user does 'mv _darcs .darcs' then it keeps working
that way everyone is happy
Schwern Or specify it in a .darcsrc file.
autrijus or a _darcsrc file.
autrijus grins 10:44
nothingmuch Schwern: that doesn't work
Schwern nothingmuch: It could.
nothingmuch because how do you know what another repos user's darcsrc says when you get/pull?
Schwern ~/.darcsrc
Global config, not per repo
nothingmuch ~/.darcs/prefs/default_path
;-)
Schwern Sure
nothingmuch or whatever
anywho, it doesn't matter 10:45
because there's still no way to tell what a user's pref was by just looking at the repo
Schwern ?
nothingmuch unless your only options are --dot-notation or --underscore-notation
Schwern I don't get it
nothingmuch in which case it's just in ~/darcs/defaults for get and init
errm
say i darcs get some/repo
darcs looks at some/repo/inventory stuff 10:46
Schwern Oh, it gets copied from the repo, right.
nothingmuch inside of _darcs by default
Schwern Then yeah, it should look for _darcs and .darcs
nothingmuch if there are only two options, _darcs and .darcs, then it works
ooh, it appears that many tests have been modified 10:47
have people been linking? =D
so anyway, i would like to write the backlink synopsis generating code 10:48
but i don't know where to point the backlinks to
someone with authority, please decide what split_test_catalog.pl should actually do 10:49
clkao Schwern: so, what do you think svk should steal from darcs?
nothingmuch and then i think we run catalog_tests.pl t | split_test_catalog.pl test_catalog/; fudge_synopses t test_catalog/ 10:50
Schwern clkao: darcs init
nothingmuch (unrecord!
clkao inplace repository you mean ?
Schwern clkao: ie. one step repository creation
nothingmuch unpull)
10:50 cnhackTNT is now known as cnhackTNT|away
Schwern clkao: The per change bit of "darcs record" 10:50
clkao: I like that I can work on several changes simultaneously and then easily commit them as seperate changes. 10:51
10:51 cnhackTNT|away is now known as cnhackTNT
Schwern clkao: The fact that "darcs send" is so damned easy. 10:51
clkao uhm, give me a perl unidiff applier.
nothingmuch ppt should have one, no?
autrijus ppt's patch.
Schwern I mean, I got going in darcs in no time. 10:52
clkao: No bloody URLs to deal with.
nothingmuch search.cpan.org/~andrewc/Algorithm-.../Apply.pod
search.cpan.org/~cade/Text-Patch-1.1/Patch.pm
(that's exactly what you want, i think)
clkao last time i checked non of them are streamy 10:53
text-patch is new
Schwern clkao: Mirroring and localizing is one step (darcs get) rather than several in SVK.
clkao Schwern: uh? i thought svk cp URL does that in one step?
Schwern clkao: One command, many steps 10:54
nothingmuch svk cp should be as simple as svn co for the same thing
Schwern Where am I mirroring from? What's the mirror path? What's the local path? Where's the checkout directory?
clkao i thought they all have sensible default?
Schwern And then there's the mirror vs local repository thing that always seemed like an exposed implementation detail to me. 10:55
clkao or we need svk cp --use-default etc
10:55 cnhackTNT is now known as cnhackTNT|away
Schwern clkao: Sure, but its not as simple as: darcs get wagner.elixus.org/~autrijus/darcs/pugs/ 10:55
Done. 10:56
clkao how do i just get latest 100 revisions then?
Schwern Who cares?
Some switch
clkao it can have 100k revision
hm.
Schwern If that's the case I'll flip the switch.
clkao looks at autrijus. the interactive stuff wasn't my idea. 10:57
Schwern Or perhaps if SVK detects a large amount of revisions it asks.
autrijus looks at clkao. it was not even possible.
clkao it can't. it's svn api restriction.
autrijus you had to mi, and sy, and sm, and co, and up
remember? :)
Schwern I'm going to be a user (ie. dick) here and say "not my problem".
autrijus maybe we can make it fetch all by default. 10:58
and local mirror by default
and use default names.
and you turn on interactive with --interactive.
nothingmuch autrijus: now that obra is in your time zone can you find another comitbiter? 10:59
autrijus k. I need to run.
brb.
clkao *sigh* wagner.elixus.org/~clkao/software.jpg 11:00
nothingmuch i hate softwre most of the time too
the exceptions are software that is for fun
and software which I need and actually works when I need it
like, err, /bin/sh
and... cat
... 11:01
malaire nothingmuch: Are you still interested in porting re_tests to pugs? (If not, I could work on it, but if you want to do it, then I'll find something else to do) 11:14
clkao after darcs get URL, how do i create another feature branch apart from primary local branch?
nothingmuch malaire: i've been too busy this week to work more than 10 minutes on anything except last night 11:15
clkao I think i can hack together a frontend to resemble all darcs commands.
nothingmuch so i did not get to it yet
i think perhaps it's better that you do it
because my timeline for the short term future does not look very good either 11:16
malaire ok..
nothingmuch sorry
malaire no problem, I like working on re-tests. I just don't want to duplicate too much work 11:17
nothingmuch ok
malaire autrijus: I read about you going to implement m:p5// -- would it be possible to get some support for $-[n] along with captures? 11:18
Jonathan_ Morning folks...if I'm going to put some tests in for something that's not working in Pugs yet, do I just use is_todo for them? 11:21
malaire Jonathan_: yes, but you can use other todo_* functions also from Test.pm, whatever seems best 11:28
clkao Schwern: is each darcs working copy exactly a mirror and local branch? 11:29
malaire to be exact, the function is todo_is not is_todo
Schwern clkao: I believe so.
clkao: I can make local commits without sending them to the repo and I have full history access.
s/repo/master/
clkao ok, but you can't branch for another feature off the darcs-get'ed working copy? have to get another one? 11:30
Schwern clkao: Its not so much a working copy as another repository
clkao: I don't know about branching in darcs. However it works I'd be able to do it.
clkao: A darcs checkout is its own repository.
clkao yes, i already know that. but what if you want another experimental branch besides the remote tree and the local tree? 11:31
Schwern clkao: I make local commits in it and, if I want, send them back to the master repo (or another repo)
clkao: I don't know about branching.
I think darcs branches by copying
"A new branch is just another local copy of the repository." 11:32
So I just make a copy of my own repostiory (which becomes its own repo) and work on that, moving patches back and forth as I wish. 11:33
nothingmuch and subetha heads around?
clkao ok.
jabbot pugs - 905 - Test ?.assuming(params) 11:37
Schwern clkao: Something else I've just discovered about darcs. When you do "darcs get www.foo.com/dir" its literally just a web spider of a dumb http server.
clkao: No web dav required.
clkao: Since all the repo information is right there in _darcs/
clkao ya. 11:43
nothingmuch damnit 12:14
forgive me everyone, for i have sinned 12:30
catalog_tests.pl now uses eval 12:31
err, not catalog_tests 12:32
the thing that creates backlinks
elmex morginig 12:44
jabbot pugs - 906 - Fix link regexes a bit 12:47
nothingmuch pod help, anyone? 14:01
hola gaal
gaal heya
nothingmuch i have a dillema about =item L<foo> 14:03
are you pod savvy?
gaal you talkin' to me? are you talkin' to me?
cause if you are, unfortunately i'm not really :) 14:04
nothingmuch ok
bummer
gaal i guess that'll be my next conversion up 14:05
theorbtwo I'm not POD-savvy either, but I can try to answer anyway. 14:07
nothingmuch =item link $link_count 14:08
well
the best thing i've come up with is
L<S04/...>
this is for two reasons
=item should be linkable
and =item should have content, otherrwise it's formatted in a funny way
because consecutive =items are smushed as alternatives 14:09
but i think that this output is overly verbose
theorbtwo Yeah, I think the only way to get reasonable output is to output HTML directly. 14:10
nothingmuch uhuh
well, anywho 14:11
for now lots of pod output is done
so i'll leave it like this, and rehaul later
theorbtwo To whit, output <super><a href="...#random" title="whatever"><sup>T</sup></a>.
nothingmuch actually i mean on the other side 14:12
theorbtwo Then same, but use D instead of T. Oh, and I forgot the id="" or name="" parameter.
nothingmuch yeah, html will make this simple 14:13
but i don't want to do it right now, i'd rather do the other stuff first and leave that til lthe end
theorbtwo Nod. 14:14
14:26 cnhackTNT is now known as cnhackTNT|away 14:27 cnhackTNT|away is now known as cnhackTNT
autrijus greetings lambdacamels! 14:39
full perl 5 regular expression support is in!
clkao crazy bastard
autrijus 'string' =~~ rx:perl5/.../
now works.
other syntax forms will be worked on later. 14:40
all captures, $1, $2 etc works.
$0 (i.e. $& in perl5) works.
with lookahead, lookbehind, lookarounds.
clkao how far is it for pugs to use p5lib? i want to write darck to wrap around svk.pm
autrijus clkao: you need to ask larry for that
it could be next week. or next month. or next year. 14:41
because he's not publishing his progress :)
there is progress, but we don't know how far.
alternatively, work on PPI->Yaml->Pugs in hackatons
alternatively, work on Inline::GHC
jabbot pugs - 907 - * perl 5 regular expressions, with ~~ an 14:47
theorbtwo Wow!
theorbtwo is very impressed.
autrijus =)
I expect a real compiler to take another day or so.
clkao is it a push -l? takes quite long 14:48
autrijus err, I mean ~~, not =~~ 14:49
malaire perlbot, nopaste?
perlbot Paste your code here and #<channel> will be able to view it: sial.org/pbot/<channel>
pasteling "malaire" at 217.119.39.217 pasted "make failure with r907" (32 lines, 1.6K) at sial.org/pbot/8363 14:51
autrijus platform?
malaire linux, x86, GHC 6.4 14:52
autrijus try go to pcre.c 14:53
and s/register //g
malaire a moment.. 14:54
no change 14:55
I still get exact same error. 14:57
autrijus thinks.
make it toplevel? 14:58
I mean
nvm, I'll just checkin a fix
15:00 cnhackTNT is now known as cnhackTNT|away
nothingmuch autrijus: i need some advice 15:02
autrijus nothingmuch: yes? 15:03
malaire: try again.
nothingmuch i'm working on the catalog_tests.pl script
it's parsing our weird link syntax now
but for linking to actually work we need to standardize how the files will be distributed
autrijus ok. just for the record, I think t/Synopsis/ no longer makes sense. 15:04
theorbtwo I think it never made sense...
nothingmuch me neither 15:05
i think we should have catalog_tests.pl t/ output_dir
where output dir has Synopsis/SXX.{html,pod}
and t/foo/test_name.{html,pod}
autrijus it made sense before we have this perl5 like layout.
t/Synopsis was the first thing :)
nothingmuch and backlinks are to L<t::foo::test_name.t>
heh
forward links look like L<Synopsis::S01/"Thing"> 15:06
theorbtwo I vote for putting the perl design docs in lib/Perl6/[ASE]*
(When installed, anyway.)
nothingmuch is agnostic
autrijus so ext/Perl6-Design/
nothingmuch anywho,
autrijus or something like that?
nothingmuch perhaps
malaire it compiled ok... what was the working syntax? =~~ or ~~
autrijus ~~
nothingmuch except that's where the originals are
autrijus tests welcome.
jabbot pugs - 908 - * try fix reg. spilling 15:07
malaire my $s = "Test"; $s ~~ rx:perl5/../ -- gives "Segmentation fault"
theorbtwo Sounds right, autrijus.
nothingmuch the generated ones should live somehwere generated
autrijus uh. wow.
segfault.
nothingmuch: yup. and persumably not under t/ 15:08
out/ or something?
nothingmuch the user tells the script
autrijus right.
nothingmuch and our website will have some path
but...
theorbtwo I get link failure -- ghc --make -I. -Isrc -Isrc/pcre -i. -isrc -isrc/pcre -static -Wall -fno-warn-missing-signatures -fno-warn-name-shadowing -o pugs src/Main.hs ... undefined references to pcre_*.
nothingmuch i'm thinking of being nasty and using Pod::SAX
or something like that
to merge the original pod stream
and our additions
i'm researching on how else to do this
(pod object model type stuff would be good) 15:09
but i haven't found much
or rather, i've found too much
autrijus hrm. 15:10
I don't neccessary think merging makes a lot of sense. 15:11
I'd much rather prefer generated html and links.
theorbtwo I'd suggest finding the approps section, then inserting the markers where they need to go.
theorbtwo is with autrijus.
nothingmuch ok
theorbtwo I'd convert the design docs to HTML, then use HTML::TreeBuilder to do what you want.
nothingmuch hmm
rg0now Hi folks! 15:12
nothingmuch theorbtw++ # nice idea
rg0now might I have a small qusetion
nothingmuch rg0now: NO!
rg0now :)
Ok, so here I go
nothingmuch we're busy, can't you see? now go away
bzZZT! 15:13
rg0now now, you confused me
theorbtwo You can have all the questions you want, but you'll never know if other people have answers to them unless you ask them.
nothingmuch ok ok fine
theorbtwo nothingmuch is attempting to make a joke.
And not doing a very good job of it. ;)
rg0now ok, I get it
so, is returning multiple arrays froma sub supported?
I mean something like this: return @first, @second; 15:14
and eating this like: my(@arr1, @arr2) = test;
where tst is the sub
at the moment, @arr1 slurps up everything
even if it should not...:-) as of S6 15:15
autrijus malaire: try again
S6 says it should not?
rg0now I think so
autrijus color me surprised.
rg0now but I can never be sure
autrijus I thought only := does that, and = always flattens. 15:16
rg0now the same should work with array refs (see the example in E6), but it does not work either
malaire autrijus: now make is failing again
autrijus I may also be completely wrong.
malaire: same register foo?
malaire yes
autrijus oh well. 15:17
time for plan B.
jabbot pugs - 909 - * deregister
theorbtwo r909 still gets link error for me.
nothingmuch theorbtwo: want to contribue something to make HTMLs from the synopses?
pasteling "theorbtwo" at 84.245.186.25 pasted "link errors" (59 lines, 3.2K) at sial.org/pbot/8364 15:18
Khisanth nothingmuch: aren't those already in HTML? 15:19
autrijus ok. let me try plan B
nothingmuch well, ont he p6 website they are
in our tree they aren't 15:20
i'd like to be able to make html -f ext/Perl6-Design/Makefile
malaire btw, I now have re_tests.t with 1581 todo-tests for rules (some using rx:perl5//), but I'd prefer to commit it only after I get some of it working (just to know the syntax is ok)
autrijus plan B is committed. try again. 15:22
theorbtwo gcc: pcre.o: No such file or directory 15:24
rg0now ok, I see now, = always flattens, I wanted := as autrijus said (not to mention that := seems to be broken now), so ignore me...:-)
autrijus theorbtwo: ugh.
malaire Linking ... gcc: pcre.o: No such file or directory
autrijus ok ok I got it :) 15:25
jabbot pugs - 910 - * plan B 15:27
autrijus try again. 15:29
hi elmex!
elmex: we have p5 regex now!
theorbtwo Well, we do, but it doesn't compile for most people. ;)
autrijus I think it does now :) 15:30
elmex autrijus: WOHOO
autrijus: that rox
autrijus: how? via ghc regex es?
autrijus elmex: no, PCRE
elmex: svn up and try it
elmex autrijus: ah, ffi interface?
autrijus yeah.
elmex cool
autrijus FFI is such a wonderful thing.
and when PGE arrives, we'll just have rsc/RRegex/PGE.hs
and voila. 15:31
autrijus parises the FFI gods
praises, even
malaire, theorbtwo: how's it compiling?
malaire Linking ... gcc: pcre.o: No such file or directory
theorbtwo Same.
autrijus svn up?
or is already latest?
I fixed Makefile.PL 15:32
if still fails, check if src/pcre/pcre.o is there.
malaire it's latest, r911
autrijus if not, see if ./pcre.o is there. 15:33
if it's in either place, copy it to the other place and rerun make.
voodoo debugging :)
malaire I moved src/pcre/pcre.o -> ./pcre.o and linking is OK 15:35
theorbtwo Ah, it needs to be coppied to pcre.o in the root dir of pugs.
malaire copied, not moved
scw why is the make command 'ghc --make -o pugs src/Main.hs pcre.o'
elmex autrijus: you're wonderful ;)
malaire and simple ~~ seems to be working :) 15:36
jabbot pugs - 911 - * link pcre.o 15:37
theorbtwo ./pugs -e '"foo" ~~ m:perl5/o/ the right syntax? 15:38
elmex autrijus: any knowledge about the speed of pugs with pcre and regexes ?
theorbtwo Because pugs doesn't think so... 15:39
Oh, rx, not m.
Now I get a segfault.
elmex with regex'es there is a lot more pugs can do ;) 15:40
autrijus elmex: every every fast.
s/every/very/
bad autocompletion. 15:41
elmex autrijus: really? compared to raw c speed?
autrijus sure. because pcre is raw c ;)
elmex autrijus: what about the haskell interface around the ffi stuff?
autrijus GHC compiles down to C.
stevan morning all
autrijus so, negligible overhead.
nothingmuch aren't haskell strings linked lists though? 15:42
elmex autrijus: well, if you put a for (long l = 0; l < MAX_LONG; l++); somewhere it wont be fast C ;)
nothingmuch or are they special case optimized?
autrijus nothingmuch: they are almost always optimized.
would be crazy if not
nothingmuch that's good
stevan nothingmuch: I did a bunch of linking last night, but I am not sure I did it correctly 15:43
nothingmuch stevan++
stevan needs to read the Synopsis again
nothingmuch yes, i saw
stevan nothingmuch++
nothingmuch it looks right
stevan cool
nothingmuch now, i'm going through trying to fudge the output a bit
stevan I think Pod::SAX is probably a good idea
event based stream parsing ++ 15:44
nothingmuch stevan: i think it's too much trouble though
elmex autrijus: slow programs can be written also in C ;)
nothingmuch theorbtwo had a good ide
stevan nothingmuch: I have never use it actually so I cant really say
but I like SAX
Coltrane is my favorite ;)
theorbtwo nm, I'll code it up if you tell me what I should work off of. 15:45
nothingmuch sax is nice, yes =)
but i like it only when I'm not being Q&D
like when i want to use SMP, or if latency is an issue, or if i need to handle big things
theorbtwo: i would like to have a reliable way to get HTML from the Sxx.pod files
stevan autrijus: I am getting a make error on OSX with the new reg-exp stuff
nothingmuch like the ones on dev.perl.org/perl6/synopsis/sxx.html
maybe with their pretty CSS stolen 15:46
theoretically this is only pod2html
but i've never gotten along with it
theorbtwo Yeah, that's what I was going to say.
autrijus it's Pod::Simple
better than pod2html.
but anyway.
stevan autrijus: if you get reg-exp working, I will to re-work File::Spec so we can actually use it
nothingmuch well, same thing as far as i'm concerned
autrijus folks, please rm pcre.o and svn up and retry building
jabbot pugs - 912 - * add pcre makefile 15:47
autrijus stevan: that'd be great.
theorbtwo rm'd all of the pcre.os, make clean, perl Makefile.PL, make...
stevan autrijus: it uses classes, but I can convert it to modules for now 15:48
without VMS support of course :P
autrijus sure :P
theorbtwo ...makes, but still segfaults on actually tying to do anything with the RE engine. 15:49
/tmp/ghc16867.hc: In function `RRegexziPCRE_zdwccall_entry': :519: warning: assignment discards qualifiers from pointer target type
autrijus that's the only warning? 15:50
theorbtwo Yep.
stevan ok, I am outta here, I will be back later on &
autrijus do you have gdb fu to see the error? :)
theorbtwo Is there any way to compile that will make gdb able to give useful stack traces?
Not really, no. pcre_stack_malloc is what bt #0 shows, but after that it's all ??s. 15:51
autrijus inking.
thinking.
hm. 15:52
can you download pcre by yourself
run configure
and compare config.h with src/pcre/config.h in pugs? 15:53
I'm trying to see a way out of requiring autoconf.
wow. the tibetan test in t/syntax/unicode.t fails 15:54
because it uses $ą¼¢
and that's a digit in tibetan
so we try to use it as $1 $2 etc 15:55
nothingmuch hah!
autrijus and that fails
nothingmuch i love for-free correctness
it makes me feel fuzzy on the inside
but not like i have hairballs, in a good way
autrijus :) 15:56
theorbtwo LOL
elmex autrijus: why a perl 6 interpreter, why didn't you chosse a VM as backend ? ;-) *G*
autrijus elmex: oh. you wasn't around earlier
I was asking what I should do today: a Compile.hs or a RRegex.hs
elmex autrijus: me doesn't change anything ;)
autrijus because I think I can do either in an afternoon
and someone says regex is more practical
so here we are.
elmex well, regex does rcok 15:57
autrijus so another afternoon :)
and we'll have a Compile.hs.
nothingmuch autrijus: you mean, err, like rules?
only a subset of them?
elmex korc... erm... kroc... ack... rokc...ar... rock
autrijus nothingmuch: RRegex is p5 regex.
nothingmuch: we have it now :)
except some people segfaults.
but I'm fixing it.
nothingmuch ah, you were asking
i thought you were asking now
elmex i love segfaults, they are so easyly to debug with gdb ;) 15:58
theorbtwo Huh? Looks byte-for-byte identical.
malaire autrijus: should I commit my re_tests.t file, even if pcre dies on some tests?
e.g. C< "ab" ~~ rx:perl5/a\(*b/ > gives pugs: user error (PCRE Regular Expression Error: a(b ^ missing ))
autrijus theorbtwo: huh. so compile your pcretest
theorbtwo: and does it secfault?
wait wait wait.
aha. that's a pugs bug.
interpolation is limited to vars
not backslashes
malaire should I commit those tests? (they are within eval '', but that error still quits the testing) 16:00
16:00 rafl_ is now known as rafl
theorbtwo Looks like pcretest works. 16:01
autrijus malaire: I fixed it. 16:03
so please commit ahead
r913 is my fix. 16:04
theorbtwo: ok. so it's FFI 16:05
malaire I'll do quick test first..
autrijus theorbtwo: can you grab repetae.net/john/computer/haskell/hsregex/
and see if _that_ works?
gaal hello and autrijus++, no **, for RRegex! 16:07
jabbot pugs - 913 - * escapes in p5 regex is passed to pcre,
autrijus :D
theorbtwo Blink, blink...
Double-#define of HAVE_REGEX_H, then a missing pcre/pcre.h... even after I install libpcre3-dev 16:08
Oh, also, tlk: command not found.
autrijus :(
wow.
that's something.
blink indeed.
try passing -L ? 16:09
err -I
theorbtwo Looks like I have /usr/include/pcre.h and it expects <pcre/pcre.h>
autrijus hand fix then.
theorbtwo Did, works. 16:11
With warnings as before.
autrijus but no segfaults. 16:13
malaire ok, few more tests coming in...
(if you plan on doing "make test", take good breath and relax -- it *will* take some time) 16:14
autrijus ok. now to find a poem to RRegex.
malaire r914 (and now I need to go off until Monday...)
autrijus A king he was on carven throne
In many-pillared halls of stone
With golden roof and silver floor,
And runes of power upon the door...
jabbot pugs - 915 - * Licenses, poetry, the usual deal. 16:17
pugs - 914 - Few tests for rules, 1581 to be exact
autrijus er.
wow.
autrijus is at a loss of words
uhm. someone planning to unTOTO the passing ones? 16:19
and comment out the p6 rules for now?
(just to preserve sanity until 6.283)
unTODO, even
I need to 1)journal 2)sleep.
theorbtwo: sorry for pugs still segfaulting for you. :-( will definitely look tomorrow 16:20
theorbtwo No prob. 16:21
Sleep well.
autrijus thanks!
nothingmuch anybody know of a cheap mkdir -p in perl? 16:36
i have some tricks that keep getting better
but i never like them
Limbic_Region there's a module that does that
gaal nothing: split on /, create all sections failing silently
or file::path, yes
jabbot pugs - 917 - * disable p6 rules tests for now. 16:37
pugs - 916 - * move p5 regex parts out
nothingmuch perlbot nopaste
perlbot Paste your code here and #<channel> will be able to view it: sial.org/pbot/<channel>
gaal (or doing an incremental -d test) 16:38
it's actually fun to implement in functional style
nothingmuch File::Path it is 16:40
gaal you get this warning too with latest r? 16:43
Chasing modules from: src/Main.hs
Compiling RRegex.PCRE ( src/RRegex/PCRE.hs, src/RRegex/PCRE.o )
autrijus what warning?
journal up, I'm going to sleep now. :) 16:44
gaal warning: assignment discards qualifiers from pointer target type
but go to sleep :)
autrijus right.
does it cause segfault?
try cd into src/pcre
and do a
gaal still compiling
Limbic_Region sleep well autrijus
autrijus perl -pi -e's/register //g' *
and see if that fixes it
if yes, commit it
or it may be unsignedness.
good night!
&
gaal ok, autrijus, but please sleep :) 16:45
nothingmuch ok, i have something to share 16:46
gaal k 16:47
whoa, many many new tests!
oh, it's just reorg -> builtins/
gaal has been out of things for so long! like, 30 hours 16:48
nothingmuch hehe
Limbic_Region right - anyone watching Pugs development will have to rethink the whole "Rome wasn't built in a day" thing 16:50
gaal nothing: is &todo_is yours? i'm getting a haskell error about it in t/rules/perl5 16:51
or did autrijus say he was deferring that? 16:52
nothingmuch gaal: i todo_is is pretty old
gaal but it looks like a haskell todo_is
nothingmuch but t/rules/perl5 is new, and malaire should know about it
gaal no sorry
nothingmuch then i really don't know
gaal i'm undercaffinated
nothingmuch i've been working on tests catalogging
so i'm a bit out of date myself ;-) 16:53
gaal it looks like a use problem, probably
let me unconfuse myself a but
bit
beh, there's a random death there, apparently it doesn't parse very well :) but there are no coordinates, i'll try binary searching for the problem. 16:57
vi++
well, that binary search would have been a little useless, since it fails already in the first test. 16:59
good thing i did a sanity test first :) (890 tests!)
nothingmuch, you wanted tyo share sometihng before? 17:00
nothingmuch gaal: err, broke it again in the meanwhile
give me 10 mins
this is with pod output backend
i'm not doing HTML yet
for HTML i'd like a more Devel::Cover type output
gaal but he went to sleep till monday. 17:01
nothingmuch who? what? where?
gaal sorry, that referred to malaire knowing about the parsefails, my scrollbar was wrongly positioned 17:03
nothingmuch ah 17:05
gaal i don't get it. earlier i mentioned a warning and autrijus recommended going to src/RRegex and s/register//ing the code. i did a grep first, and the word register only occurs in an object file, nowhere in the source. must be coming from a library?
ah, scratch that
he said src/prce, not src/RRegex
the s/// didn't help, FWIW 17:07
nothingmuch good food... mme 17:35
gaal food's a gopod idea, and my cat agrees.
18:04 Aankh|Clone is now known as Aankhen`` 18:41 cnhackTNT is now known as cnhackTNT|zZZ
elmex talk 19:02
gaal blah blah blah
elmex hmm
not much information in that sentence
gaal "0 but true" 19:03
elmex ;( 19:06
gaal what information are you seeking? 19:08
elmex entertaining talk about pugs 19:09
bd_ has a patch for elsif support - anyone interested? :)
bd.beginyourfear.com/~bdonlan/pugs-elsif.patch
elmex hmm, why no t commit it? 19:10
bd_ I don't have committer access
elmex bd_: get svn access, ask i.e. autrijus for it
bd_ So, I'm placing it here for someone with committer access to see it
gaal ooh, bd_, cool. does the patch come with tests? there are no tests for elsif, embarrasingly
bd_ gaal: Yes, it has 16 tests for if...elsif, if...elsif...else, unless..elsif, and unless...elsif...else 19:11
gaal bd_++
elmex wow
gaal tries patch
bd_ I'm a bit new as haskell and such, so feel free to point out if my style is lacking or whatever :) 19:13
at*
gaal bd_, i'm not a haskell hacker mylsef but if it runs here and passes the tests i'll check it in and the gang can review it when they wake up 19:14
bd_ okay
elmex it's so early ... 19:15
gaal not in tw 19:16
GMT+7, i think
bd_ It's 2pm where I am :)
gaal uh, i'm new to subversion/svk, how do i apply your patch? :) (i haven't had svk itself installed till now) 19:19
--my tree is pure svn 19:20
bd_ you can just use patch
the svk block is optional
gaal k
bd_ tbqh, I only know how to generate svk patchblocks, not how to apply them :) 19:21
but anyway I'll just kill my local branch once the patch is merged so I don't need to worry about mergng
merging*
gaal linking takes ages on this machine :( 19:23
nothingmuch yuck, i hate bad code 19:25
gaal r918, ++bd_
bd_ thanks :) 19:26
gaal bd_, talk to autrijus tomorrow (your evening) about commit privs
uh, your this evening that is
which will be his tomorrow
nothingmuch or obra, or bestian, or hcchien
gaal and thank *you* :) 19:27
jabbot pugs - 918 - bd_'s elsif patch
bd_ okay. In the meantime I'll look at some todo tests maybe
nothingmuch god damnit, i hate cargo culting
i hate arbitrary decisions that don't make sense
i hate artificial limitations that make utility code useless 19:28
gaal nothingmuch in a good mood :)
! the music stopped!
nothingmuch you don't want to know... =P 19:29
the good news
who wants to see the backlinks in html? 19:30
gaal nm, i'm out of that debate - don't think i understand enough. 19:31
nothingmuch it's really simple now
one second
gaal nm: i want to aggregate several comparisons in one ok() test. is there a neat way? the comparisons are deep. 19:32
nothingmuch nothingmuch.woobling.org/pugs_test_status/
gaal oh, and i'm in perl5 right now. 19:33
nothingmuch is_deeply
from Test::More
ah, you're in luckl
you don't have to port it ;-)
anyway, in that link, start browsing by synopsis
gaal "description: got the value we expected" x 1000 is kinda boring :) 19:34
(that was re: your link)
but is_deeply _is_ the test
i want to do something like
my @problems = grep { comparison failed test } @things_to_compare;
ok(!@problems) or diag (@problems) 19:35
nothingmuch gaal: tell whoever wrote the test that description is boring
nothingmuch.woobling.org/pugs_test_...tml#Blocks
Test::Deep is smarter than Test::More
and it has an eq_deeply
metaperl Test::More has is_deeply doesn't it?
nothingmuch i think it's reporting will make you happier
gaal this is quite nifty, actually, nm!
nothingmuch metaperl: appearantly that's not good enough 19:36
kungfuftr Test::Plan++ # lovely
gaal BTW: if anyone here has mingw and wants to fix pcre/RRegex there, i'd be much obliged :) 19:37
nothingmuch kungfuftr++; # good link
kungfuftr seems dead, but been a fair bit of cvs activity recently 19:38
gaal it not building there forces me to use the slow machine and violates impatience :)
kungfuftr bah
wrong chan
=0)
nothingmuch: which link?
nothingmuch Test::Plan
kungfuftr ah
gaal could never ever understand how hardcore ircers listen on more than one channel
nothingmuch damnit, bug in s06 19:39
sort o
f
kungfuftr nothingmuch: kungfuftr.com/code/mozilla/search/
also handy
nothingmuch L<Sxx/""This is a title""> doesn't work, nor does L<SXX/"\"This is a title\"">, and neither does E<> 19:40
how the hell can you link to =head "This is a title"?
kungfuftr: /me uses safari. it's isn't as flexible, but it's much faster on my aging laptop
gaal Xanadu?
kungfuftr, why not use the cool url quicksearch feature? 19:41
nothingmuch stevan: ping 19:43
gaal nm: test::deep carries two dependencies :( will we have something like it in pugs Test? 19:44
nothingmuch pugs test? err, maybe 19:45
probably not soon
it's a big thing to port
although it should be trivial to port given the current status
jabbot nothingmuch: thou speakst also ye olde English?
gaal okay, brief status on my part:
nothingmuch does jabbot come with a volume dial? 19:46
gaal after discussing the p6 command line the other day, i've been looking into Getopt::Declare
nothingmuch i don't like things that parse strings unnecessarily 19:47
they seem breakable
but this one... errm. I dunnoh. 19:48
i've always had mixed feelings, but never used it.
gaal (we decided pure bsd or gnu getopt wasn't powerful enough; and i proposed to see if G:D was powerful enough, and maybe port it to haskell. Autrijust had the brilliant idea of having pugs parse its own command line using an embedded *perl6* g:d.)
so, i wanted to test if g:d could indeed parse p5's command line, yes? 19:49
and i found that probably it could
nothingmuch hmm
gaal but both are comples
complex
and g:d doesn't have a test suite :)
so first off i've been writing a p5 command line g:d spec
and writing a test suite for that spec 19:50
nothingmuch ok, so make perlrun into a quick test suite for g:d
submit it to dconway
gaal now, p5's command line is *a good start* for a g:d test suite
heh
nothingmuch how is it passing so far?
gaal well, it's not completely *running* so far, because i wanted to organize the test around a declerative spec too 19:51
check_parse($_) for @tests, 19:52
where each $_ contains a command line, the options it expects and those that it doesn't, etc.
there are lots of bits and pieces -- eg should anythign of the comand line not be parsed? sure: everything after the options turns into the *script's* @ARGV, and so on 19:53
plus if there are parse errors, make a bid on what they should be
(btw, g:d currently diags to stderr so i'll have to wrap that or hack it to give me the erorrs where i want them) 19:54
nothingmuch Test::Output should help yyou 19:55
gaal so i need some machienery to do these tests
nothingmuch err, IMHO if you need the tools, don't be shy with dependancies
gaal ah, keep 'em coming 19:56
nothingmuch use Test::Deep
itdoes the job, and it does it *very* very well
gaal ok
tools++ of course
but it looks like if we decide to use a p6 version of getopt::declare for pugs own command line parsing, we're in for hella lot of work! 19:57
port: g:d itself
nothingmuch @tests is a { str => "-a foo -absa", declaration => "", expected => <something describing the results?> }
gaal write: tests for it
nothingmuch yes, it seems very daunting
gaal port: tools for the tests
but very much fun
actually str is no good 19:58
because getopt "really" gets an array from originating from the shell (or whoever called exec*)
so you need str => [ ]
but yeah, that's the idea 19:59
well, @test is an [ ] of those specs
yeah
nothingmuch VM: 6.07G 20:00
damnit
DAMNIT!: 5117 perl5.9.1 31.9% 50:39.35 1 13 818 619M+ 1.35M 71.7M+ 2.06G
gaal not sure i like the interface of test::output
um 20:01
is that for like perl -e 'print grep $$, `ps`' ?
nothingmuch no, that's the CPAN shell 20:02
gaal holding all of cpan's metadata
nothingmuch i don't like that
anyway, time for reboot
gaal they will so need an sqlite backend soon
nothingmuch ran out of disk soon
hide gaal: what's about the interface of Test::Output do you not like? 20:09
gaal i need something that locally captures std(err|out) into a var; right now i want to test what something() returned, THEN i'd wnat to test its return value. 20:11
er
it's prints, sorry
hide use stderr_from, or stdout_from 20:12
or ($stdout,$stderr)=output_from
gaal ah, i missed that -- thanks 20:13
cool.
what perl does this require? 20:14
hide I've only tested it on 5.8.x
gaal fer sure won't work before 5.6
hide looks like it passed a 5.6.0 test
gaal but i don't think we're targeting that right now :) 20:15
nice. okay, i'll be using it then.
hide++
hide my gaol is to add it to pugs
goal even
gaal gaol is a peculiar word, btw.
British folks will know it's pronounced like 'jail' 20:16
but that's so surprising!
integral it is?
vladtz isn it
gaal indeed it is. www.m-w.com/cgi-bin/dictionary?book...=0&y=0 20:17
theorbtwo It's been my experince that it is not, but it means the same thing.
integral was surprised that it was surprising, not the pronounciation :-P 20:18
vladtz I just got 918, lots of tests fail, are you experiencing the same? 20:20
gaal how many is lots?
there are known fails. 20:21
vladtz 15/90 scripts, 36/1500+
gaal that's what i'm getting too (more or less: my suite is still running) 20:22
i think it's "okay"; many things added lately. 20:23
vladtz well itś [probably better to let broken things be broken instead of turning them back into todoś 20:24
gaal mpd++ 20:26
musicpd.org/
stevan nothingmuch: ping 20:39
nothingmuch nothingmuch.woobling.org/pugs_test_status/
stevan yes I saw
nothingmuch++
nothingmuch it's horrible, horrible, but ugly, disgusting code 20:40
i hate Pod::Simple now
stevan who cares about the code
nothingmuch it's completely unreadable
stevan I would not expect it to be
:)
nothingmuch it is filled with $STUPIDLY_NAMED_GLOBALS
and the oddest control flow i've ever seen
gaal $STUPIDLY_NAMED_GLOBALS += $ESPECIALLY_IF_THEY_DONT_MEAN_WHAT_THEY_SAY; 20:41
nothingmuch so anyway, i'd like a see session, to clean it up a bit
before i check it in
gaal ($ESPECIALLY_IF_THEY_DONT_MEAN_WHAT_THEY_SAY is negative, you see) 20:42
nothingmuch and after the checkin
i'd like to make it use HTML directly
gaal with spiffy CSS it will be very beautifoul
nothingmuch gaal: CSS does not concern me much... we can just steal it from dev.perl.org
it probably also uses Pod::Simple
vladtz and it would be nice if you could see pass or fail at a galnce...
s/al/la/ 20:43
gaal yay, that woud be cute indeed
stevan nothingmuch: I cant do a SEE session now
why dont you check in and I will fiddle with it later
nothingmuch vladtz: errm, in the results?
stevan I promise not to hold the code against you
nothingmuch they take too long to generate
gaal and have test::harness listen on a high port and you can point your browser to localhost:l337 and see the green add up
vladtz in the overview, and in the indivdual test as well, why not? 20:44
stevan wants to start writing some oo tests soon
nothingmuch vladtz: i have a slow going pet project to generalize Test::Harness into an object that can get commands and dispatch events 20:45
which can then be wrapped with a thin daemon
thus enabling crap like this to be easier
vladtz: the correlation of the data is hard
vladtz it need not be done overnight
nothingmuch vladtz: conceptually if you have a nice way to do it, why not 20:46
but i don't see how to make the link, and not lose out on a lot
vladtz test::harness does have the info (from what I see in the test summary), combining it might be tough I agree 20:48
elmex hmpf 20:49
vladtz bye bye 20:51
gaal bye vladtz
too slow :)
elmex hmm 20:57
gaal hey elmex, what's up? 20:58
elmex dunno... i'm sitting here listening to mp3's of an old friend and my mind is chewing a problem about a private project... and i wonder whether i get sick within next days, not feelin that well 20:59
and i feel a pity that my haskell skills are so bad
nothingmuch r919 21:00
gaal there's hella lot to be doing in t/
elmex and i wonder whether there is also hell a lot to be doing in pugs itself... 21:01
besides the tests
gaal the tests are supporting very rapid development, so they're important
plus, you get to write in perl6, not haskell 21:02
:)
elmex i hate tests...
gaal so port a module
elmex i don't know what to test
echo "#!/usr/bin/push; use v5;"; cat FavoritePerl5Module.pm >> Perl6Module.pm ;-) 21:03
s/push/pugs/
gaal heh 21:04
elmex btw. for a highlevel vm, whats better, stackbased or registerbased?
gaal *shrug* outside my field of competence 21:05
integral stack based means you don't have to do register allocation which is rather tricky
elmex integral: yes, thats why i tend to think about a stack... it's easier... and maybe as fast if you do it clever...
integral there's small tricks that make stacks faster, like how you allocate them 21:06
elmex i guess i will do that..
lumi_ Hi
integral OTOH quite a few virtual machines just use chained frames of variables, which makes continuations easy
gaal hi lumi_
nothingmuch hola lumi
theorbtwo OTOH, register-based means you can apply lots and lots of research into how to optimize things on real machines, which is why parrot is register-based.
(Mostly.) 21:07
jabbot pugs - 919 - Now creates a tree of pods, with bidirec
lumi_ I have a thought about the linkin'
jabbot lumi_: thou speakst also ye olde English?
nothingmuch damnit jabbot!
lumi_ A bot digs its way towards /ignore
elmex theorbtwo: i just don't understand why parrot limits itself to a limited number of registerts
lumi_ If the test thing could output to some file: "$filename $line $pass" and then if it exists you can merge it in the html 21:08
theorbtwo I don't 100% understand it myself.
elmex ah, i know how to pass the data... lol
lumi_ I think that's the simplest way
theorbtwo Blink, blink... nm, should catalog_tests spew warnings? 21:09
elmex theorbtwo: i don't see the point with limited registers in parrot... why make a extra-layer with PASM and IMCC? it's imho not the simplest way to limit myself to few registers and then do registerallocations all myself for those... well... i'm no expert... 21:10
lumi_ Also, and not to pick any nits, but that would be middle English and not old English 21:11
theorbtwo nothingmuch?
jabbot theorbtwo: nothingmuch is attempting to make a joke.
nothingmuch theorbtwo: sorry 21:12
yes
those are links it couldn't parse
or couldn't find in the synopses
gaal i must be too sleepy. why is this a syntax error? map { "-$_" => "-$_" } split //, $args; 21:13
theorbtwo Not just those, a lot of Use of uninitialized value in substitution (s///) at /usr/share/perl/5.8/Pod/ParseUtils.pm line 59[01].
nothingmuch ah
i had to fix that
vim that file
and s/$text/$node/
so much for use strict... *sigh* 21:14
gaal beh :/
theorbtwo Edit Pod::ParseUtils ?
Err... yep, looks like it... 21:15
nothingmuch uhuh
gaal why doesn't map like me?
or: map likes me, split doesn't like map. why? 21:16
theorbtwo Hm.
Ah. Well, did you actually want that to return a hashref? 21:17
gaal hashref? why?
a list
made into a hashref, later, surely 21:18
theorbtwo Put a + in front of the first quote.
gaal but this shoudl take "asdf" and turns it into ( "-a" => "-a", "-s" => "-s" etc.
errrr 21:19
i'll be.
++theorbtwo
theorbtwo It sees the curlies, knows it can either be a hashref or a coderef, looks ahead one more token, sees that it's not a function, and assumes it's a hashref, then gets confused when it doesn't see a comma after the EXPR in a map EXPR, LIST.
The parser can't handle enough lookahead. 21:20
gaal yes, i understand -- only i usually had it in the other way around
so i didn't think of that
and the compiler error was regretfully about the "} split" 21:21
so thanks :)
theorbtwo This is why perl6 doesn't have a map EXPR, LIST form. 21:23
gaal perl6++ then # adding to the tally 21:24
elmex sad 21:30
gaal sad he has lots of $work this coming week 21:31
okay, i have a first shot at a simple "harness" for g:d tests 21:33
prolly shouldn't go in pugs yet though
jabbot gaal: thou speakst also ye olde English?
gaal middle and up, jaboot.
Whan that April with his showres soote / The droughte of March hath perced to the roote, / And bathed every veine in swich licour, Of which vertu engendred is the flowr 21:35
(etc. etc)
anyone interested in testing this further? 21:37
the intention: verify Getopt::Declare can parse perl5's command line
elmex hmm, i got over my heavy-tired-phase of the day... now i can work on for hours... 21:47
stevan nothingmuch: ping 22:11
nothingmuch pong
stevan hey, lumi's idea about integrating the tests
nothingmuch say again?
stevan there is now a feature in Test.pm to log tests
<lumi_>If the test thing could output to some file: "$filename $line $pass" and then if it exists you can merge it in the html
nothingmuch i say after HTML backend
stevan nothingmuch: yes, later 22:12
nothingmuch there is another problem
stevan but we have the test output abilities now
nothingmuch tests are not always linear
that's good
i don't know how to map line <-> test_case
because line can be N test cases
stevan ture
s/true/ture/
nothingmuch and if we go into that, well, solve the halting problem and i'll do the rest, i promise 22:13
stevan :P
nothingmuch anyway 22:15
i have to go
but i want it to be two phased
i will try to give the catalog_tests.pl script another whirr
first convert all tests to html
and all syns to html
conversion of tests to html should be clever
stevan check it in ,.. I will try to look itover tonight
nothingmuch but not too clever
i did
stevan ok
nothingmuch++
nothingmuch and then HTML::TreeBuilder or other should be used to implant backlinks in the synopsis HTML
based on where the L<> match *ends*
or right after the heading it points to 22:16
as a superscript '1 2 3', a number for each test, with <a name="name_of_test.t" >
i'll write all this at the top of the file
jabbot pugs - 920 - This script is no longer necessary 22:17
gaal is calling it a moose soon too
theorbtwo Moose?
nothingmuch hola rozallin 22:18
rozallin Hello.
gaal forum2.org/moose/
hey
nothingmuch see://woobling.org
gaal yawn 22:22
well, bye all
gaal &
rozallin Night gaal.
nothingmuch stevan: 921 22:25
adds todo
nothingmuch goes now.. *poof*
stevan nothingmuch: good night
jabbot pugs - 921 - add a list of todos and remove idiotic ' 22:27
bd_ hmm, what's the correct way to get/set the topic in a reduction rule (Eval.hs)? 22:32
findVar env "$_" doesn't seem to work in a for loop
autrijus hi /\camels! 22:51
bd_: to get a topic, evalVar "$_" 22:52
to set a topic, use enterLex
elmex autrijus: hi 22:54
autrijus greetings.
bd_ hmm, I've been doing that, but it acts a bit weird. I've been working on implementing given/when, and when the topic is set by given, when uses evalVar "$_" and gets the topic ok. But if it's in a for loop, ti only works if the expression in the when or the when's code block references $_
cm autrijus of pugs fame.
bd_ pugs> for (1) { when (1) { say "x" } }
Fail: cannot cast as Str: VError "Undefined variable $_" (Val VUndef)
pugs> for (1) { when ($_) { say "X" } } 22:55
X
pugs> for (1) { when (1) { say $_ } }
1
autrijus aye
bd_ It seems like $_ isn't in the lexical pad when the when reduction rule tries to look it up - how do I override this?
autrijus you need to extract $_.
look at AST.hs
I'll attempt a fix now 22:56
jabbot pugs - 922 - * Mark Stosberg caught a typo. 22:57
autrijus are you parsing when() as a App or Syn?
bd_ ummm, how do I do that? I'm new to the pugs codebase, and I'm not sure what extract does...
Syn
autrijus ok.
extract extracts implicit params from closures.
it turns
{ say $^x }
into
-> $x { say $x }
and
{ say $_ } 22:58
into
-> *$_ { say $_ }
bug was it doesn't know 'when' implicitly uses $_.
bd_ so how do I tell it that? 22:59
autrijus svn up and see my fix? 23:00
bd_ okay
autrijus had not really tested, but should work
in any case you can build on that :)
bd_ thanks :)
autrijus I need to run. obra arrived.
need to pick him up.
see you in a bit!
theorbtwo Later! 23:01
bd_ src/AST.hs:447: 23:03
Conflicting definitions for `imps'
In the binding group for: imps, imps, exps', vs'
[in a pristine checkout]
pasteling "nullfield" at 67.167.45.157 pasted "how do i modify this to have it where it executes the a command when the button is pressed and a diffreent command when it is released" (23 lines, 446B) at sial.org/pbot/8373 23:04
nullfield ?@?@??!?!/!?? 23:05
OMG HAX!
help?
elmex nullfield: yes, we love you
nullfield oh
how kind
theorbtwo nullfield, this channel is for perl6, that is perl5.
Try #perl
nullfield #perl - no help\ 23:06
elmex lol?
Perl6 <=> Perl Sex ?
nullfield wow! 23:07
elmex horny pointers exploit your heap ;)
nullfield lol 23:08
elmex nullfield: how can we serve you?
nullfield i told you...
i need help with binding in tk 23:09
elmex ah
nullfield i need several button
elmex i don't know binding...
nullfield each with two subroutions
*subroutines
elmex nullfield: you mean binding perl6 with tk?
nullfield one exectuted when the button is pressed...
another when it is released?
yes
bd_ I think he's referring to the paste?
nullfield no
elmex not perl5 ? 23:10
nullfield im not shure
elmex there is no real perl6 yet...
nullfield oh
i guesss perl 5
tk though
jabbot nullfield: thou speakst also ye olde English?
nullfield for gui
autrijus bd_: fixed
Khisanth elmex: basically he got no response in #perl so he is trying #perl6
nullfield jabbot ehh matey 23:11
no
elmex Khisanth: ah, thats of much logic ;)
pjcj nullfield: look up the bind manpage and search for ButtonPress and ButtonRelease
bd_ autrijus: ah, thanks
jabbot pugs - 924 - * fix typo 23:17
pugs - 923 - * Syn "when" should produce $_
bd_ hm, should all blocks define '&break' or only given? and if only given, how should 'when' break out of the block? 23:38
autrijus last? 23:42
bd_ S04 says in a loop it needs to continue to the next item in the loop, so last wouldn't be good 23:43
autrijus last has to be written last() for now
redo?
bd_ Does that work in an ordinary block? 23:44
or rather, should it?
S04 says that redo works as in perl 5; in perl 5 { print "y\n" ; redo; print "x\n" } is an infinite loop
We don't want when {} in an ordinary block or given to infinite-loop :) 23:45
autrijus aye aye :)
please dig deeper 23:46
I need to hop on MRT car now
&
pan-wireless city++
bd_ okay :)
If &break being defined in all blocks is ok, I've almost got a complete implementation :)
hm, though actually, that's slightly wrong when I think further 23:47
jabbot bd_: thou speakst also ye olde English?

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!