-Ofun: xrl.us/hxhk | pugscode.org | pugs.kwiki.org | paste: paste.lisp.org/new/perl6 or sial.org/pbot/perl6
Set by apple-gunkies on 11 November 2005.
luqui ?eval sub foo($bar?) { say $bar } foo() 01:09
evalbot_7938 bool::true
luqui ?eval sub foo(?$bar) { say $bar } foo() 01:10
evalbot_7938 Error: unexpected "?" expecting formal parameter, ":" or ")"
luqui sweet
?eval sub foo(:$bar) { say $bar } foo bar => 42 01:11
evalbot_7938 bar42 bool::true
luqui ?eval sub foo(:quux($bar)) { say $bar } foo quux => 42
evalbot_7938 Error: unexpected "q" expecting space or ")"
svnbot6 r7939 | Darren_Duncan++ | r811@Darren-Duncans-Computer: darrenduncan | 2005-11-12 17:12:00 -0800 01:15
r7939 | Darren_Duncan++ | added new utilty script util/add-test-file.sh that combines 'svn/k add foo' and 'add-svn-props.sh' into one convenient step, good for mass text file adds
Juerd The postfix ? in signatures is blessed now?
Hurrah!
luqui is rather happy about that too 01:16
dduncan the script is actually add-text-file; the commit message is typod 01:17
Juerd I wonder if the + to ! change is official too
dduncan and it was tested by adding itself
Juerd Hoping pugs is a good reference,
?eval sub foo (:$bar!) { say $bar } foo()
luqui yeah, postfix ! for mandatory
evalbot_7939 Error: No compatible subroutine found: "&foo"
luqui is blessed :-)
dduncan I like the postfix ! for mandatory 01:18
Juerd Neat!
luqui ?eval sub foo (:$bar!) { say $bar } foo(bar => 42)
evalbot_7939 42 bool::true
Juerd That's my first, ehm, signature symbol, I think it's called. Do these decorators have a name?
luqui thinks that the "no compatible subroutine found" error must be given a better message
luqui goes off to fix that
dduncan now, if only saying that would result in 42 ending up in $bar
Juerd luqui++ # fixing
dduncan I wonder ... 01:19
luqui dduncan, what do you mean?
dduncan just a minute ...
?eval sub foo (:$bar!) { say $bar } foo('bar' => 42)
evalbot_7939 42 bool::true
Juerd dduncan: Note that the first 42 is the result of "say", the second is that of foo
dduncan I had a problem with this yesterday, have to debug 01:20
Juerd luqui: Can you change the evalbot so that it does something to separate output from return value? The two being next to eachother is often confusing, especially as people use "say" a lot, not expecting the return value to be listed too. 01:21
Is the bot in the pugs repository, by the way?
theorbtwo I'm not sure what people's excuse for redundant says. Juerd, yes, somewhere under examples.
Juerd theorbtwo: "say" is very useful, as it allows you to output anything at any point, without considering call/return stacks. 01:22
dduncan the problem I had yesterday concerned "submethod BUILD" having named arguments
Juerd dduncan: Note that BUILD is called by, not an alias for, new.
dduncan yes
theorbtwo Yeah, but it gets used a lot in examples when it's completely redundant... it's probably actually because people are so used to perl -le '...' 01:23
Juerd theorbtwo: Possibly.
Still, I think both are useful.
theorbtwo Absolutely.
Juerd A little separation would probably solve this.
dduncan if BUILD is working for other people, I'll try debugging mine now
theorbtwo It'd be nice if it printed Output: ... Returned: ...
luqui Juerd, I'll look at it
theorbtwo (Or Output: ... Died: ...)
?eval die "Foo!" 01:24
evalbot_7939 Error: Foo!
Juerd theorbtwo: ... I was just typing something very similar.
theorbtwo Close enough.
Hm, though, I wonder...
?eval say "Output"; die "Error"
evalbot_7939 Output Error: Error
Juerd theorbtwo: Still, even a simple separator char would work. And be easier to implement, probably :)
Or something circumfix
Output [[Error: Error]] 01:25
Output [[4]]
(Double brackets to avoid confusion with the often used .perl)
luqui what if I omitted the return value if there was output?
theorbtwo Nah, sometimes you want to see both. 01:26
luqui then maybe you put an extra say there?
Juerd luqui: I think that the return value can be useful even if it wasn't expected to be.
dduncan the situation is that, following an invocation of Locale::KeyedText::Message.new( 'msg_key' => 'MYAPP_HELLO' ) ...
Juerd In my bash shell, I have $? as part of my prompt.
luqui fair enough
dduncan spitting out the value of the msg_key property displays both the pair key and value
Juerd Most of the time, I don't think about it. Sometimes, seeing a non-0 there warns me something was wrong, even though the output was normal.
dduncan now I'm trying to track down why the pair value wasn't extracted in BUILD 01:27
Juerd You've probably noticed this in feather's default $PS1 ;)
dduncan the signiture is ... submethod BUILD (Str :$msg_key!, Any :%msg_vars? = hash())
Juerd dduncan: How does new call BUILD? Does it use *? 01:28
wolverian is that = hash() redundant?
dduncan I use the default new(), and don't declare my own
theorbtwo Somebody remind me what the : there does?
luqui ?eval my %foo = hash(a => 1, b => 2); say %foo.keys
evalbot_7939 Error: Unknown syntactic construct: Syn "named" [Val (VStr "a"),Pos (MkPos "<eval>" 1 21 1 22) (Val (VInt 1))]
luqui ?eval my %foo = hash('a', 1, 'b', 2); say %foo.keys
evalbot_7939 ab bool::true
Juerd (Should it? How does one properly call another sub/method using the same arguments, assuming compatible signatures? Do we have argument lists in variables?)
dduncan right now, the issue is with the msg_key, a scalar 01:29
wolverian Juerd, yeah, we do.
Juerd theorbtwo: :$foo in a signature means it is a named-only parameter.
theorbtwo: The argument will be in $foo.
theorbtwo: The name is 'foo'
theorbtwo OK. 01:30
":" is named, ? or ! is required or optional.
Juerd theorbtwo: :foo($bar) means it is a named-only parameter; the argument will be in $bar; the name is still 'foo'.
theorbtwo IWFM.
Juerd theorbtwo: Yes.
wolverian Juerd, see 'Argument list binding' in S06
if that's what you meant. I'm really tired.
good night :) 01:31
Juerd theorbtwo: Positional defaults to mandatory, named to optional. Anything with a specified default value is also automatically optional.
wolverian: Good night.
theorbtwo I ought to read the changelogs of the S*, if not reread all of the S*es. 01:32
dduncan ?eval class Foo; submethod BUILD (:$bar) { say $bar; } Foo.new('bar'=>'baz');
evalbot_7939 barbaz Foo.new(("bar" => "baz"),);
Juerd All mandatory positionals must come before optional positionals, but this should be entirely natural. If not, please experiment with alternatives ;)
dduncan so, there's the bug
'barbaz' was spit out, not 'baz' 01:33
Juerd dduncan: Write a test.
dduncan okay
Juerd And one day, either you will find your test removed or changed, which would mean the bug is on your side, or the bug is on pugs's side, and you will find it fixed :) 01:34
dduncan does my code look wrong?
theorbtwo You might want to comment the test, but don't comment it out.
Juerd dduncan: I dunno. 01:35
dduncan: I find the new/BUILD thing unnecessarily complicated.
luqui yeah
dduncan I think its designed to make class inheritence easier 01:36
Juerd fails to see the need for separation.
Or the need for submethods, for that matter...
luqui but I think it has failed for that
Juerd This may be my ignorance.
luqui since usually a subclass wants to control the parameters to the superclass, not the person creating the object
dduncan one can write a manual new() instead and call bless(), such as if you want the constructor to take positional args, like in perl 5
Juerd dduncan: I'm usually more interested in the default (thus implicitly preferred or even recommended) way of doing things than in what is possible. 01:37
A lot is possible, and certainly enough is
dduncan I think the advantage of the BUILD approach is that if a subclass doesn't want to know about all the optional args a superclass constructor takes, it doesn't have to 01:38
Juerd But the language is not defined by what it can do, but by what it does.
dduncan various other advantages discussed at length elsewhere
Juerd Creating a dialect does not make the base language any better :)
dduncan suffice it to say that using BUILD is fairly standardish now, and what I'm trying to use
luqui thinks it is worth arguing 01:39
Juerd doesn't even know which things should go into new and which into BUILD.
dduncan I do know
Juerd I've never cared, and the names aren't explaining anything to me.
So I'd have to read documentation.
dduncan BUILD is what bless() calls
Juerd Which I haven't done yet.
dduncan after bless creates an object, BUILD sets its attributes 01:40
new() is called before bless
Juerd We still set attributes by executing code?
Why not has $.foo = "initial value"?
luqui Juerd, that is allowed
just not implemente
d
dduncan it is possible that you may want to do several things at init time in response to the same input or combination thereof 01:41
Juerd And why would attributes not already be available in new? Would they have to be class attributes?
luqui it turns out that attributes are not available in BUILD
dduncan object attrs aren't available in new, because the object hasn't been made yet
luqui defaults are set at the *end* of BUILD
for some stupid reason
Juerd luqui: That appears counter productive :)
luqui yep
dduncan attrs are available in BUILD, to be set
when BUILD is called, you have a new object whose attrs are all undefined, and you set them 01:42
luqui yeah
it needs a rehash
we need to simplify and optimize for common cases
Complex is a good common case
Juerd dduncan: So what? We have hypothetical variables, all kinds of implicit declaration, and even interesting forms of undef, that already know what class they belong to (think my Dog $fikkie .= new). This should be possible.
dduncan its like if in perl 5, BUILD is called in this context: sub new { my $obj = bless {}, shift; $obj->BUILD(@_); } ... thereabouts 01:44
BUILD starts from the point that the hash-replacement is blessed, when one starts to set its attrs 01:45
I like this approach because I can then use my invocant as the object even in the constructor, and not just all the other methods or destructor 01:46
luqui we know how it works. *why* is what we're contesting
Juerd Okay, fine that that is what it is, but I still don't like it or see why it's needed.
dduncan what do you propose instead ... the perl 5 way?
luqui I think BUILD is okay
Juerd luqui: Actually, I didn't know how it works :)
luqui I don't like that the parameters are forced to be named
I don't like that defaults are set at the end
I've had to work around it once or twice 01:47
and I have reasons for those
dduncan last I checked, defaults were actually set at the start of BUILD, not the end ... but Class::Std sets them at the end for perl 5 folks
Khisanth change it before it's too late! :)
luqui but I think it would be more productive to p6l them
Juerd Wouldn't it make a lot of sense to have new and BUILD have exactly the same signature?
dduncan at least that's how I read A6
new() defaults to having the same signiture as BUILD, the same as bless, if you don't declare a new() but you do declare a BUILD ... if you declare a new(), it has what you declare instead 01:48
and bless takes only named args
luqui that would be nice 01:49
Juerd I see 01:50
How does it do that? :)
Is there a syntax to take another sub's signature?
dduncan if you declare neither, then both default to taking named args matching your public attributes
luqui Juerd, arg tuple :-) 01:51
dduncan er, new() does
Juerd luqui: What syntax do you imagine in declarations?
luqui sub new (\$args) { .... $obj.BUILD(*$args) }
dduncan in my case, I make all my attributes private, and so I have to define explicit args in my constructor
I think
Juerd Aha, backslash.
wolverian Juerd, I pointed you at the relevant S06 section :)
Juerd wolverian: It's almost 3am, I'm in bed, tired, really can't stand any white background :) 01:52
wolverian: So I'm postponing any serious reading until tomorrow :)
wolverian excuse accepted! :) now I'm really going to fall asleep, as well.
Juerd In fact, I was just going to wish you all a good night
Good night :)
dduncan good night
wolverian good night!
dduncan if anyone needs another analogy, BUILD() in perl 6 is what initialize() generically means in any language 01:54
luqui the philosophy isn't the problem, it's the details
Juerd dduncan: Then the name "BUILD" would be wrong, imo.
Does it actually build anything?
Or does it initialize? 01:55
But I'm sleeping, really
afk
dduncan well, "build" is generic enough to include initialize
in the english sense of the word
theorbtwo [01:38] [theorbtwo(+i)] [4:#perl6(+n)] [Act: 1,5] 01:57
Sorry. 01:58
azuroth unrelated, but...
anyone think this warning might be causing a segfault? ;-p "/usr/bin/ld: warning: libstdc++.so.6, needed by /usr/lib/gcc-lib/i486-linux-gnu/3.3.6/../../../libpqxx.so, may conflict with libstdc++.so.5" 01:59
or can I make it un-link to stdc++.so.5 ? 02:00
luqui ? 02:01
no idea
azuroth I probably ought to be asking in #c++ :-)
luqui ?eval 02:28
?eval ()
evalbot_7939 ()
svnbot6 r7940 | luqui++ | Minor spelling/grammatical fixes. 02:39
r7941 | luqui++ | Made evalbot distinguish output from return value.
luqui ?eval 42 02:48
evalbot_7941 42
luqui ?eval say "hi"; 42
evalbot_7941 OUTPUT[hi ] 42
luqui it's ugly, but it'll do
azuroth ?eval sub foo() { say "blah"; 1;} say "a"; foo(); 02:49
evalbot_7941 OUTPUT[a blah ] 1
azuroth woot
luqui do newlines work 02:51
in irc?
oh, not really
azuroth they do in jabber :-)
rafl I'd also prefer using a jabber MUC over IRC. 02:55
luqui ?eval "it's a test bot" 03:04
luquivb "it\'s a test bot"
evalbot_7941 "it\'s a test bot"
luqui ?eval "it's a test bot, seriously" 03:05
03:05 luquivb is now known as luquivb_2
luquivb_2 "it\'s a test bot, seriously" 03:05
evalbot_7941 "it\'s a test bot, seriously"
luqui good
svnbot6 r7942 | luqui++ | Lazy nick resetting. 03:09
luqui ?eval "\$quotemeta?" 03:12
evalbot_7941 "$quotemeta?"
luqui hmm, that's a bug too
?eval "update yourself!" 03:13
03:13 evalbot_7941 is now known as evalbot_7942
evalbot_7942 "update yourself!" 03:13
luqui woohoo
azuroth that's cool 03:14
dduncan "update yourself!" 03:15
?eval "update yourself!"
evalbot_7942 "update yourself!"
dduncan oh well, timing not so good
luqui it just checks to see if it has been updated when you eval something
it updates every five minutes
that was the best I could do without getting into yucky ipc stuff 03:16
dduncan well, it won't break anything, given that it's a string literal
luqui hmm? 03:17
dduncan except that this is an interpolating string, so presumably any interpolation is done first, then if the result is whatever it does whatever
luqui right
fixing
ugh, this is more difficult than I though 03:32
t
eric256_ ?eval say "hello"; 03:38
evalbot_7942 OUTPUT[hello ] bool::true
eric256_ just wanted to see the new output ;)
Khisanth where did the extra space come from? 03:52
azuroth they're \ns I think
svnbot6 r7943 | luqui++ | Fixed '$foo'.perl (wasn't properly escaping the $). 03:58
r7943 | luqui++ | In the process, I blessed and implemented "\d[123]".
luqui ?eval "\$quotemeta" 04:06
04:06 evalbot_7942 is now known as evalbot_7943
evalbot_7943 "\$quotemeta" 04:06
luqui ?eval "\$quotemeta\{curly\}" 04:07
evalbot_7943 "\$quotemeta\{curly}"
luqui excellent
gaal autri, hello. a few questions for when you wake up :) 05:26
first, I was adding smartmatch against closures, but couldn't get literal when { } { } to work -- any idea why adding ruleBlockLiteral as an option to ruleCondPart (or a similar rule intended just for "when" didn't do the trick)? 05:27
svnbot6 r7944 | Darren_Duncan++ | r811@Darren-Duncans-Computer: darrenduncan | 2005-11-12 17:12:00 -0800
r7944 | Darren_Duncan++ | added new utilty script util/add-test-file.sh that combines 'svn/k add foo' and 'add-svn-props.sh' into one convenient step, good for mass text file adds
r7945 | Darren_Duncan++ | r818@Darren-Duncans-Computer: darrenduncan | 2005-11-12 21:14:34 -0800
r7945 | Darren_Duncan++ | added new test t/oo/named_parameters_in_BUILD.t about that BUILD() receives declared-named argument values correctly
gaal second, more genrally, i was looking for ways to increase my Parsec fu by tracing it -- not just add trace calls, but maybe put the whole thing in a debugger. couldn't find one that works though -- any ideas about that? #haskell sady was no help, but perhaps i should poll again. 05:28
luqui gaal, re: when, it's possible that we'll disallow that form anyway 05:30
in favor of when ({ }) { }
dduncan clkao, it appears that svk again sent the same commit to the pugs main server twice ... this time the pugs commits 7939 and 7944 are actually the same one, which (as with the first such error) involved adding a new file 05:32
this time, however, I have a partial terminal log that I will nopaste ...
gaal third, I'm going through TaPL now. I'm very pleased to find I can read the OCaml there. when you started pugs, did you consider OCaml? Without pugs today, I would have assumed OCaml to be more suitable than Haskell -- did you already know about the fancy modern Haskell stuff when you started off, or did you not worry about getting to need any of it? :-) (this last Q I need for the article)
dduncan that shows what happened from the moment of my second actual commit of the day, named_param... 05:33
pasteling "dduncan" at 24.108.164.7 pasted "client-side details of what happened when same commit pushed to pugs mainserver a second time" (44 lines, 2.6K) at sial.org/pbot/14266
gaal luqui: ACK. in fact ruleCondPart has a maybeParens in it so that should have worked but doesn't 05:34
dduncan seen clkao 05:36
jabbot dduncan: clkao was seen 1 days 5 hours 38 minutes 49 seconds ago
gaal luqui: if you're already here -- do you have an idea for the semantics of has = default? for example:
if there's both a has-default and a BUILD routine, does the default get installed before the routine is called? 05:37
if there's a has-default, and a BUILD routine that specifies an optional value for the same attribute, I guess that's a compilation error 05:38
..back on the when {} {} thread, of course I want to fix this but it's more important to me in the long shot to have a better understanding of parsec. ideally i'd like an AOPish tool that at runtime can give me a trace of productions tried, like RD_TRACE (I think it is? in P::RD). 05:42
or something along the lines of this: www.cs.kent.ac.uk/people/rpg/tatd2/goodFib.pdf
It's pretty darn impressive we've gotten this far without a debugger! haskell++ I suppose 05:43
luqui gaal, unfortunately, it is specced so that the defaults are initialized *after* execution of BUILD 05:50
there was some reason for this, but it wasn't a very good one :-)
gaal can't the spec be updated? 05:52
the case where I say ...new(something => undef) should probably also work (that is, something really should be undefined, not get the default) 05:53
dduncan considering that undef is a legitimate value, perhaps the test as to whether to apply to a default is if the name key 'exists' rather than if its value is 'defined' 05:56
exists in the argument list
gaal dduncan: yes, that's the idea. problem is if you apply has-defaults AFTER BUILD, it may be tricky to go back and look at the argument list. 05:57
dduncan if that happened, then you could correctly set to undef if defaults are applied after
I don't see why the argument list would have expired ... it should still be around for the has-defaults to look at
since that happens before the bless call returned 05:58
gaal well, i was just pointing out a possible corner case to nail down. it may have been obvious. 06:00
$work &
dduncan in any event, I sort of agree that defaults should be applied before BUILD 06:01
or I'm not actually sure what should happen
for now, I just set all my attributes in BUILD explicitly, whether from arguments or defaults
and don't define defaults in any other place 06:02
gaal well, I also wonder if has $.member = { somecode } should assign a closure or perhaps run the thunk on instantiaition 07:35
probably it makes more sense for it to be consistent and assign a closure. 07:36
Juerd gaal: Assign a closure, because that's what's written 11:15
gaal: But wasn't there something like "has $foo will init { ... };"? 11:16
12:02 theorbtw1 is now known as theorbtwo
gaal Juerd: where did you see that? 12:12
masak gaal: something like that is hinted at search.cpan.org/dist/Perl6-Bible/li...ure_traits 12:17
gaal masak: thanks. (and hi :-) 12:19
"will first" ?
in any case: this was not explicitly for instance data, but rather for my/our vars. 12:28
seen brentdax 12:32
jabbot gaal: brentdax was seen 30 days 7 hours 38 minutes 18 seconds ago
gaal darn
r0nny yo 12:34
azuroth hey
gaal ho
12:37 lampus_ is now known as lampus
Juerd gaal: I have no idea. 13:07
qwm so, when will we see an ISO standard of Perl 6? 13:08
Juerd Were old style +$foo named parameters optional or mandatory by default? 13:12
qwm: Hopefully never, because ISO standards are expensive.
scook0 Juerd: optional IIRC; mandatory was ++$foo 13:14
Juerd Ah, thanks. 13:15
sub foo ($pma, $pmb, ?$poa, ?$pob, ++$nma, ++$nmb, +$noa, +$nob) { ... } 13:17
sub foo ($pma, $pmb, ?$poa, ?$pob, +:$nma, +:$nmb, :$noa, :$nob) { ... }
sub foo ($pma, $pmb, ?$poa, ?$pob, +:$nma, +:$noa, :$noa, :$nob) { ... }
sub foo ($pma, $pmb, $poa?, $pob?, +:$nma, +:$noa, :$noa, :$nob) { ... }
sub foo ($pma, $pmb, $poa?, $pob?, :$nma!, :$noa!, :$noa, :$nob) { ... }
Is that the correct order in which things happened?
p => positional, n => named, m => mandatory, o => optional, a => first, b => second 13:18
Ohm, there are some mandatory named arguments mistakenly labelled 'o' 13:19
s/Ohm/Oh/
rafl is :noa! really named optional a?
Juerd No, that's one of the mislabeled. 13:20
sub foo ($pma, $pmb, ?$poa, ?$pob, ++$nma, ++$nmb, +$noa, +$nob) { ... }
sub foo ($pma, $pmb, ?$poa, ?$pob, +:$nma, +:$nmb, :$noa, :$nob) { ... }
sub foo ($pma, $pmb, ?$poa, ?$pob, +:$nma, +:$nma, :$noa, :$nob) { ... }
sub foo ($pma, $pmb, $poa?, $pob?, +:$nma, +:$nma, :$noa, :$nob) { ... }
sub foo ($pma, $pmb, $poa?, $pob?, :$nma!, :$nma!, :$noa, :$nob) { ... }
rafl OK then.
:$nma! two times for the last three lines? should be :$nmb!, I guess. 13:21
Juerd You're right
I think my IQ was divided by 2 last night.
rafl What did you do last night? 13:22
Juerd Stubbornly not sleep, while being exhausted.
liz6 wonders whether chanting "OM" to Juerd would help ;-) 13:25
Juerd OM?
rafl Juerd: Unfortunately I did the same. :-) 13:26
Juerd As in "het roer om", "Openbaar Ministerie", or something else?
liz6 en.wikipedia.org/wiki/Aum
Juerd Ah
Don't think it'd help :) 13:27
rafl: Then perhaps your IQ is also half of what it was before.
rafl Juerd: I already work about two hours to fix a segfault in some perl bindings. You're probably right. :-) 13:29
Juerd :) 13:30
svnbot6 r7946 | iblech++ | * Changes to accomodate to new (r6582) S02, S03, and S06 in src/, ext/, t/, 13:38
r7946 | iblech++ | perl5/, docs/ and examples/:
r7946 | iblech++ | * eqv is now spelled ===.
r7946 | iblech++ | * (:who($name), :why($reason)) := ... is now a syntax error, one has to
r7946 | iblech++ | explicitly use :(...) syntax to disambiguate:
r7946 | iblech++ | :(:who($name), :why($reason)) := ...
r7946 | iblech++ | * Fixed many occurances of =:= which actually meant === (but, as in the
r7946 | iblech++ | current Haskell runcore, &infix:<=:=> ::= &infix:<===>, these mistakes went
r7946 | iblech++ | unnoticed).
r7946 | iblech++ | * docs/quickref/op was in iso-8859-1; fixed. 13:39
r7946 | iblech++ | * t/oo/type.t is outdated (no current Synopsis speaks about .type and the
r7946 | iblech++ | notion of types and classes etc. is changing currently).
r7947 | iblech++ | * PIL2JS: lib/PIL2JS.pm: Do not write to "deleteme_eval.js" any longer when
r7947 | iblech++ | using JSSM (putter++!).
r7947 | iblech++ | * t/oo/attributes/undeclared_attribute.t: Improved xinming++'s test to
r7947 | iblech++ | also test for binding of undeclared attributes to be illegal.
r7947 | iblech++ | * New t/operators/binding/attributes.t: Tests for binding {public,private}
r7947 | iblech++ | {instance,class} attributes.
r0nny re 13:47
Juerd ===? 13:48
Oh my!
HELP
(NOOOOOOOOOO!)
liz6 some equals are more equal than other equals? 13:50
Juerd =, ==, === doesn't make the language any clearer.
PHP has this. It sucks.
theorbtwo That was always true... but = and == is already plenty, thankyouverymuch.
Eeep, I'm an echo. 13:51
dakkar Juerd: JS is even stanger
Juerd 14I'm an echo...
r0nny i'd rather have eqv then ===
i think eqv is much more readable 13:52
Juerd I dislike eqv too. I think this thing should be a method, not an infix op.
Or maybe not.
r0nny i tihnk infox op is ok
Juerd Ah well, at least we don't have to use this often.
dakkar is 'eqv' pointer-comparison? 13:53
r0nny it does what it should do - in a radable way
Juerd Readable, yes, distinguishable, hardly.
= versus == is doable.
== versus === is much less.
r0nny syntax-highlighting is needed for sure
Juerd The language should never depend on syntax highlighting for things to be clear. 13:54
r0nny yeah
Juerd Besides, I doubt any syntax highlighting engine can make anything useful from:
r0nny but humans tend to see it more clear if the machine hepla a bit with colors
Juerd /<before foo %bar{/<quux>/}>/ 13:55
theorbtwo I think infix op is not only OK but right. If there's no difference between the invocant and the argument, making it a method is confusing.
Juerd theorbtwo: You're right.
Juerd tends to see $foo infixop $bar as $foo.infixop($bar) though. 13:56
theorbtwo It's not, though.
It's infixop($foo, $bar)
Juerd (1.+(2)).==(3)
)
theorbtwo ?eval 1.+(2)
13:56 evalbot_7943 is now known as evalbot_7947
evalbot_7947 3/1 13:56
theorbtwo OK then.
dakkar what was that? 13:57
r0nny ?eval 1.+(3)
evalbot_7947 4/1
dakkar ?eval (1).+(3)
r0nny wtf ? why is the reult 4/1 - it should be 4
evalbot_7947 Error: unexpected "." expecting term postfix, operator, postfix conditional, postfix loop, postfix iteration, ";" or end of input
dakkar r0nny: might that be a Rational?
r0nny hmm 13:58
theorbtwo Pugs has several different sort of numbers; that's a rational. Doesn't much matter; as everybody knows 4/1 is 4.
dakkar And I think it gets parsed as <1.> <+> <(> <3> <)>
r0nny confusting
theorbtwo sub{1.+(3)}.perl
?eval sub{1.+(3)}.perl
evalbot_7947 "sub \{...}"
theorbtwo Grr. 13:59
dakkar the 'decimal dot' should be consumed by the 'number' token
theorbtwo I keep forgetting it can't do that.
r0nny usually i expect a integer and not a rational as result of addidion from integers
dakkar 1. is not an integer ;-)
r0nny hmm
?eval 1+(3)
evalbot_7947 4
theorbtwo OK then...
anyway, we're drifting.
r0nny argh
?eval 1.infix:>>+<<(3) 14:00
evalbot_7947 Error: unexpected "i" or "." expecting digit, "_", exponent, term postfix, operator, postfix conditional, postfix loop, postfix iteration, ";" or end of input
theorbtwo What should "are these the same variable" look like?
?eval 1.infix<+>(3) #, I think...
dakkar I asked: waht does exaclty 'eqv' do? pointer equality?
evalbot_7947 Error: No compatible subroutine found: "&infix"
theorbtwo dakkar: "Are these two things really different names for the same thing?"
r0nny dakkar: =:= does pointer equal
i trhink i shoud read more 14:01
theorbtwo === is the new name for =:=, I think.
dakkar argh. so C<a === b> iff the two names refer to the same container
so it's a relation on names, not values, so it cannot be a method, since methods operate on values
theorbtwo ?eval $a=1; $b=[$a, $a]; $b[0] === $b[1] 14:02
evalbot_7947 Error: Undeclared variable: "$a"
theorbtwo ?eval my $a=1; my $b=[$a, $a]; $b[0] === $b[1]
evalbot_7947 bool::true
Juerd dakkar: Methods operate on containers.
dakkar: The container is the invocant
theorbtwo (The container often falls back^W^Wredispatches to the value, though, if it doesn't handle the method itself.) 14:03
dakkar Juerd: ? an object is a value... or am I missing something? C<tie>s operate on containers
Juerd dakkar: Your terminology is off :)
theorbtwo ?eval $a=1; $b=[$a, 1]; $b[0] === $b[1]
evalbot_7947 Error: Undeclared variable: "$a"
theorbtwo ?eval my $a=1; my $b=[$a, 1]; $b[0] === $b[1]
evalbot_7947 bool::true
dakkar ok, where is the "terminology quick-reference"? ;-)
theorbtwo Blink... is that last result wrong, or is my understanding wrong?
Juerd dakkar: There are names, containers and values, originally, but Autrijus thought of a new scheme with cells instead of containers. The new one isn't in my head yet. 14:04
theorbtwo: It is wrong, or === isn't container-equivalency
dakkar Juerd: oh. I might have looked at a picture, once...
Juerd dakkar: Anyway, a "value" is the contents of a container. 14:05
dakkar ehm... $b=[$a,$a];$b[0]===$b[1] -> true is wrong, too, so
Juerd: yes, and a container is (more or less) a "memory location"
Juerd dakkar: After $foo = 3, the container named $foo holds the value three.
Operators like + use the value.
dakkar Juerd: ok 14:06
Juerd Normally, when passing an argument, you pass the container, even without "is rw". Some magic is applied to make the container read only in the sub.
dakkar where is the type annotation stored, though? As I see it, all three levels need some kind of type annotations 14:07
Juerd "is copy" makes perl copy your value from one container to another.
dakkar: In the container.
dakkar but I can't do @foo := $bar, so the names must be typed in some way 14:08
(or can I?)
Juerd dakkar: The value has type too, and there can even be two different values in a single container, but that's not something a language user should care about.
dakkar: Sure you can, if $bar is a reference to an array.
It will then make @foo an alias for @$bar.
dakkar so it has to know that @foo is an array name
Juerd @. 14:09
It's easy to think of real arrays and real hashes as references without the $ :)
dakkar yes, obvious. so we *do* have three levels of type annotation
Juerd dakkar: Well, sort of. We have more, even.
dakkar (BTW, this seems to be true of Perl5, too) 14:10
Juerd We need more words than "type"
dakkar yes. must start making them up ;-)
Juerd Or adjectives. Easier. 14:11
theorbtwo typf, typg...
dakkar lol
Juerd Variable types, container types, value types.
dakkar "variable" is a bit overloaded, don't you think? I keep using "name"
theorbtwo (Fortunately, we can use the shorter term "sigil" for the first one.)
Juerd A variable doesn't have to have a name. 14:12
theorbtwo ...unless I'm confused about the meaning.
Juerd It can have a name. Or even several.
So "name" as a replacement for "variable" is wrong.
theorbtwo (Think of the (somewhat abused) perl term "anonymous".)
Juerd But yes, there may also be name types.
dakkar Juerd: argh. what is a "variable"? (I'm deeply confused)
I have names ($foo), containers (memory locations), values (thing stored therein) 14:13
Juerd I'm wrong. "variable type" is misleading.
Should indeed be "name type".
dakkar oh, I'm relieved ;-)
theorbtwo Juerd: Can you give an example of a name type?
Juerd theorbtwo: $, @, %, &
theorbtwo: Sigils indicate name type.
dakkar theorbtwo: the name C<$foo> has scalar name type 14:14
Juerd Variable types are arrays, hashes, scalars
theorbtwo OK. But the name is "$foo", untyped -- the $ is part of the name.
Juerd Container types are Scalar, Ref, Num, Str, Dog, Cat (containers are scalar things)
theorbtwo: "foo type" != "type". 14:15
Disregarding that != is for numbers.
Value type is number, integer, float, string, reference
dakkar Juerd: "containers are scalar things"? what is the container that @a refers to?
Juerd dakkar: None.
dakkar: An array is a collection, not a container, in the jargon I'm used to.
theorbtwo So what's self in @a.foo? 14:16
dakkar Juerd: when I tie an array, don't I tie the container? (tie are trasparent to argument passing, are they not?)
Juerd A collection (aggregate) points to multiple containers
dakkar: I refuse to discuss tie, because it's so extremely complicated in Perl 5, and I don't know the semantics in Perl 6.
dakkar Juerd: good point
theorbtwo I don't think there is such a thing in perl 6. 14:17
dakkar there was something in autrijus's pictures...
theorbtwo OTOH, you can do my BTree @foo;
Juerd theorbtwo: That gives you "my @foo of BTree", i.e. scalars of BTree type, aggregated in a normal array.
dakkar and it should mean that the containers referred to by the subscripting @foo can olny contain values derived from the BTree value type, right? 14:18
theorbtwo Um... are you sure?
Juerd theorbtwo: Never.
dakkar lol
r0nny rafl: does your sixpan stuff have any usable depency solving classes ? 14:27
Juerd Fuck. 14:30
spamd hasn't been running for an hour.
Two hundred spam messages in my inbox.
r0nny omg
Juerd Good thing clamd was still running.
r0nny this many in 1 hour ?
Juerd Yep.
Numerous duplicates too
r0nny they should have a death-penalty for spammers 14:31
Juerd No. 14:32
Did you actually think about that before you typed it? 14:33
r0nny not really
theorbtwo Try doing that in the future.
r0nny but at least they should cut them from internet for 10 or 20 years 14:34
Juerd I disagree.
theorbtwo So do I.
r0nny so they should be left cappable of spamming ?
Juerd People selling drugs via the internet should be punished as drugs dealers.
Because that's what they are. The means of doing so is irrelevant.
This is half of today's spam. 14:35
r0nny hmm
gaal Juerd: if you use procmail, look at the force-rescan script in one of the manpages
theorbtwo People fradulantly claiming to deal drugs should be punished as fraudsters.
Juerd Then, there's the scams, same thing. Scamming is forbidden.
pasteling "gaal" at 192.117.178.34 pasted "(OT) force-rescan with procmail" (19 lines, 455B) at sial.org/pbot/14270
r0nny in some countryed they have desth-penalty for selling hard drugs 14:36
countryes
theorbtwo countries.
r0nny ack
Juerd If you treat all these things via existing laws, you will end up with commercial spam, legal stuff. This should be allowed via strict guidelines, and punisheb by money fines.
gaal: My server can't take it if I use non-spamd 14:37
gaal you can do that with spamd too i think, assuming the procmail recipe calls spamc.
Juerd I need qpsmtpd-forkserver because tcpserver spawning things is too much.
r0nny i think they should disallow main as adversting platform
Juerd gaal: The problem in question was that spamd died :) 14:38
r0nny eh mail
Juerd r0nny: I disagree.
r0nny since mail id for comunication, not roasting your beain with the newest ads
gaal Juerd: and is still dead? that's a problem all right :)
theorbtwo So do I.
Juerd r0nny: Direct marketing can be very useful for both seller and target audience.
gaal: No, I re-started it.
r0nny: Marketing and advertisement are communication.
r0nny i see most of it as roasting peoples brain with untrue stuff 14:39
Juerd The problem with spam is that most spam is for illegal goods and activities, or porn related.
But that should be dealt with using non-spam laws.
Sending porn to minors is illegal in many countries. If you cannot verify that someone is an adult, don't send it. 14:40
r0nny yeah
Juerd If someone is an adult, go ahead, send the porn spam
theorbtwo "people's brains". You do have a point, r0nnie, but if the problem is the untrue stuff, then make sending untrue stuff illegal, and prosecute for it.
Juerd Just keep it reasonable.
14:41 lampus_ is now known as lampus
Juerd theorbtwo: Guess what - it already is ;) 14:41
theorbtwo Juerd: I'm not sure I like that solution, but only because I like free spam.
Juerd In most countries.
theorbtwo Juerd: I'm aware.
Er, s/spam/porn/
Juerd Countries without laws are a problem.
r0nny problem iss most of these spammer asholes are very unreasonalbe - i think for these unreasonable people there should be unreasonable penaltyes
Juerd theorbtwo: There are many opt-in mailinglists for that.
theorbtwo Juerd: But if they have to verify your age, they will become non-free. 14:42
Juerd r0nny: No, even for unreasonable people, there should be reasonable penalty.
theorbtwo: "Sending to", I meant in the non-american form :)
If someone *requests* it, give it. They should be able to judge for themselves. 14:43
If a minor requests porn, by all means give it to them. The minor is at fault here.
r0nny they dont ask u - its like menthal terror
Juerd I meant "sending to" in the spam way. Unsollicited mail relating to porn.
That should not be done without verification. 14:44
theorbtwo I rather like the smell of menthol. You meant mental.
Juerd menthol terror :)
lol.
r0nny rofl
bad english is allways good for a joke ;P
theorbtwo Juerd: The problem is that, in theory, a minor is never at fault. r0nny: "always".
theorbtwo sighs. 14:45
Juerd theorbtwo: Still, one shouldn't require providers of non-real-life-interactive erotica to verify age.
theorbtwo I don't much care for that theory, and in pratice, it doesn't do any good, nor is it very true.
Juerd theorbtwo: This would be unreasonable, as it indeed destroys free porn. 14:46
This is unrelated to spam, though :)
Pull and push media should have different rules. 14:47
And opt-in is pull
And then there's still an important difference between synchronous and asynchronous. 14:48
svnbot6 r7948 | iblech++ | docs/quickref/data: Outlined the current variable name, container, and cell
r7948 | iblech++ | semantics, as they're implemented in PIL2JS and as they will be implemented in
r7948 | iblech++ | the new runcore and the new PIR compiler.
r7948 | iblech++ | * Note: Do not use the current Haskell runcore for experiments with
r7948 | iblech++ | containers, =:=, ===, etc. In the current runcore,
r7948 | iblech++ | $a = 42; $b = 42; $a =:= $b is true!
r7948 | iblech++ | * Note: Reading t/operators/binding/*, t/operators/identity.t, and
r7948 | iblech++ | t/operators/value_equivalence.t is strongly recommended.
Juerd Erotica on TV is fine late at night, not just before kiddy bedtime
But sending it by mail would be wrong at any time.
This can also be seen as a broadcast/direct difference.
E-mail isn't a broadcast medium, even though it is used like it.
$a = 42; $b = 42; $a =:= $b is true! 14:49
ahh
That explains some stuff :)
r0nny the problem is they do everything to get it torugh filters
Juerd Define "they" 14:50
r0nny porn and drug spammers
Juerd The fun thing is that the people who evade filters, are usually doing something illegal and can be punished based on that.
Spam isn't bad by definition.
I get some good spam.
The only reason I run a filter thingy is that most spam is indeed the kind that I don't want. 14:51
And that I don't mind not getting the good spam.
dakkar starts to read the documents on name-container-value stuff...
r0nny last time i got a usefull spam msg was 6 months ago
i think there should be laws, define how to use mail - so all thos damn mails with fake information are very illegal 14:52
and it should be threaded like faking documents like a testament or a contact 14:53
so the penatyes of this massive abuse the bad spammers do will automagically add up to very high ones 14:54
for example there could be a rule all ad-mails must have a valid sig, so the sender can be proven and a note in the subject 14:55
theorbtwo "penalties". I think it should certianly be far lower a penalty then for purgery in a court of law. 14:56
I'm not real sure if it should be illegal.
Note that giving a false name is not, in general, illegal.
r0nny fakind data to make users look at it should be illegal
theorbtwo (In the US, at least.)
r0nny theorbtwo: why should the penalty for people just making money if casuing financial dammage to others be low ? 14:57
s/if/of
theorbtwo r0nny: I couldn't make sense of that sentance. 14:58
r0nny actualy bad-spam-technology does huge dammage
i dont see why there should be low penalty on this
theorbtwo Because it's a fairly minor offese. 15:00
r0nny they wont stop if there is low penalty 15:01
btw - Juerd i rechecked the definition of spam - selective adversting could be usefull for the customer doesnt count as spam 15:02
theorbtwo r0nny: Whose definition?
r0: They'll stop if there's a fairly low penalty, and that penalty is more commonly assessed. 15:03
r0nny theorbtwo: the one on wikipedia
theorbtwo: a low penalty per spam mail is ok - as long, as it adds up - and since the amount is high, the resulting penalty will be, too 15:04
theorbtwo WP is fairly clear that different people define it differently.
Read the whole first paragraph, not just the first sentance.
The thing is that false advertising is illegal, dealing drugs is illegal, sending porn to minors is illegal. 15:05
(Not only that, but those are all /criminal/ charges.)
... hm, I shouldn't have parenthisized that -- it's an important point. 15:06
r0nny hmm
theorbtwo When you do any of those things, the State can charge you.
The federal government fairly recently made sending email with false header information illegal. 15:07
en.wikipedia.org/wiki/CAN-SPAM
The problem is not that things aren't illegal. 15:09
The problem is that law enforcement does things that they see as more important, and in many cases they are absolutely correct. 15:10
I'd much rather have agents working kiddy-porn rings then spammers.
OTOH, I'd rather have them tracking down major phishers then either.
...but phishing is illegal by dint of it being fraud, far more imporant the fake email headers. 15:11
wolverian (in some countries, it's illegal to send advertising at all without prior acceptance from the customer, and that acceptance must be active, not passive ("default").) 15:49
Juerd There should be penalties for sending duplicate mail with different message-id's :) 16:20
i.e. for sending mail and letting the MTA generate a message-id.
theorbtwo It's easy to set the penalty for that -- drop the message. 16:22
Juerd It's expensive to test for it :)
theorbtwo (Or reject it -- do not bounce, ever.)
Not on the reciving MTA it's not.
Juerd It means keeping records of previous messages.
Which sucks. 16:23
theorbtwo If there's no message-id, drop.
Juerd There's usually a message-id generated by the sending MTA.
It should be generated in the client, for identification of duplicate messages. 16:24
theorbtwo You could always md5+sha1 the body, and drop if it matches. You'll have minor problems with hash colissions, but shouldn't have too many. 16:25
theorbtwo wonders if there are any good hash algos of user-settable length.
Juerd sha-n 16:28
theorbtwo D'oh. 16:33
svnbot6 r7949 | autrijus++ | * luqui's \d[] patch eats the trailing space after the bracket;
r7949 | autrijus++ | change "brackets" to "verbatimBrackets" to solve this.
r7949 | autrijus++ | * moreover, fix the old bug of "&foo() &bar()" eating the
r7949 | autrijus++ | whitespace inbetween, due to the same cause.
r7949 | autrijus++ | * test for both in t/syntax/interpolation/strings.t
Juerd autrijus! :) 16:34
gaal autrijus: ping 17:23
r0nny re 18:23
19:42 bc is now known as venk
venk . 19:46
gaal Parser.hs change in last patch 19:57
# rt.openfoundry.org/Foundry/Project/...;rev2=7949
shouldn't that say "verbatimBrackets" ?
maybeVerbatimBrackets in Lexer.hs is unused.
ah: no that refers to the old bug. 20:00
Still, I'm wondering about maybeVerbatimBrackets; why is it there?
svnbot6 r7950 | gaal++ | when ({ closure }) { ... } - fix tests (they pass now) 20:23
r7950 | gaal++ | (following luqui's remark on the channel that this is the likely spelling.)
liz6 ?eval 1^..3 23:10
23:10 evalbot_7947 is now known as evalbot_7950
evalbot_7950 (2, 3) 23:10
liz6 ?eval 1..^3
evalbot_7950 (1, 2)
Juerd ?eval 1^..^3 23:11
evalbot_7950 (2,)
Juerd ?eval 0^..^0
evalbot_7950 pugs: Prelude.init: empty list
theorbtwo Anime operators!
Juerd ?eval 0^..^0; say "foo"
evalbot_7950 pugs: Prelude.init: empty list
Juerd Bug.
liz6 report! 23:12
;-)
?eval 0..^0 23:13
evalbot_7950 ()
theorbtwo Create a test.
liz6 ? eval 0^..0
?eval 0^..0 23:14
evalbot_7950 ()
liz6 ?eval 7^..^7
evalbot_7950 pugs: Prelude.init: empty list
liz6 not just for 0...
?eval 7^..^8
evalbot_7950 () 23:15
liz6 ?eval 7^..^6
evalbot_7950 pugs: Prelude.tail: empty list
liz6 ?eval 5..1
evalbot_7950 ()
theorbtwo Somehow, I fail to see the use of ^ed infix:<..>.
liz6 was just going through S03 and noticed N^..^M for the first time 23:17
Juerd I knew it exists, but think it shouldn't. :) 23:19
theorbtwo: Re the report - will perhaps do later.
liz6 ?eval 5..1 23:20
evalbot_7950 ()
liz6 isn't that supposed to generate (5,4,3,2,1) ?
theorbtwo ?eval 5..1:by(-1) 23:21
evalbot_7950 Error: unexpected ":" expecting digit, "_", fraction, exponent, term postfix, operator, "does", "but", "is", "^..^", "cmp", "<=>", "^..", "..^", "..", postfix conditional, postfix loop, postfix iteration, ";" or end of input
Juerd Though if you feel like copy/pasting for me: ok(eval "0^..^0; 1", "Empty list doesn't die");
liz6: Unfortunately, no.
liz6: Larry still isn't convinced.
theorbtwo Anyway, that's supposed to do 5,4,3,2,1.
liz6 theorbtwo: yes, I remember from the talk at Euro OSCON now... 23:22
lookin ar S03, "=:=" is now "===", is that right? 23:23
clkao r7rl
Juerd liz6: Yes
theorbtwo I think iblech patched to that effect earlier today. 23:24
liz6 ?eval $a := $b; if ($a =:= $b} say "yes"
evalbot_7950 Error: unexpected "i" expecting ";", statements or end of input reserved word
liz6 ?eval $a := $b; if ($a =:= $b) say "yes"
evalbot_7950 Error: unexpected "i" expecting ";", statements or end of input reserved word
liz6 ?eval $a := $b; if ($a =:= $b) {say "yes" }
evalbot_7950 Error: Undeclared variable: "$b"
liz6 ?eval $b = 1;$a := $b; if ($a =:= $b) {say "yes" } 23:25
evalbot_7950 Error: Undeclared variable: "$b"
xinming ?eval $.a = 1;
evalbot_7950 Error: Undeclared variable: "$?SELF"
liz6 ?eval my $b = 1;my $a := $b; if ($a =:= $b) {say "yes" }
evalbot_7950 OUTPUT[yes ] bool::true
liz6 ?eval my $b = 1;my $a := $b; if ($a === $b) {say "yes" } 23:26
evalbot_7950 OUTPUT[yes ] bool::true
liz6 hmmm... can both be right?
luqui in pugs === and =:= are the same thing 23:28
but they're not supposed to be
liz6 ah, ok 23:29

Powered by Cro and the Raku® Programming Language.

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

Thank you!