»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'p6: say 3;' or rakudo:, std:, or /msg camelia p6: ... | irclog: irc.perl6.org | UTF-8 is our friend!
Set by masak on 12 May 2015.
00:03 FROGGS left 00:06 FROGGS joined 00:07 Zhtwn left 00:08 ggoebel left
dalek c/choices: d3ef4e1 | ab5tract++ | bin/p6doc:
No longer theoretical
 00:13
00:13 msouth left 00:14 lizmat left 00:18 quietfanatic left 00:20 rurban left 00:23 lizmat joined 00:25 amurf left 00:26 amurf joined 00:27 amurf left 00:38 kurahaupo1 left, msouth joined, llfourn joined 00:44 laouji joined 00:46 msouth left 00:53 msouth joined 00:54 Zhtwn joined, bin_005 joined 01:09 cognominal joined 01:14 nys left 01:18 adu joined 01:21 cognominal left 01:22 prevost joined 01:38 BenGoldberg joined 01:46 yqt left 01:54 gtodd1 joined 01:59 Ben_Goldberg joined 02:01 dayangkun joined 02:02 BenGoldberg left 02:04 prevost left, ggoebel joined
dalek c/choices: 145185a | ab5tract++ | bin/p6doc:
Run doc output through fmt on the way to page
 02:04
c/choices: 571f437 | ab5tract++ | bin/p6doc:
Cleaner by far
02:11 rurban joined
dalek c/choices: d7ae128 | ab5tract++ | bin/p6doc:
Fix menu
 02:13
02:17 msouth left 02:22 adu left 02:26 ggoebel left
pmichaud good evening, #perl6 02:26
02:27 AlexDaniel left 02:28 noganex joined
hoelzro evening pmichaud 02:29
02:30 msouth joined 02:31 noganex_ left 02:35 Ben_Goldberg left, rmgk_ joined, rmgk is now known as Guest7371, rmgk_ is now known as rmgk 02:36 kurahaupo1 joined 02:39 Guest7371 left 02:48 adu joined 02:50 jack_rabbit joined 02:51 Zhtwn left
pmichaud TimToady: ... is "take" considered an instance of abnormal control, in the sense of being trappable by CONTROL handlers...? 02:53
*abnormal control flow
02:54 sunnavy left
TimToady presumably 02:54
at least as currently implemented
pmichaud well, I mean more "as designed"
TimToady it doesn't really have to be control flow, if it merely knows where to put the next value 02:55
pmichaud I'd prefer it not be control flow.
TimToady fine by me 02:56
pmichaud in some sense it is, in another it's not, since as far as the block is concerned execution continues immediately following the take anyway :)
it's more like an interrupt than a control flow break :)
TimToady certainly we want to maintain the right to optimize 02:57
pmichaud yes
works for me.
skids Well, someone will come looking for a coruotine construct if take stops behaving as one.
pmichaud skids: I'm not changing that. 02:58
adu TimToady: you're such a libertarian 02:59
pmichaud There are just cases where take can be smart enough to recognize that it's safe to store the value somewhere and proceed without immediately doing the control jump.
02:59 adu left, adu joined
skids Ah, good, I am all for optimizing those too :-) 02:59
pmichaud when it's not safe, it does the continuation/jump bit. 03:00
adu TimToady: you're such a libertarian
TimToady you can say that again
adu libertarian
it's a Good Thing 03:01
Liberty is what protects us from Hitler and population reductionism 03:02
pmichaud
.oO( is Hitler still a threat? I thought he was dead. )
 03:04
adu I'm 100% sure that Hitler's public double is dead
pmichaud Aha. Good to know. 03:05
adu I'm 5% sure that the original Hitler is dead
03:06 rurban left
skids m: say 2015-1889; # I'm a bit more than 5% sure. 03:06
camelia rakudo-moar 15c6fb: OUTPUT«126␤»
adu make that 45% sure 03:07
dalek ecs: a62d2ce | pmichaud++ | S07-glr-draft.pod:
More GLR discussion, plans, and notes.
 03:12
pmichaud TimToady: not sure it's a good idea, but @@ popped into my head as a possible flattener. 03:14
it's very steamroller-y
I'm afk # walk, packing 03:17
03:24 laouji left 03:25 laouji joined 03:29 laouji left 03:37 sunnavy joined
dalek ecs: e71c0a1 | skids++ | S07-glr-draft.pod:
Minor pronoun clarification

Historical note for those not up to date on GLR progress		 03:37
03:48 yeahnoob joined 03:51 prevost joined 03:54 kst left, bin_005 left 03:55 bin_005 joined 03:57 Gardner joined 04:01 laouji joined
dalek ecs: b95b2a1 | Mouq++ | S07-glr-draft.pod:
Ask about a list concatenation op
 04:04
ecs: 2a7daba | Mouq++ | S07-glr-draft.pod:
Untypo last commit
 04:07
04:16 kaare_ joined, adu left 04:17 mattp__ left
TimToady Mouq: yes, re irclog.perlgeek.de/perl6/2015-06-10#i_10729270 a deepmap is intented to retain structure, so there's that's gone buggy with some of our GLR stuff, and will need fixing 04:19
*something
Mouq Okay 04:22
TimToady jnthn: the intent of having a .parsefile in the first place is to capture the filename for use in error messages 04:23
otherwise it's rather uselessly redundant with .slurp.parse
well, .parse(slurp) or some such 04:24
dalek ecs: 4482589 | pmichaud++ | S07-glr-draft.pod:
Add answer to skids++'s e71c0a1 question.
 04:25
Mouq Part of me, after implementing it, is still uncertain about »+» creating lists of the same type as were passed, until we get coersion worked out better :/
TimToady that seems right to me 04:26
pmichaud I suspect a lot of new information will fall out of the new GLR semantics
er, phrased that wrong
I suspect we'll have a lot better picture of how it should work once the initial GLR semantics are in implementation 04:27
I'm very much liking the model that is emerging from thoughts/discussions
it feels clean, elegant, and (potentially) fast
04:31 hanekomu left 04:32 broquaint left 04:40 araujo_ left
japhb
.oO( Choose three ... )
 04:42
dalek ecs: 99aeec7 | pmichaud++ | S07-glr-draft.pod:
Conjecture names for the conjectured insert-into-list type.
 04:43
ecs: 43f7fde | pmichaud++ | S07-glr-draft.pod:
Answer Mouq++'s b95b2a1 question.
ecs: 69b6eb5 | pmichaud++ | S07-glr-draft.pod:
More answer to Mouq's example.
 04:50
04:50 araujo joined
dalek ecs: 84ff4f3 | pmichaud++ | S07-glr-draft.pod:
Add "Sublist" to the brainstorm.
 04:51
pmichaud Also "SList", which combines "sublist" but maybe sounds a bit like "slice". :-)
04:52 SevenWolf joined
TimToady the conjectured type won't replace Nil, only Empty 04:54
pmichaud correct.
Nil ends up being quite orthogonal to lists, I think. 04:56
as in, it doesn't have any special listy properties.
pmichaud adds to S07
TimToady I just was reading a partial context line, and didn't see the pre-GLR note on Nil
04:57 lkjkj joined 04:58 lkjkj left
Mouq pmichaud++ btw 04:58
dalek ecs: d419e70 | pmichaud++ | S07-glr-draft.pod:
Add note about post-GLR C<Nil>.
TimToady kinda wonders whether our new type is really just an iterator, rather than something you ask for an iterator
pmichaud it's certainly Iterable.
and in pre-GLR, all Iterators were Iterables. 04:59
Mouq Also, "Expand"? (Seq bikeshedding)
TimToady
.oO(Shoehorn)
Hoist 05:00
pmichaud I knew I'd be opening the bikeshed paint store with that commit. :)
Clip, Cut, Paste were ideas coming from the notion of film editing. :)
TimToady Splice
pmichaud I thought of Splice, but it's too close to .splice for my taste.
TimToady well, rather similar idea 05:01
pmichaud yes, but part of me feels that .splice should then return a Splice. :)
TimToady whatever it is, Foo() has to feel like Empty 05:02
pmichaud thesuarus.com only knows about the verb form of "splice", and the noun definitions I've seen think of a "splice" as the whole result of the join, and not the piece being joined
TimToady More()
pmichaud I kind of like Seq() 05:03
I didn't at first, but it's growing on me.
Mouq To bad we already used Mu :P
pmichaud Lu
TimToady Moo 05:04
we were wanting a Dog type...
pmichaud I'm also toying with the idea that the type object itself disappears in a list.
05:04 lolisa left
pmichaud like Empty does now. 05:04
m: say Empty.defined
camelia rakudo-moar 15c6fb: OUTPUT«False␤»
pmichaud so then one could just return Seq
and not have to do Seq()
TimToady but Seq doesn't feel empty
pmichaud (or whatever it gets called) 05:05
TimToady Slot
pmichaud Hole
TimToady Virus
pmichaud Gene
TimToady Intron 05:06
Slip
as in, "slip this in"
but also "slip one cog" when nothing is there
pmichaud "Welcome to Pm's Bikeshed Paint Emporium! Today we have a special running on all sorts of list-related terms" 05:07
yes, I like the double-meaning to slip
I was thinkng "slip by this one"
TimToady and it implies "sneaky"
pmichaud Sneak
(slip is better)
Mouq Pm()
pmichaud are you implying I'm empty? :-)
Mouq You just fit in so well 05:08
pmichaud I think I'll tentatively go with Slip for now, I like it better than Seq
TimToady me2
dalek ecs: 3c71890 | pmichaud++ | S07-glr-draft.pod:
s/Seq/Slip/ # better bikeshed choice for now
 05:10
pmichaud fwiw, I tried really hard to convince myself that C<Dog> could work. :-P
and perhaps C<Tac> is too much "you think that's cute today" 05:11
05:14 llfourn left, skids left
pmichaud I wonder if the result of an infix:<...> should naturally be a Slip 05:30
(idle thought)
TimToady we might implement the postulated foo(@inner): @outer using a slip of some sort too 05:33
pmichaud this feels very much like a primitive type we've been missing for some time :)
TimToady .u ､ 05:34
yoleaux U+FF64 HALFWIDTH IDEOGRAPHIC COMMA [Po] (､)
pmichaud I get a box for that char in my font. 05:35
TimToady .u ⸴ 05:36
yoleaux U+2E34 RAISED COMMA [Po] (⸴)
pmichaud same there. :)
TimToady looking for non-Texas forms :)
"raised" is cute though
pmichaud well, we want to make sure it's visually distinct from normal commas, too.
TimToady u ⨩ 05:37
.u ⨩
yoleaux U+2A29 MINUS SIGN WITH COMMA ABOVE [Sm] (⨩)
TimToady has been considering various Texas forms
05:37 prevost left
TimToady ,, or \, or ^, and such 05:37
the (,) ain't so bad except it would imply a set operator 05:38
pmichaud yup
TimToady <,>
pmichaud I conjectured @, at one point.
but too sigil/twigilish 05:39
TimToady it's kind of a shame that _ is considered alpha 05:40
as a "flat" sugar
other possible flat sugars: *() *: 05:41
would obviously have to win under LTM, but fit nicely with * for slurp
pmichaud there's always *** :-) 05:42
that's the Alaska form. :)
TimToady too bad there isn't a monster truck glyph 05:43
05:43 RabidGravy joined
pmichaud aha 05:44
the O_O operator
TimToady maybe we should use snowman for hierarchical, and comet for "flattening" :)
pmichaud O_O looks like a steamroller :)
TimToady O¯O 05:45
pmichaud I'm pretty sure someone can create a Texas monster truck
but for that matter, how about infix:<slip> 05:46
I really don't expect it to be common at all.
TimToady o̿o
pmichaud (perhaps I'm wrong about that with all of the other flattening/itemization changes we're making... perhaps it'll be very common)
TimToady I considered infix:<slip>, but while it's huffmanly ok, it just doesn't quite feel right for some reason 05:47
pmichaud or even prefix:<slip>
TimToady prefix slip is close to your one-level flat
pmichaud yes, and it fits with the other "list" and "flat" contextualizers.
and then you can do return slip;
as well as return slip()
TimToady but a reduce on the infix is probably sufficient instead of a prefix 05:48
pmichaud indeed
TimToady return Slip; ain't so bad
pmichaud yeah, but Slip.new(1,2) isn't so much fun.
It works, but....
TimToady [,,] 1,2 05:49
pmichaud I suppose we can do Slip(1,2), but that feels wrongish also.
TimToady not opposed to a slip listop
pmichaud return [slip] 1,2 feels interesting also
the "slip" almost looks parenthetical :)
FROGGS hmmmm, 'return Slip' would be identical to 'return Panties' where I am 05:50
pmichaud FROGGS: TMI
TimToady the Slips I'm familiar with are not identical to Panties...
pmichaud WTMI
FROGGS :D 05:51
good morning :o)
05:51 FROGGS[mobile] joined, FROGGS left
TimToady @x X (@y Zslip @z) feels odd 05:52
@x X, (@y Z^, @z) seems more better
or @x X, (@y Z\, @z) 05:53
so I'd like it to seem like a comma variant
pmichaud I'd be okay with ^, or \.
TimToady that's probably what bothers me about slip
pmichaud er, \,
yes, I prefer a comma variant of some sort also.
I just haven't come up with one I like better than slip. :) 05:54
I just haven't come up with one I like.
TimToady [\,] is bad though
m: .say for [\,] ^5 05:55
camelia rakudo-moar 15c6fb: OUTPUT«0␤0 1␤0 1 2␤0 1 2 3␤0 1 2 3 4␤»
pmichaud ,, seems like it could work but it scares me somehow.
TimToady yeah, looks like an error
*, maybe 05:56
pmichaud aha!
infix:<⏏> :-P
TimToady .p ❟ 05:57
.u ❟
yoleaux U+275F HEAVY LOW SINGLE COMMA QUOTATION MARK ORNAMENT [So] (❟)
pmichaud I'm going to go with "slip" for now, knowing we'll likely change it to something comma-based. 05:58
I definitely like the prefix "slip" form. 05:59
06:00 bin_005 left 06:01 bin_005 joined
TimToady the thing I like about ,, though is that it kinda implies the two commas in the upper list on either side of the bit we're slipping in, a kind of comma-hole 06:03
pmichaud put a "hole" between the commas, then. 06:05
like ,O,
but not alpha
so that it doesn't look like a typo but is obviously deliberate. 06:06
06:10 diana_olhovik_ joined
pmichaud perhaps we could do ,, and make it illegal to mix ,, and , in the same chain. 06:12
3,,4 # okay 06:13
3,4,,5 # error
3,(4,,5) # okay, but same as 3,4,5
oh! 06:14
for that matter, even if someone typoed the ,, they'd end up with the same result. 06:15
(I think)
I can probably find a case where that doesn't work.
TimToady well, 3,4,,5 would already be an error, since two list assoc have to be the same operator already 06:18
06:18 quietfanatic joined
quietfanatic Insert() was what I was thinking 06:18
06:18 llfourn joined
TimToady we've mostly settled on Slip for the moment 06:18
quietfanatic ah
06:19 Gardner left
quietfanatic p6: grammar ASDF { multi rule TOP { ^ asdf $ }; multi method TOP ($/) { $/.make: "asdf" } }; say ASDF.parse("asdf", :actions(ASDF)) 06:19
camelia rakudo-moar 15c6fb: OUTPUT«Too many positionals passed; expected 1 argument but got 2␤ in regex TOP at /tmp/tmpfile:1␤ in block <unit> at /tmp/tmpfile:1␤␤»
quietfanatic I was really hoping that would work
06:20 FROGGS joined
quietfanatic I wonder why that error message happened though? 06:20
06:20 msouth left
TimToady extra arg to 2nd TOP 06:20
quietfanatic But the signature of first TOP is (ASDF: *%_) 06:21
according to introspection
so why would anything be calling anything with two args?
RabidGravy struggling to reproduce this "smallbin double linked list corrupted" thing in a bite size chunk of code 06:22
it appears to only happen when running tests with "panda-test" 06:23
which is odd 06:24
quietfanatic And those two multis really shouldn't affect one another. Even if this is an unlikely use case, it looks like a bug to me. 06:25
RabidGravy and also makes it difficult to run valgrind on it
06:26 telex left
TimToady quietfanatic: you're probably completel freaking out the LTM analyzer by mixing a rule and a method... 06:26
quietfanatic Well I wouldn't freak out if I were it. :) 06:27
pmichaud Yes you would.
The entire regex engine is a big mess of "OMG WTF I AM SO FREAKED OUT!"
if you're wanting to put the action methods into the grammar, how about just using a closure? 06:28
FROGGS m: grammar ASDF { multi rule TOP { { say "WAT" } }; multi method TOP (*@, *%) { say "hello" } }; say ASDF.parse("asdf", :actions(ASDF))
camelia rakudo-moar 15c6fb: OUTPUT«WAT␤hello␤(Any)␤»
06:28 telex joined
pmichaud rule TOP { ^ asdf $ { $/.make: "asdf" } } 06:28
quietfanatic Yes, that's what TimToady replied offline
I recall somebody several years ago on here telling me not to do that though 06:29
pmichaud you don't get inheritance/substitutability, but I suspect it works just as well.
TimToady there was a point where make didn't work in rakudo if you used it there
but that's been fixed
FROGGS quietfanatic: yes, it becomes more dirty and dirty when your grammar grows
quietfanatic ah, maybe that wasa it
pmichaud well, putting the two methods in the grammar is going to get dirty as well :)
FROGGS true 06:30
TimToady the main reason for separating out to a separate class is that you can use the same grammar with different actions
pmichaud agreed.
quietfanatic ah, that makes sense too
TimToady such as a pretty-printer
pmichaud that, and also to make it easier to read
I always found it really hard to follow parsing when the action code was intermingled in the rules.
quietfanatic it's easier to read but there's a lot of coupling of method names and organization 06:31
pmichaud yes, I didn't find a way to avoid the name coupling.
tbh, I didn't expect it to be adopted. :) :)
TimToady yeah, and it's difficult to optimize away the double dispatch too
06:31 domidumont joined
TimToady we really need a composition operator of some sort 06:31
with a grammar and a set of actions 06:32
06:32 mattp__ joined
pmichaud 'action' keyword, perhaps? 06:32
grammar Something does SomethingActions
TimToady well, important thing is to set up some wrapping scheme so only a single dispatch is needed
FROGGS I guess the proto has some trouble coping with a rule and a method in the ASDF example here...
quietfanatic or hey, what about
FROGGS I bet it is auto-gen'd wrongish
quietfanatic grammar X is Y { 06:33
whoops, hold on
TimToady this strikes me as the same problem as wrapping role's BUILD submethods around the class's
pmichaud compose the actions directly into the grammar, and find a way to link the two.
quietfanatic grammar X is Y { method TOP { callsame; $/.make: whatever; } }
TimToady but you don't want the grammar saying what it's mixing in 06:34
quietfanatic You can just inherit the grammar and override and supercall all its methods
06:34 Klaudioh left
quietfanatic Y is the original grammar without actions. 06:34
pmichaud could be a runtime mixin
anyway, I vote post-6C :)
TimToady well, I wanna rework the fate system first, I suspect
and that probaby wants to be pre
FROGGS quick! rewrite all the things! /o/ 06:35
:D
TimToady well, I pretty much know how, since STD did it without rescanning the same token multiple times
06:35 domidumont left
quietfanatic p6: grammar ASDF { rule TOP { ^ asdf $ } }; grammar FDSA is ASDF { method TOP { callsame; $/.make: "fdsa" } }; say FDSA.parse("asdf") 06:36
camelia rakudo-moar 15c6fb: OUTPUT«Method 'to' not found for invocant of class 'Any'␤ in block <unit> at /tmp/tmpfile:1␤␤»
06:36 domidumont joined
FROGGS TimToady: btw, I'm going to make run() and shell() able to do what pipe() does right now... I just need a better name for Proc::Status, and I'm thinking IO::Proc might not be too bad 06:36
TimToady it's too late for me to bikeshed :) 06:37
FROGGS TimToady: ahh, that's what you mean (fates), yeah, I'm looking forward to that
pmichaud I consumed all of TimToady's bikeshedding cycles for the night, I think :)
FROGGS pmichaud++ 06:38
TimToady: this if on topic somewhat... a recent nqp commit of yours causes a regression: RT #125345
synbot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=125345
FROGGS this is* 06:39
TimToady yes, I know
been thinking about it, will probably clean up with adding :: and ::: support
FROGGS still giggles internally about the "I asked you if you want to take a seat" :D
ohh
nice
TimToady: I'll port it to jvm for you 06:40
TimToady and the fates cleanup also probably gets us <*...>
quietfanatic now why doesn't that inheritance one work? Another regex engine freakout?
FROGGS yeah... I still wonder if I will every use that
TimToady it's very difficult to write for anything that is not literal abbreviations 06:41
whereas it's easy for the NFA to simply return success instead of failure when it reaches as far as it can
FROGGS p6: grammar ASDF { rule TOP { ^ asdf $ } }; grammar FDSA is ASDF { method TOP($/) { callsame; $/.make: "fdsa" } }; say FDSA.parse("asdf")
camelia rakudo-moar 15c6fb: OUTPUT«Too few positionals passed; expected 2 arguments but got 1␤ in method TOP at /tmp/tmpfile:1␤ in block <unit> at /tmp/tmpfile:1␤␤»
FROGGS p6: grammar ASDF { rule TOP { ^ asdf $ } }; grammar FDSA is ASDF { method TOP { callsame; self.CURSOR.make: "fdsa" } }; say FDSA.parse("asdf") 06:42
camelia rakudo-moar 15c6fb: OUTPUT«Method 'CURSOR' not found for invocant of class 'FDSA'␤ in method TOP at /tmp/tmpfile:1␤ in block <unit> at /tmp/tmpfile:1␤␤»
TimToady surely you want to capture the return value of callsame into $/ 06:43
06:44 SevenWolf left
TimToady grammar ASDF { rule TOP { ^ asdf $ } }; grammar FDSA is ASDF { method TOP { my \m = callsame; m.make: "fdsa"; m } }; say FDSA.parse("asdf") 06:45
m: grammar ASDF { rule TOP { ^ asdf $ } }; grammar FDSA is ASDF { method TOP { my \m = callsame; m.make: "fdsa"; m } }; say FDSA.parse("asdf")
camelia rakudo-moar 15c6fb: OUTPUT«Method 'make' not found for invocant of class 'FDSA'␤ in method TOP at /tmp/o1gWyYnEgP:1␤ in block <unit> at /tmp/o1gWyYnEgP:1␤␤»
TimToady m: grammar ASDF { rule TOP { ^ asdf $ } }; grammar FDSA is ASDF { method TOP { my \m = callsame; m.CURSOR.make: "fdsa"; m } }; say FDSA.parse("asdf")
camelia rakudo-moar 15c6fb: OUTPUT«Method 'CURSOR' not found for invocant of class 'FDSA'␤ in method TOP at /tmp/9r5zfg8Qb0:1␤ in block <unit> at /tmp/9r5zfg8Qb0:1␤␤»
FROGGS m: grammar ASDF { rule TOP { ^ asdf $ } }; grammar FDSA is ASDF { method TOP { my \m = callsame; say m.WHAT; m } }; say FDSA.parse("asdf") 06:46
camelia rakudo-moar 15c6fb: OUTPUT«(FDSA)␤｢asdf｣␤»
FROGGS m: grammar ASDF { rule TOP { ^ asdf $ } }; grammar FDSA is ASDF { method TOP { my \m = callsame; say m.^mro; m } }; say FDSA.parse("asdf")
camelia rakudo-moar 15c6fb: OUTPUT«(FDSA) (ASDF) (Grammar) (Cursor) (Any) (Mu)␤｢asdf｣␤»
quietfanatic I got that far, but it doesn't return what I've made. 06:49
pmichaud sleepytime for me
be back tomorrow evening
TimToady sleep/drive well 06:50
quietfanatic You know, if regexes (as methods) simply returned what they either matched or made, this would be really easy. 06:51
TimToady yeah, always returning both is kinda silly 06:52
quietfanatic method TOP { return process(callsame); }
currently they return Mu
TimToady there are a few places in the P6 grammar that revisit the submatches after make, but not many
quietfanatic in those cases you could just save the matches inside of what you make manually 06:53
TimToady conceivably
06:57 mr-fooba_ left 06:58 mr-foobar joined 06:59 Ven joined
moritz good morning 07:00
yoleaux 11 Jun 2015 22:30Z <ab5tract> moritz: I pushed a 'choices' branch to doc.git for disambiguating, for instance, -f push
RabidGravy marnin
quietfanatic Not that I want to encourage yet another big refactor before Christmas, but having regex routines return their matches is more functional, more object-oriented, more composable, and simpler. 07:01
moritz .tell ab5tract nice. A few small comments: 1) you should probe for the 'fmt' executable before using it. 2) I don't like interactive prompts much; maybe just show the docs for all matches (much as 'man less more' does)? 3) the routine/method distinction in the user interface is weird, because Method ~~ Routine 07:02
yoleaux moritz: I'll pass your message to ab5tract.
quietfanatic and probably more optimizable.
okay I'm done for the night. Later everyone. 07:03
moritz quietfanatic: there's just one problem: those lists need to be lazy
quietfanatic which lists?
moritz and lazy lists aren't very fast
the list of possible matches
quietfanatic ah
I don't entirely understand the situation I guess 07:04
Do you mean this would require more lazy lists, or it would prevent them from working right? 07:06
well, I should really get to sleep soon. I'll be back later. 07:07
07:07 quietfanatic left
moritz it would mean more places where we need lazy lists, which are expensive 07:07
so a performance problem
pmichaud which are *currently* expensive
moritz yes 07:08
when they become faster, it might be worth evaluating the regex design again
RabidGravy does ':::' as a "package part separator" actually exist or is it being silently ignored or turned into something else? 07:09
moritz after all, the lazy list of matches/cursors is what the design docs say :-)
pmichaud regex methods return cursors, though. Making them return match objects will get expensive.
07:09 _mg_ joined
pmichaud (and returning match objects directly was pge's original model, which needed to be abandoned in favor of cursors) 07:11
RabidGravy lizmat, I don't think I'm going to be able to golf that crashiness - it doesn't even do it if I run the tests in question on their own, only with "panda-test" 07:19
but github.com/jonathanstowe/Audio-Snd...ree/crashy reliably does it with 2015.05-76-g94eed8f built on MoarVM version 2015.05 07:25
pmichaud Oh! The Slip type may also be a way to relax the need for immutable iterators! 07:30
very intriguing.
RabidGravy (and yes it does still do it with the correct "multi sub infix:<~~> ......)
dalek c/choices: cdff8ec | ab5tract++ | bin/p6doc:
Remove confusiong between routine/method
 07:31
moritz pmichaud: do you have any links to Slip types? 07:33
07:33 kaare_ left
moritz googling for it mostly shows up clothing stuff :-) 07:33
Ven waves 07:34
dalek ecs: 15083d9 | pmichaud++ | S07-glr-draft.pod:
Add discussion note about immutable iterators.
 07:36
pmichaud moritz: S07-glr-draft.pod describes the proposed Slip objects 07:37
(it's a type we just came up with today, largely replaces Empty)
07:39 g5 joined
RabidGravy (^(10)).map({$_}) isn't lazy is it? 07:40
pmichaud depends on the target 07:41
it can be lazy
07:41 [TuxCM] left
RabidGravy it's assigning to an array 07:41