<< 14-01-2015 >>

---

00:00:01	Araq	# "emit first or next parameter"
00:00:31	Araq	so we get things like #.foo<>()
00:03:55	flaviu	Ok, looks good.
00:04:43	Araq	well the '.' is special too as Nim is smart and sometimes uses '->' instead of '.'
00:07:55	*	sillesta quit (Ping timeout: 256 seconds)
00:08:04	*	Boscop_ joined #nim
00:08:41	keyle	Araq: where is the source of what you're referring to? Would love to take a look for learning purposes
00:09:02	Araq	keyle: just wait for my commit
00:09:20	*	flan3002 quit (Ping timeout: 264 seconds)
00:09:20	*	flyx quit (Ping timeout: 264 seconds)
00:09:21	*	wan quit (Ping timeout: 264 seconds)
00:09:22	*	wtw quit (Ping timeout: 264 seconds)
00:09:22	*	ldlework quit (Ping timeout: 264 seconds)
00:09:22	*	flyx joined #nim
00:10:20	*	flan3002 joined #nim
00:10:53	*	Boscop quit (Ping timeout: 240 seconds)
00:12:20	*	ldlework joined #nim
00:12:23	*	wan joined #nim
00:12:51	*	wtw joined #nim
00:16:47	*	brson quit (Quit: leaving)
00:17:07	keyle	Can someone explain in layman's terms why gcc compiles to an older nim.exe than then requires cooking with Koch to be up to date?
00:19:46	*	endou______ joined #nim
00:20:34	*	Boscop__ joined #nim
00:21:43	*	Trustable quit (Remote host closed the connection)
00:24:02	*	Boscop_ quit (Ping timeout: 245 seconds)
00:26:43	dts|pokeball	just out of curiosity, how come nim-lang.org doesnt have a favicon for the site?
00:28:57	keyle	it does <link rel="shortcut icon" href="assets/images/favicon.ico">
00:29:28	*	Boscop_ joined #nim
00:33:11	*	Boscop__ quit (Ping timeout: 252 seconds)
00:35:12	*	CARAM__ joined #nim
00:35:50	dts|pokeball	weird... chrome doesnt show one
00:36:10	flaviu	ekarlso-: ping
00:36:25	flaviu	I have the header working, but for some reason I can't get consistent results.
00:36:58	reactormonk	Is casting to int the correct way to do pointer arithmetics?
00:37:43	reactormonk	or are pointers unsigned...
00:40:05	keyle	dts|pokeball: does here, maybe try clean cache
00:40:13	keyle	or try incognito mode for testing
00:40:18	dts|pokeball	sure
00:40:27	Araq	reactormonk: nobody knows but I usually use -% and it works
00:43:56	Mat4	ciao
00:44:00	*	Mat4 left #nim (#nim)
00:48:43	*	Boscop__ joined #nim
00:50:15	flaviu	Does https://gist.github.com/863a6c7947e51d1767ec return inconsistent results for anyone else?
00:52:46	*	Boscop_ quit (Ping timeout: 255 seconds)
00:56:32	*	Guest81214 quit (Ping timeout: 245 seconds)
00:59:10	keyle	Araq: what's broken about switch?
00:59:34	Araq	keyle: the syntax.
01:00:00	Araq	'break' everywhere or else you get nice fallthrough
01:04:51	keyle	ah I see
01:12:58	*	z1y joined #nim
01:19:19	*	VinceAddons quit (Read error: Connection reset by peer)
01:24:05	reactormonk	How do I convert an unsigned integer to a signed one?
01:24:22	Araq	int(x)
01:24:53	reactormonk	mail is a cstring - is the operation let mailptr = cast[uint](ptr mail) correct?
01:25:19	Araq	that's not even correct syntax
01:25:29	reactormonk	duh
01:25:32	Araq	remove the 'ptr' and then it might work
01:25:35	reactormonk	Too much SCala I'd say.
01:26:22	reactormonk	Seems to compile in the first pass, afterwards I've got some template problems
01:30:11	reactormonk	Fixed. \o/
01:30:39	*	Ether-plays-4fun quit (Ping timeout: 264 seconds)
01:31:07	*	Ether-plays-4fun joined #nim
01:32:56	*	z1y left #nim (#nim)
01:33:48	*	z1y joined #nim
01:33:51	*	saml_ joined #nim
01:37:09	*	gmpreussner_ joined #nim
01:41:02	*	z1y quit (Ping timeout: 244 seconds)
01:59:42	reactormonk	Araq, is {.compile.} a toplevel pragma or can I use it in the scope of a proc?
01:59:54	Araq	top level
02:01:07	reactormonk	Can I combine {.exportc.} and {.closure.}?
02:01:32	reactormonk	I'm trying to figure out how to pass the seq that the functions exported to C modify
02:02:12	Araq	well nobody hinders you from using .exportc with .closure but it makes no sense I think
02:02:58	reactormonk	I could always use a global var, but the thought disgusts me
02:04:12	reactormonk	I might need to write some more extended C code.
02:04:34	Araq	you need to write *less* C code
02:04:41	Araq	you should write Nim instead
02:04:45	reactormonk	ragel produces C code.
02:05:02	Araq	fix ragel
02:05:27	gmpreussner_	:)
02:05:30	gmpreussner_	howdy all
02:06:05	ldlework	sup
02:08:29	*	z1y joined #nim
02:12:51	reactormonk	Araq, ragel is GPL. The output is not licensed at all. Might that produce problems?
02:12:59	Araq	no.
02:13:54	*	z1y quit (Remote host closed the connection)
02:14:58	*	z1y joined #nim
02:17:29	reactormonk	Araq, the ruby codegen is 4.3k LoC. No thanks.
02:18:23	Araq	but you don't need to write the backend in C with strcat. you can use Nim instead. :-)
02:18:31	*	no_name joined #nim
02:18:51	EXetoC	yeah why C
02:18:55	reactormonk	It's cpp
02:19:27	reactormonk	let's see how well c2nim works on it
02:19:50	*	JinShil quit (Quit: Konversation terminated!)
02:20:31	reactormonk	not too well.
02:23:32	Araq	good night
02:24:55	*	jefus_ joined #nim
02:26:22	*	jefus quit (Ping timeout: 255 seconds)
02:29:04	*	darkf joined #nim
02:32:18	*	reactormonk quit (Quit: Leaving)
02:41:09	*	JinShil joined #nim
02:41:29	*	Demon_Fox quit (Ping timeout: 256 seconds)
02:41:54	*	Demon_Fox joined #nim
02:46:44	*	Amrykid quit (Changing host)
02:46:44	*	Amrykid joined #nim
02:48:15	*	iivvoo quit (Ping timeout: 265 seconds)
02:48:16	*	EastByte_ quit (Ping timeout: 265 seconds)
02:48:32	*	EastByte_ joined #nim
02:48:45	*	dymk quit (Ping timeout: 265 seconds)
02:50:22	*	dymk joined #nim
02:53:48	*	iivvoo joined #nim
02:56:21	*	hguux_ quit (Read error: Connection reset by peer)
02:58:00	*	hguux_ joined #nim
03:02:19	*	BitPuffin quit (Ping timeout: 245 seconds)
03:13:40	*	Sergio965 quit (Quit: My MacBook Pro has gone to sleep. ZZZzzz…)
03:24:23	*	TylerE quit (Ping timeout: 244 seconds)
03:26:19	*	TylerE joined #nim
03:33:09	*	iivvoo quit (Ping timeout: 244 seconds)
03:33:30	dom96	https://news.ycombinator.com/item?id=8883791
03:34:43	*	nande quit (Remote host closed the connection)
03:36:35	flaviu	dom96: It'd go faster if he pulled the try-catch out of the at hot loop.
03:36:50	*	Boscop__ is now known as Boscop
03:37:17	*	endou______ quit (Ping timeout: 244 seconds)
03:37:39	*	Boscop quit (Changing host)
03:37:40	*	Boscop joined #nim
03:38:24	dom96	It seems he built the Nim version with --opt:speed
03:38:26	dom96	and no -d:release
03:38:32	onionhammer	dom96 spammers
03:39:02	dom96	onionhammer: Yeah, I need to do something about that it seems.
03:39:15	onionhammer	:\
03:39:21	onionhammer	that captcha isnt working I gues
03:39:51	*	endou______ joined #nim
03:40:08	dom96	onionhammer: it is, that user is doing it manually.
03:40:29	onionhammer	seriously...?
03:40:51	*	iivvoo joined #nim
03:41:46	flaviu	lol, he can't be bothered to read the readme.
03:42:05	flaviu	[url=http://spamspam.com]blah blah[/url]
03:42:33	dom96	onionhammer: I think so.
03:43:02	onionhammer	i bet it's a bot..
03:45:02	*	z1y1 joined #nim
03:46:23	keyle	if it's a bot you'd get hundreds of spam
03:46:53	*	z1y quit (Ping timeout: 240 seconds)
03:46:58	flaviu	Why not both? Maybe it's a human writing a bot.
03:49:10	*	dymk quit (Ping timeout: 244 seconds)
03:49:28	*	dymk joined #nim
03:49:41	dom96	I deleted that user.
03:49:50	dom96	And removed his posts.
03:50:23	keyle	good point, often they run tests over a few days to see how long the op gets them removed
03:50:28	keyle	then they unleash the scripts
03:51:36	flaviu	dom96: Have you banned the IP?
03:51:48	dom96	no
03:52:20	flaviu	You should probably also ban tor and public proxies from posting too.
03:53:54	*	z1y1 quit (Remote host closed the connection)
03:58:41	*	z1y joined #nim
04:03:11	keyle	Nim has crazy speed, not sure why that guy recommends picking Rust for performance
04:03:23	keyle	Besides, developer performance ways more than anything...
04:04:51	onionhammer	who?
04:06:39	*	saml_ quit (Ping timeout: 264 seconds)
04:06:51	*	nande joined #nim
04:08:17	*	iivvoo quit (Ping timeout: 244 seconds)
04:10:22	*	iivvoo joined #nim
04:11:19	*	brson joined #nim
04:20:41	*	Boscop_ joined #nim
04:22:12	dom96	keyle: Likely because he compiled with --opt:speed not -d:release
04:24:33	*	Boscop quit (Ping timeout: 252 seconds)
04:30:30	*	iivvoo quit (Ping timeout: 244 seconds)
04:32:15	keyle	onionhammer: was referring to https://news.ycombinator.com/item?id=8883791
04:33:22	*	iivvoo joined #nim
04:37:06	*	ARCADIVS joined #nim
04:40:58	onionhammer	ah
04:41:25	onionhammer	has anyone done the bench w/ -d:release?
04:49:35	keyle	no but it'd be sensible to reproduce his claims
04:57:53	*	iivvoo quit (Ping timeout: 244 seconds)
05:01:23	*	iivvoo joined #nim
05:03:18	*	keyle quit (Quit: leaving)
05:12:51	*	brson quit (Quit: leaving)
05:15:01	*	z1y quit (Quit: Leaving.)
05:37:38	ldlework	plz plz plz plz plz plz
05:37:45	ldlework	plllllllz, forward type declarations, plz
05:37:50	ldlework	kthx.
05:43:51	*	gmpreussner__ joined #nim
05:44:04	*	gmpreussner_ quit (Ping timeout: 244 seconds)
05:47:22	*	BlaXpirit joined #nim
05:48:22	*	z1y joined #nim
05:48:32	*	z1y quit (Remote host closed the connection)
05:49:27	*	z1y joined #nim
05:56:36	ldlework	I wonder if anyone can improve the times here: http://arthurtw.github.io/2015/01/12/quick-comparison-nim-vs-rust.html
05:59:16	*	BitPuffin joined #nim
06:03:43	*	BitPuffin quit (Ping timeout: 255 seconds)
06:10:19	fowl	ldlework, yes
06:12:02	fowl	write it imperatively
06:12:23	ldlework	not sure what that means to you
06:17:50	gmpreussner__	Rust syntax looks like crap. try to improve that :)
06:18:15	ldlework	sure, but I am interested in the performance
06:18:27	gmpreussner__	they're probably more advanced wrt optimization
06:18:27	ldlework	I've seen benchmarks beating rust, but they were simplier benchmarks
06:18:46	fowl	well theres superfluous allocation here
06:18:49	gmpreussner__	i wouldn't worry about it too much. if you're really interested, you should take a look at where exactly the time is spent.
06:21:02	gmpreussner__	it might help to look at the C code too
06:21:25	gmpreussner__	in the game of life test, unrolling some of the inner loops may work wonders
06:22:05	ldlework	sure but that's no good, who wants to unroll their own loops
06:22:20	fowl	anywhere you see toSeq can be optimized away
06:22:36	ldlework	I only count that once
06:23:02	fowl	theres a line in the life example that creates 2 seqs to get a longest-line count
06:23:18	ldlework	fowl: show him up!
06:23:36	ldlework	there's always the guy who comes by and optimizes the original benchmarks
06:23:41	Triplefox	in that kind of "match over a 2d grid" situation i like to assume a 1d array so that iteration can be all in one go...however that's a basic data design thing
06:24:06	ldlework	Sure but the Rust does the same thing
06:24:15	gmpreussner__	note that the next() loop is pretty close to Rust. most of the extra cost goes into print()
06:24:16	ldlework	I think fowl is identifying the right kind of changes
06:24:25	ldlework	the kind that come with unfamiliarity with the language
06:24:29	ldlework	not algorithm optimizations
06:24:48	ldlework	both implementations seem to do roughly the same thing
06:24:49	fowl	yes this is written by someone who loves one-liners
06:24:50	Triplefox	well, i'm not that familiar with nim :p
06:24:54	ldlework	though I can't read Rust very well
06:25:02	ldlework	Triplefox: :)
06:25:05	Triplefox	i didn't even know the syntax "for y in 0.. <iy:" existed
06:25:12	gmpreussner__	i didn't either
06:25:16	BlaXpirit	i didnt
06:25:27	gmpreussner__	i figured it was a formatting error on his blog
06:25:27	fowl	`<`(n: int) is just n-1
06:25:38	gmpreussner__	:)
06:25:46	BlaXpirit	waaaat
06:25:51	BlaXpirit	hy
06:25:52	BlaXpirit	why
06:25:56	Triplefox	does it map to high()
06:26:09	fowl	for the iterator 0 .. <10 goes from 0 to 9
06:26:13	Triplefox	cause i was using high()
06:26:45	*	iivvoo quit (Ping timeout: 244 seconds)
06:26:58	ldlework	Triplefox: wait did you write that article?
06:27:21	Triplefox	no
06:27:23	ldlework	oh
06:27:26	*	ldlework rubs eyes
06:27:39	ldlework	Also, I wonder if we would ever consider pattern matching
06:27:57	ldlework	it is admittedly nice
06:28:02	BlaXpirit	meh
06:28:14	BlaXpirit	it's only nice when you have a lot of things like Option and whatnot
06:28:20	ldlework	we already have do syntax
06:28:36	ldlework	mixing that with pattern matching results is a very readable structure is all
06:28:36	BlaXpirit	in nim it's needed very rarely
06:28:54	ldlework	I don't see why Nim would affect how often you want to match
06:29:08	BlaXpirit	well, exactly, it doesn't
06:29:19	BlaXpirit	Rust, however, almost enforces it
06:29:21	ldlework	right, so its nice to have pattern matching + do syntax
06:29:34	ldlework	because it represents the underlying semantics, visually, quite nicely
06:29:52	ldlework	one of the first things that made me go "oooh" to Rust's documentation when I first read them
06:31:02	ldlework	BlaXpirit: rust enforces it, only in so much that other languages force you to use switch
06:31:20	BlaXpirit	...
06:31:24	ldlework	BlaXpirit: except that with pattern matching you can represent certain match semantics much more sucinctly
06:31:52	fowl	dont you have to pattern match to extract information from enumerations
06:32:37	*	keyle joined #nim
06:33:11	fowl	how could you get the value held by Option<T> without it?
06:33:36	Triplefox	hmm oh i get it now, i can use high() on a seq or array, but i can use < for an integer variable
06:34:01	BlaXpirit	Triplefox, well, you can use < on high as well :p but it's kinda pointless
06:34:18	BlaXpirit	<len is a bit more useful
06:34:20	ldlework	fowl: well that doesn't seem bad really
06:34:22	fowl	high is just len-1
06:34:26	*	iivvoo joined #nim
06:34:28	fowl	so you could use < on len()
06:34:33	ldlework	I want to add a patch for first and last
06:34:35	BlaXpirit	well not really... arrays...
06:35:07	fowl	nobody uses that feature
06:35:13	BlaXpirit	low..high is a nice idiom, because 0..<len is not always reliable
06:36:15	*	BlaXpirit wonders if access to seq.low is optimized to a constant
06:36:24	BlaXpirit	probably is, just need to look at C code really
06:38:08	BlaXpirit	aand it is
06:39:20	Triplefox	i appreciate having options for dealing with the off-by-one iteration problem
06:39:40	BlaXpirit	which reminds me
06:40:33	BlaXpirit	<s.len compiles to subInt(s_89021->Sup.len, 1) :(
06:41:05	ldlework	how could it compile a sequence's length to a const?
06:41:09	BlaXpirit	what did i expect though
06:41:10	BlaXpirit	yeah
06:41:12	BlaXpirit	lol
06:41:18	ldlework	oh okay, you were confused
06:41:19	*	gmpreussner___ joined #nim
06:41:24	ldlework	I thought *I* was super confused
06:41:25	*	gmpreussner__ quit (Ping timeout: 244 seconds)
06:41:29	BlaXpirit	<5 is 4 yay
06:41:51	Triplefox	it should be able to optimize an array that isn't 0..0
06:41:58	ldlework	yeah
06:42:14	BlaXpirit	well theoretically it could use <len and actually transform the loop to use < instead of <=
06:42:15	BlaXpirit	but yeah
06:44:18	*	skyfex_ joined #nim
06:44:22	ekarlso-	any of you doign webdev with nim ?
06:44:55	ldlework	ekarlso-: seems light a nightmare right now
06:45:20	ekarlso-	ldlework: howso ? :p
06:45:24	BlaXpirit	i really think it's never going to be anywhere near the best tool for the task
06:45:42	BlaXpirit	I'm totally happy with Python in many applications, and I'm gonna keep it that way
06:45:50	ldlework	ekarlso-: comapred to the nice ecoystem in Python?
06:46:03	ekarlso-	ldlework: by all means :)
06:46:08	ekarlso-	yeah, I use mostly python too :|
06:46:16	ekarlso-	or dropwizard for that matteR*java*
06:46:18	ekarlso-	:D
06:46:31	ldlework	I mean I probably wont even consider it until there is something like jinja2
06:46:33	fowl	i decided a long time ago that i dont do web work. now i laugh when my friends complain about working with rails and javascript. suckers.
06:46:57	ekarlso-	ldlework: though the whole packages program is written in nim
06:47:00	ekarlso-	about 600 ish lines atm
06:47:02	ldlework	yeah I'm glad I've shifted away from it too
06:47:36	ldlework	ekarlso-: I think writing api's in Nim might eventually be workable
06:47:54	ldlework	but like actual web-applications? Its def not there right now.
06:48:11	Triplefox	for single components that have to scale vertically nim might be an easy sell to python coders...for the whole stack, yeah, there's other things to worry about
06:48:35	ekarlso-	some people got me to code it in nim :p
06:48:36	keyle	if jester works as advertise, nim should be a fine web server
06:48:36	ldlework	I mean, I'm a Python professional, I'm constantly resisting the urge to reach for Nim in the workplace
06:48:38	ldlework	So I dunno
06:48:44	ldlework	I mean, I basically do all my system scripting in Nim now
06:48:51	ekarlso-	ldlework: +1, I use python for openstack ..
06:49:13	ldlework	If Nim had a Fabric equivalent, it'd be curtains for Python on that front
06:49:59	keyle	as a python professional, can't you just write your python bottlenecks in nim and keep writing python?
06:50:10	keyle	so you get stable + fast
06:50:55	Triplefox	nim can also get a more straightforward story for deploying than pretty much any scripting runtime...make static binary, ship it
06:51:43	ekarlso-	anyone have suggestions on how I can get JWT into nim ?
06:52:16	keyle	java web toolkit ?
06:52:32	ekarlso-	json web token
06:53:28	*	AMorpork is now known as ZzZMorpork
06:58:06	ekarlso-	would be nice to have for tokens :|
07:01:59	*	VinceAddons joined #nim
07:07:21	*	atomic-hedgehog quit (Ping timeout: 252 seconds)
07:08:26	ekarlso-	what's the equiv of char* key = "012345678"; in nim ?
07:09:08	*	endou______ quit (Read error: Connection reset by peer)
07:10:12	Triplefox	uh, look at cstring. i forget if you need to introduce a pointer with addr() or not for that
07:10:54	ldlework	keyle: the thing is, I like /writing/ Nim more than I do Python (theoretically). Having the compiler tell me about errors before I actually try it out, is really really nice.
07:10:57	*	endou______ joined #nim
07:11:19	ldlework	keyle: I'm only slightly less productive in Nim than Python and that's jsut experience and sharp corners in a growing Nim.
07:11:27	keyle	I agree. I prefer static types
07:11:46	ldlework	performance is a secondary concern, at least for system scripting
07:11:50	keyle	coming from python though you have a huge advantage
07:11:59	ldlework	keyle: which advantage is that
07:12:00	Triplefox	static types are a little annoying because of the ... typing but otherwise a big help
07:12:04	keyle	syntax is very similar
07:12:12	ldlework	keyle: ah, a bit
07:12:26	keyle	unless you do funky stuff in nim
07:12:52	keyle	Triplefox: when you come from Java you can really appreciate the terse type defs :>
07:12:56	ldlework	its like pascal and python had a baby
07:13:03	keyle	haha
07:13:18	keyle	I picture this old man and that young girl
07:13:43	keyle	I read that python is adding type defs in the future, I'm guessing that's to help performance
07:14:04	keyle	not for the pleasure of typing more
07:15:16	Triplefox	i tend to get irritated now by functions and methods that aren't annotated at all
07:15:21	ldlework	its 'gradual typing'
07:15:41	keyle	Triplefox: can you just define the expectation by using prefix?
07:15:48	keyle	str_name
07:16:00	keyle	int_ae
07:16:02	ldlework	but what happens when you don't follow it on accident
07:16:04	keyle	int_age
07:16:13	Triplefox	it's unreliable, and now your variables are longer
07:16:14	ldlework	or you pass a prefixed var to the wrong function
07:16:19	keyle	Well you're using a dynamic language, that's to be expected
07:16:21	ldlework	the whole point is the compiler is helpful
07:16:23	ldlework	it isn't a burden
07:16:35	keyle	I agree
07:16:42	ldlework	the amount you have to type is nominal compared to running your program over and over with runtime debugging
07:16:52	keyle	but I can appreciated both dynamic and staticly typed languages for what they are
07:17:05	ldlework	I agree, I mean I came to Python because of its productivity
07:17:13	Triplefox	a big benefit of the dynamic structure is that you can write some very dense things, it's not worth throwing that away for faux-typechecks unless the codebase has gotten into serious trouble
07:17:15	ldlework	But I wonder if Nim is competitive...
07:17:19	keyle	in a large team though, dynamic languages are a pain in the arse and you end up with strict code conventions
07:17:20	ldlework	even being statically typed
07:17:35	ldlework	The most amazing thing about PYthon
07:17:39	ldlework	is the insane abstractions I can create with it
07:17:51	ldlework	using the magic of everything being an object
07:17:52	keyle	I did Groovy for a while, I can tell you it's nice, but nim is much nicer.
07:18:02	keyle	no fn@#$ JVM to start with
07:18:04	ldlework	but with Nim's metaprogramming and operator overloading, I can do a lot of the same stuff
07:18:09	ldlework	but with type safety :P
07:18:26	keyle	nim is like that tiny gun from star trek
07:18:39	*	gour joined #nim
07:18:49	keyle	doesn't look like it but it's very powerful
07:18:59	ldlework	dunno
07:19:08	ldlework	Nim looks like the average of mainstream languages
07:19:20	ldlework	Its featureset is a good collection of the feature's we've come to expect
07:19:33	ldlework	It gives us good implementations of each, and tries to make them work together well (theoretically)
07:19:44	ldlework	and then it adds some small improvements
07:20:27	ldlework	declarative type-classes, imperative ast macros, disjoint concurrency proofs, a very nice gc
07:20:54	ldlework	So when I look at Nim, and see everything I've come to expect from a fully featured general purpose language, it immediately communicates powerful to me.
07:21:41	ekarlso-	what type is "pointer" in nim vs ptr ?
07:21:57	ldlework	ekarlso-: ptr is a non-gc managed heap reference
07:22:06	ldlework	ref is a gc-managed heap reference
07:22:24	Triplefox	"pointer" is equivalent to C void* while "ptr ___" is typed
07:22:34	ldlework	oh okay
07:22:35	keyle	cool I didn't know
07:22:48	keyle	<- 7 days of nim
07:22:50	ldlework	no idea what a Nim void ptr is
07:22:56	ldlework	keyle: welcome
07:23:04	ldlework	keyle: what do you think you'll use Nim for?
07:23:20	keyle	small stuff
07:23:33	keyle	did this already https://github.com/keyle/hastebin/
07:23:40	keyle	for funsies & learning
07:23:45	Triplefox	ah, i like hastebin
07:23:49	Triplefox	if only because of the name
07:23:55	*	joecodes joined #nim
07:24:58	keyle	it was a great way to touch http post, json parsing, exec, parseopts...
07:25:17	keyle	And no JVM / CLR / crap to have pre-installed
07:25:47	Triplefox	i'm going to write a spritesheet packing tool soon and need to figure out my cli parser story soon
07:26:02	keyle	I work in a world of bloated monolithic software so, nim is like a fresh orange juice break
07:26:37	keyle	I'd love to do a text editor like joe/nano but the terminal support is ehm
07:26:45	keyle	not quite portable
07:26:56	keyle	ehm I shouldn't say that.
07:27:08	keyle	let's just say it'd take a lot more than I'd want to commit to
07:27:43	Triplefox	i think someone brought that up the other day...well, full control over terminals seems to be a pain no matter what
07:28:07	keyle	yeah Windows particularly
07:28:15	Triplefox	maybe one like "ed" or "debug" would be an easier starting point
07:28:49	*	nande quit (Remote host closed the connection)
07:28:53	keyle	i don't even know how to poll for a shortcut key on windows CMD
07:29:03	keyle	might not even be possible or deep down in winapi
07:29:47	*	iivvoo quit (Ping timeout: 244 seconds)
07:30:20	keyle	also would love to do a _modern_ alpine style email client
07:31:15	Triplefox	i liked pine, elm, etc.
07:31:24	Triplefox	maybe nostalgia
07:31:31	keyle	same here :)
07:34:29	ekarlso-	how the crap you create a pointer then ;)
07:35:55	Triplefox	.eval var a = [1,2,3]; var p = addr(a[0]); echo(p[])
07:35:59	Mimbus	Triplefox: 1
07:36:16	Triplefox	p is the pointer to index 0 of a
07:36:23	Triplefox	then i dereference it with []
07:37:26	*	iivvoo joined #nim
07:37:53	ekarlso-	Triplefox: what would be the equiv of char* key = "012345678"; then ?
07:38:10	Triplefox	hmm let's try a thing
07:38:26	Triplefox	.eval echo(cstring("012345678"))
07:38:29	Mimbus	Triplefox: 012345678
07:39:08	Triplefox	yeah, cstring coerces to char*
07:39:38	Triplefox	http://nim-lang.org/manual.html#cstring-type
07:45:10	ekarlso-	this ssl hmac ptr stuff is driving my head crazy
07:45:10	Triplefox	want me to gaze at the api
07:45:10	ekarlso-	https://gist.github.com/flaviut/38001b3b71d4c42e414f < flaviu left this for me yesterday ;P
07:45:36	Triplefox	oh boy
07:45:46	ekarlso-	;P
07:46:05	Triplefox	i wonder what a "d: cstring" is
07:47:11	Triplefox	okay http://www.openssl.org/docs/crypto/hmac.html
07:47:20	ekarlso-	:)
07:48:47	Triplefox	what is EVP_MD
07:49:06	*	gmpreussner___ quit (Ping timeout: 244 seconds)
07:50:18	ekarlso-	good q ..
07:50:24	ekarlso-	I'm fairly fresh on crypto stuff
07:50:50	ekarlso-	evp_md can be EVP_sha1(), EVP_ripemd160() etc.
07:51:01	Triplefox	there is example code here, this deserves study http://www.openssl.org/docs/crypto/EVP_DigestInit.html
07:52:18	Triplefox	it looks like here the pointer "md" is initialized by EVP_get_digestbyname...i.e. we aren't setting an address in our own code
07:52:54	Triplefox	likewise mdctx
07:56:19	ekarlso-	Triplefox: how we solve that ? sorry but when it comes to pointers etc I really am lost
07:56:41	Triplefox	we're calling API functions which allocate memory and return pointers
07:56:58	Triplefox	in that example: md = EVP_get_digestbyname(argv[1]);
07:57:11	Triplefox	and then: mdctx = EVP_MD_CTX_create();
07:57:23	Triplefox	for HMAC, there are also init functions it looks like
07:58:18	Triplefox	the nim bindings don't look complete enough to do all of this yet
07:58:48	Triplefox	(or i'm missing something about how bindings work in nim which is entirely possible)
07:59:30	ekarlso-	think i'll wait for flaviu to come around ;P
07:59:36	Triplefox	probably a good idea
08:00:49	Triplefox	grokking nim pointers, also worth it. you'll feel like a Real Systems Programmer
08:01:36	ekarlso-	Triplefox: how you mean ? :)
08:02:22	Triplefox	cause then you see the relationship between the flat memory layout and all the fancy indirection in data structures
08:04:02	ekarlso-	Triplefox: i thought c wrappers was easy ;P
08:06:29	Triplefox	when you deal with c you have to deal with low-level memory stuff, nim makes this easy by passing the problem on to you...which is good cause then there's no mismatch
08:13:55	*	Demon_Fox quit (Quit: Leaving)
08:17:43	ekarlso-	so Triplefox the "d" is the data you're gonna encrypt
08:19:12	ekarlso-	or so ..
08:19:43	*	HakanD joined #nim
08:20:08	ekarlso-	Triplefox: but theb there's the key pointer thing. how the stuffs do you create a pointer.
08:21:10	Triplefox	i've said everything that needs to be said on the topic.
08:22:03	ekarlso-	fair enough..
08:25:39	*	sdw quit (Read error: Connection reset by peer)
08:40:06	*	HakanD quit (Quit: Be back later ...)
08:40:52	ekarlso-	Triplefox: how you deref a pointer type then vs a ptr ?
08:43:01	Triplefox	.eval var mypointer : pointer; var a = 100; mypointer = addr(a); echo(cast[ptr int](mypointer)[])
08:43:04	Mimbus	Triplefox: 100
08:44:04	Triplefox	the only difference between pointer and ptr is what the compiler thinks they represent, so you cast from one to the other to tell it what to expect
08:45:20	*	Sphax joined #nim
08:45:25	*	Trustable joined #nim
08:48:46	*	Trustable quit (Remote host closed the connection)
08:49:17	*	JinShil quit (Quit: Konversation terminated!)
08:50:54	*	Trustable joined #nim
08:54:32	ekarlso-	Triplefox: but is the pointer type more low level then ptr ?
08:55:25	*	z1y left #nim (#nim)
08:55:32	Triplefox	it's "less typed". it just says that some kind of memory address is there, but not what it's supposed to mean when it's dereferenced
08:55:53	Triplefox	a ptr int means that when i dereference, i am getting an int back from that
08:56:07	ekarlso-	cause the compiler knows that it's a int...
08:56:10	ekarlso-	ok I get it
08:56:31	ekarlso-	and to convert from a "pointer" u need to cast[ptr foo] = somePointer
08:56:43	Triplefox	yeah
09:04:18	Triplefox	hmm freeimage is complaining it can't import FreeImage_Initialize
09:08:16	Triplefox	maybe binding is just out of date
09:10:32	*	dymk quit (Ping timeout: 244 seconds)
09:12:10	ekarlso-	so, what's csize then ?
09:12:32	*	dymk joined #nim
09:12:36	*	SplinterOfChaos quit (Ping timeout: 244 seconds)
09:13:01	Triplefox	i've never seen it
09:13:28	EXetoC	size_t
09:13:34	EXetoC	csize* {.importc: "size_t", nodecl.} = int
09:14:17	Triplefox	someday i'll have to bother learning how to do my own bindings...i'd like to get the stb libs in there
09:16:38	*	gokr joined #nim
09:21:51	*	sillesta joined #nim
09:22:17	ekarlso-	well Triplefox one step forward -,,-
09:24:54	ekarlso-	https://bpaste.net/show/0476a85f4005 < hmac.nim https://bpaste.net/show/ddbf0c39247b < result of nim c -r hmac.nim
09:25:07	ekarlso-	does procs need to have a diff name maybe ?
09:25:31	*	dymk quit (Ping timeout: 244 seconds)
09:27:03	*	dymk joined #nim
09:29:30	Triplefox	those errors are new to me
09:29:34	Triplefox	i don't know
09:30:15	ekarlso-	Triplefox: ok :)
09:31:07	*	Jehan_ joined #nim
09:39:59	*	dymk quit (Ping timeout: 244 seconds)
09:44:07	*	dymk joined #nim
09:57:51	EXetoC	why are you emitting includes? isn't that the source of the collision?
10:01:48	ekarlso-	EXetoC: tbh I dont know :) it's based off of what flaviu gist'd
10:05:06	EXetoC	you're essentially defining your own declarations here. omit those and see what happens
10:16:55	ekarlso-	ah ok
10:21:08	ekarlso-	EXetoC: you mean the importc stuff ?
10:23:36	EXetoC	ekarlso-: the includes
10:27:18	ekarlso-	dang EXetoC, that worked!
10:27:49	EXetoC	great
10:27:54	EXetoC	he probably just tested something
10:36:46	*	MajorTennis joined #nim
10:37:04	*	MajorTennis quit (Changing host)
10:37:05	*	MajorTennis joined #nim
10:40:57	*	dymk quit (Ping timeout: 244 seconds)
10:42:11	*	dymk joined #nim
10:42:49	*	HakanD joined #nim
10:51:23	*	JinShil joined #nim
11:06:59	*	HakanD quit (Ping timeout: 244 seconds)
11:07:27	*	HakanD joined #nim
11:07:28	*	lesshaste joined #nim
11:07:29	lesshaste	hi all
11:07:39	Araq	hi lesshaste welcome
11:07:48	lesshaste	hi Araq
11:17:03	EXetoC	hola
11:18:00	*	gour_ joined #nim
11:20:56	*	gour quit (Ping timeout: 244 seconds)
11:21:09	*	dumbrun joined #nim
11:28:42	*	gour_ is now known as gour
11:36:08	flaviu	ekarlso-: oops
11:36:14	flaviu	I gisted the wrong file
11:36:33	flaviu	https://gist.github.com/3e0b5a52fe9395fcf5c5
11:36:35	*	ARCADIVS quit (Quit: ARCADIVS)
11:38:30	*	skyfex_ quit (Ping timeout: 244 seconds)
11:39:28	*	yonson joined #nim
11:40:10	*	dumbrun quit (Ping timeout: 246 seconds)
11:40:37	*	VinceAddons quit (Read error: Connection reset by peer)
11:41:28	sillesta	good morning
11:48:21	*	dude joined #nim
11:48:45	*	dude is now known as Guest41089
11:49:18	*	Guest41089 is now known as someDude
11:51:02	*	someDude left #nim (#nim)
11:51:37	*	[dude] joined #nim
11:53:01	*	edsc joined #nim
11:53:17	*	endou______ quit (Ping timeout: 244 seconds)
11:53:54	*	endou______ joined #nim
11:55:50	[dude]	Hi everybody!
11:56:06	Araq	hi [dude]
11:56:19	[dude]	quick questions... I was trying to install c2nim through nimble just now...
11:56:30	edsc	Hi
11:56:34	Araq	yes, it's broken. we know
11:56:48	Araq	[dude]: it's hard to fix unfortunately
11:57:12	Araq	c2nim uses large parts of the compiler
11:57:39	[dude]	alrighty!
11:58:41	[dude]	was just trying to see if I can generate a libuv wrapper. I think found one someone created/generated for 0.11.x
11:58:45	[dude]	I'll give that a go...
11:58:47	Araq	[dude]: http://forum.nim-lang.org/t/738
11:58:56	Araq	describes how to setup c2nim
11:59:37	edsc	I have a question as well. i iterate over a string to parse every char in it but parseInt take a string in parameter. I would like to know what is the best way to cast it. thanks. it
12:00:16	Araq	edsc: just use parsutils.parseInt instead. that one fits a parser.
12:01:17	*	matkuki joined #nim
12:03:32	[dude]	Araq: I'm not sure if the post mentions how to compile c2nim ATM. I'm having trouble with 1. grabbing c2nim from nimble and 2. compiling c2nim via source. =(
12:04:18	Araq	[dude]: you need c2nim and nim in the same directory. nim should be 'devel' from github
12:04:33	Araq	and then you do:
12:04:44	Araq	cd c2nim
12:04:47	Araq	nim c c2nim
12:05:01	*	nercury joined #nim
12:06:05	Araq	if 'nim' is in /usr/bin where do nim's sources usually end up?
12:09:34	*	Sphax quit (Quit: ZZZZZzzzzz)
12:10:42	BlaXpirit	Araq, possibly nowhere
12:11:17	Araq	BlaXpirit: yeah sure but *if* it exists
12:11:29	BlaXpirit	could be anywhere :s
12:11:56	Araq	so the compiler has no chance to setup '$nim' properly
12:12:21	BlaXpirit	the structure of the project is quite weird
12:12:31	BlaXpirit	no wonder package maintainers don't know what to do
12:13:13	EXetoC	so, do we introduce /usr/lib/nim/compiler?
12:13:23	BlaXpirit	that would be best, I think
12:13:46	Araq	nah, I'll make the compiler's source a pseudo Nimble package
12:14:01	Araq	but I dunno how lol
12:14:22	BlaXpirit	Araq, here's one package if you care to look inside it https://www.archlinux.org/packages/community/x86_64/nim/download/
12:15:46	edsc	Araq, I was trying to know how to convert a char because each character need to be pase individualy and when I iterate I get a char and not a string which result in a missmatch.
12:15:59	lesshaste	how large is the development team these days?
12:16:46	lesshaste	I mean for nimrod
12:17:04	Araq	lesshaste: hard to say. for the compiler itself 3 people, but I still do most of the work :P
12:17:13	lesshaste	Araq, oh!
12:17:21	lesshaste	Araq, well in that case... a double hello :)
12:17:44	lesshaste	do you have a feeling for how many people program in nimrod?
12:17:47	Araq	for the stdlib there are a couple of contributors
12:17:55	lesshaste	it really sounds like very few
12:18:12	lesshaste	unless you work 18 hours a day on it :)
12:18:45	lesshaste	or 180 hours a day
12:18:57	MyMind	Araq: nim is a full time job for you ?
12:19:53	lesshaste	we need more nimrod codes on codegolf.se :) E,g, http://codegolf.stackexchange.com/questions/42215/program-a-racing-car
12:20:08	Araq	MyMind: no, I have a real job too. :-)
12:21:11	MyMind	unless you get sponsored :)
12:21:41	Araq	well depending on how you look at it, this already happened.
12:22:14	lesshaste	actually a nimrod answer for http://codegolf.stackexchange.com/questions/44278/debunking-stroustrups-debunking-of-the-myth-c-is-for-large-complicated-pro would be very nice
12:22:38	dom96	ekarlso-: What are you using hmac for?
12:22:39	MyMind	by whom Araq ?
12:24:41	Araq	MyMind: I'll announce it officially in some weeks. then I will also have to show you guys exciting new stuff.
12:24:55	MyMind	wow
12:25:04	edsc	I found my mistake, i forgot to use the stringify operator '$' :/ .
12:25:07	MyMind	that's exicting
12:25:48	MyMind	Araq: for the 1.0 release ?
12:26:09	Araq	MyMind: mostly orthogonal to 1.0.
12:26:38	MyMind	what means orthogonal besides a geometric form?
12:27:08	Araq	BlaXpirit: ok, so no compiler sources in packages, got it.
12:27:24	BlaXpirit	Araq, my test sample consists of 1 distro though :|
12:27:45	Araq	well technically that *is* the proper way to do it
12:28:00	sillesta	how does the "parallel:" on the nim-lang.org site work?
12:28:51	Araq	sillesta: at runtime it produces a barrier
12:29:16	sillesta	is it in a specific module?
12:29:49	Araq	at compiletime it tells the compiler "DSL with special rules". it's builtin but pretend to be in threadpool
12:30:03	EXetoC	can the compiler be turned into a package with the current package system?
12:31:22	BlaXpirit	EXetoC, what do you mean
12:31:35	BlaXpirit	can a folder with .nim files be turned into a package?
12:31:38	BlaXpirit	sure it can
12:32:34	*	yonson quit (Quit: Leaving)
12:33:58	EXetoC	I meant, using the current directory structure, but maybe the compiler interface should be separated from everything else
12:34:25	BlaXpirit	there is no logical place to put it in, except lib
12:34:41	BlaXpirit	and yes, i'm pretty sure you can do it as a separate package
12:34:48	BlaXpirit	but it would be an unpopular decision
12:34:59	BlaXpirit	it shouldn't be difficult to just copy a folder
12:35:00	EXetoC	that's what he wanted, but ok
12:35:19	EXetoC	I don't know what's wrong with installing it along with everything else though
12:35:41	BlaXpirit	it's not "wrong", it's just that it is not specified anywhere
12:35:55	BlaXpirit	package maintainers simply can't know that the compiler files may be needed at all
12:36:10	keyle	Araq: very excited at this prospect, 1.0 stuff
12:36:32	keyle	night
12:36:33	EXetoC	it's just a simple modification then
12:36:41	*	keyle quit (Quit: Morbo: Chitchat achieved!)
12:36:52	BlaXpirit	EXetoC, why don't you try
12:36:52	Araq	BlaXpirit: well this really only affects c2nim. the compiler is not a library yet.
12:37:17	Araq	so package maintainers do the right thing
12:37:19	BlaXpirit	in nim-git just adding cp -r compiler pkgdir/usr/lib/nim/compiler blah blah
12:37:37	BlaXpirit	piece of cake, really
12:40:26	EXetoC	that might do
12:40:30	*	yonson joined #nim
12:41:24	BlaXpirit	EXetoC, it's a very good PKGBUILD, but I did realize only recently that I need it to be devel branch
12:41:42	BlaXpirit	so i add a line git checkout devel , no problem
12:41:44	*	HakanD quit (Quit: Be back later ...)
12:42:36	EXetoC	if you say so
12:43:07	BlaXpirit	I don't think PKGBUILD should be devel, though, it's good.
12:43:37	BlaXpirit	but a problem is, the repository is too huge :o
12:44:57	EXetoC	the size was reduced at some point
12:45:34	*	Sphax joined #nim
12:48:25	*	edsc quit (Ping timeout: 246 seconds)
12:50:11	*	HakanD joined #nim
12:50:26	sillesta	github repo link to aporia IDE on nim-lang.org links to the old https://github.com/nimrod-code/Aporia
12:50:43	EXetoC	because csources was put in a separate repo, but it has to be included
12:51:12	BlaXpirit	EXetoC, there is probably a huge history of some binary file in that repo
12:51:55	BlaXpirit	i'm sure there have been investigations on this topic, no point in me guesing here
12:52:01	yonson	Hi, I´m trying to compile greetings.nim (from the nim tutorial) on my windows machine but I´m getting the following error https://gist.github.com/psychotropic/ab8e38ae7faf539b6e32
12:52:12	BlaXpirit	sillesta, but it's ok, because redirects :p
12:52:13	yonson	i´ve checked and nim is correctly added to my path
12:52:29	sillesta	yeah, just noticed it for some reason ;)
12:53:02	BlaXpirit	yonson, it may be because of spaces in directory name
12:53:08	BlaXpirit	:s
12:53:10	BlaXpirit	try to have no spaces in the path
12:53:13	BlaXpirit	and i'm out
12:53:17	yonson	hmm i´ll try
12:53:18	yonson	just a sec
12:53:51	yonson	oh well
12:53:56	yonson	that fixed it
12:54:04	yonson	thanks
12:54:19	EXetoC	BlaXpirit: there are several directories in csources that are similar
12:54:22	EXetoC	are all those needed?
12:54:22	Araq	yonson: er... that should be fixed for years now :-/
12:54:33	BlaXpirit	EXetoC, no idea
12:54:57	yonson	it does seem kind of a silly bug
12:54:57	BlaXpirit	you know a lot more than me, both about nim and pkgbuilds
12:57:10	Araq	yonson: report it please
12:59:16	EXetoC	Araq: are all 16 directories in csources needed?
12:59:56	Araq	EXetoC: of course. think of our poor solaris users.
13:00:08	EXetoC	-.-
13:01:31	Jehan_	EXetoC: You can make gzip/bzip2/xz pack them better by sorting the tar file so that equally named files are next to each other.
13:01:48	Jehan_	With xz, that becomes actually pretty small.
13:03:32	lesshaste	what do nimrod people think of rust? I am not really clear on the significant differences
13:03:52	EXetoC	*nim :)
13:04:09	yonson	Araq, i now see that my computer is using gcc 4.7.3 (old installation), maybe the bug is now fixed?
13:04:43	lesshaste	I suppose rust has lower level semantics?
13:05:28	EXetoC	Jehan_: and on windows?
13:05:47	Jehan_	lesshaste: Both languages really target different application domains.
13:05:53	EXetoC	we'd need a portable solution I think
13:06:02	lesshaste	Jehan_, could you tell me more about that please?
13:06:06	lesshaste	which domains? :)
13:06:17	Jehan_	EXetoC: On Windows, you can just package the Windows-only code in zip file.
13:07:06	lesshaste	I am a slightly odd user as I only ever write smallish scientific applications (for example simulations) that need to be fast
13:07:07	Jehan_	lesshaste: The primary difference is that Rust tries to avoid automatic memory management, while Nim's default is automatic memory management.
13:07:35	lesshaste	Jehan_, that's very interesting. I have no interest in managing memory myself :)
13:07:40	EXetoC	Jehan_: we need two repos then
13:08:48	Araq	yonson: please try the GCC that we ship with the installer (optional component)
13:08:54	Jehan_	lesshaste: Well, there are application domains where automatic memory management is an issue (e.g., much of embedded programming) or where you want to have complete control over it.
13:09:12	yonson	will do and report the results
13:09:26	lesshaste	Jehan_, right I can see that. I am being very selfish and just trying to work out what is most useful for me
13:09:48	lesshaste	Jehan_, I don't think I ever need to allocate memory. Although it would be nice to be able to use all my cores easily
13:09:56	Jehan_	EXetoC: Hmm, I would think that the repo size doesn't matter much. I was thinking more in terms of .tar.xxx or .zip files.
13:10:35	Jehan_	lesshaste: "Using all my cores" and "easily" usually don't go together in one sentence except for embarallel problems. :)
13:10:52	lesshaste	Jehan_, my problems are embarrassingly parallel in the main :)
13:11:13	lesshaste	Jehan_, mostly, run the simulation for a thousand different values of n and choose the maximum
13:11:26	Araq	Jehan_: I will argue that Nim provides more control over memory management than C++, but it's worth a blog post.
13:11:44	EXetoC	Jehan_: the repo would just be a single archive then, and it's around 300mb now
13:11:48	Jehan_	lesshaste: Ah, I see. But that should be easy in any language.
13:11:56	[dude]	Araq: Sorry, stepped away fro a second. I got c2nim to compile now. Thanks for the help. =)
13:12:05	lesshaste	Jehan_, it's a pain in C because everything is a pain in C
13:12:06	Jehan_	Araq: That'd be shooting fish in a barrel. :)
13:12:26	lesshaste	Jehan_, and python doesn't by default even allow you to run things in parallel
13:12:40	Jehan_	lesshaste: True. :) But it's not really something that would differentiate Nim and Rust.
13:12:47	lesshaste	Jehan_, ok :)
13:13:03	EXetoC	Araq: what do these poor solaris users need it for?
13:13:23	lesshaste	I basically want the highest level language possible with the largest set of scientific libraries possible that will let me write code that runs at the speed of C :)
13:13:27	lesshaste	too much to ask? :)
13:13:31	yonson	Araq, nope, I still get the same error with nim´s gcc (4.9.1 according to my computer). Should i fill an issue on github?
13:13:43	Araq	yonson: of course.
13:13:50	Jehan_	lesshaste: The reason I'm not using Rust is that I have no need for fine-grained control over memory management, basically, but that the memory management techniques of Rust leave a footprint on my code that I can avoid in NIm.
13:14:13	lesshaste	Jehan_, I am reading http://pcwalton.github.io/blog/2013/03/18/an-overview-of-memory-management-in-rust/
13:14:19	lesshaste	seems a lot of new stuff to learn
13:14:32	Jehan_	lesshaste: Libraries are still an issue for both, I think, especially scientific libraries; there it comes down to how easy it is to create bindings.
13:14:33	EXetoC	how about a csources_solaris for those few people. not that I know what the purpose is
13:15:15	lesshaste	Jehan_, right..sadly there is http://scikit-learn.org/stable/ and I don't know if other languages can bind to it
13:15:50	lesshaste	scala gets round this by letting you use java libraries I think
13:15:59	Jehan_	lesshaste: Not familiar with it; I'm currently working on computer algebra systems, which are basically their own niche.
13:16:12	*	matkuki quit (Read error: Connection reset by peer)
13:16:14	EXetoC	well, python has an C API
13:16:37	lesshaste	Jehan_, interesting... which computer algebra systems ?
13:16:37	*	matkuki joined #nim
13:16:41	Jehan_	lesshaste: Yeah, Java interoperability is Scala's great strength (and JVM dependency similarly probably its greatest weakness).
13:16:48	Jehan_	lesshaste: GAP, Singular.
13:16:52	lesshaste	Jehan_, nice :)
13:17:17	lesshaste	it's really stupid how all these impressive libraries are tied to one language
13:17:27	EXetoC	and we have nimborg, but it seems like yet another stale lib. there's a low level binding though I think
13:17:28	lesshaste	you feel that with some thought they could have made them interoperable from the start
13:17:50	Jehan_	lesshaste: That'd be nice, but it's harder than you think.
13:17:51	EXetoC	lesshaste: if indeed the libs are unique, yes
13:18:37	lesshaste	EXetoC, I think scikit learn is more or less unique. I mean there is a competing set of libraries called weka for java but there are thousands (millions?) of man hours in these projects
13:18:43	Jehan_	Araq: I actually think memory management in C++ is a big issue, but that ties into my usual argument how I believe that RAII is fundamentally broken (except for rare use cases).
13:18:46	lesshaste	it's madness to reimplement them for each language
13:19:00	lesshaste	Jehan_, I can imagine but it seems worth it
13:19:12	EXetoC	in which case I agree. too bad C is what it is though
13:19:30	Jehan_	lesshaste: A major problem is data representation.
13:20:18	lesshaste	Jehan_, a flat array to pass data in and out would seem to cover most languages wouldn't it?
13:20:44	Jehan_	You could work around that by making every type completely opaque, but then you need automatic memory management/GC and that means all libraries and the language need to agree on using the same GC.
13:20:54	EXetoC	lesshaste: I've wanted to write an interface that would allow one to easily write C libs in nim, but that might take some time
13:20:55	Jehan_	lesshaste: But would be totally inefficient.
13:21:04	EXetoC	not that it solves your particular problem
13:21:07	lesshaste	EXetoC, nice :)
13:21:07	Araq	Jehan_: I don't think you ever explained how RAII is broken. :-)
13:21:21	Araq	I think it's completely broken too, but I like to hear your thoughts
13:21:26	Jehan_	Assume that you want to represent elements of F[x,y,z] (i.e. polynomials over three variables and the field F).
13:22:04	Jehan_	Araq: Would be lengthy, I was thinking about posting it in the RAII thread, then I thought better than to stir up unnecessary trouble. :)
13:22:17	EXetoC	I have so many ideas, but I keep coming back to that one
13:22:17	lesshaste	Jehan_, ok
13:22:47	lesshaste	maybe nim could have a way of calling python code :)
13:22:48	Jehan_	lesshaste: There are a number of ways you can represent them. You may require them to have a reference to F, too, or F be implicit.
13:22:59	Jehan_	lesshaste: I think there already is a module for that?
13:23:04	lesshaste	oh ok !
13:23:13	*	valberg is now known as keyvalberg
13:23:16	Araq	Jehan_: IMO it's because there is no escape analysis whatsoever involved and c++ only has 'const' and not 'lent'
13:23:26	lesshaste	I suppose that works round this particular problem then?
13:24:05	Jehan_	Araq: Number of problems, short list:
13:24:18	Jehan_	(1) RAII does not have sane semantics for variables with static scope.
13:24:21	lesshaste	Jehan_, I suppose I have a very crude way of thinking about these things. I am imagine a library that takes in some data and returns something
13:24:44	lesshaste	Jehan_, in that case the API specifies the format it wants and it is your job to provide it
13:24:46	Jehan_	(2) RAII does not have sane semantics for concurrency (also why shared pointers in C++ essentially have the performance of a small glacier).
13:24:57	lesshaste	Jehan_, but your only problem is making sure the data you pass is in the right format
13:25:11	lesshaste	Jehan_, I think you are talking about something much more sophisticated... ?
13:25:45	Jehan_	(3) Main use case #1 for RAII is reference counting, which is slow if done using RAII.
13:26:10	Jehan_	lesshaste: Hmm, talking about two different things at the moment.
13:26:17	lesshaste	Jehan_, :)
13:26:18	Araq	the unique_ptr fraction would disagree ;-)
13:26:48	lesshaste	Jehan_, imagine I want to compute the inverse of a matrix. I pass the matrix and get the inverse. My only problem is to pass the matrix in the right format isn't it?
13:26:50	Jehan_	Araq: unique_ptr is too weak for a lot of important stuff.
13:26:58	EXetoC	time to look into it. I'd need a way to fail as soon as something wants the GC though
13:27:22	*	Demos joined #nim
13:27:39	Demos	Triplefox, ping
13:28:03	Jehan_	(4) RAII is used for open/close scenarios with local scope, which can be done as well and often better if you have higher order functions or clean macros.
13:28:36	*	keyvalberg is now known as valberg
13:29:08	Jehan_	More importantly, there are tons of other resource management scenarios that RAII does not have any real applicability for (such as an LRU pool).
13:29:40	Jehan_	(5) If you use it for resource handling, it leaves a significant footprint on your code (see, e.g., the Rule of Three, now actually the Rule of Five).
13:30:47	Araq	many disagree with (4) as well. :-) i don't though. I think this whole constructor/destructor business is blown out of proportion. You need to learn the API for the type anyway, even everything uses the precious 'new T' syntax
13:31:02	Araq	*even when
13:31:11	Jehan_	lesshaste: Well, yeah. And do you pass the matrix as a list of row or column vectors, for staters?
13:31:46	Jehan_	Araq: Constructors are a totally different story. I may actually post something in that thread, that can be done without inciting a flame war.
13:32:02	EXetoC	let's not forget to bash the name
13:32:04	Demos	I think (4) is more or less valid, it depends on what resource you are using, like I don't really like using RAII for files
13:32:10	lesshaste	Jehan_, ok but that's in the specs. I mean in my dream world that only thing the interface has to do is to code in nimrod to pass the data in the format the specs demand
13:32:26	lesshaste	Jehan_, he/she never has to do any other low level coding or coding in a foreign language
13:32:38	Jehan_	lesshaste: In my dream world, too. :)
13:32:50	lesshaste	Jehan_, :)
13:34:04	Jehan_	Demos: A major problem with using RAII for resource handling can be that hidden pointers keep a resource alive longer than you think.
13:34:17	lesshaste	Jehan_, I wonder why our dream worlds don't exist :)
13:34:17	lesshaste	?
13:34:23	Jehan_	This is particularly true in languages with closures.
13:34:30	Demos	Jehan_, this is true. For memory I think it is OK
13:34:41	*	leavengood joined #nim
13:34:44	Demos	for something like files not so much
13:34:46	Jehan_	Store a closure as a callback somewhere, and suddenly all of its environment is being kept alive.
13:35:04	Demos	and then there are things like sockets that do have open/close semantics but you really need to understand when to close them
13:35:09	*	leavengood is now known as Guest48178
13:35:12	Jehan_	That's actually #6: RAII and closures do not mix well, and closures are an important concept.
13:36:19	Jehan_	Demos: Also, importantly, sometimes closing a resource is non-trivial. RAII assumes that no error handling is needed for that.
13:36:26	EXetoC	though I can't use sequtils for example because of my requirements. oh well
13:37:38	*	JinShil quit (Quit: Konversation terminated!)
13:37:57	Demos	Jehan_, sure. But the resources that do require error handling don't really work in any kind of automatic system. I agree though
13:39:29	*	JinShil joined #nim
13:47:23	*	Guest48178 quit (Quit: leaving)
13:50:40	*	nercury quit (Quit: nercury)
13:51:21	Demos	well I gotta go, but Triplefox, I am the maintainer of the FreeImage library and would like to know if you got it working. It is totally possible that there is a missing dynlib directive or your OS may be having trouble finding the library. I use it on windows (usually static linked) and on RHEL
13:51:35	*	nerijus joined #nim
13:52:38	ekarlso-	dom96: JWT tokens
13:56:20	ekarlso-	flaviu: u around ? :)
14:00:07	*	nerijus left #nim (#nim)
14:17:44	*	Jehan_ quit (Quit: Leaving)
14:19:10	*	Demos quit (Ping timeout: 255 seconds)
14:24:16	*	seemsindie joined #nim
14:25:09	*	jefus__ joined #nim
14:27:04	*	gmpreussner___ joined #nim
14:29:11	*	jefus_ quit (Ping timeout: 252 seconds)
14:29:33	gmpreussner___	i have a C library with structs that hold function pointers for callbacks that users can register. how would i best model this in nim? is it possible?
14:29:56	*	gmpreussner___ is now known as gmpreussner
14:33:17	*	jefus__ is now known as jefus
14:37:53	jpoirier_	gmpreussner: yes. E.g. just mirror the C callback typedef in your nim wrapper, e.g. type readAsyncCbProc* = proc (buf: ptr uint8; len: uint32; userCtx: UserCtxPtr) {.fastcall.}
14:38:11	*	JinShil quit (Quit: Konversation terminated!)
14:46:00	BlaXpirit	gmpreussner, i made an example recently
14:46:00	*	ivan_ joined #nim
14:46:17	*	ivan_ quit (Client Quit)
14:46:52	BlaXpirit	sec
14:46:56	*	Demos joined #nim
14:47:18	*	ivan_ joined #nim
14:47:33	*	ivan_ quit (Client Quit)
14:47:50	BlaXpirit	gmpreussner, https://github.com/LaurentGomila/CSFML/blob/fdf58188caf49684fbeb5e951c61e92a0a6c3e82/include/SFML/Graphics/Shape.h#L39 https://github.com/BlaXpirit/nim-csfml/blob/254aecd3d1c68ba1febbafc950f1fa7f78872706/examples/shapes.nim#L43 https://github.com/BlaXpirit/nim-csfml/blob/254aecd3d1c68ba1febbafc950f1fa7f78872706/src/csfml_graphics_gen.nim#L2753
14:47:56	BlaXpirit	that's 3 huge links D:
14:48:19	*	ivan_ joined #nim
14:48:21	*	ivan_ quit (Client Quit)
14:48:47	*	seemsind1e joined #nim
14:48:55	BlaXpirit	still better than blob/master links because they rot :p
14:49:50	*	seemsind1e quit (Client Quit)
14:49:56	*	seemsindie quit (Quit: Page closed)
14:50:23	Demos	well irc is hardly a bullet proof archive
14:50:24	*	seemsindie joined #nim
14:50:39	BlaXpirit	it sure is, what with IRC logs available
14:50:59	*	HakanD quit (Quit: Be back later ...)
14:51:46	seemsindie	y
14:52:42	jpoirier_	BlaXpirit: Is {.cdecl.} okay for callbacks? I'm using {.fastcall.} for reasons I don't remember :/
14:52:57	BlaXpirit	jpoirier_, it must be consistent with what actually is in the library
14:53:10	BlaXpirit	if it's just a normal C library with normal C functions, it must be cdecl
14:53:27	BlaXpirit	and if there is actually __fastcall present in C code, then it's fastcall
14:53:32	BlaXpirit	i may be mistaken though
14:53:51	jpoirier_	BlaXpirit: k, that sounds right, thanks
14:56:23	*	darkf quit (Ping timeout: 240 seconds)
14:56:25	*	HakanD joined #nim
15:00:10	*	atomic-hedgehog joined #nim
15:00:42	gmpreussner	jpoirier_, BlaXpirit: thanks guys!
15:01:43	*	Varriount|Busy joined #nim
15:01:50	EXetoC	jpoirier_: it might've worked just because of luck
15:02:09	EXetoC	BlaXpirit: you could write a minimal example in C and then run c2nim over it, just to see what is being generated
15:02:28	BlaXpirit	lol i made half of c2nim myself already
15:03:10	BlaXpirit	i actually went as far as generating function type signatures from c code
15:03:57	BlaXpirit	that last link is autogenerated code btw
15:03:58	EXetoC	jpoirier_: or was __fastcall indeed in the code?
15:04:33	EXetoC	BlaXpirit: well it looks like correctly generated callbacks
15:04:53	BlaXpirit	sure is, cuz it even works :p
15:04:58	atomic-hedgehog	hi, I'm working with the httpclient library on the devel branch, and my program failed to compile. The issue is with the net.nim library requiring uint16 in some places. There's a cast to uint16 for one part of an expression, but the rest of it is outside the cast and getting coereced to int
15:05:03	atomic-hedgehog	https://github.com/Araq/Nim/blob/devel/lib/pure/net.nim#L1039
15:05:39	atomic-hedgehog	I've fixed it locally by wrapping the cast aroudn the entire expression and everything seems to be working.
15:05:48	atomic-hedgehog	Since I'm very new to nim, I'
15:05:56	EXetoC	BlaXpirit: ok you figured it out
15:06:00	atomic-hedgehog	m not sure if this is the right way to fix it.
15:06:07	atomic-hedgehog	Shoudl I do a pull request?
15:06:09	BlaXpirit	EXetoC, uhm it wasn't me who needed help
15:06:24	EXetoC	nvm then
15:06:38	BlaXpirit	atomic-hedgehog, this problem was brought up recently, i'm pretty sure
15:06:48	BlaXpirit	and fixes are probably being worked on
15:06:48	*	HakanD quit (Quit: Be back later ...)
15:07:13	BlaXpirit	hm i'm not sure though
15:07:43	BlaXpirit	atomic-hedgehog, yeah, look https://github.com/Araq/Nim/commit/f7ef00aa3be1b31ad2781f21852c41f55c5de1e0#diff-df7c4ae903eab01bc8c4f64f425b48c3
15:08:05	BlaXpirit	seems to have been fixed a day ago, maybe you're not up to date
15:08:11	EXetoC	people and their ancient builds :p
15:08:16	*	gmpreussner quit (Ping timeout: 244 seconds)
15:08:49	atomic-hedgehog	ha, nope, I found this two days ago. I have a small child so my time to work on stuff is really limited.
15:09:08	BlaXpirit	atomic-hedgehog, i don't understand what you mean
15:09:17	BlaXpirit	you mean you have devel from 2 days ago?
15:09:24	BlaXpirit	then update is your solution
15:09:29	atomic-hedgehog	Right, so it should be fixed in latest.
15:09:34	BlaXpirit	ok
15:09:40	atomic-hedgehog	I'll do a pull
15:09:45	BlaXpirit	good
15:12:36	*	BitPuffin joined #nim
15:16:47	*	nercury joined #nim
15:17:43	*	nodrygo_ joined #nim
15:17:55	*	gunn joined #nim
15:22:36	onionhammer	araq it would be nice to be able to just say {.compile with cpp.} at the top of a module
15:23:46	onionhammer	it works if I do something hacky like this at the top: proc CPP {.importcpp.}
15:23:48	onionhammer	:)
15:24:15	BlaXpirit	it's hard for me to imagine how one can compile different parts of a project with different compilers
15:24:33	onionhammer	because they're ABI compatible
15:24:44	onionhammer	and distinct .o files are generated per module
15:27:04	BlaXpirit	cool stuff
15:28:08	*	nodrygo joined #nim
15:28:23	*	nodrygo_ quit (Quit: Page closed)
15:29:13	*	nodrygo left #nim (#nim)
15:31:10	*	Varriount|Busy quit (Ping timeout: 246 seconds)
15:32:59	*	Demos quit (Ping timeout: 252 seconds)
15:36:50	dts|pokeball	does the nim compiler generate .o files? i thought all of that was handed to the compiler?
15:36:57	dts|pokeball	does the nim compiler generate .o files? i thought all of that was handed to the c compiler? **
15:37:01	dom96	THe C compiler does.
15:37:45	dts|pokeball	yeah, so ABI compatability doesnt matter in this case
15:40:05	onionhammer	why wouldnt it matter
15:41:16	onionhammer	oh, i thought .o files were binary
15:41:24	dts|pokeball	because its compiling down to c/c++/js/etc and then the system compiler generates the ABI compatibile file
15:41:53	onionhammer	i'm talking about compiling a C++ and a C file together into a final exe
15:43:06	onionhammer	well, linking them together i guess..
15:43:17	onionhammer	if that makes sense
15:43:34	*	nodrygo joined #nim
15:43:55	onionhammer	re http://en.wikipedia.org/wiki/Application_binary_interface
15:44:21	onionhammer	"ABIs cover details such as ... in the case of a complete operating system ABI, the binary format of object files, program libraries and so on."
15:47:16	*	[dude] quit (Ping timeout: 246 seconds)
15:51:41	*	johnsoft quit (Ping timeout: 252 seconds)
15:52:18	*	johnsoft joined #nim
15:54:31	ekarlso-	anyone interested in making a JWS library ?
15:56:07	dom96	what is JWS?
15:56:20	*	SplinterOfChaos joined #nim
15:57:55	ekarlso-	dom96: javascript object signing
15:58:01	ekarlso-	and encryption (jose)
15:58:02	ekarlso-	:D
15:58:04	ekarlso-	https://datatracker.ietf.org/wg/jose/documents/
15:58:19	dom96	Why do you need that?
15:58:33	ekarlso-	dom96: more libraries = more reasons to use nim :)
16:02:11	*	nercury quit (Quit: nercury)
16:03:49	*	ZzZMorpork is now known as AMorpork
16:07:59	*	VinceAddons joined #nim
16:08:46	*	brson joined #nim
16:09:45	ekarlso-	I love nim
16:18:12	*	gokr quit (Quit: Leaving.)
16:27:16	*	Demos joined #nim
16:37:35	*	nercury joined #nim
16:40:32	sillesta	don't we all :)
16:40:45	*	Demos quit (Ping timeout: 244 seconds)
16:44:47	*	seemsindie quit (Ping timeout: 264 seconds)
16:56:36	*	BlaXpirit_ joined #nim
16:58:47	*	BlaXpirit quit (Ping timeout: 252 seconds)
17:00:21	*	nercury quit (Quit: nercury)
17:04:09	*	gokr joined #nim
17:05:24	*	jokr joined #nim
17:10:37	*	yonson quit (Ping timeout: 255 seconds)
17:15:48	*	jokr quit (Quit: WeeChat 1.0.1)
17:16:05	*	jokra joined #nim
17:16:40	*	Demos joined #nim
17:22:16	*	jokra quit (Quit: WeeChat 1.0.1)
17:22:31	*	jokra joined #nim
17:24:20	*	jpoirier_ quit (Quit: Leaving)
17:24:29	*	nodrygo left #nim (#nim)
17:36:21	*	skyfex joined #nim
17:42:01	*	Sphax quit (Quit: CYA!!)
17:46:42	*	jpoirier joined #nim
17:47:22	*	gokr quit (Quit: Leaving.)
17:48:35	*	jokra quit (Quit: WeeChat 1.0.1)
17:49:38	*	adam_s joined #nim
17:50:35	*	jokra joined #nim
18:01:04	*	jefus_ joined #nim
18:04:15	*	jefus quit (Ping timeout: 264 seconds)
18:07:34	ldlework	If I have a library with BaseType, and in my application I have a Subtype, and a Foo and a Bar, and Foo creates a Subtype and uses library code (which takes BaseType) to pass the Subtype from Foo to the Bar, how I can I do this so that bar gets a Subtype, not a Basetype?
18:08:27	*	UberLambda joined #nim
18:08:53	ldlework	Passing the Subtype to the library seems to cause a downcasting
18:09:28	ldlework	I understand that I can upcast back to the Subtype in Bar when it recieves the BaseType from the library, but how should Bar know which Subtype to upcast to?
18:09:37	ldlework	It might be a SubtypeA or a SubtypeB, etc
18:30:48	*	Demos quit (Ping timeout: 245 seconds)
18:48:37	Araq	ldlework: usually people end up with init(x: var Base) procs, I think
18:55:23	*	rpag joined #nim
19:10:21	*	pjc joined #nim
19:13:14	joecodes	speaking of more libraries, after reading: http://blog.gopheracademy.com/go-and-ssh/ I wonder if there is an ssh library for nim.
19:13:51	def-	joecodes: haven't seen one yet
19:18:26	*	jefus_ is now known as jefus
19:30:18	*	yonson joined #nim
19:30:19	dom96	Araq: How can I catch DeadThreadError?
19:30:43	Araq	er .. you shouldn't
19:31:09	dom96	Meh. Looks like raises: [] on a {.thread.} doesn't work well.
19:31:35	Araq	ah I see
19:31:54	Araq	so we should hide this somehow like we do for AssertionError, perhaps
19:32:21	Araq	in fact
19:32:24	dom96	Maybe.
19:32:33	Araq	I can make it use assert instead of raise.
19:32:36	dom96	I put everything in the thread in a try catch and it still thinks that Exception is raised.
19:32:48	*	atomic-hedgehog quit (Ping timeout: 244 seconds)
19:33:29	Araq	the compiler is usually right. gradha had a similar problem and it was all his fault :P
19:34:34	*	atomic-hedgehog joined #nim
19:44:08	onionhammer	araq thoughts on specifying compiler with a bare pragma
19:44:29	*	rpag quit (Quit: zZz)
19:44:38	onionhammer	i mean.. besides your initial reaction of hating everything I say :P
19:45:37	Araq	ach quatsch, ich liebe jeden Menschen auf Gottes grüner Erde...
19:45:58	Araq	onionhammer: well ... as you noted
19:46:12	Araq	there is already a "simple" way to do that
19:46:25	Araq	which breaks with --deadCodeElim:on I think
19:46:56	Araq	but. the "mixed" mode has been designed so that you don't have to care about that
19:48:14	onionhammer	say I"m using a lot of try/catches or something and know that this module will perform much better compiled to cpp?
19:49:58	Araq	you document then that people should compile it with 'cpp' instead of 'c'?
19:50:04	dom96	Any ideas how to check if a stream has something in it?
19:50:25	Araq	or somehow specify it in your nake file, I dunno if that's possible
19:50:40	Araq	onionhammer: I think you can also do:
19:50:46	Araq	when not defined(cpp):
19:50:58	Araq	{.error: "compile as C++ you fool!".}
19:51:00	onionhammer	I suppose... just think it would be nicer to have a {.exportcpp.} pragma
19:51:11	onionhammer	or {.export:cpp.} or something
19:51:32	onionhammer	afk
19:51:33	Araq	yes, lots of things would be nice. we have ~400 open issues.
19:52:23	dom96	Argh. The only way would be to check asynchronously I guess.
19:54:05	Araq	dom96: posix hates that. afaict you can only 'read'
19:55:25	onionhammer	that's fair
19:58:45	dom96	Araq: Any chance I could hack something together with spawn?
19:59:02	Araq	to do what?
19:59:43	dom96	To attempt to read from a process' errorStream
20:00:00	dom96	I don't suppose that will work.
20:00:02	*	adam_s quit (Quit: Leaving)
20:00:20	EXetoC	onionhammer: then what? you compile as C and possibly get an empty program?
20:00:42	onionhammer	no if you do nim c cppmodule.nim it will use cpp anyway
20:00:47	onionhammer	(currently)
20:00:59	*	Sergio965 joined #nim
20:02:26	Araq	dom96: no, cannot work
20:02:36	dom96	oh well
20:02:43	dom96	No fancy highlighting of stderr then
20:03:00	dom96	maybe I can just match the errors with pegs
20:03:04	*	gunn quit (Ping timeout: 265 seconds)
20:03:25	Araq	dom96: json output for the compiler's error messages is in the works
20:03:35	ldlework	What's the thing in Nim where I can constrain generic type discriminators?
20:04:16	onionhammer	as part of idetools araq?
20:04:35	onionhammer	capital-I ide tools or lower-case
20:04:42	Araq	onionhammer: it'll be a general switch, not part of idetools
20:04:56	dom96	Araq: cool
20:05:00	dom96	What about idetools?
20:05:01	Araq	onionhammer: we still do cs:none for all command line things
20:05:19	ldlework	found it
20:05:21	Araq	and will keep this way
20:06:19	*	gunn joined #nim
20:06:58	*	pjc quit (Ping timeout: 246 seconds)
20:07:15	ldlework	If you have a type which is generic, do you have to also make all of its method's generic too?
20:07:20	ldlework	wait no
20:07:31	ldlework	well yeah
20:07:59	*	Sergio965 quit (Quit: My MacBook Pro has gone to sleep. ZZZzzz…)
20:08:23	onionhammer	ldlework http://nim-lang.org/manual.html#type-classes
20:08:41	*	enquora joined #nim
20:08:54	ldlework	I'm starting to feel my brain melt
20:09:56	*	yonson quit (Quit: Leaving)
20:10:39	Araq	ldlework: there are implicit generics in Nim. so when you do:
20:10:46	Araq	type Foo[T] = object
20:11:04	Araq	proc p(f: Foo) # p is implicitly generic with parameter T
20:12:02	ldlework	interesting
20:12:20	onionhammer	is that true of objects too?
20:12:23	*	t4nk680 joined #nim
20:13:15	t4nk680	Hello, I am currently trying to use sfml with nimrod. I have installed bindings with nimble.
20:13:45	t4nk680	It compiles fine, but when I try and run it can't seem to find it 'could not load: libcsfml-window.so'
20:14:22	t4nk680	I have installed csfml and sfml
20:15:13	def-	t4nk680: is the libcsfml-window.so in /usr/lib, /usr/local/lib or somewhere?
20:15:15	t4nk680	and usr/lib/libcsfml-window.so exits in /usr/lib
20:15:21	t4nk680	Yes
20:16:05	def-	that's strange. does it work if you copy the libcsfml-window.so to your current directory?
20:16:27	t4nk680	Let me just try
20:16:42	def-	ah,and is the libcsfml-window for the right architecture? might be a 32bit/64bit problem
20:17:14	t4nk680	Still does not work when 64bit version into dir
20:17:46	matkuki	t4nk680: BlaXpirit_ can probably help, he's the SFML bindings creator.
20:17:56	t4nk680	Am I compiling the nimrod program with the wrong arch?
20:17:56	BlaXpirit_	hi
20:17:59	t4nk680	Hey
20:18:31	def-	t4nk680: you can check with the "file" program
20:18:37	BlaXpirit_	t4nk680, what is your distro?
20:18:46	t4nk680	Arch linux
20:18:48	BlaXpirit_	ok
20:18:53	BlaXpirit_	csfml is broken on arch linux
20:19:00	t4nk680	Ahh ok
20:19:19	BlaXpirit_	i wrote to package maintainer today
20:19:24	BlaXpirit_	he doesnt wanna take action
20:19:28	t4nk680	Maybe using aur package
20:19:34	t4nk680	Build from source?
20:19:38	BlaXpirit_	let's go to private message
20:19:47	t4nk680	OK
20:24:45	BlaXpirit_	but damn, sure is a ton of people using arch linux
20:25:48	t4nk680	Yeah, its kinda grown recently.
20:26:29	t4nk680	People moving away from bloated distros
20:32:59	dom96	Araq: What about idetools? How's work on them coming along?
20:33:32	Araq	dom96: as usual there is more pressing work tbd
20:33:36	Araq	:-(
20:33:45	Araq	but maybe next week
20:33:46	dom96	:\
20:37:59	*	matkuki quit (Quit: ChatZilla 0.9.91.1 [Firefox 34.0.5/20141126041045])
20:38:59	UberLambda	I'm on Arch after a lot of time on Linux Mint
20:39:20	UberLambda	Too much bloat, updates could kill the system and rolling release is really nice
20:39:24	Varriount	Araq: What is Nim's behavior with regard to 'undefined behavior' in the C/C++ backend?
20:40:28	Araq	UberLambda: I'm on Linux Mint with RR and I'm still waiting for the update that makes my X-window system from crashing hard
20:40:43	Araq	*makes it stop
20:41:04	Araq	Varriount: we try to avoid it, but currently not too hard.
20:42:15	UberLambda	Araq: try installing Arch on a VM, I think you'd like it
20:42:36	UberLambda	Setting it up requires some time but afterwards it doesn't take any long manteinance
20:42:52	Araq	yeah, been there, tried it, took too much time
20:43:03	UberLambda	I found that I had spent more time fixing Mint's update issues than setting up Arch
20:43:18	BlaXpirit_	yup
20:43:22	UberLambda	In the end, you can make a nice, fully-configured system in ~3 hours
20:43:28	Araq	I don't fix my linux setups anymore
20:43:39	UberLambda	You wipe & reinstall?
20:43:48	t4nk680	Think Arch's setup is complicated? try gentoo :)
20:44:01	Araq	I live with the bugs until it becomes too hard
20:44:08	Araq	then I wipe & reinstall
20:44:12	BlaXpirit_	since i've installed arch, i haven't reinstalled. enough said.
20:44:18	UberLambda	Araq: I think this way you lose a lot of productivity
20:44:24	joecodes	is there a website like npmjs.com for searching for nim packages? currently looking for a dns lib... want to be able to async send dns questions to particular servers for each question (each question to different server).
20:44:32	UberLambda	I've been running the same install for 3 years now
20:44:51	BlaXpirit_	joecodes, i think it's in the works
20:44:52	def-	joecodes: the packages should be in http://nim-lang.org/lib.html
20:45:03	BlaXpirit_	that's for now, i suppose
20:45:25	BlaXpirit_	search for DNS returns nothing
20:45:34	BlaXpirit_	this is your time to shine, make bindings xD
20:45:40	dom96	joecodes: You can search using Nimble: nimble search dns
20:46:33	joecodes	thanks. yeah no dns. I will install nimble.
20:46:35	def-	joecodes: for DNS I used this: http://nim-lang.org/rawsockets.html#getHostByName,string
20:46:39	*	Sphax joined #nim
20:46:40	Araq	UberLambda: most of the time I'm on windows so no, I don't lose a lot of productivity ;-)
20:48:05	joecodes	thanks def- I need to send the query to the authoritative nameserver for each query, so getHostByName won't work
20:48:52	UberLambda	Araq: I just couldn't move from a UNIX-like environment once having tried it
20:49:10	UberLambda	Also, Valve pushing a lot of games to Steam for Linux made my switch almost complete
20:49:15	Araq	I hate unix with a passion :-)
20:49:28	UberLambda	Araq: I hate Windows with a passion :P
20:49:52	joecodes	I use ubuntu and rarely have probs. chose it because there are so many answers to common problems out there.
20:50:35	*	gour quit (Quit: Leaving)
20:51:41	*	t4nk048 joined #nim
20:54:13	*	t4nk680 quit (Ping timeout: 246 seconds)
20:56:11	*	BlaXpirit_ was gonna write something, but rethought
21:00:49	joecodes	alrighty, am now nimble enabled :-). Someone needs to write a golang -> nim transpiler to get all those sweet sweet libraries :-)
21:02:10	Varriount	Meep.
21:02:29	Varriount	Araq: Apparently the testers are fixed, even without my patch.
21:15:38	Varriount	Araq: And I just made a change to the 32 bit portion of the registry that should prevent the WER dialogue from stalling the tester.
21:16:13	*	Sergio965 joined #nim
21:18:06	Varriount	Hello Sergio965
21:18:12	Sergio965	Hi!
21:19:03	*	Mat4 joined #nim
21:19:17	Varriount	Hey Mat4
21:19:30	Mat4	hi Varriount and all
21:21:52	*	t4nk048 quit (Ping timeout: 246 seconds)
21:22:05	*	gokr_ joined #nim
21:22:34	*	irrequietus joined #nim
21:24:03	*	jokra quit (Ping timeout: 256 seconds)
21:39:03	*	adam_s joined #nim
21:40:17	*	jokra joined #nim
21:45:07	*	gokr joined #nim
21:47:19	*	gokr_ quit (Ping timeout: 245 seconds)
21:48:18	*	irrequietus quit ()
21:48:34	*	irrequietus joined #nim
21:54:34	*	t4nk988 joined #nim
22:01:03	t4nk988	What is an ideal way to store loaded game levels from csv in nimrod?
22:01:49	t4nk988	seq[seq[]] as in y[x] ?
22:02:37	ekarlso-	t4nk988: is ti key value ?
22:02:58	t4nk988	Represents tile type
22:03:06	*	UberLambda quit (Quit: Leaving the Matrix)
22:03:17	t4nk988	for example 1 = solid
22:03:30	*	gokr1 joined #nim
22:03:34	EXetoC	yes. y[x] will yield a seq[]
22:03:50	dom96	use an enum for your tile types
22:04:07	t4nk988	dom96: Already done that :p
22:05:06	dom96	good
22:10:58	*	Demon_Fox joined #nim
22:11:58	*	Sergio965 quit (Remote host closed the connection)
22:12:15	flaviu	t4nk988: btw, seq[seq[...]] might not be ideal, you might want to use a single seq instead. Less indirection, and you can make a wrapper so that the syntax is similar.
22:12:37	def-	flaviu: if it's fixed size at least
22:12:42	t4nk988	How would I use a single seq flaviu ?
22:13:27	t4nk988	Ahh I understand
22:13:34	flaviu	t4nk988: A single seq of size x * y, with access to idx (x, y) being (y * width + x)
22:14:01	def-	like this: https://github.com/def-/nim-unsorted/blob/master/bitmap.nim#L39-L40
22:14:06	EXetoC	I don't think it will matter, but I don't bother with 2d arrays in such cases
22:14:23	t4nk988	What do you mean EXetoC?
22:14:54	flaviu	def-: You should put a licence file in that repo.
22:15:13	def-	flaviu: i don't care about licenses, do what you want with the code
22:15:13	flaviu	Otherwise, no one can use that code.
22:15:28	def-	flaviu: isn't it public domain without a license?
22:15:45	flaviu	def-: Just the opposite. If you don't have a licence, no one is allowed to touch it.
22:16:32	*	Sergio965 joined #nim
22:16:33	*	Sergio965 quit (Client Quit)
22:16:37	t4nk988	http://schep.me/license-checker/ def- flaviu
22:17:29	def-	i don't even want to read about licenses
22:18:01	def-	the code is mostly from rosetta code, so it's their license
22:18:23	t4nk988	def-: Its not, you enter your username and it checks to see if licenses are present in your repos
22:19:17	ekarlso-	is there like a DateTime equiv in nim ?
22:19:41	def-	ekarlso-: times?
22:19:49	ekarlso-	kewl :D
22:20:07	BlaXpirit_	def-, http://choosealicense.com/no-license/
22:20:38	t4nk988	Haven't seen that before, BlaXpirit_ thanks.
22:20:51	flaviu	def-: If you wrote the code, you decide the license. Rosetta code requires you license it under the GNU Free Documentation License 1.2, but the terms of that are too complicated for my tastes.
22:21:29	flaviu	def-: If you want something permissive, http://choosealicense.com/licenses/mit/ is a good choice.
22:22:53	ekarlso-	oh well, part 1 of JWx stuf done
22:22:56	ekarlso-	tmrw more ! :D
22:23:44	flaviu	ekarlso-: were you able to figure out why the C and nim code weren't matching up?
22:24:29	ekarlso-	flaviu: ya, what EXetoC said :)
22:24:52	*	t4nk988 quit (Ping timeout: 246 seconds)
22:25:28	flaviu	ekarlso-: No, I accidentally gisted the wrong file.
22:25:32	flaviu	https://gist.github.com/3e0b5a52fe9395fcf5c5 is the correct one
22:25:51	*	jokra quit (Ping timeout: 264 seconds)
22:26:02	ekarlso-	flaviu: yeah, but I managed to make it work before that too ;)
22:26:36	flaviu	ekarlso-: My problem is this: https://gist.github.com/flaviut/8e3464016063ea83f622
22:26:53	flaviu	The code seems identical, but the output is different.
22:27:04	ekarlso-	:/
22:27:24	ekarlso-	i guess the hash should be identical ?
22:27:50	flaviu	yep
22:28:01	BlaXpirit_	ekarlso-, but what if it uses gcc?
22:28:18	BlaXpirit_	nvm
22:28:19	ekarlso-	BlaXpirit_: ?
22:28:23	flaviu	exactly the same output
22:28:38	ekarlso-	flaviu: ping when u find the reason ;p'
22:29:11	ekarlso-	what's the equiv of atomic stuff in nim ?
22:29:14	ekarlso-	or locks
22:29:47	def-	http://nim-lang.org/locks.html
22:29:56	ekarlso-	kewl
22:30:16	*	Boscop__ joined #nim
22:30:17	ekarlso-	was gonna start a metrics library also a'la go-metrics or coda hale style to be exact
22:30:36	EXetoC	units?
22:30:52	ekarlso-	EXetoC: ? :Å
22:30:56	*	nande joined #nim
22:31:41	EXetoC	nvm
22:32:49	ekarlso-	EXetoC: like u have a registry ish thing where you can register diff types of metrics..
22:33:27	*	Boscop joined #nim
22:33:27	*	Boscop quit (Changing host)
22:33:27	*	Boscop joined #nim
22:33:39	*	adam_s quit (Ping timeout: 264 seconds)
22:33:59	*	Boscop_ quit (Ping timeout: 245 seconds)
22:35:15	flaviu	ekarlso-: a database?
22:36:26	ekarlso-	flaviu: https://github.com/rcrowley/go-metrics < like that
22:36:31	*	Boscop__ quit (Ping timeout: 272 seconds)
22:37:02	*	BlaXpirit_ quit (Quit: Quit Konversation)
22:39:19	*	yonson joined #nim
22:41:11	*	mwbrown joined #nim
22:41:49	mwbrown	Anyone here to answer a quick question regarding reference types?
22:42:44	dom96	mwbrown: There is lot's of people here. You should just ask your question instead of asking to ask.
22:42:46	def-	mwbrown: hi. a few are here, what's the question?
22:43:20	mwbrown	So on the tutorial (it's been a while since I touched Nim) here: http://nim-lang.org/tut1.html#reference-and-pointer-types
22:43:35	mwbrown	It declares both Node and NodeObj, NodeObj being "ref Node"
22:43:47	*	jokra joined #nim
22:44:00	mwbrown	it's easy to go from a NodeObj to a Node via new() and then accessing it normally, but is there a way to create a tracked reference from one that was NOT allocated via new?
22:44:12	mwbrown	i.e. you have a function that takes NodeObj but you only have a Node
22:44:44	mwbrown	I am assuming that any non-reference types are allocated either on the stack or in the .bss section
22:45:51	joecodes	dom96 how to set port for jester? argument to runForever?
22:45:53	ekarlso-	dom96: hirr would u be interested in jwt ?
22:45:55	mwbrown	I tried using addr but it complained about getting a "ptr" instead of a "ref"
22:46:31	dom96	joecodes: var settings = newSettings(port = Port(1234))
22:46:59	dom96	mwbrown: There is not.
22:47:04	dom96	ekarlso-: No, sorry.
22:47:17	ekarlso-	dom96: darn
22:47:21	mwbrown	dom96: in that case should I just use ptr types for those use cases?
22:47:30	joecodes	dom96: thanks!
22:47:43	*	Sphax quit (Quit: ZZZZZzzzzz)
22:48:12	dom96	mwbrown: what is your use case?
22:48:39	mwbrown	dom96: I want to statically allocate a structure (this is aimed for embedded) and want to pass it around by reference in the safest manner possible provided by Nim
22:48:41	*	Sphax joined #nim
22:48:44	dom96	joecodes: That will soon become: settings:\n port = TPort(1234).
22:48:49	dom96	or something similar.
22:48:52	ekarlso-	dom96: u not interested in web stuffs ?
22:49:04	dom96	ekarlso-: I'm not interested in JWT.
22:49:08	flaviu	ekarlso-: Just write the thing, if people are interested, they'll use it.
22:49:18	ekarlso-	okok :P
22:49:54	dom96	mwbrown: You want to allocate it on the stack?
22:50:08	joecodes	dom96: ok. will keep an eye out for it.
22:50:14	mwbrown	dom96: Either that or in the .bss like you would a static in C
22:50:21	ekarlso-	that's the sad thing though, if I create something that people dont wanna use then doh
22:50:58	dom96	mwbrown: You should use 'var Type' I think: http://nim-lang.org/tut1.html#parameters
22:51:01	dom96	Second example.
22:51:30	dom96	ekarlso-: What happened to you creating the packages site?
22:51:37	ekarlso-	dom96: still going
22:51:43	ekarlso-	dom96: need a token mechanism..
22:51:44	flaviu	ekarlso-: don't think that way, you're just setting yourself up to be disappointed.
22:52:06	Araq	mwbrown: use 'ptr' or 'var' it doesn't really matter. 'var' is a bit safer though.
22:52:06	dom96	If you're going to use it then that's enough reason to create a library for it.
22:52:13	onionhammer	build something *you* want to use ekarlso-
22:52:23	mwbrown	dom96, Araq: Thanks a lot!
22:52:33	flaviu	Exactly. If others find it useful then great!
22:55:03	*	Sphax quit (Quit: ZZZZZzzzzz)
22:55:38	mwbrown	yup, just confirmed that a top-level var is in the bss and a var inside a proc is stack-allocated
22:55:51	mwbrown	by printing the addresses. That works for me!
22:56:54	joecodes	dom96: so each route block is not async by default? so tried something like: await resp intToStr(fib(45)) but compile error.
22:57:23	dom96	joecodes: it is, you don't need 'await' in that case.
22:57:51	Araq	mwbrown: you can write C-like code very easily in Nim.
22:58:37	joecodes	hmmm. i tried and when long request (fib(45)) runs, it seems to block all other requests. this is from jester installed with nimble install jester
22:59:23	Araq	joecodes: because it's blocks your *cpu*.
22:59:41	Araq	this is not multi-threaded, async only helps with *io*
23:00:02	Araq	use spawn when you block your cpu
23:00:18	joecodes	ok thanks.
23:00:47	ekarlso-	Araq: how would you go about calling a proc every x seconds ?
23:01:07	ekarlso-	without blocking
23:01:27	dom96	while true + await sleepAsync
23:01:59	ekarlso-	does the proc you're calling then need to be async ?
23:02:17	dom96	yes
23:02:29	dom96	'await' needs to be in an async proc
23:03:15	*	matkuki joined #nim
23:03:21	EXetoC	can one make sure that no GC allocations are made? a simple hack perhaps
23:03:43	*	BitPuffin quit (Ping timeout: 245 seconds)
23:05:12	ldlework	ekarlso-: can you paste a simple example if you get that working
23:05:14	*	atomic-hedgehog quit (Ping timeout: 245 seconds)
23:05:28	ekarlso-	ldlework: hah, testing me or ? :p
23:06:00	ldlework	ekarlso-: I'm just curious
23:06:03	ekarlso-	;p
23:06:58	ekarlso-	ldlework: lemme see
23:07:10	Araq	EXetoC: you can use --gc:none and the compiler tells you
23:08:10	EXetoC	yeah I did something wrong
23:11:03	*	matkuki quit (Quit: ChatZilla 0.9.91.1 [Firefox 34.0.5/20141126041045])
23:12:46	*	Sphax joined #nim
23:13:05	*	Sphax quit (Client Quit)
23:15:36	*	Hakaslak joined #nim
23:16:34	ekarlso-	dom96: I was wondering if we could have nimble at a point have a register package call ?
23:16:44	dom96	sure
23:17:23	ekarlso-	and a release also (version etc)
23:20:51	*	mwbrown quit (Ping timeout: 252 seconds)
23:21:04	*	irrequietus quit ()
23:21:43	*	Demon_Fox quit (Quit: Leaving)
23:21:54	*	adam_s joined #nim
23:24:11	*	sillesta quit (Ping timeout: 256 seconds)
23:24:43	Varriount	Hm.. Should I put in the "Common Conventions and Practices" guide to use something like koch for complicated make-like configurations?
23:25:05	Varriount	I mean, the compiler does it, and it works fairly well...
23:26:40	fowl	koch should be a nakefile
23:31:56	onionhammer	you should work on nimlime :p
23:35:11	Varriount	onionhammer: A) What should I do, B) I would, if you can do my math homework for me.
23:37:47	EXetoC	seems like a fair tradeoff
23:37:49	flaviu	Varriount: pfft, math.
23:38:32	flaviu	who needs that?
23:39:26	*	adam_s left #nim ("Leaving")
23:45:24	onionhammer	Varriount semi-kidding. The main thing that needs to happen is idetools
23:51:07	EXetoC	flaviu: noobs