<< 04-05-2015 >>

---

00:15:24	*	vendethiel joined #nim
00:36:33	*	vendethiel quit (Ping timeout: 240 seconds)
00:46:28	*	filwit joined #nim
00:48:59	*	gyeates quit (Ping timeout: 245 seconds)
00:51:33	*	gsingh93 quit (Ping timeout: 276 seconds)
00:51:54	*	xanadulives joined #nim
00:52:44	*	vendethiel joined #nim
00:54:33	reactormonk	Araq, can't move it over, the type dependencies go haywire.
00:55:01	Araq	reactormonk: couse you can, you only need to move the 'var' to there
00:55:28	reactormonk	Araq, nope, it depends on the Suggest type, an when I import suggest in msg I get compiler/ast.nim(630, 11) Error: undeclared identifier: 'TLineInfo'
00:55:58	reactormonk	ah fuck it, let's move that one over too
00:56:56	reactormonk	nope, can't move it over. I need ast.nim because I use TSymKind which requires msg.nim TLineInfo, so the import gets nasty. Wanna go down that route?
00:57:23	reactormonk	I don't.
00:58:39	Araq	good night
00:58:43	reactormonk	kk
00:58:47	reactormonk	o/
01:00:58	*	lakurei quit (Quit: lakurei)
01:03:35	*	TEttinger joined #nim
01:12:33	*	xanadulives quit (Read error: Connection reset by peer)
01:14:31	*	vendethiel quit (Ping timeout: 264 seconds)
01:28:55	filwit	dom96: you around?
01:29:57	filwit	dom96: just wondering if Aporia is deliberately stripping whitespace from blank lines at save for some reason and, if so, why?
01:36:05	fowl	That sounds like a great feature
01:43:32	filwit	fowl: eh? it's annoying when trying to add a line between indented code and have to re-tab first
01:44:08	def-	filwit: that's because there's lots of trailing whitespace in the Nim repo and it's slowly getting removed now
01:44:17	def-	I'd guess that it can be disabled in the settings
01:44:36	filwit	it's not like it's going to save much HDD space or something by removing the spaces.. I could understand something like intelligently cropping whitespace to surrounding indent levels or something
01:44:54	filwit	def-: yeah i see, i'll look for a setting (think i already did, but maybe i missed it)
01:45:25	def-	filwit: It's quite annoying for some people. Sometimes I want to append to a line and suddenly I'm at an unexpected position because of trailing whitespace
01:46:40	filwit	def-: i can't tell if you mean it's annoying the way Aporia is now, or if it's annoying without it deleting whitespace.
01:47:04	def-	it's annoying that there are up to 60 chars of trailing whitespace in some lines
01:47:20	filwit	oh yes, i see
01:47:41	filwit	but that's easy to strip out.. stripping whitespace from blank lines doesn't make sense IMO
01:49:21	*	saml_ joined #nim
01:50:48	*	darkf joined #nim
02:20:31	*	zzach quit (Ping timeout: 255 seconds)
02:20:53	*	zzach joined #nim
02:27:35	*	banister quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
02:29:02	*	BitPuffin|osx quit (Ping timeout: 272 seconds)
02:29:53	filwit	does anyone know if/where there's been discussion on seq len/xlen/add recently? I'm wondering why 'len' does nil-checks but not 'add', also I dislike 'xlen'.. it should be 'uncheckedLen' or 'rawLen' or something with an 'uncheckedAdd()' to go with it.. but maybe I'm out of the loop and these things have already been discussed?
02:35:22	*	jefus quit (Ping timeout: 255 seconds)
02:37:58	def-	filwit: I remember Araq mentioning something about that. I think the plan is to add nil-checked add as well. I guess the x prefix is for convenience, writing "uncheckedLen" is a bit long
02:38:35	*	gsingh93 joined #nim
02:39:08	fowl	it is planned
02:39:14	fowl	maybe safeAdd/add could just be switched
02:40:00	filwit	it would make sense to switch those since len now defaults to 'safe'
02:40:39	filwit	'xlen' isn't a huge deal, though I always prefer things be clearly labeled (and just alias them if you use it often
02:41:03	filwit	either way, i guess len/xlen and add/xadd would work. thanks for the info
02:43:48	*	jefus joined #nim
02:43:56	*	ddl_smurf quit (Quit: ddl_smurf)
03:03:03	filwit	does anyone know how to iterate over the fields of a type from within a macro?
03:03:36	def-	filwit: i'd pass it to a proc and build it there I guess
03:04:29	filwit	well unfortunately I am not passing a instance, but the actual Type, eg: `impl Foo: ...` where impl is the macro and 'Foo' is the type
03:05:18	filwit	'getType' and 'fields' only seem to work on instances.. which I really don't want to alloc an instance just to get type field info
03:05:42	fowl	filwit, yea
03:07:05	fowl	filwit, https://gist.github.com/fowlmouth/3c74409c30533bb0a5b4#file-showty-nim and https://gist.github.com/fowlmouth/9b9010397ad5fe4b9872#file-xeq-nim
03:08:09	filwit	awesome fowl, you rock! lots of code there for me to pick through
03:11:25	filwit	err, your code doesn't appear to work, fowl
03:11:27	filwit	for instance: https://gist.github.com/fowlmouth/3c74409c30533bb0a5b4#file-showty-nim-L125
03:11:42	filwit	when I try to do this, I get "Error: node has no type"
03:11:56	fowl	filwit, ah i will update it
03:12:18	fowl	filwit, oh if you invoke it from a macro you have to use showtyperec
03:14:06	filwit	fowl: i'm not even calling the macro, but that line I linked to doesn't even compile for me
03:14:39	filwit	i just copied the first to lines of your 'showType' macro, but it doesn't work
03:19:28	*	saml_ quit (Quit: Leaving)
03:22:01	fowl	filwit, dunno what to tell you, it works for me
03:22:07	fowl	i just updated nim
03:24:57	filwit	so did i (0.11.1), odd...
03:26:02	filwit	oh pfft.. my mistake
03:26:15	filwit	looking at the wrong thing.. sigh..
03:26:16	fowl	it has to be stmt (or "typed" as its called now)
03:26:45	fowl	lol
03:29:54	reactormonk	anyone know where exported symbols of a module are stored?
03:29:56	filwit	fowl: sweet it works! thanks
03:32:07	fowl	good
03:34:02	filwit	doesn't seem to work in {.immediate.} macros though.. hmmm...
03:34:29	filwit	though i suppose i need to figure this out, since that pragma is being depreciated anyways
03:35:12	filwit	nevermind, was easy to do (just removed the pragma, and made the 'body' `untyped`)
03:35:48	fowl	it wont where there aren't types (immediate macros, untyped (eager) expression)
03:37:04	*	gyeates joined #nim
03:48:25	*	gsingh93 quit (Ping timeout: 264 seconds)
04:03:23	*	vikaton quit (Quit: Connection closed for inactivity)
04:27:51	*	NimBot joined #nim
04:28:50	*	solidsnack joined #nim
04:33:08	reactormonk	Araq, should semIdentDef return the same symbol in case there's two global usages? or shouldn't it be called twice in the first place?
04:36:08	*	vendethiel joined #nim
05:04:47	*	NimBot joined #nim
05:05:27	*	solidsnack quit (Quit: My Mac has gone to sleep. ZZZzzz…)
05:27:46	*	gyeates quit (Ping timeout: 272 seconds)
05:52:55	*	vendethiel quit (Ping timeout: 264 seconds)
05:53:40	*	solidsnack joined #nim
06:15:03	*	vendethiel joined #nim
06:37:26	*	vendethiel quit (Ping timeout: 272 seconds)
06:52:11	*	Trustable joined #nim
07:06:01	*	yglukhov_ joined #nim
07:11:02	*	yglukhov_ quit (Ping timeout: 246 seconds)
07:12:48	*	vendethiel joined #nim
07:16:00	*	yglukhov_ joined #nim
07:16:31	*	gokr_ joined #nim
07:23:17	*	solidsnack quit (Quit: My Mac has gone to sleep. ZZZzzz…)
07:34:19	*	vendethiel quit (Ping timeout: 265 seconds)
07:36:34	*	OnO joined #nim
08:22:06	*	NimBot joined #nim
08:26:41	*	bjz joined #nim
08:26:57	filwit	"morning" gokr :)
08:27:09	gokr	:)
08:28:33	gokr	Just verified that the guy helping me with my taxes has gotten more time from the swedish IRS - phew. Otherwise last day today ;)
08:28:52	gokr	filwit: So doing anything fun?
08:29:35	filwit	hmmm.. not exactly "fun".. working on an event/object-scene system right now
08:29:52	*	ddl_smurf joined #nim
08:30:33	*	akiradeveloper joined #nim
08:31:54	filwit	getting it to load arbitrary Nim types from a file right now (can't use marshal, not enough options), and auto-wire up events/object-links of known "game objects" (those that use my 'impl' macro to declare events).
08:33:40	*	JinShil joined #nim
08:33:48	filwit	pretty fun actually, i enjoy this kind of thing.. and now that `getType()` actually works with typedesc inside macros (i need to scan the records of the types in order to distinguish references to known game objects) things are going great.
08:36:39	filwit	gokr: you?
08:36:43	*	bjz quit (Ping timeout: 264 seconds)
08:37:36	gokr	Mmmm, we are working on integrating Urh3D in our system. So I have some issues to work on
08:39:24	gokr	When you write "load arbitrary Nim types" - you refer to serialization?
08:39:59	filwit	yes
08:40:30	gokr	So you wrote your own?
08:40:44	filwit	well i'm writing that part of it now, yes
08:42:00	filwit	i've it working with marshal already, but I need to do some special processing on the instances before storing them (their ref-type records of known "game object" types are treated specially)
08:46:42	OnO	oh, good morning everyone
08:47:48	filwit	morning
08:48:00	*	akiradeveloper quit ()
08:49:17	*	vendethiel joined #nim
08:51:40	filwit	it's pretty great that you can use procs as global variable generators for generic parameter sets... I'm not sure if that's possible in C++ or not, but it's really useful.
08:52:35	gokr	I looked at marshal - is that the "only" serialization we have in stdlib?
08:52:55	gokr	Just wondering if we have more
08:53:21	filwit	well the only one that does all the heavy lifting for you. You can use typeinfo yourself which is easy.
08:54:49	filwit	but yeah, marshal should get some updates too.. and I really don't like the `$$` name personally... it's too rare a procedure to justify such a easy-to-miss-cause-it-looks-just-like-$ name.
08:55:36	gokr	I read through it - and its easy code. But ... no special hooks or support for migrating stuff. It does handle cycles right? I thought I saw that.
08:56:14	filwit	i haven't tested for cycles actually...
08:56:39	gokr	It seems to keep track of everything stored - and writes out a pointer int instead if already written.
08:57:49	filwit	it does "deep" serialization.. always, which is my problem with it.
08:58:07	gokr	Right, no "hooks"
08:58:23	gokr	So you can override what to store.
08:58:44	filwit	right
08:59:29	filwit	i'm not familiar with other serialization libs.. i did a tad of it with C# a few years ago, but I can't remember what features where available
08:59:42	filwit	and a little with D too
09:00:55	*	bjz joined #nim
09:03:36	gokr	Pharo has a very good one these days called Fuel.
09:04:14	gokr	I just clicked through the original presentation of it - one thing it does in order to be fast is to cluster instances by type.
09:05:04	*	milosn quit (Quit: leaving)
09:05:54	filwit	well for my purposes using typeinfo and macros directly works just fine.. it's really pretty easy to serialize Nim objects yourself
09:06:45	filwit	i found this out awhile ago and it's really cool:
09:06:45	gokr	So you store the actual type too?
09:06:45	filwit	proc enabledItems(T:typedesc, eventName:static[string]): var seq[T] = var data {.global.} = newSeq[T](); data
09:07:14	filwit	what do you mean by "store the type" ?
09:07:39	gokr	I mean... what about objects and inheritance etc?
09:08:13	gokr	If you always know the exact type of every field - and don't bother with migration - then sure, no need to store the type.
09:09:27	filwit	hmm.. that doesn't help me understand what you're asking really.. i don't know what you mean by "migration" exactly (though I can guess)
09:09:52	gokr	Migration as in loading data into a program in which types have changed.
09:10:06	gokr	Say objects that now have more or less fields etc
09:10:41	filwit	i see, well that's not really applicable to my situation (I don't think)
09:11:10	gokr	And if you have an object that has a field of type Fruit - which can ref an Apple or Banana, how do you store it?
09:11:51	gokr	I presume you need to store somehow the type of the fruit - or else, how do you know what to create when you load?
09:12:13	filwit	basically I'm trying to store a game's "scene", which is just a list of arbitrary Nim types registered as a "game object".. each scene is entirely self-contained (game objects shouldn't ever reference other's outside a given scene, at least not for initialization)
09:13:19	*	vendethiel quit (Ping timeout: 264 seconds)
09:13:31	filwit	and I load the entire scene at once.. each object instance in the scene has an index, and each object reference to any other game object in that scene will store reference by that index
09:15:07	filwit	so the step's I'm planning are: Allocate each instance in scene -> wire together known game-object references by index -> wire events for known game object's (and their known events) -> ready to execute
09:16:24	OnO	need help: how do I get T type's name having arg of T : typename
09:16:30	filwit	^ steps for loading from a scene*
09:16:57	filwit	OnO: repr(T)
09:17:58	OnO	Error: internal error: expr(skType); unknown symbol
09:18:05	OnO	type User = object
09:18:05	OnO	echo repr(User)
09:18:17	BlaXpirit	OnO, so it's a macro?
09:18:56	OnO	I have a: template `.`(T: typedesc[Model], f: string): ORMStmt =
09:19:00	filwit	OnO: well this works: proc sayName(T:typedesc) = echo repr T
09:19:03	OnO	I want turn T into its type name
09:19:06	OnO	a string
09:19:54	filwit	err.. sorry, you need to make a const equal 'repr(T)'
09:19:54	OnO	filwit: hmm... crashes here
09:19:57	BlaXpirit	OnO, typetraits.name
09:20:10	filwit	OnO: yeah, it's a bug, my bad. Do this instead
09:20:20	filwit	proc sayName(T:typedesc) =
09:20:25	filwit	const name = repr T
09:20:28	filwit	echo name
09:21:01	BlaXpirit	does not appear to work either
09:21:09	BlaXpirit	typetraits.name just works -_-
09:21:16	*	vendethiel joined #nim
09:21:28	OnO	yes... typetraits!
09:21:29	OnO	thanks
09:21:36	filwit	OnO: wtf.. that doesn't seem to work either.. i was just doing this, one sec
09:23:57	filwit	but yeah, typetraits is probably best anyways
09:24:05	OnO	is there any way to prevent symbol from module to be declared without suffix, I want use typetraits.name but not name
09:24:22	OnO	since name collides with mine
09:24:41	BlaXpirit	from typetraits import nil
09:25:24	BlaXpirit	this is the annoying part about nim. import nil only works well for modules that don't have any classes
09:25:50	OnO	BlaXpirit: ok, works, thanks
09:28:28	OnO	Araq: did I understood correctly that you are working on making seq[T] concat operation work on compile time?
09:30:30	BlaXpirit	but it works normally!
09:33:20	OnO	I am sorry, what?
09:34:11	*	Ven_ quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
09:40:37	*	RegisterOn joined #nim
09:40:54	RegisterOn	sup
09:41:07	filwit	hi
09:46:39	RegisterOn	is it possible to cross-compile nim programs? e.g. make freebsd binaries from osx or windows?
09:47:01	RegisterOn	i understand it's a c++ compiler underneath, so maybe it's just how i pass flags to that?
09:47:30	filwit	RegisterOn: http://nim-lang.org/0.11.0/nimc.html#cross-compilation
09:48:07	gokr	RegisterOn: Personally I would just use Vagrant or something.
09:49:57	filwit	on second thought.. i may just use marshal to do all serialization, then just scan that tree and register known 'game objects' after allocation (to wire state/events and trigger setup)
09:52:41	filwit	i think i can do it if I just append a typename to each root instance
09:53:42	filwit	er.. duh, that's obviously what marshal was intended for, now that I think about it..
09:54:39	gokr	Its indeed a common pattern - get it into memory first, shape it up.
09:55:08	dom96	filwit: I don't see why removing empty lines of whitespace would be a problem.
09:55:15	OnO	RegisterOn: for Linux cross compiling from OSX I use http://crossgcc.rts-software.org/doku.php
09:55:25	OnO	also for windows
09:55:26	filwit	yeah.. for some reason I was thinking marshal wasn't going to work because it didn't store the names or something.. but I think it does, so I'm not sure where I got that idea
09:55:39	gokr	The names of?
09:56:18	filwit	dom96: it's only annoying to have to retab to the correct place when trying to add code into the middle of a procedure
09:56:30	filwit	which happens a ton, so yeah
09:57:49	*	HakanD joined #nim
09:57:57	filwit	def mentioned it was to remove excessive whitespace from projects, which is cool, but IMO it should leave empty lines alone. Anyways, I wasn't just trying to complain about something, it's not a huge deal or anything.. i just didn't know if you'd done it intentionally or not.
09:58:07	dom96	I usually just press enter and the next line is at the correct indentation level.
09:58:30	dom96	I think there is an option to disable it though.
09:59:42	*	HakanD_ joined #nim
09:59:46	filwit	i looked through raw options and didn't see anything, but maybe it was just missing from the existing name (i noticed "scrollPastEnd" was in rawPrefs and not prefs). I'll look through the source code later and find out
10:03:15	*	HakanD quit (Ping timeout: 256 seconds)
10:03:58	filwit	dom96: any idea what the option would be called? (looking at aporia/cfg.nim right now)
10:04:55	dom96	hrm, I guess there isn't one. Sorry.
10:05:20	*	vendethiel quit (Ping timeout: 252 seconds)
10:05:54	filwit	no problem, do you know where it's handled in aporia.nim by any chance? I'll look into adding one.
10:07:55	*	antranigv joined #nim
10:13:32	filwit	nevermind, found it in utils.nim:normalize
10:15:15	*	HakanD_ quit (Quit: Be back later ...)
10:19:51	antranigv	is there a newsgroup for nim? (:
10:20:20	repax	antranigv: There's a forum.
10:20:23	repax	http://forum.nim-lang.org/
10:21:18	RegisterOn	thanks, seems like cross-compiling works just fine
10:22:18	filwit	dom96: okay, I changed it to keep whitespace of empty lines, but I'll add an option for it before I make a PR... I want to add de-indent on backspace in 2-or-more-whitespace chars too while I'm here
10:23:09	filwit	make that `tabcount`-or-more-whitespace chars of course
10:23:38	dom96	antranigv: there is also a mailing list http://nim-lang.org/community.html
10:24:18	RegisterOn	any well-known software written in Nim out there?
10:25:11	dom96	filwit: yes, add an option for both.
10:25:19	filwit	well known? I don't know of any. But have you seen def-'s NES emulator? it's pretty cool (and runs in the web too)
10:25:29	filwit	dom96: k
10:26:31	RegisterOn	anyway, well done guys, so far Nim looks really good for systems programming
10:26:34	filwit	RegisterOn: plus there's the Urho3D bindings as well, which is pretty cool
10:26:42	RegisterOn	i'll check it out
10:26:56	dom96	the nim forum is written in Nim too
10:27:03	RegisterOn	nice
10:28:11	*	HakanD_ joined #nim
10:30:07	*	Ven_ joined #nim
10:34:23	*	ingsoc joined #nim
10:35:53	*	ingsoc quit (Client Quit)
10:36:05	antranigv	nice :) the only reason I like newsgroups more than forums is the web (: so yay for the mailing list
10:36:16	*	banister joined #nim
10:36:47	RegisterOn	wait... what?
10:41:36	antranigv	in short. i love doing everything in alpine (:
10:42:14	RegisterOn	hah, just went back to mutt from alpine
10:51:20	*	vendethiel joined #nim
11:02:14	*	Kingsquee quit (Quit: Konversation terminated!)
11:06:11	antranigv	RegisterOn: out of topic. which one do you like better?
11:10:59	RegisterOn	antranigv: i like both, but mutt is just more configurable for the things I do, and it's has started to get some serious maintenance love lately, accepting my patches quickly
11:11:18	RegisterOn	antranigv: http://dev.mutt.org/hg/mutt/
11:11:34	RegisterOn	that's right, 11 hours ago :)
11:13:56	*	vendethiel quit (Ping timeout: 256 seconds)
11:14:16	antranigv	o~ nice (: did you write the patch in nim?
11:14:31	RegisterOn	heh, alas
11:29:49	*	TEttinger quit (Ping timeout: 245 seconds)
11:40:21	def-	http://hookrace.net/blog/nim-binary-size/
11:40:40	def-	^ New article, hope it's fun to read
11:41:01	RegisterOn	by you?
11:41:36	def-	yes
11:42:23	RegisterOn	awesome, i'm going to use Nim in a freestanding kernel project, just the stuff I love reading about :)
11:44:18	*	banister quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
11:44:34	jefus	i've yet to actually dive into nim, but indeed that was a fun and interesting read; thanks def-
11:47:54	RegisterOn	are there plans to provide a pure standard networking library (only wrappers now)?
11:48:10	*	seymour_ joined #nim
11:49:34	dom96	RegisterOn: have you seen the net module?
11:50:06	RegisterOn	dom96: ah, missed that, thanks!
11:50:36	*	seymour quit (Ping timeout: 240 seconds)
11:50:40	RegisterOn	oh, there's an asyncdispatch/net too
11:50:41	RegisterOn	niiiice
11:54:14	RegisterOn	ugh, seems like brew only has 0.10.2, i hope this gets updated soon
11:54:36	def-	RegisterOn: yes, they have an issue about it I saw yesterday
11:54:49	RegisterOn	yeah?
11:55:00	def-	https://github.com/Homebrew/homebrew/pull/39272
11:55:19	RegisterOn	aha, thanks!
11:55:31	def-	Where they say they want to wait for 0.11.2, which should should be released today if Araq hasn't changed plans
11:55:55	RegisterOn	Araq: plan on? :)
11:58:07	*	Ven_ quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
12:07:28	*	iamd3vil joined #nim
12:26:45	FDGFGFGDFD	http://www.reddit.com/r/programming/comments/34tb7r/nim_binary_size_from_160_kb_to_150_bytes/cqxvm1j
12:28:52	Araq	RegisterOn: yes.
12:34:17	thotypous	FDGFGFGDFD: nice article :D
12:35:48	FDGFGFGDFD	its from def- , i just wrote the stupid comment ;)
12:39:38	OnO	Araq: just to ensure that you've said yesterday evening, are you working on compile-time folding (concat) of seq[T]?
12:39:56	Araq	OnO: yes but not for 0.11.2
12:40:12	Araq	OnO: try --implicitStatic:on with your code, it might help
12:40:39	OnO	okay.... here's my experiment updated: https://gist.github.com/nanoant/ce1428eee5d009fdf3eb
12:41:05	r-ku	def-: thats pretty cool article. was interesting read
12:41:14	OnO	I am still not sure whether I should use templates or macros for generating SQL + argument variables seq from stmt
12:41:40	OnO	templates seem to be more natural, but so far I can't make strings fold this way
12:42:02	Araq	you're doing it wrong ;-)
12:42:03	ddl_smurf	you probably can't do everything with templates, but what you can you should
12:42:49	*	OnwardEuler joined #nim
12:42:59	OnO	Araq: tell me more :)
12:43:38	Araq	you're building some AST from Nim's AST via templates, that's not necessary, you only need a single toSql macro
12:43:57	Araq	or maybe not, but that's what I would try
12:44:07	OnO	yes, this is what I was asking :> so I'll try with macros
12:44:09	gokr	def-: Fun article
12:44:13	Araq	macro toSql(x: untyped)
12:44:43	OnO	just one Q everything that comes as "s" to macro(s:stmt) is already pre-optimized?
12:45:08	OnO	I mean calling toSql(v == 1 + 1) I will get AST for 1+1 or just 2
12:45:09	OnO	?
12:45:31	Araq	that's what --implicitStatic:on tries to do and what I'm working on
12:46:19	Araq	but well for 'untyped' you won't get this
12:46:33	Araq	but it's not necessary for this I think
12:48:04	*	vikaton joined #nim
12:48:25	r-ku	gokr: poke. just so you know.. urhonimo github repo wasnt synced in a month. maybe whenever you are bored and have nothing else to do..? ;)
12:49:48	*	kulelu88 joined #nim
12:51:39	*	NimBot joined #nim
12:51:46	kulelu88	vikaton: yes.
12:51:50	kulelu88	let me find it for you
12:51:59	vikaton	thanks
12:52:21	kulelu88	vikaton: https://github.com/idlewan/nawak
12:52:23	RegisterOn	but is it nimble? :D
12:52:43	kulelu88	the guys who work on Nim don't seem to be too eager to make nim a web-ready language
12:52:58	kulelu88	the scope is probably too narrow for just web stuff
12:53:12	kulelu88	so they prefer to stay on a gaming/systems kind of level
12:53:20	Araq	IMO def- did destroy the point of web framework benchmarks yesterday night
12:53:22	vikaton	o
12:53:40	vikaton	Araq, what did he say?
12:53:43	OnwardEuler	Araq: how so?
12:54:29	kulelu88	and they may also be shooting themselves in the foot by making Nim web-ready. Python and Ruby are just so much easier to use for web stuff and it kind of seems irrational to choose nim/go/rust for your small webapp
12:54:48	*	iamd3vil left #nim ("Leaving")
12:55:54	vikaton	I benchmarked Rust vs Crystal yesterday https://gist.github.com/Vikaton/3102c16ee14ef3d76244
12:56:01	vikaton	and I was looking forward for benchmarking Nim
12:56:24	vikaton	though maybe I should wait since dom96 informed me that jesters isnt quite optimized yet
12:56:48	*	EXetoC joined #nim
12:57:14	kulelu88	vikaton: I think the ideal benchmark would be to handle +1 million requests per second?
12:57:17	dom96	kulelu88: I certainly want to make Nim web ready.
12:57:43	vikaton	kulelu88: this is a 512 mb RAM vps, kinda hard to get it to hit 1mil
12:58:10	vikaton	though it does show a rough image between the 2
12:58:49	kulelu88	dom96: even with a genius like yourself behind Jester, the momentum seems to be with Go for fast webapps
12:58:56	*	HakanD_ quit (Quit: Be back later ...)
12:59:08	dom96	kulelu88: Hah. I wouldn't consider myself a genius but thank you :)
12:59:16	vikaton	I really dont like Go
12:59:18	dom96	I guess the biggest problem is that I have very little time nowadays.
12:59:32	kulelu88	what's Crystal?
12:59:38	Araq	he said he got 1-2 requests per second when his article was the top at hacker news and reddit
12:59:39	vikaton	I would def go with Crystal rather than Go, think of it as a blazing fast rails app
13:00:11	vikaton	kulelu88: compiled Ruby (basically, though not literally)
13:00:32	Araq	not 1000 requests per second, not 100, not 10, but perhaps 2.
13:01:18	vikaton	Araq, Def-'s website is personal is it not?
13:01:41	EXetoC	2 :p
13:01:59	kulelu88	a game like agar.io in Nim would be the ultimate test to see if it can compete with C++
13:02:08	EXetoC	oh
13:02:12	kulelu88	I've never heard of Crystal before. I know Pypy
13:02:46	vikaton	kulelu88: its barely 1 year old
13:02:47	pigmej	kulelu88: pypy is 'python' not the same group
13:03:28	*	HakanD_ joined #nim
13:03:29	vikaton	Crystal/Nim/Rust/C++/C are all in the same window of performance
13:03:40	antranigv	I am amazed that github recognizes Nim, but not Oberon-2 :)
13:03:43	kulelu88	pigmej: I know Pypy is python +P I'm just trying to say that I've never heard of compiled Ruby before
13:04:17	kulelu88	at least you guys welcome these discussions. Try asking these questions in Go :D
13:04:21	pigmej	kulelu88: pypy is not 'compiled' python
13:04:26	thotypous	antranigv: there is a linguist repository for recognizing languages which accepts PRs, so basically it is collaborative
13:04:36	vikaton	Go is before my time
13:04:43	kulelu88	it's JIT and 'faster' than Cython (apparently)
13:04:56	pigmej	kulelu88: cpython,
13:05:00	vikaton	Crystal is AOT
13:05:03	pigmej	cython is also different language
13:05:13	kulelu88	aah yes. i'm confused again
13:05:17	pigmej	kulelu88: the thing is, you can't compare apples and oranges :)
13:05:46	*	mpthrapp joined #nim
13:05:55	pigmej	pypy have awesome JIT, it's starting to have STM, it's generally very cool project
13:06:04	kulelu88	pigmej: true. that's why Im no longer fighting it. I'll use Python for now and dream of a nim-py-go blazing fast, easy to write, web language someday
13:06:10	pigmej	and rpython does the job very well in other langauges (ruby, php)
13:06:39	vikaton	whats
13:06:52	vikaton	nim-py-go blazng fast?
13:07:38	kulelu88	vikaton: in simple terms, convulated bullshit +)
13:07:40	kulelu88	=)
13:08:44	vikaton	kulelu88: if u want that, maybe this will interest u, https://github.com/dhruvrajvanshi/Moonshine
13:09:59	kulelu88	would it be possible to write the underlying workings of a web-framework in nim and use Python on top of it for the basic stuff?
13:10:21	kulelu88	something like: https://github.com/StarfruitStack/guava
13:12:12	*	banister joined #nim
13:12:32	thotypous	kulelu88: does python act like a glue language for plugging C modules in guava?
13:12:47	thotypous	imho python is not the most adequate language for this kind of task
13:13:40	thotypous	lua is usually better as a glue for lowlevel code
13:13:49	kulelu88	I think the core of the work is written using the Python-C API and this framework is rest-like, so you write what I would call the "fluff" code on top
13:14:05	kulelu88	I could be wrong
13:14:12	kulelu88	Always assume I am wrong
13:14:15	kulelu88	=)
13:18:08	Araq	kulelu88: the point of compiling things is not only speed, it's also security
13:18:24	Araq	Nim has no 'eval', Python does
13:19:18	Araq	Nim tries to prevent SQL injection attacks via its type system, Python doesn't even have a type system
13:19:32	antranigv	Araq: agree (:
13:21:08	thotypous	well, one could argue the same for java or .net :P
13:22:32	Araq	thotypous: Java and .NET are heavily used for web development but have reflection which weakens these systems in practice to be hardly better than dynamically typed, ymmv
13:23:25	thotypous	(i purposefully said .net in order to include F# in the list :P)
13:24:03	vikaton	could Nim implement n eval command tool/
13:24:15	vikaton	'nim eval 'echo "hi"`
13:24:20	*	OnO quit (Quit: My iMac has gone to sleep. ZZZzzz…)
13:24:34	thotypous	if you link to the compiler and to libtcc, it would be possible
13:25:55	Araq	yes and there is also os.execShellCmd anyway. but you can use Nim's effect system to prove "this section of code doesn't exec external stuff"
13:27:16	thotypous	which gets us to the point that, unless you strictly enforce W^X memory protection, you can treat the cpu as being as vulnerable to "reflection" as the java or .net VMs ;D
13:28:27	Araq	er ... no. why do you think that?
13:30:07	thotypous	yes, you are right, not really with nim, unless there is a bug in unsafe code
13:31:07	*	vendethiel joined #nim
13:36:20	pigmej	kulelu88: the thing is if you want fast python 'api' then just use it
13:36:34	pigmej	you will need uwsgi / nginx in front anyway, so ...
13:36:37	pigmej	it's C level,
13:46:26	RegisterOn	that said, reflection is a wonderful tool for writing serialization and ORM frameworks :)
13:47:06	thotypous	metaprogramming ftw
13:47:11	RegisterOn	while in the C world, stuff like s11n has to resort to pretty lame linker tricks
13:47:27	*	darkf quit (Quit: Leaving)
13:47:32	Araq	RegisterOn: these things are better done with a macro system
13:47:55	OnwardEuler	Are there any examples of how to use the socket libraries?
13:48:33	RegisterOn	Araq: how can you discover facts about types at runtime with a macro system? Bear with me, I'm a total Nim noob :)
13:48:53	RegisterOn	from time to time, that's mighty useful
13:49:07	RegisterOn	load in plugins dynamically, discover stuff
13:49:27	Araq	you don't, that's the point. instead you generate specialized serializers at compile-time with the type knowledge you have at compile-time
13:49:38	RegisterOn	heh
13:56:16	*	vendethiel quit (Ping timeout: 240 seconds)
13:59:12	dom96	OnwardEuler: What would you like to do?
14:00:32	dom96	The net docs really do need more examples.
14:00:35	OnwardEuler	dom96: I'm trying to make a simple client/server based app
14:00:40	dom96	The asyncnet docs have an example.
14:00:51	dom96	of exactly that
14:01:04	dom96	http://nim-lang.org/asyncnet.html
14:01:19	OnwardEuler	Awesome, thanks!
14:02:18	OnwardEuler	I hadn't even looked at the asyncnet pages... that's exactly what I needed =)
14:02:20	*	banister quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
14:09:36	*	ddl_smurf quit (Quit: ddl_smurf)
14:11:17	*	banister joined #nim
14:14:27	*	gsingh93 joined #nim
14:16:21	*	saml joined #nim
14:18:12	kulelu88	proc is the same thing as func right?
14:18:45	BlaXpirit	kulelu88, func does not exist
14:18:49	BlaXpirit	yet
14:19:19	kulelu88	but isn't proc equivalent to a function?
14:19:27	BlaXpirit	but yes, you can use `proc` for procedures that return values
14:19:48	thotypous	as in "equivalent to the concept of function in other languages"?
14:19:49	thotypous	yes
14:19:51	*	superfunc joined #nim
14:19:54	BlaXpirit	long story short, proc is a function
14:22:08	*	gsingh93 quit (Ping timeout: 244 seconds)
14:22:14	*	jsudlow joined #nim
14:22:17	kulelu88	in the case of optimizing jester, where is the optimization lacking? In the algorithms or in the use of outdated features?
14:22:49	thotypous	is jester slow at all? i think it is pretty fast
14:23:06	superfunc	haven't been around in a bit, have typeclasses changed much, or are they still mostly experimental?
14:23:28	BlaXpirit	superfunc, they work better
14:23:30	Araq	superfunc: they kind of work now and are called 'concept's now
14:23:39	superfunc	cool
14:23:43	superfunc	also, hey Araq
14:24:00	*	kulelu88 superfunc should be known as superproc
14:25:01	*	milosn joined #nim
14:26:00	OnwardEuler	dom96: that example in the docs won't compile, says I've got an Error: unhandled exception, ivalid format string [ValueError]
14:26:17	*	Ven joined #nim
14:28:04	*	HakanD_ quit (Quit: Be back later ...)
14:28:15	dom96	OnwardEuler: works for me on Windows.
14:28:37	*	HakanD_ joined #nim
14:28:38	OnwardEuler	I'm on windows, latest nim and everything
14:28:38	*	HakanD_ quit (Read error: Connection reset by peer)
14:28:52	*	mhi^ joined #nim
14:30:17	*	ChrisMAN joined #nim
14:30:48	dom96	OnwardEuler: weird
14:30:48	OnwardEuler	" Error: unhandled exception: invalid format string [ValueError] "
14:30:59	dom96	Is the compiler crashing?
14:31:06	OnwardEuler	No line info or anything, could it be a macro thing?
14:31:19	*	BitPuffin|osx joined #nim
14:31:22	Araq	more likely your config is borked
14:32:01	*	Jehan_ joined #nim
14:32:29	OnwardEuler	I'll take a look at my cfg
14:34:27	filwit	def-: nice article
14:34:48	OnwardEuler	Araq: I'm pretty sure I'm using the default config
14:35:25	OnwardEuler	Should I look for anything in particular?
14:35:25	def-	Araq: any idea how I can set codegenDecl on the declaration of a proc only and not the definition? I'm trying {.codegenDecl: "$# $# $# __attribute__((noreturn))".} because {.noReturn.} doesn't seem to do that
14:35:26	*	gokr quit (Quit: Leaving.)
14:35:51	OnwardEuler	oh, dom96: yeah, I'm pretty sure the compiler's crashing
14:36:13	Araq	yeah well... that doesn't work, def- :-/
14:36:30	def-	Araq: alright, too bad. would have saved another byte ;)
14:36:48	dom96	OnwardEuler: Can you compile hello world?
14:38:24	OnwardEuler	dom96 - Nope that crashes too, same error
14:38:46	OnwardEuler	should have tried that sooner, I only just installed 0.11
14:41:02	OnwardEuler	reinstalling, will report back
14:44:18	*	gsingh93 joined #nim
14:48:14	OnwardEuler	dom96, Araq: Don't know what happened, reinstalled, works now. Thanks!
14:48:36	dom96	OnwardEuler: That's good to hear.
14:49:06	superfunc	I wonder what happened
14:50:23	*	HakanD_ joined #nim
14:50:58	Araq	bbs
14:57:02	def-	t
15:04:11	*	Ven quit (Disconnected by services)
15:04:32	OnwardEuler	dom96: though telnetting into the server yields some weird results hahah
15:04:51	*	Ven_ joined #nim
15:06:01	*	OnO joined #nim
15:10:48	*	gyeates joined #nim
15:21:13	kulelu88	dom96: do you have enough time to write an app similar to this one: http://tangowithdjango.com ?
15:22:44	dom96	kulelu88: I am planning on writing something similar for Jester. Whether I have the time... nope, at least not right now with exams coming up...
15:23:16	*	ddl_smurf joined #nim
15:25:42	*	ingsoc joined #nim
15:26:58	kulelu88	dom96: perhaps you can skip the tutorial part and write the app only. I'd like to see a proper crud app in jester that i can do cookie-cutter work with
15:28:08	dom96	kulelu88: Isn't the forum already a good enough example app?
15:28:31	kulelu88	dom96: not really. there needs to be a simpler tutorial using basic crud workflow
15:28:46	kulelu88	a forum is a big application with lots going on
15:28:55	*	mhi^ left #nim (#nim)
15:29:36	*	ddl_smurf quit (Quit: ddl_smurf)
15:29:39	*	jholland joined #nim
15:30:09	dom96	sure
15:30:22	*	aboisvert joined #nim
15:31:48	kulelu88	but if you're busy your studies should take priority, not some silly webapp :P
15:38:22	*	superfunc quit (Ping timeout: 246 seconds)
15:41:00	*	vendethiel joined #nim
15:45:51	*	OnO quit (Quit: My iMac has gone to sleep. ZZZzzz…)
15:49:42	HakanD_	def-: your article at front page of HN again \o/
15:53:02	*	Ven_ quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
15:53:35	*	aboisvert quit (Quit: aboisvert)
15:54:56	*	ingsoc quit (Quit: Leaving.)
16:00:12	*	aboisvert joined #nim
16:06:04	*	vendethiel quit (Ping timeout: 245 seconds)
16:09:48	*	OnO joined #nim
16:10:34	*	ddl_smurf joined #nim
16:11:25	Araq	please test: http://nim-lang.org/download/nim-0.11.2_x32.exe
16:11:33	Araq	please test: http://nim-lang.org/download/nim-0.11.2_x64.exe
16:11:47	Araq	please test: http://nim-lang.org/download/nim-0.11.2.zip
16:13:28	Araq	bbs
16:15:14	*	UberLambda joined #nim
16:19:18	Jehan_	Araq: Doesn't build for me.
16:19:48	Jehan_	Well, the initial build.sh run works, but "bin/nim cc koch.nim" does the following:
16:20:03	Jehan_	Eh, ./koch boot -d:release
16:20:06	Jehan_	compiler/sem.nim(19, 43) Error: cannot open 'plugins/active'
16:20:10	BlaXpirit	try update csources
16:20:39	BlaXpirit	Araq, all my code works. regression fixed.
16:20:59	Jehan_	BlaXpirit: That's directly from the nim.zip that Araq wanted tested.
16:21:14	Jehan_	So, updating csources is sort of beside the point.
16:21:14	BlaXpirit	oh...
16:21:25	BlaXpirit	yeah, i just checkout devel :/
16:23:18	Jehan_	BlaXpirit: So do I normally. :)
16:23:36	*	ddl_smurf quit (Quit: ddl_smurf)
16:25:01	BlaXpirit	i thought build.sh is all that's needed
16:25:43	Jehan_	Ah, it looks like the plugins directory is simply not included in the zip file.
16:26:25	Jehan_	BlaXpirit: Yeah, that gets the compiler running, but I wanted to test it, and tried to do that by having it build itself.
16:26:37	*	solidsnack joined #nim
16:27:02	BlaXpirit	so... the compiler can't build itself but will work anyway?
16:27:43	BlaXpirit	nim from archive works well
16:28:35	Jehan_	BlaXpirit: Yeah, because part of the compiler sources are missing.
16:32:11	Jehan_	It's nice how fast a codebase can progress when the author can actually work fulltime on it. :)
16:33:16	*	solidsnack quit (Ping timeout: 255 seconds)
16:33:55	*	vendethiel joined #nim
16:37:32	gyeates	q
16:37:56	*	gsingh93 quit (Ping timeout: 264 seconds)
16:39:59	OnwardEuler	haha are we getting a new release already?
16:40:26	BlaXpirit	yes
16:40:44	OnwardEuler	sweet =)
16:45:43	*	DKordic left #nim ("ERC Version 5.3 (IRC client for Emacs)")
16:56:00	*	BitPuffin|osx quit (Ping timeout: 252 seconds)
16:57:16	*	dalarmmst quit (Ping timeout: 240 seconds)
16:57:20	*	solidsnack joined #nim
17:02:09	def-	Araq: runs fine for me, but some tests fail because of missing files still, for example tests/testdata/jsontest.json
17:02:38	ldlework	Why new release?
17:03:06	def-	ldlework: 0.11.0 broke existing code with a risky change
17:03:20	ldlework	ouch
17:05:22	OnO	help pls, why: macro where(T: typedesc[Model]): stmt {.immediate.} = accepts any type as 1st param, not only Model and subclasses?
17:05:38	OnO	if I remove {.immediate.} then it behaves as desired
17:06:22	ldlework	OnO: that's what immediate means
17:07:12	ldlework	OnO: immediate macros are applied lexically. it doesn't know what the type of the thing you're passing to your macro is.
17:08:01	OnO	I would expect that only for expr/stmt arguments, NOT all arguments
17:08:07	OnO	it does not make sense
17:08:57	OnO	acutally I have: macro where(T: typedesc[Model], st: expr): stmt {.immediate.} =
17:09:29	fowl	Immediate is getting removed
17:09:30	OnO	I want it match Model.where(any non-evaluated expression)
17:09:32	ldlework	OnO: there's no typing involved in an immediate macro
17:09:35	ldlework	fowl: woah really?
17:09:43	fowl	Yea
17:09:48	ldlework	why?
17:09:58	fowl	You'll use expr/untyped insteas
17:10:36	OnO	ldlework: again, I understand {.immediate.} can apply to stmt/expr aguments, but no reason it should apply to all
17:10:51	ldlework	OnO: it applies to the entire macro
17:11:00	ldlework	its a different type of macro application -all together-
17:11:26	OnO	ldlework: okay, but does it make sense? I don't see any
17:12:23	ldlework	OnO: immediate macros are lexical. If you can't write your macro only making lexical transformation, you need a non-immediate macro.
17:13:07	ldlework	That it supports syntax such as type classes is unfortunate, but it sounds like the whole thing is being removed from the language.
17:13:22	*	gsingh93 joined #nim
17:13:24	OnO	ldlework: I understand that, but you are not trying to understand me.... immediate should apply only to stmt/expr arguments of macro/template, NOT all arguments
17:13:38	ldlework	OnO: no
17:13:56	ldlework	a macro applied immediate *can't possible know* *any type information*
17:14:05	ldlework	regardless of what kin of parameter is is.
17:14:08	ldlework	it *can't* know
17:14:47	ldlework	if I understand correctly, anyway.
17:15:47	ldlework	(its applied for type information is established by the compiler)
17:15:51	ldlework	before*
17:15:57	OnO	ldlework: sorry, but I completely don't follow that, how it can't know, compiler is just deciding if it is going to parse AST or not prior passing it to macro
17:16:19	OnO	but this makes sense only for stmt/expr arguments
17:16:37	OnO	why NOT parse argument of type typedesc[Model] !?
17:17:06	BlaXpirit	immediate macros completely ignore types
17:17:45	fowl	Oh well the idiot who wrote that part of the compiler couldn't figure it out and that's something we live with. Some excuse about not wanting the compiler to have to do a hundred passes
17:17:46	ldlework	OnO: because that's what ________,.-> immediate <-.,________ means
17:18:26	ldlework	"Apply this macro, before any type information has been figured anywhere in the program, lexically, no questions asked."
17:18:29	*	fowl left #nim ("Leaving")
17:18:54	BlaXpirit	good explanation
17:19:23	*	HakanD_ quit (Quit: Be back later ...)
17:19:37	OnO	okay, I understand that any argument to immediate macro is just NimNode and any type spec is silently ignored
17:20:13	OnO	the problem I have is that cannot bind such a macro to certain type, becaue it is all or nothing
17:20:30	BlaXpirit	yup
17:20:35	OnO	:(
17:21:00	BlaXpirit	i think this will be solved with changes mentioned by fowl
17:22:03	BlaXpirit	but OnO, maybe you don't need different macros
17:22:25	BlaXpirit	do you realize that you can put a big if/case around it all and...
17:22:49	OnO	I need: macro where(T: typedesc[Model], st: expr): stmt {.immediate.} = <- where T is parsed existing subtype of Model and st is non-parsed expression
17:23:01	BlaXpirit	why do you need it
17:23:23	OnO	to type User.where(@name = somename)
17:23:38	OnO	and know what was the 1st T type
17:23:49	BlaXpirit	riiight... making a global "where" macro that works for everything is not nice
17:23:53	OnO	because where macro will eventually lead to constructing such object
17:24:15	OnO	BlaXpirit: that's right :)
17:24:18	*	JinShil quit (Quit: Konversation terminated!)
17:24:30	OnO	it is very very not nice ;)
17:24:39	BlaXpirit	i ran into this very recently
17:25:13	OnO	I think it should be the arguments annotated as {.immediate.} not whole macro
17:25:24	OnO	macro where(T: typedesc[Model], st: expr {.immediate.}): stmt =
17:25:40	OnO	so st is unparsed, but everything else is
17:25:44	BlaXpirit	[:20:58] <BlaXpirit> i think this will be solved with changes mentioned by fowl
17:26:02	BlaXpirit	who knows, it might even work already
17:26:30	OnO	oh... fowl can you elaborate this?
17:26:47	OnO	is this what "untyped" is for?
17:28:11	BlaXpirit	i have good news for u, OnO
17:28:35	OnO	yess.... it works with untyped
17:28:38	OnO	sweeeeet
17:28:41	BlaXpirit	OnO, https://bpaste.net/show/0b4084a32fc4
17:29:08	OnO	awesome.... {.immediate.} must die!
17:29:54	ldlework	OnO: congrats!
17:30:11	yglukhov_	Hello. Can anyone help with vararg forwarding please? Details: http://forum.nim-lang.org/t/1186
17:31:11	BlaXpirit	yglukhov_, varargs might not even work at all in templates
17:32:06	BlaXpirit	yglukhov_, https://github.com/Araq/Nim/issues/1862
17:32:20	BlaXpirit	oh wait.. this is about immediate, sorry
17:33:19	BlaXpirit	but how do you expect to pass an array of expressions???
17:35:09	Jehan_	OnO: Immediate will be deprecated eventually.
17:35:33	BlaXpirit	i dont know why everyone isn't talking about this "untyped"
17:36:07	*	filcuc joined #nim
17:36:08	OnO	huh, I am almost there with compile type expr to SQL :)
17:36:19	OnO	I miss compile time toLower and toUpper for strings
17:36:57	yglukhov_	BlaXpirit, simplified my example.
17:37:35	Jehan_	yglukhov_: There's an outstanding issue with varargs[expr] for templates.
17:37:57	Jehan_	I'll have to look closer at it, but that might be it.
17:38:33	BlaXpirit	yglukhov_, replied to simplified example
17:38:56	*	OnwardEuler quit (Ping timeout: 240 seconds)
17:42:30	yglukhov_	BlaXpirit, but that's not fair. I want to expand varargs so that $ will be called again on every argument, when passed to echo. But with your suggestion that's not the case, right?
17:42:53	BlaXpirit	just call it yourself
17:43:08	*	posix4e joined #nim
17:43:28	posix4e	Howdy. I am new to nim. I am curious why I should consider nim vs rust
17:44:19	BlaXpirit	there are enough such comparisons.
17:44:30	posix4e	I felt they only hit the surface
17:44:42	posix4e	Would it make sense to write an OS in nim?
17:44:52	Jehan_	Different goals, really.
17:44:55	posix4e	Is the runtime libc?
17:45:18	Araq	somebody is doing just that. moved from Rust over to Nim for OS kernel development.
17:45:31	Araq	iirc he also blogged about it
17:45:31	BlaXpirit	https://news.ycombinator.com/item?id=9049698
17:45:52	Jehan_	Nim attempts to bridge the gap between the expressiveness of a high-level language such as Python and the performance of a system-programming language such as C.
17:46:09	BlaXpirit	what
17:46:14	BlaXpirit	what language doesn't nowadays
17:46:30	Jehan_	BlaXpirit: Umm, plenty?
17:46:50	Jehan_	Rust aims more at being a more rigorous and disciplined alternative to C++.
17:47:47	Jehan_	BlaXpirit: Go, for example, aims for simplicity over expressiveness.
17:48:23	posix4e	Jehan_: But go isn't really comparable to rust right? More java or python or something
17:48:49	posix4e	aren't going to write a memory allocator in go
17:48:50	Jehan_	posix4e: Not really, either.
17:49:05	Jehan_	It's not like Rust, but it's not like Java or Python, either.
17:49:23	BlaXpirit	worst of both worlds
17:49:24	Jehan_	It's still meant for code that's pretty "close to the metal".
17:49:38	Araq	posix4e: iirc they rewrote their GC and memory allocator in Go
17:49:55	posix4e	Araq: I think it's still using libvc
17:49:56	posix4e	libc
17:50:24	posix4e	So let's talk gc
17:50:30	posix4e	it seems as though there's a rt gc in nim
17:50:37	posix4e	I have no idea why no one else does that
17:50:44	posix4e	sans azul
17:51:06	Araq	there is also Jamaica and a couple of others
17:51:21	posix4e	What if i set the max pause time to 10ms
17:51:33	Araq	what if you don't?
17:51:35	posix4e	am i going to be trading a ton of perf?
17:51:40	posix4e	Araq: sorry two part question
17:54:45	Araq	posix4e: if the GC can put up with the number of allocations that you do, or rather the amount of garbage the mutator produces, there is no perf overhead
17:55:16	Araq	in fact, it's more like: there is no perf overhead, but if the time slice is too short, it starts to leak
17:55:46	Araq	well not really "leak" but memory usage grows without boundary
17:55:57	Araq	and you either die with OOM
17:56:04	Araq	or you need to give it more time
17:56:09	*	gyeates quit (Ping timeout: 240 seconds)
17:56:30	Araq	Jehan_: can you test the download again please?
17:56:42	Jehan_	Araq: Second.
17:57:04	BlaXpirit	when to expect git tag and merge to master?
17:57:08	OnO	okay I am getting closer, how do I convert typedesc into NimNode of nnkSym type?
17:57:24	OnO	not found any proc for that in macros
17:57:34	r-ku	newNimNode()?
17:57:53	Araq	depends on the details
17:58:40	OnO	#proc `typ=`*(n: NimNode, typ: typedesc) {.magic: "NSetType".} <- commented in macros
17:59:42	OnO	macro where(T: typedesc[Model], st: untyped): stmt = newCall(newDotExpr(T, bindSym"execQuery"), newLit(genSQLQuery(T.repr, st, args)))
17:59:47	Jehan_	Araq: Hmm, still missing plugins/active
18:00:01	Jehan_	Eh, second.
18:00:05	OnO	however newDotExpr(T, bindSym"execQuery") is invalid, I need to turn T into NimNode
18:00:11	Jehan_	Download messed up, may have used the old zip file.
18:00:13	OnO	but no idea how
18:01:09	*	filcuc quit (Read error: No route to host)
18:01:23	*	filcuc joined #nim
18:02:32	Jehan_	Araq: Okay, compiler builds now.
18:02:34	*	filcuc quit (Remote host closed the connection)
18:02:48	Jehan_	And tests seem to be performing as before.
18:03:45	*	gsingh93 quit (Ping timeout: 240 seconds)
18:04:43	OnO	okie newIdentNode(T.repr) seems to work
18:06:48	Jehan_	Hmm, longish discussion about null references on HN.
18:06:54	*	dalarmmst joined #nim
18:10:54	posix4e	Well I have to admit I can't imagine times when i'd choose go over nim then. If for no other reason then the GC.
18:11:46	Araq	posix4e: well Go has a shared memory GC, Nim uses thread local heaps instead. there are plenty of different tradeoffs made in these languages
18:12:55	posix4e	Araq: that's very nice of you. It'll be interesting to see how go gc perf changes with a concurrent collector.
18:13:57	Araq	posix4e: well if Go becomes too good at one day, we'll likely grow a codegen for it ;-)
18:14:10	posix4e	that's funny
18:14:27	posix4e	Maybe I should be thinking of nim as a meta code gen language
18:15:01	posix4e	I mean when I first heard about nim-> c I was all like "What's so special c++ did this for years" But perhaps i didn't get it
18:15:03	*	stefantalpalaru joined #nim
18:15:25	*	dalarmmst quit (Ping timeout: 255 seconds)
18:17:11	*	dalarmmst joined #nim
18:18:43	stefantalpalaru	Hi! What's the best way to use Ident nodes as keys for a table? I need this for a macro that may be called from multiple modules and the idents may be procedures with the same name for which the string representation I get with $ is the same. The probmlem with using NimNode as key is that there's no defines hash proc for them.
18:19:57	Araq	stefantalpalaru: dunno, have you considered a different approach? I like to get away from these compileTime global vars which break modularity
18:20:09	BlaXpirit	stefantalpalaru, i think ident nodes don't carry any magical extra identification than the names themselves
18:20:19	Araq	and by different approach I mean staticExec plus static marshal calls
18:20:47	BlaXpirit	Araq, sometimes it is necessary to have state.....
18:21:27	Araq	also you can get a different "name" for the proc via getType()
18:22:11	BlaXpirit	this getType seems like the be-all-end-all
18:22:18	stefantalpalaru	I don't see how. I need to create and use a new var for each targeted function from 2 different macros that may be called in any order. My plan is to have a global (compile time) table with function idents as keys and these new gensymed variables as values.
18:23:32	BlaXpirit	Araq, are you seriously going to keep documentation at old version?
18:23:33	*	brson joined #nim
18:23:55	BlaXpirit	I only accidentally found out that it moved
18:24:56	*	dalarmmst quit (Ping timeout: 240 seconds)
18:25:27	yglukhov_	Another template related question. How do I do type matching here: https://gist.github.com/yglukhov/9a2a66074b67053a53da#file-gistfile1-nim-L6-L7
18:26:31	BlaXpirit	inb4 someone says that the answer here is also getType
18:26:51	Araq	yglukhov_: you can use 'T is seq' instead but it's futile, typetraits already exists and there is also macros.getType to get real type inspection facilities
18:27:00	BlaXpirit	called it
18:27:52	*	solidsnack quit (Quit: My Mac has gone to sleep. ZZZzzz…)
18:29:27	stefantalpalaru	I was looking at nim-lang.org/macros.html instead of nim-lang.org/0.11.0/macros.html and didn't find what getType. Not cool.
18:30:03	BlaXpirit	that's what i said too
18:30:08	Araq	funny that you say that, I'm removing the old stuff
18:30:27	BlaXpirit	you can't just 404 all of the links..........
18:30:43	stefantalpalaru	do a HTTP redirect instead
18:30:58	BlaXpirit	problem is it's not so easy
18:31:08	stefantalpalaru	or point them to the most recent version
18:31:13	BlaXpirit	some dynamic logic would be best
18:31:22	BlaXpirit	and it is definitely not an option to not have a link to latest docs
18:31:24	Araq	they do point to the most recent version
18:31:37	dom96	there needs to be a nim-lang.org/latest/
18:31:43	dom96	and or better yet
18:31:47	dom96	nim-lang.org/docs/
18:31:49	Araq	dom96: I thought you did that?
18:31:50	*	gokr joined #nim
18:31:52	dom96	just put the docs there.
18:31:56	dom96	not yet
18:32:04	Araq	-.-
18:32:08	dom96	I think you should put docs in nim-lang.org/docs/
18:32:16	Araq	ok
18:32:29	stefantalpalaru	they don't point to the most recent version unless you just removed getType: http://nim-lang.org/macros.html#getType,NimNode
18:32:33	BlaXpirit	that's all fine but what's going to happen to old links?
18:32:50	dom96	As far as I can see redirecting all those links is not trivial.
18:33:04	BlaXpirit	[:30:56] <BlaXpirit> problem is it's not so easy
18:33:05	*	gyeates joined #nim
18:33:43	BlaXpirit	well if some dynamic logic can be applied, then in case of 404 under root, it can look inside 0.11.2 folder instead
18:33:55	stefantalpalaru	it's nginx, how hard can it be to write some redirection rules?
18:34:16	thotypous	http://nim-lang.org/0.11.0/documentation.html <-- fun, language manual points to /0.11.0/0.11.0/manual.html
18:34:24	*	aboisvert quit (Quit: aboisvert)
18:34:30	*	dalarmmst joined #nim
18:34:42	BlaXpirit	uh, thotypous, i don't think that address is supposed to be used
18:34:59	*	solidsnack joined #nim
18:35:14	*	OnO quit (Quit: My iMac has gone to sleep. ZZZzzz…)
18:35:15	Araq	thotypous: it's called a "relative" path
18:35:22	dom96	stefantalpalaru: There is a lot of docs.
18:35:35	BlaXpirit	[:33:41] <BlaXpirit> well if some dynamic logic can be applied, then in case of 404 under root, it can look inside 0.11.2 folder instead
18:35:40	dom96	Guess I can write a script to do it for me.
18:35:42	def-	dom96: make a script to make a rewrite rule for each? this should be a one-time thing i guess
18:35:53	dom96	yep
18:35:56	BlaXpirit	...............
18:36:12	Araq	google should be smart enough to learn about the new structure within minutes
18:36:16	def-	and then google will be happy once they see that the pages moved and index the new ones
18:36:17	*	OnO joined #nim
18:36:41	Araq	google only needs to follow the links properly ...
18:36:44	def-	Araq: maybe if you removed the old ones
18:36:46	BlaXpirit	ok but could you not make hundreds of redirect rules?
18:36:57	Araq	that's a bizarre webcrawler
18:36:59	dom96	perhaps we should just let Google figure it out.
18:37:02	def-	BlaXpirit: that sounds hacky, to try redirecting everything
18:37:42	BlaXpirit	so hundreds of redirect rules is the proper way now
18:38:51	*	gyeates quit (Ping timeout: 276 seconds)
18:39:45	*	posix4e left #nim (#nim)
18:39:45	*	aboisvert joined #nim
18:39:49	*	posix4e joined #nim
18:39:54	flaviu	def-: Your articles are always nice to read :)
18:39:58	*	posix4e left #nim (#nim)
18:40:04	def-	flaviu: Great to hear, thanks!
18:41:00	antranigv	def-: yes, nice article (:
18:41:11	*	brson quit (Ping timeout: 256 seconds)
18:47:24	*	havenwood joined #nim
18:54:50	gokr	filwit: Hehe, full war on hackernews ;)
18:56:43	filwit	gokr: haha
18:57:19	filwit	gokr: i wasn't trying to start a ton of discussions, but now I've spent a couple hours just responding to everyone..
18:57:22	gokr	Funny that... every article about Nim ends up discussing nil and Rust's memory model.
18:57:29	gokr	Yeah
18:58:44	filwit	well it's the main significant practical difference between these two up-n-coming systems languages.. makes sense it's the hot topic
19:01:54	*	solidsnack quit (Quit: My Mac has gone to sleep. ZZZzzz…)
19:02:04	antranigv	I hate when people start compairing languages.
19:03:34	filwit	it's an unavoidable consequence of our pattern seeking minds... on the plus side, that same ability allows us to write these languages in the first place :)
19:04:47	Araq	bbl
19:05:48	*	Matthias247 joined #nim
19:06:22	OnO	okie macro implementation of expr to SQL works now like a charm, thanks for great "untyped" type: https://github.com/nanoant/nim-orm/blob/master/orm.nim
19:07:05	ldlework	nice
19:07:22	ldlework	OnO: working on an ORM?
19:08:00	OnO	yup, something like Ruby's but most work to be done on compile time
19:08:09	OnO	and lazy evaluation if possible
19:08:16	flaviu	http://goo.gl/forms/W9XDgby8a4 < quick 5 second poll about API syntax.
19:09:46	filwit	flaviu: can't decide between #2 and #5..
19:10:27	flaviu	filwit: I can't help you there, #2 and #5 are different for everyone :)
19:10:48	flaviu	I was hoping that would prevent bias from people choosing options closer to the top.
19:11:30	filwit	well i certainly don't like the other ones.. if that's any help.. maybe i'll write that in 'other'
19:11:48	*	gsingh93 joined #nim
19:11:58	flaviu	Sounds good to me.
19:12:02	BlaXpirit	the problem with this poll is people don't see any merit at all in a few of the options
19:12:18	BlaXpirit	but there is a reason for them to be like this. the "anycrlf" part is rare
19:12:34	BlaXpirit	and some of the options are just consistent with PCRE's syntax
19:12:53	BlaXpirit	(?i) can appear in the middle of the string
19:13:05	BlaXpirit	and (*ANYCRLF) is a syntax that is already supported by PCRE
19:13:46	BlaXpirit	and (?i) too
19:15:02	flaviu	filwit: My point was that I don't know which answers #2 and #5 are. Everyone has a different #2 and #5.
19:15:40	*	BitPuffin|osx joined #nim
19:15:41	filwit	oh crap, lol.. then you really wont understand what I wrote in the 'other' box...
19:15:48	flaviu	yep.
19:16:11	filwit	re(r"..", "..") and re"..".opts".."
19:16:39	flaviu	I've fixed it.
19:16:44	filwit	^ those where my choices.. other's where really obscure and left me wondering what the actually expr was doing
19:17:22	flaviu	Well, responses have tapered off a bit. https://docs.google.com/spreadsheets/d/1y_Co5OV3MIkH9l4MFe-F9-ouB9gIhP8aixGz3yy8pqk/edit?usp=sharing if anyone is interested.
19:17:57	flaviu	If you're going to answer the poll, I'd appreciate it if you don't click that link until afterwards to avoid bias.
19:18:16	*	gsingh93 quit (Ping timeout: 240 seconds)
19:20:37	*	iamd3vil joined #nim
19:21:41	stefantalpalaru	What's the best way to insert NimNodes before the top level node passed to a statement macro? Would it be a problem if I create a new StmtList, make the previous top node a child and return that?
19:22:01	BlaXpirit	stefantalpalaru, that last part is good
19:22:06	stefantalpalaru	ok
19:22:11	BlaXpirit	i'm pretty sure
19:22:55	OnO	is it possible to inject some module into other module? or pass a parameter to some module?
19:23:00	*	mischanix joined #nim
19:23:09	BlaXpirit	you wish
19:23:48	*	mischanix quit (Client Quit)
19:23:52	BlaXpirit	there is no way to pass settings to modules, the only way is to pass compiler flags
19:24:03	BlaXpirit	and inject... isn't that just include?
19:24:40	BlaXpirit	oh you mean it in the context of configuring it
19:28:41	*	mischanix joined #nim
19:44:41	*	gsingh93 joined #nim
19:44:49	*	iamd3vil quit (Ping timeout: 264 seconds)
19:49:09	flaviu	Can someone with access to the website add a symlink from /latest/ to whatever the latest release is?
19:50:40	*	brson joined #nim
19:51:54	flaviu	Also, what exception do I throw for a NPE?
19:52:32	BlaXpirit	for a what
19:53:34	flaviu	Well, I need to throw an exception if I get a null parameter.
19:57:41	*	gsingh93 quit (Ping timeout: 256 seconds)
19:57:53	*	mpthrapp quit (Remote host closed the connection)
20:00:47	*	posix4e joined #nim
20:01:10	posix4e	Araq: so nim's gc is conservative-ish?
20:02:03	stefantalpalaru	why am I getting a linking error from the tables module? nimcache/tables.o: In function `HEX24_77242': tables.c:(.text+0x592): undefined reference to `HEX24_51813'
20:02:12	*	mpthrapp joined #nim
20:03:08	def-	stefantalpalaru: strange, works here. tried removing nimcache?
20:03:14	BlaXpirit	that might just be a compiler bug
20:03:36	stefantalpalaru	yes, I removed nimcache
20:08:59	flaviu	AccessViolationError is basically a NPE, right? ;)
20:10:19	flaviu	Hmm, InternalError would also be nice.
20:12:59	*	BitPuffin|osx quit (Ping timeout: 256 seconds)
20:13:16	*	repax_ joined #nim
20:15:04	*	repax quit (Ping timeout: 246 seconds)
20:16:08	dalarmmst	Is it not possible to compare uints?
20:16:19	def-	dalarmmst: import unsigned
20:16:49	dalarmmst	Thanks
20:16:58	dalarmmst	Where do I find this info?
20:17:06	*	mpthrapp quit (Remote host closed the connection)
20:17:07	dalarmmst	I'm reading the nim tutorials
20:17:18	dalarmmst	But I didn't see it there. Maybe I missed it
20:17:44	stefantalpalaru	I found the cause of the error. I was printing an empty table at compile time with unstringifiable values. This worked, printing {:} but apparently generated that undefined function.
20:18:49	def-	dalarmmst: Looks like it's missing in the tutorial. I'm suggesting to make unsigned operations available by default because everyone runs into this problem at some point
20:19:02	dalarmmst	I agree
20:19:45	OnO	doing newNimNode(nnkCall) with .add(newIdentNode("fetch")), if fetch is a proc then it works, if fetch is an interator I get Error: undeclared identifier: 'fetch'
20:19:50	stefantalpalaru	and deprive Nim users from finding out about Araq's strong dislike of unsigned integers? :-)
20:19:59	OnO	how to return an iterator from macro?
20:20:33	*	mpthrapp joined #nim
20:22:06	*	Jehan_ quit (Quit: Leaving)
20:22:48	BlaXpirit	OnO, yeah, calling an iterator is indeed invalid
20:23:25	BlaXpirit	a normal iterator can only be used in the context of a `for`
20:23:38	BlaXpirit	not sure if that means it's impossible to return from a macro
20:23:39	OnO	okie so: for u in macro(....): echo u <- how do I return iterator in macro?
20:24:09	OnO	trying to do a call in macro, but it does not work
20:24:16	*	CARAM__ quit (Ping timeout: 265 seconds)
20:24:30	OnO	I guess iterator has different AST, but what ? :)
20:25:30	*	CARAM__ joined #nim
20:25:32	BlaXpirit	appears it doesn't have different ast
20:27:33	OnO	hmm... but it is not a nnkCall
20:29:40	BlaXpirit	yeah, it definitely doesn't work, and I don't see a way to make it work
20:29:48	BlaXpirit	maybe i can suggest some alternatives
20:30:45	*	vikaton quit (Read error: Connection reset by peer)
20:30:45	*	NhanH quit (Read error: Connection reset by peer)
20:30:47	*	CARAM__ quit (Read error: Connection reset by peer)
20:31:01	BlaXpirit	maybe you can settle for returning just the iterator without calling it in advance
20:31:16	BlaXpirit	maybe you can use a closure iterator, no need to shy away from them
20:32:22	flaviu	Hmm, how about some sort of InvalidUnicodeError?
20:32:55	*	Kingsquee joined #nim
20:33:28	*	NhanH joined #nim
20:33:38	*	CARAM__ joined #nim
20:34:00	flaviu	I'm not sure that belongs in nre, it might be better in the unicode module.
20:34:53	BlaXpirit	standard library never checks for invalid unicode
20:36:00	*	zahary joined #nim
20:36:29	*	vikaton joined #nim
20:36:47	*	Kingsquee quit (Client Quit)
20:36:48	flaviu	I know, but I don't want there to be two different unicode exceptions.
20:40:17	*	vikaton quit (Client Quit)
20:40:30	*	vikaton joined #nim
20:41:19	*	aboisvert quit (Quit: aboisvert)
20:42:10	*	aboisvert joined #nim
20:44:54	*	gsingh93 joined #nim
20:45:15	*	UberLambda quit (Quit: Leaving)
20:49:59	BlaXpirit	dom96, 0.11.2 tags to devel, master is not updated
20:50:46	Araq	flaviu: if you do if x.isNil: raiseNilException() you're doing it wrong
20:51:03	flaviu	Araq: I know, but PCRE does it for me.
20:51:26	Araq	ah ok
20:51:26	flaviu	"PCRE_ERROR_NULL" on man pcreapi.
20:51:55	Araq	does it do PCRE_ACCESS_VIOLATION too? ;-)
20:52:08	*	ddl_smurf joined #nim
20:52:09	Araq	PCRE_INVALID_POINTER
20:52:39	BlaXpirit	google now sends to 404. just great.
20:52:46	flaviu	Well, PCRE_ERROR_BADMAGIC and PCRE_ERROR_UNKNOWN_OPCODE are other indications that a pointer is bad.
20:53:41	BlaXpirit	please someone make redirects from /system.html etc
20:55:27	BlaXpirit	even for me here http://blaxpirit.github.io/nim-random/random.html the links broke
20:55:37	BlaXpirit	(imports section links)
21:00:12	*	mpthrapp quit (Remote host closed the connection)
21:01:13	dalarmmst	How do I make a type T have it's data inlined in another type X? in C++ this would be struct X { T t; }, instead of struct X { T* t; }
21:02:38	def-	dalarmmst: just don't make it a ref type, then that happens in Nim
21:03:25	dalarmmst	Nice, thanks!
21:04:29	*	solidsnack joined #nim
21:05:09	*	OnO quit (Quit: My iMac has gone to sleep. ZZZzzz…)
21:09:07	*	solidsnack quit (Ping timeout: 255 seconds)
21:13:50	vikaton	does reem come here often?
21:15:21	*	jsudlow quit (Remote host closed the connection)
21:15:23	Araq	vikaton: why is that an important question?
21:15:45	vikaton	Araq: it isnt, I havent seen him and wanted to talk to him
21:18:19	*	OnO joined #nim
21:28:43	*	banister is now known as banisterfiend
21:32:35	*	bjz quit (Quit: My MacBook Pro has gone to sleep. ZZZzzz…)
21:35:45	OnO	FYI: https://github.com/Araq/Nim/issues/2656 iterators with vaargs arguments and {.closure.} cause invalid code :(
21:38:22	BlaXpirit	damn
21:40:19	*	biz joined #nim
21:40:26	biz	hello
21:40:42	*	biz is now known as Guest91971
21:40:57	BlaXpirit	hello
21:41:07	*	Guest91971 is now known as bizn
21:41:15	bizn	biz was registered :(
21:41:17	bizn	what's up
21:41:48	bizn	well you see i came to irc to ask for a bit of help
21:41:56	bizn	i need help communicating with skype api Skype4Com
21:42:01	bizn	zero idea where to start
21:42:14	BlaXpirit	uh i think they're gonna remove it
21:42:29	bizn	so useless to mess with?
21:42:36	bizn	will be outdated in a month~ i assume
21:42:37	bizn	?
21:43:13	*	zahary quit (Quit: Leaving.)
21:43:57	BlaXpirit	hmm skype api was supposed to be killed 1.5 years ago...
21:44:03	bizn	lmfao
21:44:14	bizn	what's with all the current skype bots?
21:44:18	bizn	manual possibly?
21:44:22	*	zahary joined #nim
21:44:26	bizn	clownfish still works and that's one of the main ones
21:51:28	*	Kingsquee joined #nim
21:53:16	*	vendethiel quit (Ping timeout: 240 seconds)
21:54:50	*	milosn quit (Ping timeout: 256 seconds)
21:58:05	*	bizn quit (Quit: Page closed)
21:59:14	*	Senketsu quit (Quit: Leaving)
21:59:57	stefantalpalaru	is there any way to get "ref T" from an instance y of T without doing "var x: ref T; new(x); x[] = y" ?
22:01:36	flaviu	stefantalpalaru: Of course! Pull that code into a procedure!
22:01:37	flaviu	https://gist.github.com/1ae1d7deac4c57d0a84c
22:02:26	stefantalpalaru	I meant without the copying step
22:03:48	stefantalpalaru	something simple like getting a ptr with addr. don't we have something like that for refs?
22:04:11	*	BlaXpirit is now known as BlaXpirit_
22:04:14	*	BlaXpirit_ is now known as BlaXpirit
22:05:25	def-	Ha, with this I can get the binary 1 byte smaller: https://github.com/Araq/Nim/pull/2657
22:06:16	*	BlaXpirit quit (Remote host closed the connection)
22:06:34	*	BlaXpirit joined #nim
22:06:47	*	BlaXpirit quit (Client Quit)
22:07:01	*	wtw quit (Ping timeout: 264 seconds)
22:10:27	flaviu	Is the noReturn pragma really necessary?
22:11:43	renesac	w/o it your binary may be 1 bit bigger!
22:12:02	*	yglukhov_ quit (Quit: Be back later ...)
22:12:07	renesac	well, I guess there is some use for it, as the three main compilers have some sort of support for it
22:12:21	renesac	*three main C compilers
22:13:23	flaviu	If people want it, then they should use C.
22:13:47	flaviu	Because if they get to the level where that sort of stuff is important, then they are effectively writing C in Nim anyway.
22:14:21	renesac	don't araq always says that you can use Nim as a better C if you want?
22:14:22	renesac	:D
22:15:32	def-	flaviu: noreturn allows optimizations, even for totally normal code that just never returns. i don't see why we should have a noReturn pragma, but not pass that information to the C compiler
22:15:35	renesac	http://stackoverflow.com/questions/10538291/what-is-the-point-of-the-noreturn-attribute
22:18:03	flaviu	"need not save and restore any volatile state around the call" - write it in assembly if a function call is too much overhead for you.
22:18:04	flaviu	"dead-code eliminate any code that would otherwise follow the call" - Too minor of an advantage.
22:18:17	flaviu	I'm sure that the compiler could infer that it'd never return from the code.
22:19:07	renesac	not if it don't have the full source of the called code, and even then I'm not sure if it can be infered
22:19:56	def-	of course it can't infer that. Why would the compiler know whether a syscall kills your program (for example)
22:20:01	renesac	it is a pragma, an obscure pragma that most people don't need to know
22:20:39	gokr	"I'm sure that the compiler could infer that it'd never return from the code." - doesn't that sound like the stopping problem? :)
22:21:13	gokr	Or "Halting problem" I mean.
22:21:18	flaviu	gokr: Sure, but methods tend to be short.
22:21:33	flaviu	And I'm assuming there is no non-local control flow like exceptions.
22:21:33	*	BitPuffin|osx joined #nim
22:23:04	flaviu	https://codesearch.debian.net/results/__attribute__((noreturn))/
22:23:11	flaviu	I'm not convinced noreturn is useful.
22:24:14	flaviu	It seems that the only programs that have any mention of noreturn are development tooling, and that's not because they actually use noreturn.
22:24:26	*	yglukhov_ joined #nim
22:24:29	*	brson quit (Quit: leaving)
22:26:09	renesac	the cost of keeping it in nim seems very low, even though the benefit may be low too
22:28:04	renesac	you didn't search right
22:28:17	renesac	https://codesearch.debian.net/results/noreturn/page_0
22:28:50	*	yglukhov_ quit (Ping timeout: 256 seconds)
22:28:58	renesac	2088 pages
22:29:32	flaviu	hmm, looks like everyone defines a macro for their noreturns.
22:29:48	flaviu	fair enough then.
22:30:12	def-	flaviu: yeah, because it's not platform independent. same thing as I did in the PR. In C11 noreturn is finally in the standard
22:37:20	*	Jesin quit (Quit: Leaving)
22:44:19	stefantalpalaru	how do I convince Nim do type convert every member of a tuple literal in an assignment? I get "Error: conversion from (int, int, nil) to f8_args_tuple is invalid" for a tuple defined as "type f8_args_tuple = tuple[a: int32, b: int64, c: pointer]". If I convert individually each of the members, it works.
22:46:53	*	Jesin joined #nim
22:51:14	*	Matthias247 quit (Read error: Connection reset by peer)
22:54:57	*	gsingh93 quit (Ping timeout: 240 seconds)
22:59:06	*	solidsnack joined #nim
22:59:43	filwit	stefantalpalaru: 'int' in Nim is a pointer size, and it does not implicitly convert to either int32 or int64 IIRC. So you'd need to pass (int32, int64, nil). So somethinglike (0'i32, 0'i64, nil)
23:00:43	stefantalpalaru	as the error says, I'm passing (int, int, nil) into a (int32, int64, pointer)
23:01:19	stefantalpalaru	the compiler fails to automatically convert the first int to int32 and the second to int64
23:01:20	filwit	so (int32(a), int64(b), c)
23:02:08	stefantalpalaru	I know the workaround, but I'm working on a macro and it would make my life a lot easier if it would just convert each tuple member
23:03:15	filwit	that isn't really a 'workaround'.. you have to explicitly convert an int -> int32/64.. however you can make this happen implicitly with a converter
23:05:47	*	Jesin quit (Quit: Leaving)
23:05:48	filwit	converter toIntIntPtrTuple(t:tuple[a, b:int, c:pointer]): tuple[a:int32, b:int64, c:pointer] = (int32(a), int64(b), c)
23:07:26	flaviu	Isn't this basically a problem of contra/covariance?
23:08:29	flaviu	Hmm, not really I guess.
23:09:37	*	gsingh93 joined #nim
23:10:46	*	aboisvert quit (Quit: aboisvert)
23:11:22	*	Trustable quit (Remote host closed the connection)
23:11:38	stefantalpalaru	I'm dealing with generic tuples
23:11:47	*	OnO quit (Quit: My iMac has gone to sleep. ZZZzzz…)
23:11:56	filwit	you can define a generic converter
23:12:23	stefantalpalaru	for any tuple with any number of members which have any combination of types?
23:12:35	*	aboisvert joined #nim
23:12:58	filwit	i doubt it.. but i also doubt you really want to do that
23:13:30	filwit	but you could make a macro which did that
23:13:33	stefantalpalaru	I do. I keep proc parameters in hose tuples.
23:14:35	stefantalpalaru	I'm taking a generic proc and transforming it into a proc that takes a tuple ref cast into a pointer and in that tuple ref are the original parameters
23:16:11	stefantalpalaru	can I look inside a tuple and see its member types from a macro, given only the tuple ident?
23:16:26	*	asdf joined #nim
23:16:36	filwit	sounds like you want a macro or template. Note, you can run through tuple vars by index, someTuple[0]
23:16:54	asdf	how can I get an ast of some code at runtime
23:18:47	filwit	asdf: const x = getAst someTemplateCall()
23:18:57	filwit	should work i believe
23:19:00	filwit	'getAst' is in 'macros'
23:20:21	filwit	stefantalpalaru: you should be able to do what you want with macros, or maybe even directly inside your proc at runtime via tuple len/indexing.. either way, I can't much help without seeing more context, and I need to work on stuff now, sorry.
23:22:00	asdf	I need to do this at runtime though, as in read in a string that represents nim code and get its AST
23:24:18	flaviu	asdf: That's a *little* more complicated ;)
23:25:32	asdf	flaviu: haha I know, it doesn't need to be easy
23:25:44	filwit	asdf: if you mean read arbitrary strings at runtime (from actual runtime vars), you basically looking at using the compiler as a dependency (which is actually possible).. but yeah, it's not so straight forward, and unless you're writing a REPL, probably not needed.
23:26:16	flaviu	well, if you want to parse nim code, then you have to link to the compiler like filwit says.
23:26:22	filwit	if you look at your problem though, you might find you just need to process string literals into Nim ast, which is much more straight forward.
23:26:25	flaviu	see compiler/parser, compiler/ast.
23:27:34	asdf	alright, in I go
23:29:24	k1i	any idea if/when a shared, GC'd heap will make its way into nim?
23:31:49	*	solidsnack quit (Quit: My Mac has gone to sleep. ZZZzzz…)
23:31:51	k1i	btw the nim manual is 404ing
23:31:51	flaviu	I don't think it'll happen, but there are assignment operators.
23:31:57	flaviu	http://nim-lang.org/0.11.0/manual.html#type-bound-operations-operator
23:32:04	k1i	ah, nvm, relocated
23:32:42	*	solidsnack joined #nim
23:33:40	*	elbow_jason quit (Quit: Leaving)
23:34:36	renesac	k1i: the boehm GC option collects shared memory
23:34:55	renesac	it seems
23:35:09	k1i	without boehm
23:36:43	renesac	maybe sometime post 1.0?
23:36:47	renesac	don't hold your breath
23:45:10	*	TEttinger joined #nim
23:45:27	ldlework	hi asdf !
23:45:34	*	Jesin joined #nim
23:45:41	ldlework	wait a minute.. you're not the asdf I know.
23:56:59	asdf	ldlework: sorry to disappoint
23:58:34	ldlework	asdf: the other asdf is happy to hear you exist
23:58:52	ldlework	as a real person, rather than a forgotten registered nick
23:59:25	*	solidsnack quit (Quit: My Mac has gone to sleep. ZZZzzz…)