pugscode.org/ | nopaste: sial.org/pbot/perl6 | pugs: [~] <m oo se> (or rakudo:, kp6:, smop: etc.) || We do Haskell, too | > reverse . show $ foldl1 (*) [1..4] | irclog: irc.pugscode.org/
Set by TimToady on 25 January 2008.
spinclad say vray. 00:03
pugs_svn r21346 | ruoso++ | [smop] YAY! Finally the memory leaks in test/14 are solved. I still need to implement .^!methods().shift and push for the test to pass, but the memory leak being solved is a giant step already. test/15 now only leaks the two char* based Str, but I should work on it now 01:18
pugs_svn r21347 | ruoso++ | [smop] fixed memory leak in test/15. We are all clean again 01:26
ruoso valgrind++ 01:27
ruoso sleep & 01:32
s1n Auzon: if some of the spectest files that are fudged have fudged section that are actually passing, should they not be fudged anymore? the comment indicates it's "unspecced" 03:22
Auzon s1n: all unneeded fudge should be removed. Which file is this? 03:23
s1n range.t
Auzon which one?
There's one in S02 and one in S03
s1n test 4..7 at least, in S02 03:24
in fact, there are many tests that can be unfudged, so i need to setup a t/ checkout with my svn permissions so i can commit the changes
Auzon Sounds great :) 03:25
But yes, they should be unfudged and a comment above them should be listed
s1n how do i keep a manual checkout in my parrot checkout? i tried a symlink but that caused grief
Auzon Try removing parrot/languages/perl6/t/spec and checking out pugs/t/spec 03:27
s1n just the spec folder?
Auzon Yes, because that's all Rakudo checks out in the tests.
and put pugs/t/spec in place of the other one, in case I wasn't clear with that 03:28
s1n so move t/spec out, checkout pugs/t/spec to t/? 03:29
Auzon Yeah.
s1n bleh, i don't know my way around svn well enough, how do i authenticate with it? 03:31
i.e. how do i handle the password?
Auzon It should ask you automatically when committing 03:32
s1n just provide username during co and password during ci?
Auzon both during ci, since you don't need it to co 03:33
It should cache them the first time, which is why I don't remember either ;) 03:34
s1n i'm used to using cvs thru ssh, makes things much easier
okay, so back to the problem, what does it mean that it's "unspecced"? the tests pass but the comment indicates there's no reason for it to 03:36
Auzon Unspecced means that it's not defined in the spec
s1n should i mark it with a TODO or not unfudge it? 03:37
Auzon Go ahead and unfudge but put a # NOTE: .perl on Range is unspecced
s1n is that the format that is typically used "# NOTE: "? 03:38
Auzon # NOTE or # XXX 03:39
unspecced is the important thing.
s1n what's the triple X mean?
Auzon eye catching attention getter
s1n oh heh 03:40
i usually use FIXME or TODO since they tend to get highlighted by most editors
Auzon It's also easy to grep for
s1n i'll go with the XXX, thanks 03:41
Auzon The p6 highlighting on vim catches XXX but not NOTE
I think we prefer XXX
s1n yeah, looks like pmichaud is using that too
Auzon: what's the easiest way to fudge a single test file so I can test just what I changed? 03:45
Auzon You can edit spectest_regression.data to only hold that test and then revert it. 03:46
I usually do this, though:
../../parrot perl6.pbc `~/pugs/t/spec/fudge rakudo testfile` 03:47
Auzon (I'm assuming you're on Linux) 03:47
s1n yeah 03:48
aight, bedtime, i'll finish this up and check it in tomorrow, thanks 03:50
ingy [particle]: hai 05:31
moritz_ Auzon, s1n: echo testfile > t/localtest.data; make localtest 06:44
"testfile" is assumed relative to t/spec/ 06:45
mathr2 the year by year returns given this fund's financial report do not match the annual returns given in the monthly performance report. any idea why this might be? 07:31
moritz_ because the reporting software is b0rked? 07:32
mathr2 that is your best guess?
moritz_ anyway, if it's not written in Perl 6 it's off topic here
mathr2 i can show you the documents if you are interested.
i can show you the documents if it might help. 07:33
moritz_ is any component of your system written in Perl 6? 07:35
mathr2 why were the americans not outraged when they found out that their civil servant ted kennedy was bypassing the system when he had his brain tumor? 07:41
what incentive do men like ted kennedy have to improve the system when they themselves are exempt from it?
araujo ? 07:42
mathr2 if your politicians are not going to utilize the very system they are supposed to be serving,
if they are free to live in a bubble world above yours
then they are not your civil-service politicians but instead your rulers. 07:43
ted kennedy BYPASSED your health system when he had his brain tumor.
why did you americans not FORCE him to utilize the common system? 07:44
HE WOULD HAVE DIED.
At least in the corporate world executives have performance incentive salaries. 07:48
moritz_ politics are also off-topic in #perl6
araujo there is a channel for it
mathr2 They are TIED to their vehicle of employment. They have an INCENTIVE to it.
araujo go there
mathr2 But these Ted Kennedy politicians are out of this world!
That man would have DIED if it used the system he gave to you! 07:49
kane_ hard to argue with a bot...
moritz_ ingy, TimToady: if you are around, could you please ban mathr2?
I think we need more ops here :/ 07:54
pmurias ruoso: hi 07:55
pmurias pugs: &*say.("hi\n"); 08:13
p6eval pugs: OUTPUT[hi␤␤]
pmurias pugs: $*OUT.print("hi\n"); 08:13
p6eval pugs: OUTPUT[hi␤]
pmurias pugs: $*OUT.say("hi");
p6eval pugs: OUTPUT[hi␤]
pugs_svn r21348 | moritz++ | [spec] clean up :todo that's now handled by a fudge 08:45
pmurias moritz_: what kind of info should i put into the pixie HACKING file? 08:55
pmurias pixie: say 3; 08:56
p6eval pixie: No output (you need to produce output to STDOUT)
moritz_ pmurias: structure of the expected AST, a (very) short summary of available syntax, how to test/debug it 08:58
pmurias: and maybe a short note where to start with simple things
pugs_svn r21349 | pmurias++ | [pixie] renamed brew compile
pmurias moritz_: would you be satisified with writing the action in, er, perl5? 09:00
moritz_ pmurias: probably, erm, yes ;-) 09:00
pmurias available syntax means the syntax which elf or pixie parses? 09:01
s/elf/STD_red/ 09:02
moritz_ well, the question is moot if the actions are writtin in perl5 anyway
I meant the Perl 6 subset that I could use for actions - haha ;-)
pmurias moritz_: i'll get working on santiy-04 to rembemer what i blocked on last time 09:08
ruoso pmurias, hi... 09:15
ruoso just passing around before shutting down the computer to go to work...
pmurias, have you seen that I did all the plumbing ;) 09:16
pmurias ruoso: yes 09:36
ruoso: nice work
s/work/job with the plumbing 09:46
pugs_svn r21350 | pmurias++ | [pixie] 09:50
r21350 | pmurias++ | initial HACKING
r21350 | pmurias++ | fixed to work with refactored elfX
pmurias moritz_: pixie should now work 09:52
moritz_ pmurias: OK, I'll take a look 09:53
pmurias moritz_: re simple things you can choose a simple bit of syntax, and backtrack if it gets to complicated ;)
moritz_ pmurias: that's exactly the answer I didn't want ;-) 09:54
pmurias s/to/too/
moritz_ for a beginner it's hard to know which features can be implemented without having to implement a whole bunch of other features first 09:55
pmurias you can just swith to the other features
* switch
;)
pmurias moritz_: it's hard to judge it for me to 09:56
moritz_ that's a better answer, albeit not very helpful either ;-) 09:57
pmurias but you can tell me what you want to chose, and i'll see how hard it is
moritz_ ok, I'll do, but I don't know if I'll find tuits today 09:58
I should really be learning stuff for my exams :(
pmurias physics? 09:59
moritz_ yes
araujo moritz_, just remember everything that goes up, has to come down 10:00
moritz_ ? 10:01
ruoso off to work & 10:17
pmurias ruoso: why can't REFERENCE and RELEASE be handled by MESSAGE? 14:31
ruoso hmmm... I thought that as it would be called all the time, it would be nice to have it as part of the responder interface 14:31
pmurias so it's an optimalisation 14:32
?
ruoso yes... it could be made inside MESSAGE as well... but it looked too pessimistic to have to do an additional compairision instead of just invoking a function pointe 14:34
pmurias it can always changed when our method calls are as faster than function pointer invocation ;) 14:41
ruoso hehe 14:48
pugs_svn r21351 | moritz++ | [irclog] removed obsolete Net::IRC based bot, and updated README a tiny bit 16:44
pmurias * it can always be changed when our method calls are faster the function pointer invocation 17:05
pmurias wonders why his english ends up so horrible on irc 17:06
spinclad optimize them enough and it's bound to happen
(both english and faster calls)
moritz_ lol
pmurias TimToady: can i add a use v6 to Cursor.pm and STD.pm so nicer error are thrown if perl5 attempts to run them? 17:55
TimToady um Cursor.pm is p5 17:57
actually, there is no Cursor.pm right now
or shouldn't be 17:58
that Cursor.pm is way obsolete
pmurias there is one for sure in the repo i'm using
TimToady STD.pm shouldn't need it because it starts out with 'grammar', but if you add one, be sure to make gimme5 remove it. :) 17:59
it's ancient
and wrong it any number of ways
we need to backtranslate Cursor.pmc to a P6 Cursor.pm that can produce Cursor.pmc via p6-to-p5 translation 18:00
pugs_svn r21352 | pmurias++ | [STD] another [@list] hack to make chaining operators produce proper ast
pmurias TimToady: we need a proper p6-to-p5 translator 18:01
TimToady indeed, that's the next step; replace gimme5 with a STD-based translator
well, not replace, but overlay 18:02
since we need to keep gimme5 for bootstrapping
also need to do some refactoring of namespaces so that "use Perl" means use the language, not the grammar 18:03
maybe "use STD::Perl" pulls in the grammar
or maybe that's backwards...
anyway, we need to distinguish the operations from the standpoint of the user 18:04
they can't both be "use Perl".
pmurias it's the same issue with use Python, and other language changing modules 18:05
TimToady indeed
which is why we need to figure out the policy now
possibly even by having a "use" variant that does the other thing, whichever that is 18:06
moritz_ use Perl :syntax; ?
TimToady blech
I mean a different keyword even 18:07
[particle] like 'load'
TimToady lang Perl;
[particle] lang++
TimToady except
that ordinary "use" is also expected to mutate langauges
*languages
so it seems like "use Perl" should be a mutator, and the derives from could perhaps just be intuited from the "is" declarations 18:08
so "grammar STD::Perl is STD::Cursor" would automatically pull in STD::Cursor without a use, maybe 18:09
moritz_ do we have `require' in perl 6?
TimToady though we had trouble with "use base" because of htat
yes
moritz_ why not just make require Perl; load the Perl module, and 'use Perl;' be the mutator? 18:10
pmurias require works at runtime
moritz_ that's a good reason ;) 18:11
TimToady the problem with auto-using or auto-requiring is that you don't know if there's really a module of that name out there or some other module is expected to define the base class, which is why "use base" fails silently if it can't require 18:14
on the other hand, class names used in Perl 6 are generally required to be defined first 18:15
so maybe we can just autouse the class if it's not predefined
pmurias we can always use "use Perl::Grammar"
TimToady that's not right
"Grammar" should be top level 18:16
otherwise what is the name of Grammar::Perl::Regex?
or STD::Perl::Regex, or whatever we end up with
(I like the psychology of using STD for the standard language, but it doesn't generalize) 18:17
pmurias use STD::Python?
TimToady if it *is* the STD Python, that's fine :)
but I think we'll have people fighting over the STD space in that case
TimToady (though only if they implement Python in the Perl namespace :) 18:18
since otherwise it's just "use Python" as a black box
pmurias also we will likely also have a Perl grammar for using in regexes, for matching if something is in Perl 18:20
TimToady if we have other languages *choosing* to implement their parsers in STD space, then maybe it really is a Sexually Transmitted Disease :)
pmurias: I don't follow that
are you referring to the language inside regex closures? 18:21
that's just an anonymous language to the regex, pulled out of $+LANG
pmurias something like: if /<Perl.comp_unit>|<Python.comp_unit>/ { say '$_ is in either Perl or Python' } 18:22
spinclad ( 'use Perl <Grammar Actions>' ? ) 18:23
TimToady currently written something like / [:lang(Perl) <comp_unit>] | [:lang(Python) <comp_unit>] / 18:24
pmurias the current STD would choke on syntax errors if run on something which is not a perl string
TimToady no, "use Perl" should mean exclusively that you want to use the language, not the grammar, I think
you've just said that a parser will choke on anything it doesn't recognize, or did you mean something more? 18:25
spinclad so 'use Grammar::Intercal' and 'use Grammar::Intercal::Actions' ?
s/Grammar/STD/ 18:26
:)
pmurias TimToady: yes
will normal regexes would just fail 18:27
TimToady what's "normal"?
pmurias s/will/while/ 18:28
TimToady currently STD would handle the "normal" regex above if you define the Python grammar
[:lang(Perl) <comp_unit>] | [:lang(Python) <comp_unit>] /
that's why I put slashes around it
but it'll probably end up :lang(STD::Perl) instead 18:29
next question is whether it's STD::Perl::Regex or just STD::Regex 18:30
since "STD" already implies a P6 namespace
pmurias want the perl6 comp_unit die when it encouters a syntax error
TimToady then other languages could use STD::Regex without having to type the dreaded P word
pmurias * won't 18:31
TimToady sure, but after a :lang(STD::Python) it won't be calling perl 6's <comp_unit>
Auzon perl6: my $tracker = 0; for (1..5) { next if 2 < $_ < 4; $tracker = $_;}; say $tracker;
p6eval elf 21352: OUTPUT[Undefined subroutine &GLOBAL::next called at (eval 121) line 6.␤ at ./elf_f line 3861␤]
..pugs: OUTPUT[5␤]
..rakudo 29525: OUTPUT[Could not find non-existent sub if␤current instr.: '_block25' pc 207 (EVAL_14:66)␤]
moritz_ rakudo does neither "next" nor "if" as an statement modifier 18:32
Auzon The test says it should be 3, and I don't see why.
moritz_ uhm, shouldn't it be 5? 18:33
Auzon Yeah. I agree.
pmurias by normal i meant like things in Regex::Common, which just fail if the don't match something instead of throwing an error
TimToady perhaps it should be unless
Auzon TimToady: That's what the comments seem to indicate
moritz_ or just '$tracker = $_ if 2 < $_ < 4' 18:34
then it's clear without any break() logic
Auzon It's specifically testing next.
moritz_ s/break/next/
oh
ok ;-)
Auzon t/statements/next.t ;)
Sorry, didn't give you context there
moritz_ thought it was a test for chained comparison ;-)
somehow I don't think it's a very good test at all ;-) 18:36
TimToady pmurias: so perhaps the relationship of comp_unit with panic is special
Auzon moritz_: It's the first linked test below "Loop Statements" in S04: perlcabal.org/syn/S04.html#Loop_statements
lambdabot Title: S04
TimToady rather than scoping the parse failure to //, scope it to the comp_unit 18:37
I'll have to think about that
ordinarily you might just write / [:lang(STD::Perl) <comp_unit>] / or / [:lang(STD::Python) <comp_unit>] / currently 18:38
but the | form would do longest token matching across both languages :)
pmurias don't i have to wrap it in try {} to catch die's 18:39
TimToady the p5 impl of STD is actually using eval {} and die to do backtracking control 18:40
so it's merely a matter of trapping it correctly
it pays to know the scope of one's failure :)
moritz_ or the scope of one's death ;-) 18:41
TimToady well, in this case, the scope of what you're trying to kill 18:42
fortunately, "close" only counts in horseshoes and hand grenades
unfortunately, nukes are a type of hand grenade
and at the moment, <panic> is more of a nuke... 18:43
TimToady we need something more like an Orion Drive 18:43
rindolf Hi all.
Hi TimToady
pmurias hi rindolf 18:44
rindolf Hi moritz_
Hi pmurias
pmurias: I biked today.
Was kinda late.
moritz_
.oO( chainsaw fail() ? )
rindolf But I saw a group of Brazilians in the end of the Park playing and singing.
They rocked.
Wish I could record them.
pmurias rindolf: don't you carry your cellphone around ;) 18:50
?
rindolf pmurias: I do. 18:51
pugs_svn r21353 | Auzon++ | [gsoc_spectest] moved next.t into spec. Fixed a test case in it.
pmurias TimToady: will the perl6->perl5 translator go through an AST stage or directly to perl5? 18:54
TimToady pmurias: well, even gimme5 uses more of an AST approach for the regex parsing, since it blesses nodes that can have methods called on them 19:11
rindolf pmurias: I thought about calling my mom. 19:15
pmurias: but recording using it would probably be awful quality.
And not sure it has a recording.
I heard about a girl who got arrested for photographing the Transformers movie using her cellphone. 19:16
Do they really consider it copyrights-infringement?
moritz_ #perl6-soc in 1/4 h 19:20
spinclad no, they really consider it off-topic... 19:22
but i've been there myself now and then 19:23
Auzon .oO(testing an infinite loop? Seems like a bad idea to me)
moritz_ Auzon: we'll might to have to test for infinite junctions soon :/ 19:24
Auzon How can they be infinite? any(1 .. *) ? 19:25
moritz_ Auzon: see p6l, thread "Complex planes" 19:26
Auzon: and yes, through being lazy (and perhaps special cased) 19:27
Auzon Ah, ok
Makes sense :)
pugs_svn r21354 | Auzon++ | [gsoc_spectest] smartlink cleanup 19:28
spinclad 2pi * (any(Int) + 1/4) 19:29
suggests the extension any(Type) 19:30
(s/2pi/2*pi/, of course) 19:31
Auzon What would that stringify or numify to?
masak liked 2pi :)
Auzon masak: my constant 2pi = 2 * pi; # ;)
masak hehe 19:32
Auzon Or define pi as a postfix operator
spinclad postfix:<pi>
moritz_ spinclad: any(Type) might not be computable for subsets
pmurias rindolf: intellectual property is arbitrary, but i find it hard to imagine her facing serious consequences 19:35
spinclad in some uses, it might not need to be generable, only matchable: is(acos(1), 2pi * any(Int)) perhaps
rindolf pmurias: yes.
pmurias: but it's ridiculous.
pmurias it's hard to destinguish a photograph from a recording 19:36
spinclad but that (matching 2pi * any(Int)) is of course numerically ill-defined (what epsilon?) 19:38
s/what epsilon/to within &/
masak it's r29526 on my box, and I can't get rakudo running 19:48
can anyone confirm this? 19:49
running ./perl6, I get "Class 'Perl6Sub' already registered!"
moritz_ tries
../../parrt perl6.pbc works fine, and ./perl6 segfaults 19:50
Auzon masak: That usually means you need to rebuild parrot.
masak :/
I just did
moritz_ with a "make realclean"?
Auzon with make realclean?
moritz_ hehe ;)
masak moritz_: no. I'll try that
Auzon: no. I'll try that :) 19:51
everything works now -- thanks 19:55
ok, say I want to read a line from STDIN. do I use =<> ? 19:56
my $a = =<>; # gives me a strange "Lexical '$expr' not found" error 19:57
masak is rakudo in error here, or should I go read up on a given Sxx ? 19:59
moritz_ my $a = =$*IN 20:00
masak ah.
moritz_ <> is an empty < ... > list, I think
but I'm not sure
or $*IN.readline
masak my $a = =$*IN # gave me the same error, plus a segfault :/ 20:01
speckbot r14563 | larry++ | [S04] another whack at defining consistent closure semantics
masak ./perl6 -e 'say "hello world"' # gives same error, segfaults
something is in a state of relative b0rkenness here. 20:02
moritz_ open a ticket, if you feel like
masak I feel like. hold on.
./perl6 -e '1' # error, segfaults 20:03
./perl6 -e '' # enters REPL mode!
now I have two distinct bugs to report :)
moritz_ heh, ../../parrot perl6.pbc -e 0 also goes in REPL mode 20:04
masak perl 5 doesn't
moritz_ so it seems to do a perl5-like truthness test on the argument to -e
masak truthness test? 20:06
seems to me it checks the length of the argument after -e
moritz_ if ($ARGV[-1]) { .... } 20:07
false for '', '0'
true for "foo", "1", "0;" etc.
masak ah 20:08
would `if (defined $ARGV[-1])` help? 20:09
ah, EOBATTERY. seeya. 20:13
moritz_ (just for backlog: not quite, unless the '-e' is stripped from @ARGV first) 20:17
pmurias TimToady: can i delete expect_tight_infix? 20:21
spinclad r14563++ # much clearer 20:34
TimToady pmurias: yes, you can delete it 20:39
pugs_svn r21355 | pmurias++ | [STD] remove expect_tighter_infix 21:08
speckbot r14564 | larry++ | typo from Brandon++ 22:58

Powered by Cro and the Raku® Programming Language.

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

Thank you!