openfoundry down, fallback: svn.perl.org/perl6/pugs/trunk/ | -Ofun: xrl.us/hxhk | pugscode.org | pugs.kwiki.org | paste: paste.lisp.org/new/perl6 or sial.org/pbot/perl6
Set by autrijus on 4 November 2005.
autrijus theorbtwo: also if you are working on Cgen, misc/Blondie/ may be inspirational 00:01
RTS = runtime system, which carries the ability to eval string 00:02
theorbtwo That'd be... hard.
I was thinking that'd be pretty well dead last -- is it needed for simple stuff?
Right now, I'm at can't-start-yet stage, because I can't get PIL for Prelude.pm. 00:03
autrijus hm. 00:05
theorbtwo Yeah, I probably shouldn't consider this a blocker yet.
autrijus try taking the FIRST away and desugar it?
theorbtwo Odd; there's only one FIRST and I just changed it, but I still get the same error. 00:10
autrijus did you rebuild pugs? 00:11
I think it's taking its own compiler space FIRST
since prelude is inlined
now it really shouldn't do that.
theorbtwo Oh. I modified a copy of prelude.
Hold on a sec, let me try another trick or two first. 00:12
OK, renaming the file and changing the module declaraton at the top don't help, will try editing the main copy and recompiling. 00:13
Oh -- I'm not using a quite-current SVN snap; should I change to the fallback repo?
svnbot6 r6477 | autrijus++ | * Squash an empty import warning for precompiled preludes.
autrijus change to fallback please
I may have a fix for you in a bit
(shower, bbiab)
dduncan actually, what I think I'll do first is ... 00:14
an in-memory database that can be dumped to and reloaded from disk in the form of perl code
it won't be very fast, but it should be suitable as a proof of concept 00:15
theorbtwo Eh, dduncan?
dduncan the perl code wouldn't be very complicated, just hashes of arrays of ...
I'm just saying what I'll probably do prior to writing a MySQL client 00:16
theorbtwo Oh!
I forgot that you were writing a DBI.
dduncan the easiest thing to do in my mind is just make a database in perl
talking to other ones can come after this is known to be working
Pugs' testing needs should be largely met either way 00:17
Juerd Will it be a straight port, or are interface suggestions welcome? 00:18
If the latter, please study DBIx::Simple
dduncan suggestions are always welcome, though I have an idea what I want to do ... it will be influenced by existing things but not be a straight port of them 00:19
theorbtwo Allo, scook0. 00:20
dduncan the native interface will be a SQL AST, so I don't have to parse SQL internally, but a SQL parser can be wrapped around it for people that prefer that way of doing things
scook0 hi
dduncan away to the mailbox ... 00:21
theorbtwo dduncan: I'm not sure if that'd mean more work or less.
I suspect that means many backends are going to have to put the AST back down into actuall SQL. 00:22
autrijus theorbtwo: I have a fix for you now. will commit in a bit 00:23
theorbtwo How do you svn switch from the native repo to the perl.org one? Plain switch wants to contact the original repo, --relocate complaigns about mismatching uuids... 00:24
autrijus svn can't do that -- need svk
theorbtwo Oh. 00:25
theorbtwo just rm -rfs the directory, and will reget from the right repo with svk.
dduncan just to clarify, my very first perl 6 database related thing that works will be completely self-contained; it *is* the back-end ... 00:26
for very simple relational databases, coding up an implementation is actually quite easy, easier than interfacing to other dbs
mainly this is a proof of concept, to test that a front end works correctly, while limiting the number of points of possible failure, for testing 00:27
autrijus theorbtwo: you can get svk with just syncingc the [h]ead 00:28
dduncan off to the mailbox
theorbtwo feels like a rank n00b. 00:30
svnbot6 r6478 | autrijus++ | * theorbtwo notes that FIRST {} blocks did not
r6478 | autrijus++ | desugar into anything that passes PIL compilation,
r6478 | autrijus++ | as it's useing a Stmt in LValue position.
r6478 | autrijus++ | Fixed by desugaring using ternary if.
theorbtwo HTTP Path Not Found: REPORT request failed on '/perl6/!svn/bc/1/pugs/trunk': '/perl6/!svn/bc/1/pugs/trunk' path not found
autrijus this is with [h]?
theorbtwo Yep.
autrijus very weird.
maybe try -3 or something
theorbtwo First-time using, took defaults for everything. 00:31
autrijus default is [a]
theorbtwo Well, everything except that. 00:32
Hitting up-enter from the shell prompt is now working, though.
autrijus mm, can't reproduce it here :/
ok...
then svk pull from it
to get the latest change 00:33
that should fix prelude comp for you
theorbtwo (Didn't ask me how much to get, though, and seems to be getting everything.)
autrijus I'll go shower for real :)
theorbtwo Have fun.
nothingmuch autrijus: ping 00:40
autrijus nopong 00:47
(about to sleep, so try to be brief :))
...timeout? 00:56
oh well. good night lambdacamels, see you tomorrow :)
autrijus waves &
nothingmuch not yet 00:57
theorbtwo Goodnight, autrijus.
autrijus nothingmuch: for pads chip is the guy you want to ask 00:58
nothingmuch one short thing
i skimmed the doc and wasn't sure
autrijus I merely confirm the sanity, and the ability for codegen to gen it
nothingmuch are they statically resolved? or dynamically at runtime by hashing?
because if they could be made into an offset table, that would be performance happy
autrijus you are asking impl, not spec, then
the spec doesn't say really what :outer will be turned to by PIR 00:59
nothingmuch just whether the spec seems to be vague enough to allow it
autrijus yeah
nothingmuch ah, ok
autrijus impl is leo's domain.
& 01:03
rafl_ wants to comment on S22 but is too lame to check out yet another repository.. :-/ 02:00
autrijus: Could Module::ScanDeps be extended to give me the required version if a script does 'use Foo::Bar 1.00;'? 04:36
'-Ofun: xrl.us/hxhk | pugscode.org | pugs.kwiki.org | paste: paste.lisp.org/new/perl6 or sial.org/pbot/perl6' 05:06
nothingmuch obra: ping 05:50
obra: what size delay should I expect from replies to rt.cpan.org? please answer with my nick so I can backlog 05:51
obra hey 05:54
nothingmuch: we greylist. so at least 5 minutes. but depends on your MTA 05:55
nothingmuch: also, we go through perl.org for spamfiltering
nothingmuch ah
in that case i won't suspect failure yet 05:56
thanks =)
good night
06:03 rafl_ is now known as rafl
dduncan well, it appears that the usual server is up again 06:41
geoffb *yawn* 08:19
gaal say "Hello, world!" 08:22
xinming ?eval multi sub infix:<+> ( $a, $b ) { "$a $b" }; "good"+"afternoon"; 08:36
evalbot_7842 "good afternoon"
christopher Why can't I write [=] ($a, $b, $c) 3; to initialize a whole array? 10:36
svnbot6 r6479 | autrijus++ | r7891@not (orig r7842): rafl | 2005-11-05 07:02:24 +0100 10:41
r6479 | autrijus++ | r19201@ata: rafl | 2005-11-05 06:53:22 +0100
r6479 | autrijus++ | * My ideas on S22.
r6479 | autrijus++ | r19202@ata: rafl | 2005-11-05 07:01:46 +0100
r6479 | autrijus++ | * Some more S22 thoughts.
christopher sorry, that should be [=] $a, $b, $c, 3; same question. 10:47
autrijus iblech: svnbot6 can point back to openfoundry 10:50
christopher: because assignment is not (yet) an infix operator.
hm, wait, but it is :)
autrijus looks 10:51
looks like an easy fix... fixing 10:52
christopher autrijus: so it's not wrong; just a bug? or unimplemented?
autrijus fixed 10:53
will commit in a bit
christopher: write a test for it?
in t/operators/reduce-metaop.t 10:54
christopher: do you have a committer bit?
christopher cool, thanks. No, I can't commit. 10:55
autrijus okay, your email addr please
(for sending the committer bit) 10:56
christopher autrijus: malon at mit dot edu 10:57
autrijus sent. welcome aboard :) the svn repo at svn.openfoundry.org/pugs is back -- please add yourself to AUTHORS and commit a metaop.t
christopher Do you recommend [=] ($a, $b, $c)>>.foo 3 to initialize member variables at once? 10:58
autrijus other than the slight problem of >>.method not yet implemented, sure, why not 10:59
though the vanilla
.foo = 3 for ($a, $b, $c)
is maybe more readable
svnbot6 r7841 | stevan++ | Perl6::ObjectSpace : me-- 11:00
r7842 | rafl++ | r19201@ata: rafl | 2005-11-05 06:53:22 +0100
r7842 | rafl++ | * My ideas on S22.
r7842 | rafl++ | r19202@ata: rafl | 2005-11-05 07:01:46 +0100
r7842 | rafl++ | * Some more S22 thoughts.
r7843 | autrijus++ | [Massive backport from svn.perl.org now openfoundry is back]
autrijus :D 11:01
(excess flood)++ # spared the channel of a total replay
svnbot6 r7845 | renormalist++ | Emacs Mode for Perl6: 11:07
r7845 | renormalist++ | The formerly forked "cperl6-mode.el" is now reintegrated
r7845 | renormalist++ | into cperl-mode.el based on its latest version 5.10.
rafl Hello renormalist 11:15
nothingmuch obra: ping 11:23
rafl autrijus: Was it intentional to remove my thoughts on S22 or did that happen by mistake? 11:24
autrijus rafl: it happened by mistake 11:26
I did a co --import from svn.perl.org
so it overwrote the only openfoundry change meanwhile
mea culpa, fixing
done :) thanks for noticing 11:27
rafl Thank you
svnbot6 r7846 | autrijus++ | * rafl's thoughts on S22, accidentally removed during merge from svn.perl.org. Sorry. 11:29
renormalist hi rafl 11:30
I just have to go with my wife and child around the block. Will be back in 1h. :-)
rafl renormalist: OK, see you.
svnbot6 r7847 | renormalist++ | - Forgot contact and usage. (cperl-mode) 11:34
r7848 | autrijus++ | * Parse for [=] per request from christopher. 11:40
r7848 | autrijus++ | However, the lvalue context is not properly restored, so
r7848 | autrijus++ | it does not currently work yet.
theorbtwo So [=] $a, $b, $c, 5 would set $a..$c to 5? 11:44
christopher why not just $a?
sorry, I was going on to the next question.
yes. But if you did ([=] $a, $b, $c, 5) = 3; then $a alone would become 3.
autrijus that is correct. 11:45
though this borders on obfu :)
theorbtwo The last one certianly does, at least.
autrijus brb... *ponders more about PIL2 * 11:46
svnbot6 r7849 | iblech++ | Usual svn props. 11:52
gaal ?eval sub foo(Int $p1) { } foo("this should be an error, not a coercion, right?") 11:53
evalbot_7848 Can't exec "./pugs": Permission denied at examples/network/evalbot//evalhelper.p5 line 46.
gaal ?eval sub foo(Int $p1) { } foo("this should be an error, not a coercion, right?") 11:54
evalbot_7849 undef
gaal ?eval sub foo(Int $p1) { say "got: $p1" } foo("this should be an error, not a coercion, right?")
evalbot_7849 got: this should be an error, not a coercion, right? bool::true
gaal That can't be right.
What *ought* to happen? 11:56
svnbot6 r7850 | malon++ | Tests for assignment as a reduction operator [=] 11:58
gaal ?eval sub foo($p1) { say "alive" } foo(1, 2, 3) 12:00
evalbot_7850 Error: No compatible subroutine found: "&foo"
gaal good.
?eval sub foo($p1) { say "alive" } foo(1) # sanity 12:01
evalbot_7850 alive bool::true
theorbtwo Woo, prelude PIL (p5), pretty-printed. (Thank you, demerphq, for data-dump-streamer, which BTW gets deep recrusion warnings dumping that.) 12:22
autrijus gaal: this is same as 13:33
my Int $p := "Foo";
I'm pretty sure that it triggers coerce:<as>
and as such should be a warning and 0 13:34
not "Foo"
otoh until the "types" milestone, I don't really have a rigorous understanding of static vs runtime typing rules :) 13:35
theorbtwo: nice to hear!
svnbot6 r7851 | autrijus++ | * Repair PIR codegen; use "new .Continuation" instead of 14:10
r7851 | autrijus++ | the old style "newsub" op. Suggested by Leo.
Juerd One thing I've noticed is that once non-Perl6 things start taking time again, it's very hard to follow development, even superficially... 14:12
I skip most of the p6l discussion, the irc chatter, the commit logs.. 14:13
Thank God (That's Autrijus) there's Autrijus's journal
autrijus :D 14:14
to be fair, luqui and obra and allison has all been blogging as well. 14:15
# planet.pugscode.org
Juerd I know, but:
- I'm mostly interested in non-technical updates, as I don't understand things anyway, at the current level 14:16
autrijus nodnod
Juerd - Luke's journal is hard to grok because it's almost verbatim (I read it sometimes, but not much stays in my head)
- IRC logs are nice for archival, not for reading :) 14:17
This leaves your journal as my favourite way of staying more-or-less up-to-date
autrijus :) 14:20
there's the weekly summary, though
theorbtwo wonders if anybody has created a livjournal feed from that yet.
Juerd autrijus: They're hardly summaries. 14:21
TBH, one can learn almost as much by looking at Subject: lines
theorbtwo Nope.
autrijus I don't know... it helps me a lot :)
Juerd They used to be very good. Now, they read more like indexes than summaries 14:22
theorbtwo I'm off for a walk; later. 14:23
Juerd Me too 14:24
autrijus same, actually :) 14:29
rafl renormalist: ping 14:31
svnbot6 r7852 | autrijus++ | * inline Continuation support is broken on switch and CGP cores 14:34
r7852 | autrijus++ | in Parrot -- dies with "Illegal register number"; switch to
r7852 | autrijus++ | computed goto core for -B Parrot.
r0nny_ yo 15:20
15:33 wilx` is now known as wilx
Steve_p Hey...where might the most up to date docs for say() be? 15:34
Perl6::Bible? 15:35
masak Steve_p: well, &say is just &print with a newline thrown in 15:36
i'm not sure if there's anything more to document :)
Steve_p There's a bit more to how it works with objects, especially IO-based object IIRC 15:37
masak Steve_p: I can't find it in S29 15:39
Steve_p Hmmm
OK, I'll follow what's in Perl6::Say 15:40
r0nny_ ping? - is there a wiki for duscussion of special perl6 module collections ? 16:01
rafl What are special perl6 module collections? 16:02
r0nny_ casue i got a nice idea for a data serialisation/deserialisation framework should be cappable of handling multiple languages
languages as in xml, yaml and so on
rafl Isn't yaml already a language that's suitable to serialize almost anything? 16:04
r0nny_ yeah
but xml is widely used, too
also it should be cappable of replacin the .perl method
rafl You can serialize xml with yaml.
r0nny_ where is the need of serializing xml with yaml, if the app needs xml ??? 16:05
and in webstuff they use xml for async comunication 16:06
so a serializer cappable of xml isnt the worst idea
even if xml is bad vodoo, its common (unfortunately) 16:07
rafl Where is the need of serializing something with xml when the app needs yaml?
r0nny_ there is ne need
so u dnet need to use this posibility
u just need to be cappable to do 16:08
bacasue the next gerneration of web will massively use xml
Khisanth hmm is throwing gasoline on yourself and setting yourself on fire really a good idea just because it is popular to do so? :) 16:09
rafl So what's special now? Simply write an XML serializer.
r0nny_ Khisanth: i know what u mean
Khisanth yes, what makes it special? 16:10
r0nny_ pecial as in give the people forced to put themself on fire fire-resistent clothes 16:11
rafl Is that desirable? 16:13
r0nny_ u dont need to give it to people put themsolf on fire by their own will 16:14
Khisanth but it doesn't make it special 16:15
r0nny_ well - if u do next-gern webapps, u have to deal with xml - so i'd like to see different ways of serializing data with the same api
hmm
rafl I don't see why an XML serializer is good for that. But well.. simply write it and check it in. Lets see what it can do afterwards.
r0nny_ maybe special just was the wrong word
it should be one of the modules shiped directly with perl6 16:16
rafl Write it and ask those guys who are responsible for the Perl 6 distribution. 16:18
r0nny_ ok 16:19
Khisanth that sounds like a pretty far off discussion to take place
r0nny_ then im going to investigate how to use perl6 to get informations on its data structures
17:26 dbrock`` is now known as dbrock
Juerd Today I bought and read (skimmed) Perl Best Practices 18:07
Most of the rules I already practice
Some I find very silly or even a bad idea
Some inspired me and will probably change how I code 18:08
And there are some rules that I find contradictory
But it's a good book, anyway
geoffb Juerd, I think you just had the Standard PBP Response (tm). :-) 18:12
Juerd I'm a bit afraid that people who have read this book will never love Attribute::Property because it breaks at least 3 rules on the outside and at least a dozen on the inside 18:18
But none of the arguments for the rules it breaks apply... (That is, the outside rules. The source could be much clearer, of course.) 18:19
geoffb ?eval my ($q, $r) = (10 / 7, 10 % 7); 18:23
evalbot_7852 (\10/7, \3)
geoffb ?eval my ($q, $r) = (int(10 / 7), int(10 % 7));
evalbot_7852 (\1, \3)
geoffb ?eval my ($q, $r) = (int(10 / 7), int(10 % 7)); "$q $r"
evalbot_7852 "1 3"
geoffb ?eval my ($a, $b) = (10, 7); my ($q, $r) = (int($a / $b), int($a % $b)); "$q $r" 18:24
evalbot_7852 "1 3"
geoffb ?eval my ($a, $b) = (10, -7); my ($q, $r) = (int($a / $b), int($a % $b)); "$q $r"
evalbot_7852 "-1 -4"
geoffb ?eval my ($a, $b) = (-10, 7); my ($q, $r) = (int($a / $b), int($a % $b)); "$q $r"
evalbot_7852 "-1 4"
geoffb ?eval my ($a, $b) = (-10, -7); my ($q, $r) = (int($a / $b), int($a % $b)); "$q $r" 18:25
evalbot_7852 "1 -3"
geoffb good, at least that matches P5 . . . 18:26
svnbot6 r7853 | autrijus++ | * Parrot 0.3.1 doesn't allow set_addr for Closure, so use 19:27
r7853 | autrijus++ | newclosure instead.
r0nny_ how can i get the hash conains the list of member-name=>menber-value ? and the data-type ? 19:37
autrijus elaborate? 19:38
r0nny_ it tells it doesnt know this one 19:39
autrijus post some code example?
r0nny_ rafb.net/paste/results/FRdIwR73.html 19:40
whad did i wrong ? 19:42
anyone ? 19:43
autrijus er
I did not mean ".elaborate" 19:44
you want to dump the attributes
is that the idea?
r0nny_ yeah
autrijus say $baz.perl;
is one
if you want it in hash notation
say %$baz.perl
r0nny_ i want them in another way
autrijus is another
if you want to control it 19:45
for %$baz.for %$baz.kv -> $k, $v { say "$k: $v" };
err
for %$baz.kv -> $k, $v { say "$k: $v" };
something like that
r0nny_ i want to try writing a generic data serialisation framework can read/write perl data, yaml, yml, and so on
autrijus try the "for" above
gaal souldn't $baz.kv work too?
s/s/sh/ 19:46
r0nny_ it wont do
ok - how can i get the data type ?
autrijus gaal: no, that would be calling kv on the obj 19:47
which doesn't have such method
the % is a cast
r0nny_: you can use .ref
r0nny_ direct method ?
gaal oh it's an *object*
ok
autrijus r0nny_: yeah, like $v.ref 19:48
r0nny_ hmm 19:49
now i need to find a way to deal with cyclic data
rafl I guess you can use =:= for that. 19:52
r0nny_ =:= ?
rafl See S03 19:53
r0nny_ nice 19:55
gaal I found two amusing ROT-13 pairs today. One is amusing because it's English: "irk".rot13 => "vex" 19:58
the other is amusing because it's Perl: "gnat".rot13 => "tang" :-)
r0nny_ hmm 19:59
does anyone know a good way to find cyclic data structures ? 20:00
Khisanth there are a module or two that does that 20:01
r0nny_ wich ones ?
(and are they perl6) 20:02
Khisanth search.cpan.org/~lds/Devel-Cycle-1....l/Cycle.pm
rafl r0nny_: Simply remember all things you've seen so far and use =:= to check whether the underlying variable is the same. Then you can replace that with a reference to the first occurence. This doesn't discover cycles, but it breaks them up. 20:04
r0nny_ hmm 20:05
i think i will take a look at the perl5 modules, and port them 20:06
masak i have leafed through A05 again, and (again) i realize that perl6 rules are way cool 20:09
is there an implementation somewhere?
(even a half-finished one)
r0nny_ afaik the recent pugs version is cappable of them
(if u use a recent parrot) 20:10
masak r0nny_: ok, i'll give it a try
rafl Look at t/rules 20:11
What is the 'but' keyword for? 20:23
autrijus S12 20:25
"does" is destructive mutator, "but" is the nondestructive (functional) one 20:26
svnbot6 r7854 | autrijus++ | * Accomodate Parrot trunk's PGE::Hs change (r9802) 20:30
r7854 | autrijus++ | to capture the PGE_String form in Matches.
autrijus PGE shift/reduce parser support is now in Pugs. 20:31
obra ooh
Khisanth but is it in Parrot yet? :P
autrijus sure is, because that's where we got it from :)
pugs> ('a*' ~~ /<p6rule>/)<p6rule><expr><type> 20:32
"postfix:*"
('a*' ~~ /<p6rule>/)<p6rule><expr>[0]<value>
"a"
etc
you can walk the tree in p6
just as in PIR 20:33
Khisanth oO
what is ('a*' ~~ /<p6rule>/)<p6rule><expr><type> suppose to mean? 20:34
autrijus the match object returned by matching the string 'a*'
against the builtin named rule <p6rule>
is a match object that has a named "p6rule"
which contains an expr submatch, of type "postfix:*" 20:35
with one positional part, a literal "a"
the <p6rule> matches a perl6 rule string :)
Khisanth well I was mainly confused by the <p6rule><expr><type> part :) 20:39
now it just looks like it's accessing a hash?
autrijus yeah
that's exactly what it is 20:40
?eval 1 ~~ /1/
evalbot_7853 Error: cannot cast from VUndef to Pugs.AST.Internals.VCode (VCode)
frederico ? say "test"; 20:44
liz6 frederico: you mean: 20:45
?eval say "test";
evalbot_7853 test bool::true
liz6 ?
frederico yes
liz6, I'm only want to see the bot working. thanks
svnbot6 r7855 | autrijus++ | * Much improved pretty printing for VMatch objects. 21:19
r7855 | autrijus++ | This allows a convenient dump of a Match tree in the
r7855 | autrijus++ | Pugs interactive shell. Try this for some AST fun:
r7855 | autrijus++ | pugs> 'ab*|c' ~~ /<p6rule>/
r7855 | autrijus++ | pmichaud++ for doing all the shift/reduce parser work
r7855 | autrijus++ | behind this feature. :)
autrijus hmm, L~R broken Makefile.PL 21:26
by introducing File::Spec ;)
oh well, I guess I'll still keep those. 21:27
svnbot6 r7856 | autrijus++ | * Parrot embedding was broken by Limbic_Region's File::Spec 21:31
r7856 | autrijus++ | change in Makefile.PL -- repaired the confusion between dot
r7856 | autrijus++ | and commas, although File::Spec is still kept.
autrijus okay, PGE and eval PIR on both embed parrot and external parrot passed
just in time for Parrot's monthly release :) 21:32
svnbot6 r7857 | gaal++ | Article for The Perl Review (WIP). 21:33
r7857 | gaal++ | Your suggestions are welcome! My intended audience: Perl 5 programmers who
r7857 | gaal++ | haven't necessarily been following the Perl 6 process. The main focus is
r7857 | gaal++ | getting people interested with things that work now, without overwhelming
r7857 | gaal++ | theoretical discussion.
rafl autrijus: Great. parrot packages for Debian are also ready so I can release the next pugs in Debian with perl5 and parrot embedding. :-) 21:34
autrijus rafl: wonderful :) 21:36
ooh, gaal article
liz6 where in the tree can I find that article? 21:37
eric256 ?eval 'ab*|c' ~~ /<p6rule>/
autrijus apparently docs/articles/tpr.pod
eric256: I don't think evalbot is running on a machine with rules support
?eval 'ab*|c' ~~ /<p6rule>/
evalbot_7857 Error: cannot cast from VUndef to Pugs.AST.Internals.VCode (VCode)
eric256 how do i get rules support on feather? i manage the build there 21:38
liz6 hmmm.. I don't seem to have a docs/articles...
autrijus liz6: svn up
eric256: perlcabal.org/~autrijus/tmp/tmp.txt for output
eric256: you need trunk parrot
rafl eric256: Ah, good. How about migrating from 'make; make install' to the Debian packages of pugs and parrot? 21:39
autrijus accessible somewhere in PATH
that's all
oh right, feather is debian, talk to rafl, never mind me :)
eric256 lol
autrijus liz6: svk pull, even
eric256 rafl are the debian packages avail every 15mins?
autrijus there's a debian/ file in pugs 21:40
directory, even
rafl eric256: No, but it's not a problem to migrate your compile scripts to compile and install Debian packages. Where can I find them?
gaal hi 21:42
eric256 but if Debian packages aren't up to date then whats the point of converting the scripts? 21:43
/home/eric256/auto
gaal please tell me I'm not the only one who accidentally selects an entire web page in the browser because I hit ^AA to switch to a shell... 21:44
svnbot6 r7858 | autrijus++ | * wizards.p6 - optional params are now ($param?), not (?$param).
geoffb eric256, I believe rafl is saying "build fresh debian packages during build cycle, and install them" 21:45
eric256 ahh. that would make more sense.
geoffb eric256, :-)
svnbot6 r7859 | gaal++ | ?$optional -> $optional? in tpr.pod too 21:47
eric256 looks like i just needed to export the path to parrot.. so i tried that and we'll see how it works out next build ;)
autrijus gaal: I never did that because firefox interprets "A" as a search for me
(and sometimes I have the autoselect extension on) 21:48
eric256 gets impatient and forces a build. ;)
autrijus mugwump: you okay if I rm docs/class/ ? 21:49
it's a wonderful start, but by this time it's outdated
eric256 so how does 'ab*|c' ~~ /<p6rule>/ generate and object anyway? and does that mean we can make rules that build objects somehow?
gaal I should just map ^Aa to alt-tab or something. or better, raise the shell window.
autrijus eric256: yes and yes. 21:50
it generates and object by running PGE in parrot land, which yields parrot objects, which are serialized, which are then imported to pugs land
gaal autrijus: what's outdated?
autrijus gaal: docs/class/
gaal ah
autrijus eric256: however we are currently only treating parrot objects as hash
which is just as well -- we don't yet have obj meth call support in embed anyway
rafl eric256: Yes, that's what I mean. Where are those scripts? 21:51
autrijus fortunately, pmichaud's design has it so all attributes are straight data
so it's not that hard to walk the generated hash
rafl eric256: We don't need to use the packages if you don't like that, but fresh packages for Debian every 15mins are a good thing though. 21:52
autrijus eric256: to write a rule that builds object, write a method that returns objects -- but this requires either a working parrot embedding with objcall support (like we did for p5)
or a working parrot target backend (which depends on lexpad)
or porting PGE to perl6
(or to hs, that is)
eric256 ralf that would be fine i think. as long as they don't take too long to build. its on feather at /home/eric256/auto
autrijus all are interesting directions, but I'll do them after OO/PIL2 :) 21:53
rafl eric256: They take approximately as long as a normal build and install plus some seconds to put everything in packages and install them.
eric256 autrijus...sounds cool . like we'll be able to right a dataformat rule and just matching it will build the object or hash at least
autrijus eric256: yes. although with rules alone you can only build object trees, not walk or transform them 21:54
eric256 still sounds cool. alot different than just straigh old regexp 21:55
autrijus to do the walk/transform part, luqui ported the hs UUAG system to perl5 on CPAN as Language::AttributeGrammar -- he also started porting it to p6, and allison ported it to parrot 21:56
to me, as an avid XML::Twig user, AG feels like Anything::Twig ;)
eric256 ?eval 'ab*|c' ~~ /<p6rule>/
evalbot_7859 Error: cannot cast from VUndef to Pugs.AST.Internals.VCode (VCode)
eric256 why is that error different than the interactive one?
autrijus ?eval 1 ~~ 2
evalbot_7859 bool::false
autrijus ?eval 1 ~~ // 21:57
evalbot_7859 Error: cannot cast from VUndef to Pugs.AST.Internals.VCode (VCode)
autrijus ?eval 1 ~~ m:p5//
evalbot_7859 pugs: *** cannot cast from VUndef to Handle (VHandle) at Prelude.pm line 197, column 24-45 Prelude.pm line 221, column 13-25 Prelude.pm line 278, column 5-54 Prelude.pm line 61, column 30-59
autrijus ?eval 1 ~~ m:p5://
evalbot_7859 Error: unexpected ":" expecting word character, block construct, term postfix, operator, ">>=:=<<", ">>eqv<<", ">>!=<<", ">>==<<", ">><=<<", ">>>=<<", ">>~~<<", ">>!~<<", ">>eq<<", ">>ne<<", ">>lt<<", ">>le<<", ">>gt<<", ">>ge<<", ">><<<", ">>><<", "\187=:=\171", "\187eqv\171", "\187!=\171", "\187==\171", "\187<=\171", "\187>=\171", "\187~~\171", "\187!~\171", "\187eq\171", "\187ne\171", "\187lt\171", "\187le\171", "\187gt\171", "\187ge\171", "\18
autrijus sigh
I think the rx_ macros putter made in Prelude.pm is somehow not working in evalbot
it seems to be an evalbot thing not a pugs thing
eric256 how do i tell if it has parrot linked (or whatever you call it)
autrijus you can use ldd 21:58
eric256 autrijus i get the same error on the command line in interactive mode
autrijus but either way should work -- we don't really need embed parrot
eric256 ohh you just need it in the path while running..i though i needed it in the path while building
parrot executable is in the path on feather 21:59
whooo... feather pugs is at 7755...100 commits behind? 22:00
?eval $?PUGS_VERSION
evalbot_7859 \"Perl6 User\'s Golfing System, version 6.2.10, October 10, 2005 (r7859)"
rafl autrijus: Do we have an svk mirror parrot on feather?
autrijus rafl: we do 22:01
see perlcabal.org/~autrijus/
rafl "Let me know if you'd also like mirrors for ponie, perl 6 design repo, and/or parrot."
autrijus er sorry.
p6 design repo is //mirror/p6 22:02
I don't think we have parrot, no. maybe just use svn for that?
eric256 eval bot isn't running the general pugs build on feather../me goes to investigate 22:03
rafl I think it would be useful to have a parrot mirror. Some users have parrot svn copies in their homes.
autrijus okay. 22:04
eric256 hmm svk on it locked for somereason
autrijus I'll set one up
rafl autrijus: 4 as far as I see. I'd be the 5th.
autrijus eric256: sync.pl updates every 60 secs
rafl autrijus: Thanks.
eric256 autrijus: on the autobuild co svk said locked, i had to do svk cleanup, now it should work 22:06
autrijus that's fine... I'm downloading parrot svk boot
eta 2min -- brb
eric256 just realizes...does that mean i could pass a perl6 file to that rule and get something cool and usefull back? 22:10
Juerd Poll:
Given an email link and a contact form, which would you use?
liz6 s/and/or/ ? 22:11
eric256 contact form
Juerd liz6: No, both are an option
eric256 hates email links because they always open outlook instead of gmail.. ;)
Juerd eric256: ... 22:12
eric256: Get Firefox and the extension that uses gmail for mailto: links
Your vote is counted regardless, though :)
liz6 contact form
eric256 i especial hate when they are label "contact us" but its an email link 22:13
liz6 eric256: agree
Juerd Okay, thanks for your votes. Any specific reasons?
Why does a contact form feel better?
liz6 because it will guide you in your question
and feels more "professional" to me...
eric256 i just don't like having my browser open an external program without asking.. and like liz is suggesting when it does i just stare at it with no idea what exatcly to say (most the time). 22:15
Juerd It's funny, though, that most people say they prefer contact forms (I don't), while in my own experimentation, I've received more feedback providing just an email address.
eric256 of course maybe thats a problem with my browser and not the link. dunno. ;)
rafl eric256: Why aren't the pugs modules installed on feather?
eric256 pugs modules?
Juerd rafl: Just the binaries are copied. 22:16
liz6 juerd: providing an email address is always good
eric256 providing an email adress.. and proving a mailto: linke are differnt, i don't mind the former
Juerd liz6: Certainly. But the question here was which one you'd use, if both options were present. 22:17
gaal Juerd: I always prefer an email, which I use with my mail client to send the message. This way I have a copy for future reference.
Juerd So far, it's almost 50/50 across channels
rafl eric256: FindBin.pm, etc.
Juerd: I know. I want to know the reason for that.
Juerd So I think it's safe to say that anyone who has a professional website should have both.
autrijus eric256: you can, once we transcribe Pugs.Parser into something PGE understands
Juerd rafl: I didn't want to pollute the system without packages
eric256 Juerd: definitly
liz6 juerd: yep 22:18
autrijus eric256: although, pmichaud says he'll have the basic expressions done in a couple weeks
in which case you can do "'some p6' ~~ /<perl6>/'
eric256 oh so p6rule isn't complete yet?
autrijus p6rule is complete
but it only matches a "rule"
not a "perl 6 program"
Juerd rafl: The debian packages weren't there yet when I built this script. I'm sure that patches are welcome, but remember that it must be automated.
eric256 ohh.. /me thought p6rule = rule to parse p6
but that makes since
autrijus yeah, it's a bit confusing :)
dduncan so, unless there are any major commits pending in the next hour, I'll smoke the current Pugs now 22:19
Juerd Perl 6 Rules!
That alone sounds MUCH better than "Perl 6 Regexes!"
eric256 dduncan FYI... feather smokes every 6 hours
autrijus "Perl 5 already Regexed the world. Perl 6 will Rule the world"
dduncan is it a darwin box
rafl Juerd: OK. 22:20
eric256 nope ;)
dduncan then this isn't completely redundant
anyway, its possible that all the darwin smokes over the last couple weeks were me ... I've been doing it every 2 days or so, including last night
eric256 dduncan. definitly not. just wanted to make sure you knew. since it takes quite a while
Juerd rafl: eric256 can get sudo access to dpkg if needed
dduncan smoke takes about 50 mins on my machine 22:21
Juerd I have no idea how sudo would work non-interactively, though
dduncan now that it seems to be precompiling Test and/or using both cpus
before that, it took about 100 mins
eric256 both cpus...show off
dduncan more likely its just the prelude though ...
rafl Juerd: It will. I don't use interactive questions using debconf or somethng like that. Even if I would it could be told to use the defaults and not to ask. 22:22
Juerd: I only need to wait for the parrot mirror..
dduncan last I checked, despite setting number of CPUs to 2 in the config, top only showed a single Pugs running at once
I will recheck that now
right, it is smoking now ... 22:23
and top shows two perl processes but only a single pugs process running
is setting cpus to 2 in the config supposed to do 2 pugs processes? 22:24
svnbot6 r7860 | rafl++ | r19238@ata: rafl | 2005-11-05 23:21:11 +0100
r7860 | rafl++ | * Debian: build-depend on libparrot-dev instead of libparrot$(SOVERSION)-dev to
r7860 | rafl++ | have less work on new parrot releases.
dduncan smoke killed and will restart ... not due to that commit though 22:26
but it will be included
I meant, when smoke_concurrent=2, are there supposed to be 2 pugs processes? 22:29
smoke begun again ...
rafl Juerd: Hm. How can we install packages using sudo non-interactively? 22:31
Juerd rafl: That's for whoever wants to do it, to find out
r0nny_ rafl: sudo can be set up, not to ask for a password
Juerd r0nny_: That won't happen, though. 22:32
rafl Indeed.
Juerd rafl: An alternative is to use a suid binary, catered to a specific need, limited to a specific group.
rafl Juerd: How about allowing eric or me to use only 'dpkg -i' without a pw?
Juerd rafl: I wouldn't know how. If you do, make it happen :) 22:33
rafl User_Alias AUTOBUILDERS = eric256, rafl \n Cmnd_Alias DEBINSTALL = dpkg -i \n AUTOBUILDERS ALL = DEBINSTALL should work 22:39
eric256 if regexs..i mean rules.. build these objects isn't that going to severly slow down the process?
rafl Juerd: Not sure if Cmnd_Aliases can have whitespaces
Juerd: Oh, a NOPASSWD: DEBINSTALL at the end is missing. 22:40
Juerd: Look at /usr/share/doc/sudo/examples/sudoers
svnbot6 r7861 | autrijus++ | * Work around PGE's bug in not grokking zero-width rules 22:48
r7861 | autrijus++ | (reporte to p6[ic]), by special casing that // matches
r7861 | autrijus++ | always at position 0 with length 0. It's useful anyway
r7861 | autrijus++ | because split(//, $string) can be fast without launching
r7861 | autrijus++ | Parrot.
r7861 | autrijus++ | * Also, fix m:g// and m:P5:g// support so they won't loop
r7861 | autrijus++ | on zero-width matches. (Previously s:g// and s:P5:g//
r7861 | autrijus++ | was fixed in the same fashion.)
autrijus (monthly release)++ # forces me to fix parrot interop bugs every month 22:50
Juerd rafl: No, sorry, I'm not really into looking at things right now. If you have root access and can make things happen, please do so. I'll learn what you did when you've done it 22:54
rafl Juerd: OK.
autrijus rafl: svk ls /parrot/mirror/trunk 22:57
of course, you need to cp or ln-s ~autrijus/config to your ~/.svk/
rafl autrijus: Thank you. 23:02
rafl installed parrot Debian packages and removes /usr/local/bin/parrot if there are no objections. 23:04
autrijus: /parrot/mirror/truk 23:07
autrijus: Err. path /parrot/mirror/trunk does not exist. 23:08
D'OH. Typo.
autrijus ...
:)
rafl pugs depend on drift now, doesn't it? 23:24
autrijus no it does not. 23:26
rafl autrijus: This happens on svk pull when using your svk mirror, btw for parrot or pugs: Permission denied: Can't open '/home/autrijus/.svk/parrot/db/revprops/9765.tmp': Permission denied
autrijus rafl: fixing
Juerd autrijus: What do you use tcpxd for?
autrijus Juerd: SubEthaEdit
Juerd ok 23:27
autrijus Juerd: see://perlcabal.org/ works iff that happens
so I leave it running
Juerd I can't run mac osx inside linux for some reason
And my new Linux installation is so shiny and stable that I don't want to reset the uptime :)
autrijus we can do gobby instead :) 23:29
which only requires one single port
much easier to forward
Juerd Oh, I don't mind about the tcpxds
Just wondered
What is gobby?
autrijus subethaedit for gtk that works. 23:30
gobby.0x539.de
?eval 0x539
evalbot_7861 1337
autrijus ;)
Juerd Is it not compatible with see?
autrijus no, because see shuffles serialized objc objects
or was it cocoa objects. 23:31
Juerd Ah
Are there reasons to stick to see?
It's a bit platform limited :)
autrijus well, for one, see has cursor highlighting in scrollbar and edit area 23:32
rafl autrijus: I get this error when running /usr/bin/perl -Iinc util/drift.pl src/Pugs/PIL2.hs-drift > src/Pugs/PIL2.hs:
autrijus for another, gobby/osx is very hard to build
rafl runhugs: Error occurred
ERROR - Unable to open file "/home/rafl/autobuild/pugs/util/../../DrIFT/src/DrIFT.hs"
autrijus rafl: svk up
r7862 should cure that
Juerd autrijus: Ok
autrijus also, permission problem should be fixed now
rafl autrijus: Works, thanks. 23:35
svnbot6 r7862 | autrijus++ | * regen PIL1/PIL2 using latest DriFT.
autrijus cool 23:37
autrijus discovers properly idented AST tree is a good way to inflate journal's length 23:38
(just posted today's)
autrijus praises caffeine-free sugar-free coke 23:40
I wish they have that in taiwan :)
svnbot6 r7863 | autrijus++ | * VICTUALS: note that caffeine-free diet coke is preferred 23:41
r7863 | autrijus++ | over caffeineful ones for me.
geoffb . . . known as 'CFDC' amongst those overwhelmed by so many coke variants (I can currently think of nearly a dozen variants . . . and I may be missing some) 23:43
rafl feather is soo slow currently.. :-( 23:51
autrijus' darcs and eric's ghc are stealing both cpu power and memory from me.. :-) 23:52
svnbot6 r7864 | autrijus++ | * When generating precompiled prelude, strip away comments
r7864 | autrijus++ | and doc sections while preserving number to speed up parsing.
autrijus rafl: renice :) 23:53
rafl I thought about apt-get install slay :-) 23:54
autrijus oh wow, very useful bofh tool
rafl Indeed. One of my favourites..
geoffb autrijus, in r7864 comment, did you mean "while preserving *line* number"? 23:55
autrijus steps away, very slowly, from rafl
geoffb heh
autrijus geoffb: yes, oops
geoffb ok, cool.
autrijus brb, shower 23:57