»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'p6: say 3;' or rakudo:, or /msg camelia p6: ... | irclog: irc.perl6.org or colabti.org/irclogger/irclogger_logs/perl6 | UTF-8 is our friend!
Set by moritz on 22 December 2015.
00:08 Ben_Goldberg joined, Ben_Goldberg left 00:09 BenGoldberg joined 00:10 BenGoldberg_ left 00:15 nadim joined 00:18 _nadim left 00:20 vendethiel joined, _nadim joined 00:21 nadim left 00:26 Actualeyes joined 00:31 znpy joined 00:32 tmch left 00:35 telex left 00:39 kurahaupo joined, kurahaupo left 00:40 spider-mario left 00:43 dvinciguerra left, vendethiel left 00:48 telex joined 00:50 bjz left 01:12 cognominal left 01:14 vendethiel joined 01:18 perturbation left 01:37 Fatalnix_ left 01:38 Fatalnix_ joined 01:40 bjz joined 01:44 kalkin- joined 01:45 ilbot3 left 01:47 ilbot3 joined 01:48 kalkin-_ left 02:00 BenGoldberg left 02:02 BenGoldberg joined 02:16 perlawhirl left, vendethiel left 02:19 er2 left 02:26 kid511 left 02:48 noganex_ joined 02:51 noganex left 03:01 sno left 03:15 bjz_ joined, TeamBlast joined 03:16 bjz left 03:21 BenGoldberg left 03:37 cdg joined 03:48 cdg left 03:58 kaare_ joined 03:59 woosley joined, woosley left 04:03 bjz_ left
skids m: my @s = (supply { my $l = 5000 + $_ * 100; emit [*] $l..^$l+100; done } for 0..^10); my $s = [*] await(@s); say now - BEGIN now; 04:05
camelia rakudo-moar d0c675: OUTPUT«0.064998␤»
skids m: my @s = (supply { my $l = 50000 + $_ * 1000; emit [*] $l..^$l+1000; done } for 0..^10); my $s = [*] await(@s); say now - BEGIN now;
camelia rakudo-moar d0c675: OUTPUT«(signal ABRT)»
skids m: my @s = (supply { my $l = 500000 + $_ * 10000; emit [*] $l..^$l+10000; done } for 0..^10); my $s = [*] await(@s); say now - BEGIN now;
camelia rakudo-moar d0c675: OUTPUT«7.65016807␤» 04:06
skids the middle one reliably gets an Abort on my local machine as well.
04:09 bjz joined
Hotkeys skids: i.imgur.com/o94Yusy.png 04:09
skids Hrm... 04:10
I wonder if it's a magic number depending on machine speed/cores. 04:12
Or -- are you on Linux? 04:13
Oh wait -- try it by -e not in the REPL. 04:15
Hotkeys I am on the windows 04:16
oh
still workin' okay i.imgur.com/t2bobPT.png 04:17
but I am also on windows so
/o\
skids interesting that REPL is 2 slower. 04:18
Hotkeys yeah 04:19
much faster on the 500000 one 04:20
i.imgur.com/0cvR6Fh.png
oops
forgot to up the other numbers
about 2 seconds faster on the big one 04:21
skids For me the 500000 one aborts in the REPL.
Maybe something to do with handling of the "BEGIN now". 04:22
(the REPLY being slower, I mean)
Oh, actually I can't get an abort without doing the 'BEGIN now' part. 04:25
Hotkeys the repl does it in 10 and the -e does it in 7.8 04:32
for me
also I thought the idiom was now - INIT now
is BEGIN the better way to do it?
skids Hrm dunno.
Yeah I guess INIT is better. 04:36
well anyway, time for bed 04:40
04:42 skids left 04:45 znpy left 04:48 Cabanossi left 04:50 Cabanossi joined 04:59 bjz_ joined 05:00 bjz left 05:15 khw left 05:41 ponbiki joined 05:50 TimToady left 05:52 TimToady joined 06:04 ZoffixWin left 06:07 labster left 06:15 labster joined
sortiz m: say ?IterationEnd; # Why True? 06:16
camelia rakudo-moar d0c675: OUTPUT«True␤»
06:23 jameslenz left 06:28 ZoffixWin joined 06:41 CIAvash joined 06:48 domidumont joined
timotimo why care about the details of IterationEnd? its only quality is its identity 06:52
06:53 domidumont left, domidumont joined
timotimo i thought we would've put IterationEnd into RakudoInternals by now 06:54
06:56 firstdayonthejob joined
kaare_ If I 'panda install App::GPTrixie' should I be able to find the bin/gptrixie somewhere? 07:04
teatime kaare_: I ran into this the other day
kaare_: you probably need to do 'panda rehash' afterward
timotimo you mean "rakudobrew rehash" 07:05
i think?
m: say uniname(0x180e)
camelia rakudo-moar d0c675: OUTPUT«MONGOLIAN VOWEL SEPARATOR␤»
teatime heh yes I guess so. 07:06
and I guess that assumes the use of rakudobrew.
so, I still don't know how it works in the general case.
timotimo in the general case, you'll have put the folder that binaries land in into your $PATH
like panda tells you at the end of bootstrap, iirc
teatime ok. 07:07
maybe I should make a PR to add 'rakudobrew rehash' to the rakudobrew README. it befuddled me for quite a while.
timotimo could be a good idea, yeah
07:16 jameslenz joined 07:18 fireartist joined 07:37 firstdayonthejob left 07:39 atweiden left 07:43 fireartist left 07:44 fireartist joined, Actualeyes left 07:49 _nadim left 07:50 _nadim joined
sortiz timotimo, If Iterable is a public API, IterationEnd should be. I understand with should be tested by identity, can't be other way. But as a Mu, I expect to be False :) 07:56
07:56 CIAvash left
sortiz s/with/why/ 07:57
07:57 wamba joined
timotimo it's a defined Mu, though 07:59
m: say so Mu.new
camelia rakudo-moar d0c675: OUTPUT«True␤»
08:03 Actualeyes joined
sortiz Yep, at first that surprises me, but it's irrelevant, though 08:03
timotimo Iterelevant :) 08:04
08:08 sivoais left 08:12 sivoais joined 08:14 darutoko joined 08:29 ufobat joined 08:30 sno joined 08:36 rindolf joined
masak antenoon, #perl6 08:44
teatime wb. 08:45
masak needs regular scheduled downtime, much like other alie^Whumans
anyone here good at database design (or wants to discuss it regardless of skill)? 08:48
I have a design I'm about to commit to, and I'd like for someone to check if it's sane...
teatime can you link it? I'm curious, if unskilled. 08:49
masak yeah. about to put together a gist.
nine masak: shoot 08:50
masak basically this: gist.github.com/masak/b357e469b40403672839 08:53
(I don't know how to best transmit a database schema, but I think that gets the idea across)
teatime you could use explicit foreign keys. 08:55
and auto-increment IDs
masak both of those are good ideas. consider it done.
teatime and do you have to store JSON rather than create more DB structure
that's kindof a red flag, but may make sense for your specific case. 08:56
masak (I'm actually going to go with postgres, not sqlite3. but I wanted an environment to type my commands into to double-check them)
teatime: yes, that red flag is why I brought it up here.
and it's borderline off-topic, except that my web application is running Perl 6.
teatime: the JSON blob is basically saying "hands off, DBMS, I'm a black box and schemaless!" 08:57
which is a red flag until proven innocent, kind of
teatime yes. 08:58
08:58 TEttinger left
masak and that's where I'd like to be further enlightened, too. I know about the normalization forms, and I basically know why they are good and desirable. 08:59
but under which circumstances should and shouldn't I use a JSON blob like this in a database design?
what's the underlying principle that makes this a red flag, and how can I check whether I qualify?
DrForr AKA "I could have any random cr*p in me, have fun guessing what my data layout might look like!" 09:00
09:00 ninjaaron joined
masak yes, I appreciate that point. 09:01
teatime one benefit of RDMS is auto-checking referential integrity. kindof like how in a typed language the compiler can detect more errors than in a loosely typed language. if you use blobs like that, you at least lose that benefit for that data.
masak the Move table basically quickly descends into being... CouchDB.
teatime yes, if you're just going to store schema-less JSON blobs, might as well use a nosql DB.
ninjaaron is there a simple way to do "my @x = `ls -l`" in Perl6? 09:02
masak nine: qx[ls -l]
er, ninjaaron
sorry, nine :)
moritz ninjaaron: my @x = qx/ls -l/.lines
masak ah, moritz++ saw the need for .lines which I missed :) 09:03
ninjaaron moritz, masak Thanks! Does that grant a shell to the command?
Timbus yo.. just a headsup. postgres has json specific support: www.postgresql.org/docs/9.3/static/...-json.html
masak teatime: it seems to me that if I don't have *references* in my JSON -- and I don't -- then I don't need to feel bad about losing checking for those.
teatime: remains then checking data types. 09:04
Timbus not that i'd suggest using it, but its there
teatime masak: and there's also checking on- exactly, the types of data, and existence/non- of items.
masak Timbus: why wouldn't you suggest using it?
Timbus teatime is handling that argument for me
teatime also I dunno how efficient variable-length field lookups are, but I don't know much about SQL.
Timbus :P
masak teatime: checking types/existence is a benefit that might do me some good in the future, yes. 09:05
teatime there are also disadvantages to laying everything out in SQL.
masak teatime: but I have to weigh it against the cost of increased schema complexity.
right.
teatime exactly.
masak I have three types of moves.
teatime you could always use an ORM :)
masak so if I started laying this out, I'd have to model a sum type in SQL. 09:06
yes, I could use an ORM, of course :)
it doesn't sound more attractive than "serialize the stuff into JSON"... :)
moritz you could not use an ORM, of course :-)
teatime that's a fair enough assessment.
masak .oO( <intuitionist> you could not not use an ORM, of course :) ) 09:07
teatime in general, though, you may find that thinking about your data model up-front is time well spent, regardless of whether it's to create an SQL schema or not.
09:07 ninjaaron left
masak oh, no doubt 09:07
teatime while we're on the subject. is there any mature / non-painful JSON schema tool? 09:08
masak it's that kind of thinking that led me to storing the *moves* rather than the board state
Timbus www.compose.io/articles/is-postgre...-database/ here ya go btw, masak. nice overview and stuff
moritz teatime: for Perl 6, or in general?
teatime in general.
moritz jsonschema is mature, but not painless 09:09
teatime the last time I tried to make use of actual "JSON Schema" I eventually gave up, it was terrible.
I don't remember in what way, just that it was infuriating.
sortiz masak, I'm interested in the details_json structure: what's a "move".
? 09:10
Timbus rjbs (?) made a json schema tooly thingy.. which looked okay.
masak sortiz: detailed here: github.com/masak/nex/blob/master/README.md
Timbus was it rx? yes. google says it's rx 09:11
masak sortiz: besides those two move types, there's also "swap", "resign" and (I guess) "timeout"
teatime a generic schema language w/ a validator that could do both JSON and YAML would be sweet. even if it only handled the most basic types.
Timbus thats the one.
teatime I mean it could even do Python/Perl/etc. aggregate datastructures. 09:12
09:13 perlawhirl joined 09:14 RabidGravy joined
teatime Timbus: looks like what I asked for :) 09:17
09:17 Actualeyes left
sortiz masak, ok, so for every move type, the data to store is different; so json seems good to me. The other option is to create a table per type, per move store the type and the row-id of the specific entry. I don't see the need. 09:18
masak right; that's what I meant by "sum type" 09:22
it seems over-the-top to me too
if I could imagine some situation where I'd get returns from having all that structure in the database... but I can't 09:23
timotimo to be fair, postgresql outperformn many "nosql" DBs out there in some benchmarks :)
sortiz Yep, RDBMS are good for *regular tabular* data, otherwise nonsql are better options.
masak right; great.
so the underlying principle is "follow the normalization principles for regular tabular data; disappear into a blob when the data gets heterogenous" 09:24
that's the kind of rationale I was looking for :)
timotimo the thing with nosql is that you almost never have something truly schemaless
masak indeed; I *could* make this into a schema
it's a sum type, and I know how to model one in SQL 09:25
timotimo for this use case, i'd say a json field is fine
masak cool.
sortiz timotimo, Yes, but many *other* "nosql" outperform any SQL solution.
Sometimes a simple KV store can suffice. 09:26
RabidGravy with Sofa I'm working toward being able to do a Document Object Mapping thing, where your class definitions are the schema
also see rethinkdb which is "NoSQL" but has some relational and schema capabilities 09:27
Hotkeys m.imgur.com/xY2xDxo?r 09:29
Oops wrong chan
timotimo poor doge 09:30
masak such scorpion. much regrets.
teatime oh god, Hotkeys, that's terrible.
masak poor scorpion 09:34
Woodi hi today :) 09:35
masak .oO( always jam yesterday or jam tomorrow, never jam today... ) 09:37
timotimo suppe.tobold.cc/post/681686396/Image - have a pleasant animal picture/gif to compensate 09:39
DrForr No boom today. Boom tomorrow.
masak teatime, Timbus, DrForr: thank you for your input on schemas. that was valuable to me.
timotimo big badaboom
teatime thank you.
for the eyebleach.
timotimo YW
:3
DrForr Or "*Raspberry*. Only one man would *DARE* give me the raspberry."
timotimo Whatever Starrr! >:( 09:40
Woodi masak: about: CREATE TABLE Game(id INTEGER PRIMARY KEY); looks like perfect but players ids or just description could be added
masak Woodi: yes, of course. 09:41
I guess I like to focus on one slice of the problem space at a time. ;) 09:42
there's also no mention of the game type, or the board size.
or timestamps for the moves. 09:43
DrForr Go, by chance?
timotimo not quite
it's called nex and lives on github under masaks user
masak github.com/masak/nex
DrForr You have my interest :) 09:44
masak I already have a working game engine, in 189 LoC Perl 6.
timotimo DrForr: so far you know it's a game that has moves to it ... so ...? :D
DrForr Ah, got it. (having played Hex...)
Woodi masak: as with 99% things, schema depends on how you use it. even redundand columns are usefull. eg. you can have column with values that are in json blob but you use them a lot. a bit like extracting some data from raw data in data mining
masak now I "just" need to expose it on the web so I can play it with a friend...
timotimo Woodi: why not just use a view for that?
Hotkeys Nice masak
nine masak: note that PostgreSQL also supports inheritance of tables. That could work for your different move types, too.
masak nine: intriguing -- even if I don't believe I'll make use of it :) 09:45
nine: url?
Woodi timotimo: for no-players-ids thing or second one ? 09:46
masak Woodi: are you familiar with the distinction between OLTP and OLAP?
nine masak: what I don't like about your use of JSON is, that your JSON does actually have very limited options for the structure. So it's far from schemaless. Question is: is it imaginable that you'd ever want to query only parts of this JSON-data or use parts of it in a where clause or an aggregate?
timotimo Woodi: for whatever's in the json blob
Woodi masak: not from my head, but I'm sure wiki will have short definition for them :)
masak nine: actually, for my use of the JSON I might as well use a comma-separated list 09:47
nine: the only thing JSON gives me over that is descriptive key names
nine masak: www.postgresql.org/docs/9.5/static/...herit.html
Woodi timotimo: becouse it is blob and situation is that that data are black boxed. but some part can be copied to some column, if usefull
masak Woodi: my point is that the shape of the schema depends a lot on whether you're doing mostly-reads or mostly-writes on the database 09:48
nine masak: then you can as well use PostgreSQL's arrays for these lists and get Perl 6 arrays back from your queries.
AFAIK DBIish already supports that
timotimo i don't understand why you would copy that data out instead of just using a view, though
DrForr If I were back in the US I'd have access to my Hex strategy book.
masak Woodi: your advice wrt denormalization is good for mostly-reads, but bad for mostly-writes :)
nine: ooh, excellent
09:48 labster left
masak nine: so in the end I had an XY problem, and you managed to suggest the right solution. awesome. 09:49
nine++
nine masak: I think your biggest problem is that PostgreSQL really has excellent support for all the possible designs. Even if you want to select only part of the JSON or need an index on it, you can do that.
masak sounds like a nice problem to have, at least in the long run 09:50
DrForr: would that be the Cameron Browne book, by any chance?
Woodi masak: right. but you need to analize that data later and parse that blob anyway :) 09:51
DrForr The name rings a bell, let me look.
masak www.amazon.com/Hex-Strategy-Making-...1568811179
DrForr Yes, it is. 09:52
masak that book is... decent. nothing wrong with it. but at some point I felt that we had a much bigger corpus online than was collected in that book.
at www.hexwiki.org/ to be exact.
still hoping to bring that site back at some point... :/
timotimo MadcapJake: i've had an expert friend tell me what the difference between VAO and VBO is. interested? 09:53
Hotkeys After reading the Nex readme I was like "this sounds like go" but then I read the wiki and it isn't go
masak Hotkeys: connection games have quite a different feel than Go.
DrForr No, it's not.
Woodi timotimo: I assume parsing json is not sql db job. so: no possibility to create view
masak Hotkeys: in Go the main resource is territory. in Hex and Nex, your main abstractions tend to be chains and connections. 09:54
Hotkeys Ah
timotimo postgres has json and jsonb fields along with operators to operate on them, and it can also create views and indices from json blobs
sortiz Btw, what's the best/fastest perl6's JSON engine? Need to use that in DBIish for the JSON column types.
Hotkeys Provably json::fast
masak provably!
Hotkeys Probably*
tadzik provably too 09:55
masak .oO( it has "::fast" in the name. QED! )
Hotkeys Yeah I realized that I can't just leave that typo :p
Masak++
Actually isn't there a json::faster
masak *groan*
timotimo there was, but it got nommed by JSON::Fast
masak I'd skip directly to json::fastest
DrForr I keep meaning to finish writing my Kamisado solver. en.wikipedia.org/wiki/Kamisado - There's apparently a 43-move forced win.
timotimo json::faster only provided a to_json that was fast, while json::fast only provided a from_json that was fast 09:56
Hotkeys Ah ok
sortiz timotimo, there are in core? 09:57
masak DrForr: by the looks og images.google.com, looks like a fun game (with no sense of color scheme) :P
timotimo sortiz: there are what in core?
sortiz json::faster and/or json::fast 09:58
timotimo no, both in the ecosystem
DrForr masak: If you're coming to NLPW I'll bring my set.
sortiz timotimo, There are plans for something in core? 10:00
timotimo not for json, as far as i know.
sortiz ok, thks 10:01
timotimo we do have to_json and from_json in rakudo, but that's not per perl6 spec
m: say from_json('["hi sortiz", "how are you"]')
camelia rakudo-moar d0c675: OUTPUT«5===SORRY!5=== Error while compiling /tmp/rB9ZFrs6bn␤Undeclared routine:␤ from_json used at line 1. Did you mean 'from-json', 'to-json'?␤␤»
timotimo m: say from-json('["hi sortiz", "how are you"]')
camelia rakudo-moar d0c675: OUTPUT«[hi sortiz how are you]␤»
timotimo m: say from-json('["hi sortiz", "how are you"]').perl
camelia rakudo-moar d0c675: OUTPUT«$["hi sortiz", "how are you"]␤»
Hotkeys How do those compare to the functions in json::fast 10:02
timotimo they are slower 10:03
10:03 araujo joined 10:04 araujo left 10:05 araujo joined
sortiz Umm, oki. 10:05
10:06 araujo left
timotimo i think they were brought in for supporting older variants of our compunitrepo spec 10:07
10:07 araujo joined
sortiz timotimo, and seems buggy! 10:07
timotimo or maybe we still need it for that?
masak DrForr: I'll not be at NLPW, sadly :/ 10:08
sortiz m: my @a = from-json('["hi sortiz", "how are you"]'); say @a.elems
camelia rakudo-moar d0c675: OUTPUT«1␤»
10:09 _nadim left, araujo left
timotimo yeah, it returns an itemized array there 10:09
masak DrForr: have to be much more selective now that @family > 0
timotimo m: my @a = from-json('["hi sortiz", "how are you"]'); say @a.list.elems
camelia rakudo-moar d0c675: OUTPUT«1␤»
timotimo uhm...
10:09 araujo joined, araujo left, araujo joined
timotimo m: my @a = from-json('["hi sortiz", "how are you"]'); say @a.perl; say @a[0].perl 10:10
camelia rakudo-moar d0c675: OUTPUT«[["hi sortiz", "how are you"],]␤$["hi sortiz", "how are you"]␤»
timotimo ah, yeah
when you assigned the itemized list to the array, you got an array with a single element in it
that's why .elems and .list.elems gave 1
sortiz Yep, the itemized return is the bug. 10:11
10:11 _nadim joined
timotimo IMO it could move to Rakudo::Internals 10:14
DrForr masak: I understand, no worries.
Woodi anyway, "schemaless" things are recipe for disaster or unhappines at least :) usually stable schema skeleton is known and free form data can be put into "extension". I think never protocols are designed with ways to extend them 10:20
newer* ?
teatime Is it safe to assume that this syntax for regexes is not available in rakudo? en.wikipedia.org/wiki/Parser_Gramm...#Internals 10:24
for defining regexes/grammars, not Perl operators. 10:25
timotimo the operator precedence parser is a special feature built in a grammar, but it's not a feature of perl6 grammars in general 10:26
teatime forgive me, I don't understand. 10:27
do you mean, operator precedence is something you have to explicitly/manually implement for a given grammar?
if so, is there many a module to help?
s/many/maybe/
timotimo you understood correctly 10:28
we don't seem to have an OPP in the ecosystem yet 10:29
teatime I really enjoy this perl6 feature. I should probably learn more about parsing in general.
tadzik OPPa grammar style
masak teatime: I implemented that once, if you're interested
timotimo for 007? :)
masak for GGE 10:30
teatime masak: sure. I've seen one or two methods.
masak oh yeah, and for 007
teatime: strangelyconsistent.org/blog/what-y...nt-without
perlawhirl does p6 have an equivalent to PERL5LIB ?
tadzik yes, it's called PERL6LIB
teatime there's supposedly a PERL6LIB, but the doc said it may go away
masak the doc is wrong 10:31
or at least alarmist
teatime I think there are lots of wrong perl6 docs out there :) 10:32
perlawhirl yeah i tried PERL6LIB but it still doesn't find my module
masak try harder! :)
teatime: yes, but there's an excellent team here working quite hard on making it less wrong all the time.
RabidGravy yeah it definitely works as half the modules tested on travis use it 10:33
teatime masak: no offense intended.
perlawhirl right... commencing try harder :D
teatime and there's nothing you can do about blog posts from e.g. 2006 anyway.
masak m: class C { method foo {}; method bar { self!foo() } }; C.new.bar 10:34
camelia rakudo-moar d0c675: OUTPUT«5===SORRY!5=== Error while compiling /tmp/zJP3e7gf8_␤No such private method 'foo' for invocant of type 'C'␤at /tmp/zJP3e7gf8_:1␤------> 3 { method foo {}; method bar { self!foo(7⏏5) } }; C.new.bar␤»
masak this error message leaves something to be desired. the most likely problem is that the author wanted to make `method foo` private, but forgot/failed somehow. 10:35
teatime: oh, none taken. and I'm actually not part of that team, even though I've told myself many times that I should put some tuits on that. 10:36
10:40 dvinciguerra joined
teatime masak: do GGE::Exp / GGE:: OPTable still exist/work ? 10:41
10:42 spider-mario joined
teatime 'Tests: 43 Failed: 42' hehe, nice. 10:43
masak teatime: was gonna say 'probably not'
this is where I also hand you that all-uppercase NO WARRANTY part of the license :P 10:44
teatime thank you for the inspiration nonetheless
masak hold on, I'll link you to the working one in 007 :) 10:45
RabidGravy well, I just tested 48 of my modules and they all pass
masak github.com/masak/007/blob/00fac05d...#L289-L399 10:46
actually, github.com/masak/007/blob/00fac05d...#L218-L399 10:47
in 007 things happen in two levels, sort of.
because EXPR is basically `rule EXPR { <termish> +% <infix> }` 10:48
we handle infix precedence resolution in `EXPR`, and prefix/postfix precedence resolution `termish`
Perl 6 is not so lucky, since the precedence of infixes and prefixes/postfixes can mix
teatime perhaps I am just being overly ambitious, given my compsci-fu. 10:51
kaare_ teatime, timotimo: Thanks for panda/rakudobrew help. 10:52
timotimo sure :)
masak teatime: I for one encourage you heading down this road :) 10:54
I've had a lot of fun on it over the years
10:58 FROGGS joined 11:01 Actualeyes joined
sortiz m: use Test; is \(:id(1), :name('First')), \(:name('First'), :id(1)); # I don't see that before! 11:04
camelia rakudo-moar d0c675: OUTPUT«This representation (VMIter) cannot unbox to a native string␤ in sub is at /home/camelia/rakudo-m-inst-1/share/perl6/sources/C712FE6969F786C9380D643DF17E85D06868219E (Test) line 136␤ in block <unit> at /tmp/p8rcCdmq55 line 1␤␤»
masak looks like a bug of some kind 11:09
11:17 vendethiel joined 11:19 edehont joined
teatime <![{]> # }], you're welcome vim 11:19
lolol, I suspect will use this trick. 11:20
timotimo yeah :| 11:22
jnthn Guess it needs to learn about Perl 6 char classes :)
teatime yeah. there's a perl6 syntax definition but hard to tell how old it is 11:24
and/or how old the typical distro vim release is
11:24 kid51 joined
timotimo that has recently (well, maybe a year ago now) gotten a really nice overhaul 11:25
but i don't know if that version is shipped with your vim
it is on github, though. as "vim-perl" i think?
RabidGravy yeah 11:27
teatime installed. we'll give it a shot.
11:41 vendethiel left
RabidGravy today I am hating the interaction of roles and BUILD :( 11:45
11:46 Begi joined
timotimo i think many people aren't quite happy with that, either 11:47
RabidGravy is there any way at all to ensure the role's attributes get initialised in the presence of a BUILD in a consumer without the consumer doiing anything special? 11:50
timotimo you can probably call $?PACKAGE's BUILD or some trick like that 11:51
RabidGravy but the consumer has to do that 11:52
timotimo not if you override the BUILD via the role?
RabidGravy role's BUILD doesn't get called at all 11:53
timotimo darn :|
lizmat RabidGravy: even if it's the only BUILD in town
RabidGravy oh, if the role has a BUILD and the consumer doesn't then it's fine 11:54
11:54 kid51 left 11:55 vendethiel joined
RabidGravy but if the role is part of a public interface then you can't guarantee that to be the case 11:55
lizmat I guess we will need some compose trickery, because if the client has a BUILD, the role's build is never really composed, right ? 11:56
timotimo .o( will we create a ROLEBUILD that gets special magic added to it? ) 11:57
jnthn We've sometimes speculated that submethods will compose in such a way that all of them get called 12:09
Probably a 6.d feature, which will be made 6.d+ only by passing an option to the .^compose method 12:10
(As in, having the compiler do so)
teatime masak: what is the purpose of 007? Exercising features of Perl6? Personal learning/playing?
jnthn teatime: So far as I know, research/exploration of macros/slangs to inspire an implementation of these features in Perl 6. 12:11
And probably some amount of other reasons :) 12:12
teatime hrm, you know what would be sweet (and might exist): an aggregator of perl6 blogs. 12:15
planet.perl6.org ? :)
tadzik perl6planet is like that
yeah
lizmat pl6anet.org 12:16
timotimo we could totally redirect to pl6anet.org from planet.perl6.org 12:17
to be fair, the "Blogs" link in the footer of perl6.org already links to the pl6anet 12:19
12:23 sufrostico joined 12:26 Josh4all joined
MadcapJake timotimo: yes, do tell! 12:27
12:34 perlpilot joined
RabidGravy right, having reviewed all the options I have concluded that there is no hack I can do in a role that will ensure its attributes will get initialised whatever the consumer does so am going for the "document in bold" option 12:41
[Coke] ORM-- # my 2¢ on a backscroll
RabidGravy I think that could be simplified to "software--" without the need for backscroll 12:42
stmuk_ RabidGravy: does github.com/stmuk/MoarVM/commit/eba...ae0d4b0fc3 fix your FreeBSD 8 issue? 12:48
12:48 Actualeyes left
Begi m: say 'yes !' if 'a' ~~ ('a', 'b', 'c'); #how can I do ? 12:48
camelia ( no output )
12:48 Fatalnix_ left
jnthn m: say 'yes !' if 'a' (elem) ('a', 'b', 'c'); # one way :) 12:50
camelia rakudo-moar e5174b: OUTPUT«yes !␤»
RabidGravy or
m: say 'yes !' if 'a' ~~ any('a', 'b', 'c')
camelia rakudo-moar e5174b: OUTPUT«yes !␤»
Begi Okey, thanks ! I've forget the 'any' :-° 12:51
RabidGravy stmuk_, let me see
12:57 cdg joined
timotimo MadcapJake: so, a VAO combines configuration options, so to speak, of multiple VAOs into one object that you can bind in one step. sadly, it seems like you can't use a VAO without writing a tiny little pixel shader to work it, as opposed to VBOs, for which there are sufficiently many functions to set up all kinds of stuff 12:57
stmuk_ does the clang 3.4 src/core/interp.o wait :/ 12:58
timotimo hah
MadcapJake timotimo: neat, so stick to the VBOs ;)
timotimo ideally, we'd bind all functions necessary to work shaders, though 12:59
sortiz RabidGravy, In DBIish I have the same problem with the roles. :-( 13:02
RabidGravy stmuk_, it appears to help yes 13:03
stmuk_ ty
RabidGravy but yeah that thing after interp.o taking forever is a pain, I've abandoned building on FBSD10 on a VM here more than once 13:07
13:07 Actualeyes joined, Ven joined
stmuk_ clang 3.6 at least (which is in ports) works fine 13:08
there are probably NC problems with using a diff compiler to the system one however, hopefully FB 11 has a more recent cd lang 13:09
^ recent clang
13:12 Sgeo left, Josh4all left
masak teatime: what jnthn said. 13:15
teatime: 007 emerged as a basically-necessary side project to prototype Perl 6 macros in something more "fluid" than Rakudo (and where the consequences of failed experiments are smaller) 13:16
(and feel free to take the "fluid" as meaning "I understand the 007 code base better than the Rakudo code base") :P 13:17
teatime :) 13:18
as a much-smaller-scope language parser, it is a super-useful example.
13:24 perlawhirl left 13:26 edehont left 13:28 Ven left 13:29 skids joined 13:38 mullagain joined 13:41 ufobat left
rudi_s Is this expected? 13:42
m: sub foo(:%args) { say %args.perl }; foo :args(a => 1); foo :args(a => 1, b => 2);
camelia rakudo-moar e5174b: OUTPUT«:a(1)␤Type check failed in binding %args; expected Associative but got List ($(:a(1), :b(2)))␤ in sub foo at /tmp/DPyNLmRhpz line 1␤ in block <unit> at /tmp/DPyNLmRhpz line 1␤␤»
rudi_s Do I have to call it with foo :args(%(a => 1, b => 2)) or is there a better way?
13:42 zakharyas joined 13:43 molaf joined
masak I use private methods in Perl 6 more than I thought I would 13:43
rudi_s: you do have to do that, to a first approximation
jnthn It's expected 13:44
masak rudi_s: `a => 1, b => 2` is definitely a List and not an Associative
jnthn m: sub foo(:%args) { say %args.perl }; foo(:args{a => 1, b => 2});
camelia rakudo-moar e5174b: OUTPUT«{:a(1), :b(2)}␤»
jnthn That's the shortest way to force a hash
rudi_s Yeah. Automatic "conversion" would've been nice in this case, but I can live with it.
Thanks.
jnthn Well, when you can be explicit about the hash in the same number of chars... :) 13:45
RabidGravy or "foo(args => { a => 1, b => 2 })"
rudi_s Yeah, {..} is really nice. Thanks.
13:46 ecocode joined
teatime you could also take a list of pairs as varargs, and make it into a hash yourself at the beginning of the function? would that be a bad idea? 13:49
I thought arbitrary/not-validated variable-length kwargs was built-in
masak they are 13:51
m: sub foo(*%_) { say %_.perl }; foo(oh => "hai", how => "are you?")
camelia rakudo-moar e5174b: OUTPUT«{:how("are you?"), :oh("hai")}␤»
jnthn We just generally don't confuse a list of pairs with a hash, 'cus they ain't the same thing. Something somewhere has to say it wants to make a hash of them, 'cus it's an information-losing operation
masak even in Perl 5 they are not the same thing 13:52
14:09 sue joined
[Coke] jnthn++ for the work on the heap analyzer. 14:14
14:14 Roamer` joined
masak .oO( the ultimate heapster ) 14:14
sortiz m: my $f = sub :: { (1,2,3,4) }; say $f(); # So far, so good 14:16
camelia rakudo-moar e5174b: OUTPUT«(1 2 3 4)␤»
14:16 pmurias joined
sortiz m: my $f = sub :: { (1,2,3,4) }; for $f() { say $_ }; # What! 14:17
camelia rakudo-moar e5174b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/KyjgNPrCLc␤This appears to be Perl 5 code␤at /tmp/KyjgNPrCLc:1␤------> 3my $f = sub :: { (1,2,3,4) }; for 7⏏5$f() { say $_ }; # What!␤»
sortiz m: my $f = sub :: { (1,2,3,4) }; for $f.() { say $_ }; # Works, LTA
camelia rakudo-moar e5174b: OUTPUT«1␤2␤3␤4␤»
Roamer` Hi, is it expected that I can't do a ~~ match within a grammar's action method? 14:18
m: grammar a { token TOP { \w } }; class Actions { method TOP($/) { if "abc" ~~ /b/ { say "yep" } } }; a.parse("w", :actions(Actions))
camelia rakudo-moar e5174b: OUTPUT«Cannot assign to a readonly variable or a value␤ in method TOP at /tmp/eCiUnztAtu line 1␤ in regex TOP at /tmp/eCiUnztAtu line 1␤ in block <unit> at /tmp/eCiUnztAtu line 1␤␤»
masak sortiz: congratulations, you managed to trigger a p5 warning false positive!
moritz Roamer`: yes, because you specified $/ as a parameter, which makes it read-only 14:19
Roamer` moritz, ah, it works if I rename the parameter. Thanks!
moritz Roamer`: if you name it $m or so, a smart-match which sets $/ is possible again
sortiz rakudobug, I hope ;.)
masak, if it were only a warning, but is a compiler error :-( 14:23
masak sortiz: arguably the p5 compat error overreaches a bit here, yes 14:24
sortiz: in order for it to be p5 code, it'd have to be `$f` without the `()` after
sortiz m: my $f = 1; for $f { .say }; # and this works well. 14:25
camelia rakudo-moar e5174b: OUTPUT«1␤»
masak oh! 14:26
the compiler thinks that the parens are for the Perl 5 list!
$ perl -Mstrict -wE 'my $f; for $f() {}' 14:27
syntax error at -e line 1, near "() "
but that still doesn't make sense. seems there has to be whitespace there.
sortiz m: my \f := sub :: { (1,2,3,4) }; for f() { say $_ }; # and this 14:28
camelia rakudo-moar e5174b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/M4xOpe7wia␤Variable '&f' is not declared␤at /tmp/M4xOpe7wia:1␤------> 3my \f := sub :: { (1,2,3,4) }; for 7⏏5f() { say $_ }; # and this␤»
jnthn The form f() always looks up &f (unless there's some type f, in which case it's a coercion type) 14:29
m: my &f := sub :: { (1,2,3,4) }; for f() { say $_ }
camelia rakudo-moar e5174b: OUTPUT«1␤2␤3␤4␤»
jnthn m: my \f := sub :: { (1,2,3,4) }; for f.() { say $_ }
camelia rakudo-moar e5174b: OUTPUT«1␤2␤3␤4␤»
sortiz jnthn, In my use case f (or $f) is in fact an object with a defined CALL-ME method 14:30
diakopter m: class f {}; my &f := sub :: { (1,2,3,4) }; for f() { say $_ } 14:31
camelia rakudo-moar e5174b: OUTPUT«Method 'shortname' not found for invocant of class 'Perl6::Metamodel::CoercionHOW'␤ in block <unit> at /tmp/JrxCb7whpD line 1␤␤»
diakopter You're a CoercionHOW!
14:32 telex left
jnthn sortiz: $f.() will also work as a workaround for now, while the Perl 5 warning is toned down some 14:32
I guess making it insist on whitespce before teh ( would help
14:33 sufrostico left 14:34 telex joined, khw joined
sortiz jnthn, ok. At the moment I'll live with that. ;-) 14:39
14:40 fireartist left 14:45 sufrostico joined, znpy joined 14:46 ptolemarch joined
[Coke] m: role Foo[$x] { }; say Foo[42].WHO.^name 14:49
camelia rakudo-moar e5174b: OUTPUT«VMNull␤»
masak VMNull!? :)
teatime if cwd has lib/ and resources/ and t/, 1) how do you run the tests in t/, and 2) how do you allow use'ing your module and accessing your resources as-if installed? is PERL6LIB="." or something sufficient? or can you tell panda to install-editable like you can w/ python virtualenvs? just looking for a pointer to the basic dev workflow. 14:50
jnthn Interesting answer. :) A CurriedRoleHOW really has no meaningful .WHO, though I guess we could shove some throw-away Stash on there
[Coke] rt.perl.org/Ticket/Display.html?id=123488 14:51
RabidGravy teatime, I usually just "panda-test"
[Coke] it used to segfault back in 2014, so, progress.
jnthn Yeah
teatime oh, and 3) how do you make scripts in ./bin/ able to find modules in ./lib/ 14:52
jnthn I guess desired behavior is perhaps an empty Stash
Though if you access the .WHO here, it's hugely likely that you're actually very confused about how things work :) 14:53
masak getting VMNull is a bit like the VM coming to the small convenience store that you run, going "nice little store you got here -- would be a real shame if something were to... happen to it", and then just leaving 14:55
teatime is this really not something all of you do like every day? 14:56
RabidGravy: I appreciate that, v. helpful.
jnthn masak: It's the "you deserved a segfault but I did better than that" value :)
masak yes, that's what I mean :) 14:57
RabidGravy teatime, as to the bin thing, once they are installed they don't need anything special, to test I tend to just do "perl6 -Ilib bin/foo"
teatime oh right, I forgot -I works for other-than-REPL.
14:57 spider-mario left
moritz right, what RabidGravy said 14:57
hoelzro it works for REPL now too 14:58
teatime hoelzro: since the last couple of days?
hoelzro since...Thursday?
hoelzro forgot when he merged
teatime k, excellent!
14:59 dbohdan left
RabidGravy the %*RESOURCES also just works if you specify the -I or PERL6LIB as the CUR it uses is clever enough to work it out 15:00
masak I tend to just `export PERL6LIB=lib` in the terminal tab where I do development :) 15:08
no need for -I after that
15:10 maybekoo2 joined 15:12 ufobat joined
[Coke] RT: 1298; SEGV: 11; LTA: 129; WEIRD: 14; NOM: 7; GLR: 7; PERF: 13; JVM: 58 15:16
15:29 zakharyas left
teatime since I abuse virtualenvwrapper for all of my projects (not just python), I shall add a post-activate hook to set PERL6LIB :) 15:31
15:40 Ven joined
ufobat i want to use a "is rw" on an optional paramter.... but it is not supposed to work, isn't it? 15:43
15:43 sue left
jnthn m: sub foo($x is rw = 42) { }; foo() 15:45
camelia rakudo-moar 27dca5: OUTPUT«5===SORRY!5=== Error while compiling /tmp/4dYN8jz_Mm␤Cannot use 'is rw' on an optional parameter␤at /tmp/4dYN8jz_Mm:1␤»
jnthn Seems it's outright refused
You'd have to write something like
ufobat i thought to myself that with a default value it actually makes sense 15:46
jnthn m: sub foo($x is raw = 42) { }; foo()
camelia ( no output )
jnthn But note that then you can't be sure $x is assignable
Might be easier to write a pair of multi subs 15:47
One that takes an rw thing, one that doesn't
ufobat omg! yeah of course
thanks for the hint 15:48
i am going to read the "is raw" thingy, i dont know it yet
*being curious*
15:49 jameslenz left
sufrostico teatime: I run the tests with prove -e perl6 15:49
teatime: and for the test files... I usually add the line «use lib 'lib'« 15:50
teatime: use lib 'name-of-the-folder-with-my-private-library'; 15:51
teatime does it make any sense to mix perl5 and perl6 code in the same repo and/or lib/ directory?
15:55 cognominal joined 15:58 sue joined
awwaiid teatime: depends on what you're doing, but I imagine that it would be mostly one or the other 16:00
ufobat if i do the multi method way, and in one method i want to call the other with the additional is rw parameter, do i use samewith(..oldparams.. $rw-int) or callwith(..oldparams.., $rw-int) 16:01
lizmat is working on the Perl 6 Weekly
so let me know of anything you would like to see in the weekly!
jnthn lizmat: I'd like to see a worse pun that the one in my blog post this week :D 16:02
*than
timotimo it's hard to pun "bunny" to "punny"
lizmat
.oO( very punny )
 16:03
awwaiid lizmat: I don't have it posted, but on April 16 in Baltimore we'll have a short intro-to-Perl6 tutorial from myself and more advanced nativecall/async tutorial from [Coke]. dcbpw.org
jnthn Damn, that place name makes me hungry... :)
lizmat yeah, the DCBaltimoreW is on my list :-)
awwaiid lizmat++ # lists!
MadcapJake reCaptcha added to sixbug, now just the final gmail smtp integration and this should be good to launch! 16:05
diakopter MadcapJake: probably have it relay to other individuals (such as your self) who can also shut it off in case of emergency (in case it gets spammed) 16:06
ufobat if i do the multi method way, and in one method i want to call the other with the additional is rw parameter, do i use samewith(..oldparams.. $rw-int) or callwith(..oldparams.., $rw-int) 16:07
oups sorry
this keystroked was supposed to go to my emacs :)
MadcapJake diakopter: yes agreed, reCaptcha should be enough, if anyone wants to help they can take a peak through the code it's on github (pushing latest work after lunch though) 16:08
16:09 cdg_ joined, cdg left
hoelzro o/ #perl6 16:11
MadcapJake still planning on adding the ability to shortcut using the smtp impl and just get a redirect mailto, then I want to add a redirect to the actual ticket after submission, and I also still have to finish issue templates too. But the basics are there and ready to be used!
MadcapJake off to lunch &
hoelzro how do people feel about easter eggs in rakudo? I know we have this: 16:12
m: say "Life, the Universe, and Everything".WHY
camelia rakudo-moar 27dca5: OUTPUT«42␤»
lizmat hoelzro: could you explain in a few lines the state of the REPL since you merged your work ?
hoelzro sure! 16:13
basically the higher-level features of the REPL (line editing, tab completion, etc) are now implemented in Perl 6, not NQP
this will allow new users to read, understand, and modify the REPL itself more easily
and make the lives of those of us who would like to extend it a little easier 16:14
it also works with -I now
and it saves your history to ~/.perl6/rakudo-history or RAKUDO_HIST
16:15 Dunearhp joined
lizmat hoelzro: thanks! 16:16
dinner&
hoelzro anyway, the reason I ask about easter eggs is in a fit of feeling silly, I added this: hoelz.ro/files/lod.png 16:18
and I was wondering if I should include it in rakudo, or if that's just too silly for the compiler
dalek Iish: f1e36d2 | (Salvador Ortiz)++ | t/ (3 files):
Add missing 'use'
 16:23
Iish: 07226c7 | (Salvador Ortiz)++ | lib/DBDish/ (2 files):
Pg/mysql: Propagate conn params
16:23 espadrine joined 16:30 espadrine left 16:32 spintronic joined, TEttinger joined
dalek Iish/data-sets: 1ed8466 | (Salvador Ortiz)++ | / (4 files):
WIP/RFC: A new DataSet interface for DBIish

See added t/50-DataSet.t for details		 16:36
16:47 spintronic left, spintronic joined 16:51 espadrine joined
MadcapJake anyone know if I can just save an SSL certificate or should I request it freshly every time I need it (to send data to an SMTP server) 16:56
16:56 sufrostico left 16:58 Ven left
spintronic Why can't rakudobrew build panda correctly? I keep getting "Cannot find Shell::Command" 17:02
Begi spintronic : with "$ rakudobrew build panda " ? 17:04
17:05 spintronic_ joined 17:06 sue left
spintronic_ I'm running on Centos 6.7. Maybe it's too old... 17:06
17:07 pnu joined 17:09 spintronic left
moritz spintronic_: which rakudo did you build? 17:11
Hotkeys moritz: spintronic_ I've been getting that too
17:13 sufrostico joined
Hotkeys Something broke a couple weeks ago in panda (Shell::Command?) That won't let it build on windows for me 17:13
17:16 spintronic joined
spintronic I'm running the latest rakudobrew and perl6 17:16
17:20 Relsak joined
spintronic something breaks in panda all the time. My local perlmongers say just wait and try building again in a couple of days. 17:27
pnu masak, about heroku and (data) persistence... yes, you'd want to use postgresql, redis etc. for any storage. Lookup heroku _addons_ for those two. Dynos' local storage doesn't survive even restarts/scaling; dynos (and build process) are completely isolated. Only things shared are the slug (code image created by the build), heroku config (copied to system env
on startup) and the attached addons.
17:30 mullagain left, jameslenz joined, Dunearhp left
pnu .. for manual testing and seeing how your slug looks like you can do "heroku run bash". That too will be a separate new dyno, not a "live" server. 17:31
MadcapJake spintronic: you could try nuking your install 17:33
17:34 abaugher left
spintronic_ Ok I'll try that 17:34
17:35 abaugher joined
spintronic do we still download rakudobrew from github.com/tadzik/rakudobrew ? 17:42
Begi spintronic: yes, and just follow the installation instructions 17:43
spintronic ok rebuilding moar now...
RabidGravy yeah, you may also try simply "rakudobrew self-upgrade; rakudobrew nuke moar; rakudobrew build moar; rakudobrew build-panda" 17:44
pnu RabidGravy, masak, I think 12factor.net/ describes pretty well the ideas how/why heroku lifecycle goes as it goes. 17:45
spintronic yeah I did but nuke moar didn't seem to work.
So I'm starting all over again.
nuke jvm did work.
17:45 cpage_ left
spintronic Is zef working for anyone yet? 17:46
Hotkeys Zef and panda both worked for me on Linux last time I tried 17:49
But I can't get panda to build on windows
spintronic:
I ended up just building an older version of panda
You can specify the commit you want to build with rakudobrew
spintronic Hotkeys: building panda now. If it doesn't work I'll go with an older rev. Which one are you useing? 17:50
Hotkeys I don't remember and I'm not at my computer right now
If you look at the commits 17:51
Its the one before stuff was squashed into shell__command
I think
On Feb 22 17:52
There might be a later one that works
Not sure
MadcapJake Does Nil work for nativecall subs as NULL?
hoelzro MadcapJake: for calling? 17:53
MadcapJake yeah for calling a native sub, I want to pass NULL
hoelzro I've always just used the type object for the required type 17:54
spintronic Hotkeys: oh ok. Well it the latest still doesn't work.
hoelzro ex. something_that_takes_a_null_string(Str)
MadcapJake will that work with a repr('CStruct') class?
Hotkeys Yeah
I think panda occasionally pulls shell command into an include 17:55
MadcapJake spintronic: also take a look at the paths it tries to find shell::command in
Hotkeys And I think it just hasn't for a month now
MadcapJake: I tried putting a newer version into its local copy of shell::command like last week and it refused to work 17:56
I suppose I should file an issue
MadcapJake I think the problem is that panda uses shell:command to build itself so it's inside panda's build dir somewhere 17:57
Hotkeys It is
MadcapJake strange
Hotkeys Its in ext/Shell__Command
I tried replacing that with a new version and it didn't help any 17:58
MadcapJake strange
does rakudobrew build-panda give you a list of paths it searches for it in? 17:59
so looking at bootstrap.pl right now (in panda repo) 18:00
why no «use lib 'ext'» github.com/tadzik/panda/blob/41888...trap.pl#L5 18:01
that might be enough to make it work actually
18:04 spintronic_ left, spintronic__ joined
MadcapJake tadzik: shouldn't bootstrap.pl in panda have «use lib 'ext'» instead of the paths to each of the modules internal lib dirs? 18:05
18:09 Ven joined
moritz MadcapJake: but the modules aren't directly unlder ext/ 18:09
MadcapJake aren't they?
18:10 Relsak left
moritz MadcapJake: no, they are under ext/File__Find/lib/File/Find.pm and the likes 18:10
MadcapJake the module files you mean
hmm, trying to figure out why Hotkeys and spintronic were getting "cannot find shell::command" error 18:11
DrForr This.. is interesting. Subjecting a Crust servr to 15 concurrent users under siege(1) for a single request works. 16 hangs consistently. 18:12
I don't suppose 'grep -r 15' is going to find anything useful, but here goes :) 18:13
18:22 cpage_ joined 18:24 cdg_ left 18:25 cdg joined
RabidGravy DrForr, perhaps it is the number of threads available to the scheduler? 18:25
18:26 Ven left
DrForr Entirely possible. 18:26
RabidGravy this would be easily provided, just set $*SCHEDULER to a ThreadPoolScheduler with a smaller number
proved
DrForr Hadn't really yet looked into the threading mechanism, but I'll scribble a note for a few minutes hence :) 18:27
18:28 dha joined
DrForr I also managed to trigger an error in Crust that I can't replicate. 18:29
18:29 cdg left
RabidGravy is it possible that it is failing to close the client connections in a timely fashion and is getting deadlocked before it is able to do when it reaches the max number of threads 18:30
(I haven't looked in the code)
DrForr Nod.
Its behavior *after* trying to handle 16 threads is also reminiscent of a deadlock. 18:31
diakopter notably, these 15 would be in addition to the main program thread, right? 18:32
DrForr Power of 2, I don't believe this to be a coincidence :)
18:37 dbohdan joined
dbohdan A person on Hacker News asks, "What would you use to build a super simple web app?" and says "I don't care if it's in Ruby, PHP, Golang, Perl 6..." 18:39
news.ycombinator.com/item?id=11374752
Perhaps you could recommend a Perl 6 approach to making a simple web app to him/her. :-) 18:40
DrForr Funny you shoud mention that :) 18:41
Hotkeys Are you making a simple web app DrForr 18:42
dbohdan: I'm just surprised they mentioned perl 6 as an option to be honest
perlpilot too (happily surprised)
RabidGravy well if I wanted a super-super-simple web app I'd just hook up a shell script to inetd ;-)
diakopter to be fair(-ish), it may be the "absurd final option" 18:43
Hotkeys we need something like django or flask for p6
RabidGravy do feel free to write it :)
Hotkeys >_>
perlpilot diakopter: except that it's not. It's in between Golang and JS
spintronic__ MadcapJake: I'm trying to move the shell::command library now to see if panda can find it.
diakopter oh
Hotkeys I was gonna say "We can call it Oyster" but apparently that's a thing 18:44
diakopter it's hard to find things that aren't already things
Hotkeys perhaps "Chaos" as a play on the butterfly effect (see: camelia)
but chaos might be a bad word for a thing 18:45
diakopter camelot?
cameliot 18:46
perlpilot camelion ;)
MadcapJake Crust is pretty nice already, a sugary wrapper around it would make for a super simple webapp experience (there's got to be a good name there too: crust + sugar)
Begi For what's the name you're looking for ?
dbohdan My approach to naming tiny HTTP frameworks is "${language}http" if that isn't take. :-)
diakopter Crust sounds like something for Rust
Hotkeys lol
RabidGravy MadcapJake, I think that's precisely what DrForr is doing
Hotkeys ah 18:47
dbohdan *taken
DrForr Aroo?
RabidGravy Boop boop sheboop
dbohdan "Tea"? 18:48
DrForr files a Crust bug and returns to testing...
dbohdan (Tea's species name is Camellia sinensis.)
Hotkeys nice
DrForr Ooo, I might have to rename mine to Marzipan. 18:49
MadcapJake some sugar + crust ideas: pie, bundt, canoli, truffle 18:50
18:51 espadrine left
diakopter CARBanana 18:51
RabidGravy (Creme) Brulée
dbohdan "Truffle" would clash the JVM: wiki.openjdk.java.net/display/Graa...Guidelines 18:52
*on the JVM
MadcapJake bummer, I think canoli is my favorite (mostly cus they're my favorite treat) 18:53
DrForr Hotkeys: github.com/drforr/perl6-App-prancer for an idea.
18:54 spintronic__ left
MadcapJake DrForr: looks like I need to take it for a spin again! 18:54
Woodi DrForr: is "glaze" taken ? :) 18:55
masak if it's crust and sugar you want, just call it "cougar" :D 18:56
Begi O_o
perlpilot Woodi: link.springer.com/chapter/10.1007%2...03655-2_95 not quite the same space, but close
Woodi perlpilot: damn :) 18:57
DrForr Hrm. 9 transaction/sec on a rather overworked VM, holding up at 300k/sec transfer.
Woodi: No idea. I'd go for Krispy::Kreme but would run afoul of branding laws. 18:58
And now I'm wishing my NLPW trip went through Heathrow.
Woodi DrForr: if you want something more chocolte-y then ganache maybe ? :)
DrForr And the wish just passed, as I remember how much I despise it.
(Heathrow, not KK) 19:00
19:00 cpage_ left
Hotkeys I think I might try to make Inline::Factor 19:01
because Factor is dope
MadcapJake DrForr: I'm noticing the same freezing at >15 connections on Crust 19:02
19:03 cpage_ joined
DrForr MadcapJake: Yeah, I haven't dug into the code yet, just scoping out performance in general. 19:03
19:04 labster joined
MadcapJake I'm only getting 3 Req/Sec on my machine (just using it right on my rig though) 19:05
DrForr I'd guess if I had to that it's assuming that the thread pool is infinite so it simply spawns what it needs, and when resources run out it deadlocks trying to allocate the new threads rather than puts requests in a queue.
MadcapJake oops that's per thread! I'm getting 29 req/sec in total
DrForr Better than I'm doing here, but my VM is somewhat overworked.
MadcapJake I was a little worried that my dev rig was slower than a VM ;) 19:06
DrForr I could put it on my linode, but that puts me at the tender mercies of my network connection.
19:06 dvinciguerra_ joined
MadcapJake Hopefully, moritz will let me put this up on the p6c server 19:07
DrForr Even locally I'm seeing ~700k/sec and ~20 req/sec.
19:07 dvinciguerra left
DrForr Doing a longer run right now, we'll see what the results are. 19:07
MadcapJake Wow I am only getting 300kb/sec locally
19:08 sue joined
MadcapJake Every once in a while I am getting these core.settings errors in my log 19:08
DrForr Heh, I just coredumped siege, FD table overflow.
MadcapJake broken pipe
i'm using wrk
HTTP::Server::Tiny keeps throwing this at me hastebin.com/ejisizodim.txt 19:09
DrForr Regardless my original problem with Crust not being able to handle >1 connection per client seems to have vanished. 19:10
MadcapJake: Without looking I"m guessing that's what I filed on GH.
MadcapJake no "broken pipe" errors on MoarVM repo 19:11
DrForr You're right, that's a different error. 19:12
MadcapJake I submitted on there, not really sure what's going on, just happy that HTTP::Server::Tiny seems to just trudge right on through it :) 19:14
19:15 dogbert joined 19:16 dogbert is now known as Guest73352
Hotkeys oh neat 19:18
DrForr 18.5 transactions/sec, 800K/sec, concurrency 12.42. Not great, but better than I'd expected.
Hotkeys NativeCall can do cpp
I have no idea how to do an inline thing 19:19
fun learning adventure time
DrForr Hotkeys: Inline::Scheme::Guile :) 19:20
Hotkeys Yeah I've been perusing all the inline modules
19:20 yqt joined
Hotkeys does guile use cpp :D 19:20
19:21 kaare_ left
Hotkeys it seems the answer is no 19:21
timotimo well, gnu people do their glibc stuff
which is "their own" object system
DrForr It'll let you link in c libraries, probably c++ as well FAIK. 19:22
Hotkeys nativecall does c++ apparently
so that's fun
MadcapJake Hotkeys: no docs though, but you can also just create a C wrapper of C++ functions
timotimo yeah 19:23
Hotkeys the base of Factor's compiler is written in c++
so I assume that's what I need to use
DrForr Inline::Scheme::Guile builds a C helper library if it helps.
19:24 domidumont left
DrForr Hotkeys: I should also point out I'm testing with a custom Prancer app that spits out about 100K of HTML per query. 19:25
Hotkeys ah
I found factor.hpp
this seems like what I need
maybe 19:26
Guest73352 Hi #perl6, anyone up for a simple/stupid newbie question? 19:27
I have a problem with SetHash 19:28
diakopter probably! give it a try
Guest73352 m: my $numSet = (0..10).SetHash; $numSet<5> = False; say $numSet.keys.sort; # why is the 5 still there?
camelia rakudo-moar 27dca5: OUTPUT«(0 1 2 3 4 5 6 7 8 9 10)␤»
Begi i've a newbie problem with LWP::Simple, I guess. My header doesn't work : gist.githubusercontent.com/Emeric5...request.p6 19:29
Any idea why ? thanks ! 19:30
[Coke] m: my $numSet = (0..10).SetHash; $numSet.perl.say
camelia rakudo-moar 27dca5: OUTPUT«SetHash.new(5,7,9,4,8,3,0,1,6,2,10)␤»
[Coke] m: my $numSet = (0..10).SetHash; $numSet<5>:delete; say $numSet.keys.sort;
camelia rakudo-moar 27dca5: OUTPUT«(0 1 2 3 4 5 6 7 8 9 10)␤»
[Coke] m: my $numSet = (0..10).SetHash; $numSet{5} = False; say $numSet.keys.sort; 19:32
camelia rakudo-moar 27dca5: OUTPUT«(0 1 2 3 4 6 7 8 9 10)␤»
[Coke] ^^
5 NE "5"
<5> is "5". {5} is a raw 5.
Guest73352 Aha, thanks Coke++ told you it was simple :-) 19:33
19:33 ptolemarch left
MadcapJake m: (SetHash.new(1..10) (^) SetHash.new(5)).keys.sort 19:33
camelia ( no output )
MadcapJake m: (SetHash.new(1..10) (^) SetHash.new(5)).keys.sort.say
camelia rakudo-moar 27dca5: OUTPUT«(1 2 3 4 6 7 8 9 10)␤»
Guest73352 Ah, nice 19:34
MadcapJake Guest73352: doc.perl6.org/language/setbagmix#infix_%28%5E%29 19:35
Guest73352 Thx MadcapJake, will take a look
MadcapJake I love a good chance to use those operators :) If I wasn't worried that you'd leave IRC, I'd have taken the time to use the unicode version (⊖) xD 19:36
diakopter m: (SetHash.new(1..10) (^) SetHash.new(5)).keys.sort.say 19:38
camelia rakudo-moar 27dca5: OUTPUT«(1 2 3 4 6 7 8 9 10)␤»
diakopter m: (SetHash.new(1..10) ⊖ SetHash.new(5)).keys.sort.say
camelia rakudo-moar 27dca5: OUTPUT«(1 2 3 4 6 7 8 9 10)␤»
19:39 buharin joined
diakopter m: SetHash.new(1..10 ⊖ 5).keys.sort.say 19:39
camelia rakudo-moar 27dca5: OUTPUT«(1 2)␤»
MadcapJake So I'd say 34 reqs/sec is pretty respectable (especially because right now we're--artificially?--restricted to 15 connections)
m: (SetHash.new(1..10) (^) 5).keys.sort
camelia ( no output )
diakopter where does that rank on www.techempower.com/benchmarks/ 19:40
MadcapJake m: (SetHash.new(1..10) (^) 5).keys.sort.say
camelia rakudo-moar 27dca5: OUTPUT«(1 2 3 4 6 7 8 9 10)␤»
MadcapJake diakopter: thanks was looking for something like this!
MadcapJake is really tempted to try Ur/Web
RabidGravy I think I am going to set fire to my computer, then all the other computers in the world 19:41
MadcapJake please leave mine out of this!
RabidGravy I have just spent *hours* typing in the docs for this module, got distracted, forgot I hadn't saved and exited without saving 19:42
MadcapJake oi! pretty much the worst thing in the world right there. I feel your pain.
japhb RabidGravy: autosave FTW?
DrForr MadcapJake: Yep, respectable. I may dig into the queueing aspect later in the week, but bedtime waits for no person. 19:43
diakopter the fastest Perl one on there is 5.9% the performance of the fastest
(for the fortunes benchmark) 19:44
has someone written a fortunes benchmark for rakudo?
MadcapJake diakopter: their payload is quite a bit smaller than mine (several CSS files, icon fonts, a few external CDN resources) so I'd be curious to try to match their benchmark and see how it goes! 19:45
diakopter MadcapJake: sounds like you're the right person for it! 19:46
MadcapJake diakopter: It doesn't look too bad really 19:47
diakopter I had thought jnthn's blog posts from earlier this year were building toward a simple http server 19:48
(and maybe they still are ;)
MadcapJake The best one out there right now is HTTP::Server::Tiny (I think)
I think it's the only one (that I know of) that can send binary data :P 19:49
19:52 spider-mario joined, sufrostico left, o_sutur joined
MadcapJake I can't seem to make this debug env variable go away! 19:54
19:55 musiKk_ joined 19:56 perlawhirl joined 19:57 sufrostico joined
geekosaur ? 19:58
MadcapJake aha I know what's going on here 20:00
HTTP::Server::Tiny puts the HST_DEBUG environment variable in a constant, so the no matter what I do to the HST_DEBUG in my shell, it's already been set when it compiled 20:01
20:03 dha left 20:05 m0li joined 20:06 kmwallio left 20:07 kmwallio joined 20:08 cdg joined 20:10 cdg left, cdg joined 20:14 cdg left, cdg joined 20:18 darutoko left
MadcapJake RabidGravy: can I change the $*SCHEDULER to one with a higher max_threads? 20:20
masak m: say $*SCHEDULER.clone(max_threads => 32) 20:21
camelia rakudo-moar 27dca5: OUTPUT«ThreadPoolScheduler.new(initial_threads => 0, max_threads => 32, uncaught_handler => Callable)␤»
RabidGravy yeah 20:22
masak dunno how to install that new scheduler in $*SCHEDULER, though
MadcapJake right that's what I want to do
diakopter m: $*SCHEDULER.=clone(max_threads => 32)
camelia rakudo-moar 27dca5: OUTPUT«Cannot modify an immutable ThreadPoolScheduler␤ in block <unit> at /tmp/12KVD0Gqey line 1␤␤»
RabidGravy just
diakopter m: $*SCHEDULER.:=clone(max_threads => 32)
camelia rakudo-moar 27dca5: OUTPUT«5===SORRY!5=== Error while compiling /tmp/VIda26nwvT␤Malformed postfix call␤at /tmp/VIda26nwvT:1␤------> 3$*SCHEDULER.:7⏏5=clone(max_threads => 32)␤»
RabidGravy m: my $*SCHEDULER = ThreadPoolScheduler.new(max_threads => 32)
camelia ( no output )
RabidGravy before you do anything that requires the scheduler 20:23
diakopter m: say $*SCHEDULER.clone(max_threads => 9*9*9)
camelia rakudo-moar 27dca5: OUTPUT«ThreadPoolScheduler.new(initial_threads => 0, max_threads => 729, uncaught_handler => Callable)␤»
RabidGravy (or anything that you want to be affected thus)
diakopter m: say $*SCHEDULER.clone(max_threads => 9**9**9)
camelia rakudo-moar 27dca5: OUTPUT«(timeout)» 20:24
diakopter ETOO....JUST....TOO
masak don't immanentize the eschaton 20:25
diakopter Immanuesch 20:26
MadcapJake it doesn't work just putting it in my app's script, would probably need to be in the server module
diakopter one would think TreadProolSchreduler would lazily allocate 20:27
MadcapJake the `my` means that the scheduler is probably only limited to my script's scope and any sub routines I call inside it. the way Crust works, it EVAL's your script, so it would need to be higher that Crust's EVAL calls 20:29
diakopter mod_perl did eval, very early on
MadcapJake still not working :( 20:32
20:32 labster left
MadcapJake RabidGravy: I've tried placing that in both HTTP::Server::Tiny and Crust::Runner but i'm still getting 0 requests with anything over 15 connections 20:32
20:33 perlawhirl left, ecocode left
MadcapJake ok i got strange results now :P 20:35
If I place it in Crust::Runner.run(), it gives me 6 requests/sec! Why is it lower!? o_O
20:40 pmurias left 20:41 labster joined 20:43 itaipu joined, ptolemarch joined, Begi1 joined
RabidGravy because you don't have that many cores in your computer so it takes longer waiting to sort them out? or something. I'm not au-fait with the minutae 20:44
20:44 Begi left, Begi1 is now known as Begi
kmwallio So I did my int32 $blah = 1; but when I pass it into a nativecall method, I'm getting "Native call expected argument that references a native integer, but got P6int" 20:44
20:46 rindolf left
kmwallio the code I'm trying to run is: gist.github.com/kmwallio/8565b36d1fe9e2b18401 20:47
20:48 geraud joined
MadcapJake kmwallio: you're call is missing the final argument, just has a type 20:48
kmwallio isn't that how you pass null? 20:49
or no...
MadcapJake oh yeah that is
kmwallio or maybe I should pass 0 since it's an int...
Xliff kmwallio, couldn't hurt to pass 0 just to see.
MadcapJake if it expects NULL then you should pass the type not 0
Xliff MadcapJake, he's already passing the type and getting an error. 20:50
kmwallio creating another variable set to 0 make it seem to run happy
MadcapJake passing the type won't make the method think you gave it a P6int though
so it must be the other two arguments causing the error
weird then i guess i'm wrong, could be some sort of reportable bug 20:51
20:51 buharin left, bjz_ left
Xliff Hrm. OpenCL not in Ubuntu? 20:51
kmwallio not by default
you need to install some packages
MadcapJake Xliff: gist.github.com/rmcgibbo/6314452 20:52
20:52 rindolf joined
Xliff Well, hell. Probably won 20:53
kmwallio packages.ubuntu.com/search?keywords...chon=names
Xliff Probably won't work inside a VM, then.
*sigh*
tadzik MadcapJake: everything in panda's bootstrap.pl runs by black magic
Xliff When I have to fill out a form to download a file, my interest wanes. 20:54
MadcapJake tadzik: lol seems so!
tadzik MadcapJake: every "why" can be answered with "really, we tried"
kmwallio Xliff: if you have an extra graphics card, you might be able to do a PCI passthrough
Xliff kmwallio, I do. Don't know if I can do that using Virtualbox, tough.
MadcapJake tadzik: I've had that shell::command not found error, but I don't remember how I fixed it, probably just reinstalled everything :P 20:55
Xliff s/tough/though/
tadzik MadcapJake: it may be that things got saner these days and a lot of these things is not needed anymore
kmwallio Xliff: www.virtualbox.org/manual/ch09.htm...assthrough
not sure if it's worth it though
tadzik with precomp being in place and stuff being generally relocatable it may be completely unnecessary these days even
MadcapJake tadzik: can you change where modules are installed now?
tadzik I have some time allocated for the toolchain stuff 20:56
Xliff kmwallio, yeah. Well I also have dual-boot into KUbuntu on this box, so I may try again next time I am on the other side.
tadzik MadcapJake: that's not my responsibility anymore
Xliff Still. I hate it when I have to fill out a fscking form to download a file.
MadcapJake tadzik: ah, someone (nine?) mentioned that it was almost there a few days ago 20:57
tadzik MadcapJake: yeah, it's his child now :) I think panda can still overwrite it with DESTDIR iirc
MadcapJake Xliff: what kind of form is it? no-looky-at-sourcy? no-writing-your-own-similar-sourcy?
no-suey-in-courty-for-anythingy? :P 20:58
Xliff No. It's the gimme-your-email-before-downloady type. 20:59
In other words: "download this file so we can spam you!"
MadcapJake tadzik: eventually, I plan on writing a gx-p6, but I need to be allowed to do local module installs for it to work
20:59 Begi left
MadcapJake Xliff: lol 20:59
Xliff MadcapJake, do you have intel_sdk_for_ocl_applications_2013_xe_sdk_3.0.67279_x64.tgz 21:00
tadzik MadcapJake: oh, it's definitely possible'
I think DESTDIR is what you're after
21:00 Begi joined
tadzik I also think sergot wrote some thing that installs dependencies in the app directory, quite like Carton 21:00
21:00 Begi left
MadcapJake tadzik: I'll take a look, cus that's exactly what I'm trying to do 21:00
tadzik MadcapJake: github.com/sergot/bamboo 21:01
MadcapJake Xliff: nope, I've never delved into OpenCL (just barely getting my feet wet in OpenGL!)
Xliff *shudder&
*shudder* OpenGL *shudder*
MadcapJake lol it doesn't help that I have basically zero experience with matrix math xD 21:02
kmwallio MadcapJake: I saw your OpenGL stuff and was going to try using it, but OpenGL is limited to 4x4 matrices 21:03
I also tried the latest gist, and the animation is pretty smooth
MadcapJake kmwallio: what would you need bigger matrices for? I think you're supposed to use VertexArrayObject's to manage multiple sets of them 21:04
kmwallio MadcapJake: for markov chains, you can use a transition matrix which has rows and columns equal to the number of nodes 21:06
Xliff I used to love matrix math. I was good at it. Like.... 25 years ago.
MadcapJake kmwallio: woah that would be quite a huge matrix then, right? 21:07
kmwallio yeah :/
Xliff Now my matrix neurons have shriveled.
kmwallio So for github.com/kmwallio/p6-Lingua-EN-Summary , it takes a little over a minute to generate summaries for some of the samples 21:08
the Dr. Jekyll and Mr. Hyde takes ~10 minutes
but then again, my implementation probably isn't that great
but if you use the GPU, it's pretty instance
instant* 21:09
MadcapJake I've been trying to delve into neural networks a bit (didn't realize how much math is at the core of them :S ) but I noticed that most people say "we *could* do each node at a time but that would mean lots of concurrency so instead we use matrices!" has anyone ever tried to just do the concurrent approach? Didn't have any luck finding anything. But I
instantly though, "hey that could be a neat Perl 6 experiment"
s/though/thought/
kmwallio: are there any C libaries that implement GPU-based math? we could wrap one up. 21:10
teatime "Recursive descent allows you to hook in custom-written bottom-up logic at every top-down choice point, and it is a technique which is completely understandable to programmers with little or no training in parsing theory. When dealing with recursive descent parsers, it is more useful to be a seasoned, far-thinking programmer than it is to be a mathematician." this gives me hope, heh.
kmwallio developer.nvidia.com/gpu-accelerated-libraries
21:11 FROGGS left
masak teatime: that quote reminds me of how action methods work. 21:11
MadcapJake probably most of them are C++ though
teatime MadcapJake: exactly.
er, masak
MadcapJake one extra char before you tab-complete teatime ;) 21:12
masak teatime: because rules naturally match from leaves up to the whole program, an action method has all the AST results of its descendants available.
kmwallio MadcapJake: Neural Nets are pretty cool. Google and Microsoft (and everyone) are switching to them for everything 21:13
teatime masak: perl6 grammars seem to be the perfect combination of top-down and bottom-up to make them user-friendly and require no comp sci skills to use effectively. maybe.
kmwallio they improved part of speech tagging and translation, when they used to use markov chains
masak a grammar is just a funny class
teatime masak: perl6 regexes then.
kmwallio also text to speech now uses deep neural nets
MadcapJake kmwallio: very interesting stuff, I just need to take some serious time to learn more about them. I probably need to read a few books instead of expecting 10 minute youtube videos to explain it all to me :P
masak teatime: but I'd be lying if I said writing grammars isn't an art, one that's not quickly acquired 21:14
teatime: the grammar in 007 is definitely 007's weakest point right now. it still fails in quite brittle ways sometimes.
geekosaur remembers when neural nets were discovered... and then dropped because nobody could work with them. the state of the art had to catch up first...
kmwallio MadcapJake: it depends though, if you using a library, the youtube video is probably good. If you want to know what's going on under the hood, you probably need a deeper look
teatime masak: luckily my intended use should be much much simpler than 007.
kmwallio geekosaur: how old were you... 21:15
kmwallio was born after neural nets were discovered
21:15 wamba left
kmwallio :P 21:15
MadcapJake kmwallio: true, that's the thing, I'm looking to do something with them in Perl 6 and those videos just give surface-level stuff that focuses on (typically Python) APIs
teatime I've looked at parser generators for other languages, and perhaps I just haven't looked at the right ones, but I haven't found one that looks nearly as painless as perl6.
kmwallio Yeah, python has all of the toolkits, and even nodejs is getting some now
diakopter *boggle* 21:16
geekosaur oh actually they were discovered much earlier, first described in 1943. 1972 (I was still a kid but one who read a lot...) was what I was thinking of, and was pretty much the end of the "old way" 21:17
teatime also suggestions for mature languages that make parsing extremely natural/easy are welcome. not LISP. or haskell.
but perl6 has been so much fun and my project is so non-critical that I think I will just stick w/ that. 21:18
geekosaur came back in the 1980s but didn't really take off until the late 90s-ish
kmwallio geekosaur: my professor in college said Neural Nets were a nice toy, but weren't useful for "real machine learning/ai"
and then a few years later, there was the resurgence 21:19
and new advancements in deep neural nets
:/
diakopter kmwallio: I find it hard to believe that python has an advantage to nodejs in terms of number and quality of parsing libraries
since nodejs has 4.5x as many published packages/libraries as python 21:20
teatime diakopter: I think he was talking about AI tools. and there's a lot of utter crap in npm...
diakopter there's a lot of utter crap in all of them 21:21
geekosaur yeh
moritz until I see numbers that suggest otherwise, I'm going to assume that 90% of everything is crud.
kmwallio I'll be introducing utter crap to the Perl 6 modules :D
Acme::UtterCrap
:P
jk
mst ooh, you're implementing an npm client?
kmwallio my js to perl 6 compiler? 21:22
diakopter transpiler you mean
I'd be hard-pressed to find 10% of usableness in the Perl 6 Ecosystem.. 21:26
I assume anyway.. I haven't looked closely lately.
geekosaur looks like they hit walls several times and gave up.. and a few years later would realize they could solve the problems that made them stop, and started working with them again until they hit the next wall. (this is even continuing although the walls now are not complete show-stoppers) 21:27
geekosaur sighs... daily loss of connectivity to bouncer has begun again
one of the current issues with neural nets is that unlearning something is unreasonably expensive. have seen some stuff on re-chunking things so it's easier to "subtract out" unwanted things 21:28
21:28 skids left 21:30 Khisanth left
kmwallio geekosaur: wasn't IBW Watson delayed or something because it accidentally "read" urbandictionary? 21:31
IBM*
then there was also Tay from a few days ago... 21:32
geekosaur there's a reason one common class of science fiction involves AI becoming sentient and deciding that humans are unredeemable >.> 21:34
21:34 arcetera left
kmwallio is it possible with NativeCall to typedef things? 21:34
because cl_int is just int32, but I don't want to lose track in my head somewhere 21:35
21:35 musiKk_ left
moritz kmwallio: constant cl_int = int32 21:35
kmwallio moritz: thanks 21:39
21:43 Khisanth joined 21:49 ptolemarch left 21:56 Xliff left 22:02 Khisanth left
spintronic /join #abcl 22:03
22:03 kid51 joined 22:05 nowan left 22:07 nowan joined 22:08 Xliff joined
MadcapJake kmwallio: though not GPU-based, this one has machine-specific optimized libraries: www.netlib.org/blas/ 22:10
hey actually: developer.nvidia.com/cublas 22:11
22:12 sufrostico left 22:21 sue left 22:22 Sgeo joined
kmwallio MadcapJake: thanks, I'll look into that after work 22:23
22:23 Sgeo left 22:28 Begi joined, Khisanth joined 22:32 Sgeo joined, Khisanth left 22:33 Khisanth joined 22:34 spider-mario left 22:35 spider-mario joined, sufrostico joined 22:38 Xliff left 22:42 Khisanth left 22:46 dvinciguerra__ joined 22:47 dvinciguerra_ left, Xliff joined, nadim joined 22:50 skids joined, _nadim left
masak 'night, #perl6 22:54
tony-o later gator 22:55
22:55 perlawhirl joined 22:56 itaipu left 22:57 Khisanth joined 22:58 perlawhirl left 23:01 perlawhirl joined 23:03 Begi left, arc__ joined 23:04 arc__ is now known as arcetera, arcetera left, arcetera joined 23:06 rindolf left 23:07 Khisanth left 23:11 raiph joined 23:20 Khisanth joined
lizmat got distracted after dinner: P6W will come tomorrow 23:20
good night, #perl6!
23:30 itaipu joined, Khisanth left 23:39 avuserow joined
avuserow hi #perl6 folks, I've been out of the loop for a while. Is it possible to use NativeCall with a C++ library? A blogpost or some other reference would be greatly appreciated. 23:40
yoleaux 3 Feb 2016 04:26Z <AlexDaniel> avuserow: gist.github.com/AlexDaniel/1e2d1c50963d37c5d43a
23:41 kid511 joined
avuserow AlexDaniel++ # ALL of the cute ways 23:41
23:42 kid51 left
perlawhirl haha wow! that's from a while back... i remember i came up with one of those 23:42
teatime does panda know how to upgrade modules? maybe --force install will at least do it? 23:46
perlawhirl avuserow: i was sure someone asked that a few days ago... sure enought: irclog.perlgeek.de/perl6/2016-03-25#i_12238141 23:48
timotimo yes, teatime. ideally a newer version of the module would actually have a newer version, so --force isn't necessary
23:49 Khisanth joined 23:50 labster left
MadcapJake So panda install X just automatically installs the very latest version of module X? 23:55
RabidGravy yes
anyway bedtime
toodles
captian-adequate exit 23:57
Oops... Sorry about that.
Meant to close my terminal. ha ha
MadcapJake feels there should be an upgrade command that can be tweaked to only upgrade to certain newer versions

Powered by Cro and the Raku® Programming Language.

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

Thank you!