Think twice before running "make install" for Pugs | moritz.faui2k3.org/irclog/ | pugscode.org | sial.org/pbot/perl6 | ?eval [~] <m oo se> | We do Haskell, too | > reverse (show (scanl (*) 1 [1..] !! 4)) | "Perl 6 Today" video from YAPC::Asia: xrl.us/v6op
Set by agentzh on 6 May 2007.
00:22 nekokak joined 00:48 Teratogen joined
Teratogen so is Perl 6 out yet? 00:48
diakopter Teratogen: yes, I think it *just now* escaped. I'll go hunt around for it... it's gotta be around here somewhere. 00:53
dduncan maybe you'll catch it by Christmas 00:54
00:58 nipotaway is now known as nipotan 01:16 stef__ joined, ben_ joined 01:17 ben_ joined 01:18 ben_ joined 01:19 ben_ is now known as bennett 01:21 cognominal joined 01:29 justatheory joined 01:31 justatheory joined 01:39 agentzh joined
agentzh @tell fglock it'd be cool if you have a look at my questions here: moritz.faui2k3.org/irclog/out.pl?ch...-10#id_l23 thanks! :) 01:40
lambdabot Consider it noted.
agentzh moritz: the irclog has some oddity while processing long messages like the above: 01:42
moritz.faui2k3.org/irclog/out.pl?ch...-11#id_l17
lambdabot agentzh: You have 2 new messages. '/msg lambdabot @messages' to read them.
Title: IRC log for #perl6, tinyurl.com/yt8xrr
agentzh @messages 01:43
lambdabot moritz said 10h 49m 21s ago: you could prefix each paragraph of the synopsys on percabal.org/syn/ with an anchor and a link to that anchor, just like the irclogs
moritz said 10h 47m 20s ago: if you then provide a mapping from line numbers to paragraph numbers, the irclog could automagically turn things like S02:2775 into a link
agentzh moritz: i think it's a bit tricky. maybe i should prepend to each paragraph an anchor for every line composing it. 01:44
line-to-paragraph mapping on your side is difficult. 01:45
01:46 [particle] left, [particle] joined
agentzh Hmm, regarding the implementation of smartlinks.pl, it's sufficient to add placeholders containing lineno info to the POD AST and replace them with proper <a name="..."> afterwards. 01:48
lacking time lately, will try it out weeks later :) 01:50
&
01:50 agentzh left, cognominal joined
Teratogen I want to propose a new Perl 6 operator 01:52
<<==*==>>
Aankhen`` I wonder how come there's no Perl6::Perldoc::To::Html. 02:04
02:08 stef__ joined
Tene Teratogen: you're just... proposing a sequence of characters? 02:11
Aankhen``: nobody has written one yet.
Aankhen`` Tene: Yes, I know, but strangely enough there's an Xhtml. 02:16
I'm trying to summon enough courage to write a ::To::Html. It's just a little intimidating when I look at the code for ::To::Xhtml. 02:17
Oh, wait, first I was just going to fiddle with smartlinks.pl 02:18
Hmm. Gonna follow this up myself, but in case anyone already knows the answer: can LGPL code be incorporated into the Pugs source tree? 02:23
02:23 weinig_ joined
Aankhen`` Well, LGPL-2.1 appears to be included under LICENSE, so I'm going to guess it can be. 02:24
02:41 mako132_ joined 02:45 REPLeffect joined
REPLeffect Is something wrong with planetsix.perl.org? 02:46
I try to load it and it connects, but never fully loads.
(that sounded stupid) 02:47
I go to the page, and get a connection, but the page load never completes.
Aankhen`` REPLeffect: Seems to be working alright here. 02:53
REPLeffect: Can you ping it without a problem?
REPLeffect Yeah. Must be a Firefox problem
exacerbated by dialup, probably :-)
Aankhen`` Awww. 02:54
REPLeffect Loads fine in Konqueror (albeit slowly)
03:00 offby1` joined 03:08 SubStack joined 03:22 mako132_ joined 03:43 ggoebel joined 03:58 japhb joined 04:04 dduncan joined 04:10 ggoebel joined 04:12 ggoebel joined
clkao summon ingy 04:23
04:29 offby1` is now known as offby1 04:33 REPLeffect joined 05:06 BooK joined 05:08 buetow joined 05:10 REPLeffect joined 05:40 REPLeffect left 05:43 awwaiid joined 05:46 kunwon1 joined 05:51 dj joined 05:52 dj joined 05:54 dj joined 05:57 REPLeffect joined 05:58 TimToady joined 06:05 buetow joined 06:08 REPLeffect joined, REPLeffect left 06:09 japhb joined 06:12 REPLeffect joined
Aankhen`` Er, what do I use to upload something to ~agentzh/public_html on feather? 06:16
06:23 REPLeffect joined 06:25 SubStack joined 06:29 lisppaste3 joined 06:30 IllvilJa joined 06:31 REPLeffect joined 06:40 REPLeffect joined, marmic joined 06:48 iblechbot joined 06:50 riffraff joined
riffraff hi 06:50
spinclad hi ho 06:51
Tene hi ho hi
spinclad Tene: you need a longer nick, so those will line up 06:52
07:05 REPLeffect joined 07:12 franck__ joined 07:22 mjk joined
Tene yeah... 07:23
moritz Aankhen``: there ist perldoc2xhml in Perl6::Perldoc (or some such) that parses pod6
Aankhen`` moritz: Yes, I know, and I've used it. :-) I was wondering how come there's no HTML equivalent. 07:25
And, like I said, trying to work up the courage to write one myself.
moritz you mean html4?
you can still pipe it through `tidy' and tell tidy to produce any html format you want ;) 07:26
Aankhen``: re LGPL: of course you can import it into the pugs tree, afaict lgpl is a "non-poisoing" license... 07:32
and even it was poisoning, it would just affect the file it is linked to, not the whole repository 07:33
Aankhen`` Ah, I see. 07:35
I wasn't sure, so I wanted to check.
Then I saw it under LICENSE/ so I figured it should be okay. 07:36
And yes, I do mean HTML 4.
07:37 shady joined
moritz what do you need that for? 07:37
the HTML 4 I mean 07:38
dduncan LICENSE/ is just a convenient place for all licences used by parts of the distro can go, so that each component in the distro doesn't need its own copy 07:41
07:41 sapir joined 07:42 jisom joined
Aankhen`` moritz: Same reason why you'd need any POD convertor. 07:45
moritz: HTML makes a lot more sense than XHTML.
dduncan: Ahh.
dduncan: Welp, anyway, I figured if it's under LICENSE, it must be sanctioned, at any rate. ^_^ 07:46
dduncan LICENSE/ doesn't indicate sanctioned
it just says what people are using
if you want another licence for your part, add the copy of that licence to LICENSES and use it 07:47
Aankhen`` Ergh.
Nah, it wasn't for my part, it was someone else's code that I needed.
dduncan on a tangent, I will mention that I'm using straight GPL for my most important components ... QDRDBMS 07:48
I may possibly even be the only person whose ext/ is plain GPL 07:50
sapir I just checked out r16241... "perl Makefile.PL" dies with something about load_all_extensions but "./Makefile.PL" works... any idea why? 08:08
(I had to change the first line of Makefile.PL to /usr/bin/perl , so it's definitely the same perl executable) 08:09
moritz "perl Makefile.PL" works for me 08:10
dduncan for some reason, the Makefile says: #!/usr/local/bin/perl
do you have a perl there too? 08:11
sapir no, I had to change it
Tene sapir: run 'which perl'
sapir /usr/bin/perl 08:12
dduncan I don't have a perl under local
sapir the error it gives me is: Can't call method "load_all_extensions" on an undefined value at inc/Module/Install.pm line 128.
dduncan the makefile has long worked, though I didn't try it in the last few days
Tene sapir: do you have a /usr/local/bin/perl ?
sapir Tene: no
dduncan actually, I remember now that I always invoke my pugs makefile using /path/to/my/perl Makefile.PL 08:13
and the latter has extra modules installed
it is also a perl 5.8.8, while my system perl would have been 5.8.6
I am now re-running /Volumes/Programming160/perl58 Makefile.PL ... 08:15
it completed without errors
sapir, did you get the error while running the makefile, or after using make or something on the result? 08:16
sapir from Makefile.PL
running "./Makefile.PL" works and I can then make etc. 08:17
which is working so far
dduncan well, no problems here
sapir so I'm ok... but I don't get why it should make a difference how I run it
dduncan fyi, one reason I run my own perl with extra modules is because make smoke et al has more dependencies 08:18
but I would expect plain Pugs to work with a pristine perl
it bundles its own special build system for one thing
including the part giving you the error 08:19
sapir I seem to still have Perl modules lying around from older Perl versions... I'm trying to update them now 08:22
08:40 JanK_ joined
JanK_ could someone help me with a small regexp problem? 08:41
i want to match every space in a string, but those between two ". for example the regex should not match the space between moo and cow but all other spaces 'foo bar asdf="moo cow" chicken'
08:42 BooK joined
moritz JanK_: the easiest is a two-pass solution: first remove all "..." 08:42
JanK_: the problem is that with multiple "..." pairs a regex has a hard time to decide if it's outside a quoted string or not 08:43
JanK_ moritz: i need this to split the string on the matched spaces.
an other soloution would be to temporary replace the spaces in quotes, but i dont knwo how to do this either 08:44
moritz JanK_: what about splitting on all spaces, and rejoin those inside a quoted string?
JanK_ moritz: yeah, thank you that could be a solution
i'll try this 08:45
moritz or you write a simple tokenizer
09:19 chris2 joined 09:29 elmex joined 10:15 avar joined 10:31 ofer joined 10:35 gugod joined 10:41 dduncan left 10:42 cognominal joined 11:03 nferraz joined 11:09 cognominal joined 11:10 iblechbot joined 11:17 IllvilJa joined 11:18 fglock joined 11:24 nipotan is now known as nipotaway 11:59 xinming joined 12:33 myavuzselim_ joined 12:55 justatheory joined 12:59 rlb3_work joined 13:02 nferraz_ joined 13:53 marmic joined 14:11 ruoso joined 14:17 JanK_ left 14:19 rindolf joined 14:32 avico joined 14:34 avico joined 14:50 cj joined
cj whee. 14:50
morning folks
gnuvince Good morning 14:51
moritz heja 14:54
14:55 veritos joined, amnesiac joined 15:13 f0rth joined 15:15 rodi joined 15:39 TimToady joined 15:41 lichtkind joined
rindolf Hi all! 15:48
Is this a bug:
perl -le 'my $r = qr/^\-{79}$/m; print +(("Hello\n" . ("-" x 79) . "\nYourkok\n") =~ /$r/)?"Yes":"No"'
[particle] yes, in your code 15:51
avar You're matching a string that ends in kok\n to a regex that is exactly 79 -
[particle] print returns 1
avar [particle]: note the + 15:52
[particle] oh, i missed the + sign....
avar :)
rindolf avar: yes, but it's multi-line.
avar: notice the /m 15:53
[particle] i'm not sure / modifiers work with qr
avar they do
[particle] how about qr/(?m).../
oh
avar urgh 15:55
rindolf Oh! I need a \n? 15:56
Or something
avar avar@Arsia:~$ perl -le 'my $r = qr/^\-{79}$/m; print +(("Hello\n" . ("-" x 79) . "\nYourkok\n") =~ /$r/)?"Yes":"No"' 16:03
No
avar@Arsia:~$ perl5.9.5 -le 'my $r = qr/^\-{79}$/m; print +(("Hello\n" . ("-" x 79) . "\nYourkok\n") =~ /$r/)?"Yes":"No"'
Yes
a
16:04 mncharity joined 16:08 bernhard joined
jrockway ?eval undef eq undef 16:10
evalbot_r16241 Bool::True
jrockway wishes for a modifier on eq to make that false 16:11
16:11 nferraz_ joined
PerlJam jrockway: why? 16:11
[particle] like NaN? 16:12
avar undef has never behaved like NULL
it's explicitly set to an undef value, at least in p5.. 16:13
mncharity undef but unique ? :)
jrockway well, it should be undefined, not false :)
admittedly, the current behavior is good for actual use (well, with no warnings 'uninitialized'...) 16:14
but it feels wrong to me :)
obra seen Juerd
avar when would undef eq undef => Bool::False be useful?
obra @seen juerd
lambdabot juerd is in #perl6. I last heard juerd speak 1d 4h 51m 46s ago.
obra @seen audreyt
lambdabot audreyt is in #haskell and #perl6. I don't know when audreyt last spoke. 16:15
avar during theoretical circlejerks where the topic is definedness?:)
Juerd obra: pong 16:16
obra Juerd: hi. I got poked bout trying to help debug the commitbit perf issues on feather 16:17
But I...don't even have a feather account.
What's the right thing for me to be doing?
Juerd Heh, let's get you one then :)
obra What would you like from me?
Juerd First you send me an email in which you pretend you're genuine and stuff :)
obra [email@hidden.address] Including my ssh pubkey?
Juerd You make up a full name, and state it, together with the required username 16:18
That mailaddress is fine
You can mail your pubkey, but I'll set your password to "changeme" anyway.
Feather is low security.
Can't combine security with giving anyone access :)
obra ::) 16:19
sent
Juerd Feather has 90 users, of which I think I met only 15 or so :)
obra :)
kolibrie almost feels like he's met Juerd, but really he hasn't 16:20
Juerd obra: You're set
kolibrie: Care to elaborate? :)
16:21 veritos joined
kolibrie Juerd: just been in the chatroom for so long I feel like I know you as well as some people I've met, but thinking back, we've never met in real life 16:21
Juerd Heh, interesting. 16:22
veritos heya
obra Juerd: thanks. I'll try to poke around. Something will likely need to be done as root. should I irc or mail or what when I know?
Juerd obra: Sure
obra ok
Juerd Do my /msg's ever arrive? 16:23
Ah, they do ):
eh
:)
17:19 GabrielVieira joined, polettix joined 17:20 fglock joined
mncharity Can adverbs be meaningfully repeated? eg, foo :v:v; being distinct from foo :v:v:v; ? There is the implication not (ie, use of bar(*%adverbs), but I don't immediately see an explicit statement. 17:33
?eval sub f(*%a){say %a} f :v:v; 17:34
evalbot_r16241 OUTPUT[v1ā¤] Bool::True
mncharity ?eval sub f(*%a){say %a} f :v(2):v(3) 17:35
evalbot_r16241 OUTPUT[v3ā¤] Bool::True
17:36 SubStack joined
TimToady mncharity: if the parameter they are binding to is @v, then it gets all the :v values. 17:38
we probably need some tests for that 17:39
would you like to write some?
but it's also quite likely that it is not specced good 17:41
actually, it's at S06:463 17:42
the specs don't show any smartlinks to that section, so there probably are missing tests 17:44
especially since it was a fairly recent decision
mncharity :)
TimToady (where recent means "last year or so")
rindolf Hi TimToady
TimToady hi rindolf 17:45
rindolf TimToady: what's up?
TimToady there are missing tests, shock!
"There it was, gone completely!"
or at least, missing smartlinks 17:46
it's just an oddity of English that something missing can be said to exist... 17:48
17:48 jisom joined 17:50 buetow joined 17:52 ruz joined, Psyche^ joined
polettix TimToady: in Italian we say pretty much the same 17:58
TimToady probably and Indo-European thing... 17:59
*an
17:59 oceanlab joined 18:09 Psyche^ is now known as Patterner 18:10 veritos joined 18:22 Belaf joined 18:33 REPLeffect joined
obra Juerd: ping 18:42
TimToady: The commitbit issues on feather are, unfortunately, fascinating. 18:43
Juerd pongish 18:44
Juerd loves Friday evenings at work :(
obra Juerd: I'm seeing slightly weird things out of the apache logs on Feather. I'd like to 1) get the apache2 instance actually logging requests and 2) add time-taken to the access log for apache1. Is that ok? 18:45
Juerd obra: 1) Sure, 2) Add that to new log files, please, because it'd make the existing logs incompatible with the standard 18:46
obra (basically, I'm seeing requests that take 2 minutes of clock time not show up in the apache log until 2 minutes in...and see firebug reporting 1.67s" for the request.
Juerd obra: Note, when debugging, that access logs only include *completed* requests and are sometimes useless for segfaults and such. 18:47
(In this case, mod_perl with an early handler can be a life saver)
obra right. What I'm trying to see is whether the request is being handed off by the apache1 instantly or sat on for 2 min. 18:48
Juerd Aha 18:50
Perhaps server-status can be of help too
18:55 Southen joined, mj41 joined 19:05 larsen_ joined
rindolf Juerd: can you install links or elinks on feather? 19:10
Juerd: please?
19:24 the_dormant joined 19:33 REPLeffe1t joined 19:51 fglock joined 19:55 the_dormant joined
obra PSA: svn is down while juerd does a dist-upgrade 20:00
it'll be back as soon as that's done
fglock obra: hi! 20:04
lambdabot fglock: You have 1 new message. '/msg lambdabot @messages' to read it.
obra hey fglock! how's things? 20:06
fglock agentzh: I think <alpha> and <ident> are built-in subrules, because I've seen them in examples
[particle] yes, they are, like <ws>
Tene ?eval 'foo' ~~ m/<ident>/
evalbot_r16241 Match.new(ā¤ ok => Bool::True, ā¤ from => 0, ā¤ to => 3, ā¤ str => "foo", ā¤ sub_pos => (), ā¤ sub_named =>ā¤ { "ident" =>ā¤ Match.new(ā¤ ok => Bool::True, ā¤ from => 0, ā¤ to => 3, ā¤ str => "foo", ā¤ sub_pos => (), ā¤ sub_named => {}ā¤ )ā¤ }ā¤)
fglock obra: I'm a bit under burn out lately, but I will be ok in a few days 20:09
20:09 veritos joined 20:11 fglock joined, IllvilJa joined
obra fglock: sorry to hear about hte burnout 20:11
20:12 SubStack joined
[particle] fglock: maybe a sunburn will help 20:13
veritos fglock/[particle]: huh ? 20:14
fglock agentzh: re $<a> := (<b>) - it creates $<a><b>
[particle] hint... go outside! enjoy life
fglock sure - I've got the weekend ahead
[particle] fglock: dev.perl.org/parrot/trunk/languages...c/abc-2.pg 20:15
s/dev/svn/
veritos ooh, nifty. 20:16
moritz wow 20:19
20:19 fglock joined
fglock hmm - cgi-irc in feather is failing 20:19
veritos try irssi. 20:20
it's a fairly nice client
fglock [particle]: is this p6parrot already?
veritos: I'm on irssi now :)
veritos fglock: oh...
i was actually considering writing a small perl implementation. 20:21
moritz veritos: of what?
[particle] fglock: it's a first stab at syntax
veritos moritz: IRC
obra fglock: cgi-irc may be on the apache2 process
which is down while juerd does a system upgrade
[particle] it doesn't parse--note the opp syntax is not done
veritos moritz: yes, i know we have a lot, but mine would be faster.
moritz veritos: you mean another cgi interface? 20:22
veritos: or a pure perl irc client for the desktop?
rindolf Hi moritz, obra
veritos moritz: no, pure perl for desktop
hey rindolf
fglock veritos: ah - I thought you would implement a perl6 :)
moritz hi rindolf ;)
veritos fglock: v6-Py?
moritz hey, cool idea - let's implement perl6 20:23
oh, wait
[particle] fglock: you seem to have cornered the market on that
fglock veritos: how is it going?
[particle]: sorry?
veritos fglock: i've sorta scrapped it for now, i have a lot of schoolwork
maybe i should implement rules first for the dynamic grammar
[particle] implementing perl6--you've done it a few ways already :)
fglock [particle]: that's what I've been wondering about, 20:24
rindolf Hi veritos
veritos: what's up?
moritz: what's up?
veritos rindolf: just chatting, nothing else. 20:25
moritz rindolf: lot of learning for the exams :(
fglock [particle]: it doesn't seem to have made much difference
moritz rindolf: or at least I'm supposed to :)
[particle] fglock: if you have ideas on what an opp syntax should look like, speak up 20:26
rindolf moritz: which exams?
fglock [particle]: did you take a look at STD?
[particle] yes
moritz rindolf: master exams 20:27
[particle] translating from that to pge is the problem
moritz physics
rindolf moritz: physics... 20:28
moritz: I wasn't too fond of physics in Uni.
moritz rindolf: nothing to be scared of ;)
rindolf: it's all a matter of taste
rindolf moritz: I wasn't scared of it.
moritz I like computer science very well, but I decided it's nothing I want to do all my life
offby1 *gasp* 20:30
rindolf moritz: I studied Elec. Eng.
My first Usenet message - groups.google.com/group/rec.puzzles...f0912f3944
lambdabot Title: Gates &amp; Angels - A Tougher Version - rec.puzzles | Google Groups, tinyurl.com/233ml9
20:31 dduncan joined
obra Juerd: upgrade done? 20:31
offby1 God bless lambdabot for tiny-ifying long URLs. 20:32
moritz who wrote lambdabot? 20:33
Juerd obra: No
offby1 rindolf: even if I _knew_ where my first Usenet post was, I wouldn't tell anybody :-p 20:34
Juerd rindolf: 1994; impressive.
My first usenet posting was in 1997
dduncan afaik, I have never posted to usenet 20:35
my first internet communication was email, in 1997
and the first time I used the web was in 1995
or late 1994
offby1 finds the old "Four Yorkshiremen" film clip
fglock [particle]: one thing that bugs me is if "proto 'postfix:++'" is actually redefining the perl6 grammar, not bc's ? 20:36
20:36 bonesss joined
Juerd March 20, 1997 20:37
My first internet communication was IRC
In fact, IRC was the first TCP protocol I used successfully :)
[particle] fglock: can't that be fixed with namespaces?
fglock [particle] it is lexically redefining the global table
[particle] grammar foo;
Tene Juerd: where did you get your IRC client?
TimToady fglock: that's just the old pge notation, from before we understood that distinction 20:38
that's part of what will change in the upcoming refactoring
Juerd For a while (I was 12 years old) I had no idea there was more than this. IRC fit the description perfectly: communicating with other people, exchanging files, looking up information (infobots!)
So for all I knew, for a few months, IRC *was* the internet.
Tene Heh.
Juerd Tene: It came with the free dial up trial account 20:39
20:39 REPLeffect joined
moritz Juerd: and nobody postet http:// URLs that you were curious about? 20:39
Juerd I have no idea what the name of that client was
moritz s/postet/posted/
Juerd moritz: That's how I eventually learned there was more.
moritz: But no, very few people posted that.
A "corporate website" was still called "hompeage", and was typically a tiled background image of the brand logo, and a dozen pages in a horrible color scheme. 20:41
fglock TimToady: hmm - you mean postfix:* does change the perl6 grammar, and pge is being refactored? 20:42
Juerd This was in 1995
moritz Juerd: for most companies that's still the case :(
[particle] pge is being refactored to use the new syntax
Juerd I still coded in Quick Basic :)
moritz Juerd: well, not _so_ bad
[particle] i forget how far along it is, but we now have p6regex (old) and perl6regex (new) syntax 20:43
eventually the migration will be complete
20:44 lichtkind joined
[particle] ...but right now so many things depend on pge, we need to do a step at a time (i forgot about those syntax changes, which is why my initial response was confused) 20:44
PerlJam isn't that what branches are for? :) 20:45
[particle] why branch when you don't have to?
only trunk is smoked, and it's one less merge to do
until we get a good set of smoke utils, we're not going to do much branching, i'm afraid 20:47
TimToady postfix:* is illegal, if you meant the * literally. postfix:<*> is reserved for talking about the * operator in your current language that you're running. you can name your rules anything other than that, but postfix:sym<*> is recommended for rules that parse symbolic operators
postfix:foo works for non-symbolic operators 20:48
moritz you can use barewords as posfix operators?
TimToady and is taken to be short for postfix:sym<foo>
fglock I mean postfix:* as postfix-anything
TimToady ?eval 1i
evalbot_r16241 0.0 + 1.0i 20:49
moritz and how do you apply them to variables?
with unspace?
TimToady ?eval my $x = 1; $x.i
evalbot_r16241 Error: No such method in class Int: &i
TimToady NIY
will work as soon as STD is the parser though 20:50
fglock how do you talk about postfix in the current grammar, as opposed to postfix in the current language?
[particle] my Ordinal $x = 3; # could be the same as my $x = 3rd; # if postfix:sym<rd> is defined
obra Juerd: I see no deb/dh/apt processes 20:51
TimToady if you say <postfix> you're calling all the postfix:* rules in your grammar
using * in your sense
fglock so it's a... multi?
TimToady but if you say postfix:<*> you are specifically not referring to one of those rules, but current language 20:52
Juerd obra: It fails in several places and I'm a bad multitasker
TimToady yes, many of the rules in STD are multis--all of the ones that 'cheat' constructs a rule for at the end.
obra Juerd: ok
Juerd removes aspell and ispell 20:53
Ugh, broken packages :(
offby1 packages don't break. People do.
Juerd I never said they were breaking.
TimToady the constructed postfix rule refers to 5 postfix:* rules
fglock and "sym" is like the old "KEY" in <%hash> 20:54
TimToady the first one of which is postfix_i
PerlJam is still trying to wrap his head around "calling _all_ the postfix rules"
TimToady much like, but the rule can set $<sym> to whatever it likes
so that the eventual name of the operator in the eventual language becomes postfix:{$sym} 20:55
lichtkind hi juerd
TimToady so you can write one rule that parses various operators
STD does that with sigil casts, for instance
PerlJam oh, I see. it's lookup-tableish 20:56
TimToady PerlJam: the point is that some of the operators are defined by standard Perl, while others by the user, and they all have to have the same standing in the parser 20:57
[particle] rule dispatch
PerlJam right, I'm getting it (pardon my slowness :) 20:58
20:58 lichtkind_ joined
TimToady so when the grammar says <postfix> it knows all of the postfix operators defined at this point, and doesn't privilege the standard ones over the user-defined, or vice versa 20:58
fglock ah, I just found it in S05
20:59 japhb joined
TimToady (the cheated version isn't that smart, of course, since it's not doing longest-token really) 20:59
[particle] heads off to lunch
PerlJam needs to re-read S05 at the very least. 21:02
moritz thought about translating it to German to be forced to read it carefully ;) 21:03
PerlJam I've read it carefully once already. But it's changed a bit since then ;) 21:04
spinclad thought of writing a parser for it to force another process to read it carefully
TimToady if only we could find someone to write it carefully, we'd be home free... 21:06
moritz TimToady: do you think it's not carefull right now?
PerlJam heh
TimToady so far I've only rewritten one synopsis to stand on its own without assuming a bunch of Perl 5, and that's S03. 21:07
PerlJam Well, S05 doesn't have the words "conjecture" or "conjectural" nor does it have the word "maybe" anywhere within its text. I'd S05 is doing okay :-) 21:08
s/I'd/I's say/
Grr.
TimToady a synopsis that doesn't adequately mark its conjectures is even more hopeless :)
Tene s/Grr/Hi, I'm PerlJam/
TimToady and, in fact, S05 has two instances of "Conjecture". 21:09
PerlJam ah, my search was case sensitive when I didn't mean for it to be.
and there's one instance of "Maybe" that looks like it should be "May be" 21:10
TimToady looks like there are currently 14 Conjectures in the design. 21:11
21:11 tewk joined
fglock weekend & 21:11
TimToady have fun
fglock I will :) 21:12
moritz S05 mentions user defined modifiers - any idea how that might look like? 21:30
21:46 ruz_ joined 21:56 weinig joined
ingy hi all 21:58
moritz hi ingy ;)
ingy I just bound libyaml to perl. see YAML::LibYAML 21:59
when this work matures, I recommend it replace syck for pugs dev
hi moritz
avar ingy: why is it better than syck? 22:00
ingy because it was written exactly to the YAML1.1 spec
syck was completely adhoc and is pretty buggy 22:01
and afaik has no native unicode support
libyaml is the canonical Python implementation and has been ported to Ruby 22:02
it also is a streaming (think SAX) parser/emitter
syck is not 22:03
syck forces you to build everything in memory
whereas libyaml could theoretically process an endless yaml stream 22:04
Tene cat /dev/randomyaml | libyaml --parse
ingy cat /dev/randomyaml | libyaml --parse | /dev/phonecall --random-number 22:08
avar nice 22:09
ingy I'm following audrey's lead and implementing the whole stack over libyaml's parser/emitter in C 22:11
instead of Perl, so that it should be very fast
22:12 prism joined
Juerd Debian, f you. Just !@#$%# pick some sane defaults and don't block to wait for my input 22:12
Or group the questions together and ask them all at once, instead of intermittently.
japhb Juerd: part of that is package maintainers who still don't use debconf. Debconf questions "should" come all at once at the beginning, and you can change your question level such that it takes defaults most of the time. 22:14
22:14 nperez joined
Juerd japhb: I've never had debconf questions all at the beginning. 22:15
japhb Packages that aren't up to snuff usually irk me, but then I recall that I don't have packaging skillz either, so every packager has to start *somewhere*.
22:15 wolverian joined
japhb Now, base/essential packages that are handled idiotically really do irk me. 22:15
Juerd: perhaps different definitions of "beginning". Clearly it doesn't know what to ask until the packages are downloaded and unpacked, but it should be able to ask before doing the installs. 22:16
22:17 wolverian joined
wolverian Juerd, feather's locales seem borked. want me to investigate? can't see if you're doing something already there :) 22:17
Juerd wolverian: I disabled it 22:18
wolverian oh. well, that explains it..
Juerd wolverian: I've backupped your old config so I can restore it easily
wolverian well, it's just that I don't have an UTF-8 locale anymore.. 22:19
Juerd japhb: Except it unpacks each package individually, just before installing it.
wolverian s/an/a/
Juerd wolverian: For some reason that I have yet to figure out, feather is very slow with some things, and has extremely high loads even when doing very little. 22:20
At this moment, a load of 11 for just dist-upgrading. Has been at 19 even.
Disabling locales helped. A little.
wolverian Juerd, hm. well, UTF-8 locales are very resource hungry when sorting and such. some locales are, anyway.
Juerd If anyone has ideas...
22:21 Limbic_Region joined
japhb hmmm, /me wonders if I've not come across debconf behaving badly because I so rarely do really huge upgrades (I normally ride testing) 22:23
Juerd I have a lot of huge upgrades, because I run Debian Sid almost everywhere 22:24
"Unstable" is sometimes surprisingly more stable than "testing" :) 22:25
Even though it has many more updates
wolverian I've been trying ubuntu a bit, leaving servers to autoinstall security ugprades and otherwise letting them be for the six month release cycle 22:27
it's .. peaceful.. mostly.
anyway, no idea about the load. :( 22:28
Juerd I'm currently considering evaluating ubuntu for servers.
I've used it on my laptops for more than a year now, and am pleased in general, but not about major upgrades.
wolverian at least they've now added a text mode release upgrader 22:29
(do-release-upgrade)
kind of telling that it is needed at all. :) 22:30
Juerd It's silly to have such things gui only.
wolverian yes.
Juerd It's usually easier to build some commands and then a frontend, than to combine things and do it both monolithically.
wolverian iirc the backend is separated, they just didn't have a text frontend until now. 22:31
22:40 prism joined
japhb Juerd: I don't run unstable primarily because when unstable does get horked, it's a MESS to clean up. On the downside, when someone uploads a key package like say glibc every week, never allowing it to fulfill the testing waiting period, a LOT of packages get blocked. And when glibc finally moves to testing, it's almost like a release upgrade. 22:42
japhb just wants to shake the glibc packager and say "PLEASE STOP UPLOADING! TWO WEEKS, THAT'S ALL I ASK!" 22:44
22:48 mjk joined
Yaakov ingy! 22:58
23:00 nperez joined
Juerd japhb: Such a mess hasn't happened to me in the past 3 years. 23:03
japhb: But indeed such situations have existed
japhb Juerd: once burned, twice shy and all that. In my case, more like "several times burned, shy for a long time" 23:04
23:05 ruz_ is now known as ruz
japhb If not for the aforementioned "caffeinated packager single-handedly blocks testing" problem, I'd be perfectly happy in the testing world. 23:05
23:24 avar joined
moritz in S05: 23:33
/ ( foo | bar ) d <~~$0> / # matches using rule associated with $0
The last is equivalent to
/ ( foo | bar ) d ( foo | bar) /
isn't there a 'd' missing at the end?
eg / (foo | bar ) d ( foo | bar ) d/ 23:34
obra Juerd: looks done. did it error out again? 23:38
japhb moritz: why? The d is outside the parentheses marking $0's capture bounds, right?
moritz japhb: right 23:39
obra hm. commitbit feels a lot faster 23:40
moritz it _is_ faster 23:41
obra oh good :)
I wouldn't call it _fast_
moritz it doesn't time out on nearly every request as before
obra But now, hopefully, I have time to actually finish the new version ;)
moritz can somebody re-enable trac? 23:42
deleting a pending author in commitbit is a noop 23:45
23:49 SubStack joined
japhb obra: so what did you change? Or is it simply because of Juerd's upgrades? 23:49
obra japhb: I reindexed and cleaned the db 23:52
japhb obra: ah. I'm surprised we were large enough to have performance problems from the DB .... 23:54
23:59 araujo joined

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!