<< 13-04-2015 >>

---

00:03:12	cazov	ahh, yeah, they were introduced in vista and they do look to be C++. pretty unfortunate but not very surprising
00:03:53	cazov	MS doesn't really seem to care much about people who develop for windows and don't use VS
00:04:07	Varriount	cazov: Although, that's the only occurence of a C++ api set that I've seen
00:04:33	Varriount	I can only hope that someone realized the folly of creating a C++ API for an operating system
00:04:34	*	HakanD___ quit (Quit: Be back later ...)
00:05:30	cazov	yeah, i dunno. hopefully
00:08:49	cazov	it looks like they have some other stuff that's c++ -or- CLR, like the iTV api and GDI+
00:09:09	Varriount	Ugh
00:13:51	*	bcinman quit (Quit: My Mac has gone to sleep. ZZZzzz…)
00:27:00	*	bcinman joined #nim
00:31:49	*	pregressive joined #nim
00:37:47	*	saml_ joined #nim
00:39:41	*	brson joined #nim
00:43:47	*	Varriount quit (Ping timeout: 276 seconds)
00:44:04	*	Varriount joined #nim
00:55:45	*	lepot joined #nim
00:57:20	lepot	can someone please give me a hint as to why I can't assign to self.drinks in this example? https://dpaste.de/wJPO/raw
01:02:55	lepot	anyone used nim before?
01:04:46	flaviu	lepot: Parameters are immutable.
01:05:24	lepot	flaviu: thanks for replying. so I can't have something like an instance variable?
01:05:45	flaviu	In lots of languages, every type is a pointer, but in Nim you have to explicitly make it a pointer.
01:06:01	flaviu	lepot: Sure. `type Person = ref object`.
01:06:32	lepot	flaviu: oh! awesome! thanks!
01:06:46	lepot	flaviu: I'll go back and give the docs a closer read
01:10:28	*	bcinman quit (Quit: My Mac has gone to sleep. ZZZzzz…)
01:15:56	*	lepot left #nim (#nim)
01:20:16	*	MightyJoe quit (Ping timeout: 240 seconds)
01:24:25	*	MightyJoe joined #nim
01:24:27	*	vendethiel joined #nim
01:32:56	*	TEttinger quit (Ping timeout: 240 seconds)
01:33:40	*	bcinman joined #nim
01:36:47	*	MightyJoe quit (Ping timeout: 250 seconds)
01:37:15	*	HakanD___ joined #nim
01:40:15	*	MightyJoe joined #nim
01:42:00	*	HakanD___ quit (Ping timeout: 256 seconds)
02:00:30	*	brson quit (Ping timeout: 252 seconds)
02:02:53	*	brson joined #nim
02:14:46	*	brson quit (Quit: leaving)
02:35:11	*	BitPuffin quit (Ping timeout: 252 seconds)
02:36:24	*	darkf joined #nim
02:42:04	*	pregressive quit ()
03:11:56	*	onionhammer joined #nim
03:26:10	*	HakanD___ joined #nim
03:30:51	*	saml_ quit (Remote host closed the connection)
03:31:01	*	HakanD___ quit (Ping timeout: 264 seconds)
03:36:20	*	gsingh93 joined #nim
03:58:41	*	endragor joined #nim
04:08:04	*	vendethiel quit (Ping timeout: 272 seconds)
04:08:38	*	Senketsu quit (Quit: Leaving)
04:09:33	*	vendethiel joined #nim
04:14:30	*	Senketsu joined #nim
04:30:31	Joe-T	is there a built in iterator for
04:30:36	Joe-T	strings to chars?
04:32:53	*	ChrisMAN quit (Ping timeout: 248 seconds)
04:36:02	fowl	Joe-T, items
04:37:07	Joe-T	ah thanks, that worked
04:37:26	Joe-T	I assumed it didn't as I tried to echo it and syntastic complained
04:38:26	MagusOTB	how do I make a generic that takes an array?
04:40:45	MagusOTB	I have this http://pastebin.com/vnYSJqNh and I don't know how to make a generic proc that the last line will call.
04:41:58	*	HakanD___ joined #nim
04:46:14	*	HakanD___ quit (Ping timeout: 252 seconds)
04:55:59	*	vendethiel quit (Ping timeout: 276 seconds)
04:57:52	fowl	MagusOTB, proc test[T: not openarray](val: T)
04:59:05	MagusOTB	oh cool, thanks
04:59:22	MagusOTB	is there docs on other thing syou can do to constrain types in generics?
05:02:08	fowl	http://nim-lang.org/manual.html#type-classes
05:38:46	*	gsingh93 quit (Ping timeout: 244 seconds)
05:39:27	*	wink-s quit (Quit: Page closed)
05:42:46	*	HakanD___ joined #nim
05:46:56	*	bcinman quit (Quit: My Mac has gone to sleep. ZZZzzz…)
05:47:04	*	HakanD___ quit (Ping timeout: 250 seconds)
06:25:31	*	bcinman joined #nim
06:31:07	*	awesomo4000 joined #nim
06:35:26	*	bcinman quit (Quit: My Mac has gone to sleep. ZZZzzz…)
06:38:13	*	bcinman joined #nim
06:38:42	awesomo4000	anyone know how to compile nim itself against musl-libc
06:49:27	fowl	awesomo4000, i guess you would do it similar to this http://www.schipplock.software/2015/02/static-linking-with-nim.html
06:49:46	fowl	try passing those options to koch
06:49:58	*	endragor quit (Ping timeout: 245 seconds)
06:50:35	*	adam12 quit (Ping timeout: 244 seconds)
06:51:53	*	adam| joined #nim
06:55:04	*	endragor joined #nim
06:55:55	*	Kingsquee quit (Quit: Konversation terminated!)
07:02:39	*	Ven joined #nim
07:03:23	*	vendethiel joined #nim
07:05:30	*	endragor_ joined #nim
07:05:41	*	HakanD___ joined #nim
07:07:49	*	Ven quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
07:08:37	*	endragor quit (Ping timeout: 248 seconds)
07:14:37	*	Ven joined #nim
07:18:52	*	HakanD___ is now known as HakanD
07:21:46	*	bcinman quit (Quit: My Mac has gone to sleep. ZZZzzz…)
07:24:31	*	endragor_ quit (Remote host closed the connection)
07:26:16	*	vendethiel quit (Ping timeout: 240 seconds)
07:26:56	*	Trustable joined #nim
07:29:18	*	bjz joined #nim
07:39:58	*	Ven quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
07:40:32	*	BlaXpirit joined #nim
07:46:06	*	Ven joined #nim
07:48:34	*	HakanD_ joined #nim
07:51:37	*	HakanD quit (Ping timeout: 256 seconds)
08:07:52	*	coffeepot joined #nim
08:08:37	*	bjz quit (Quit: My MacBook Pro has gone to sleep. ZZZzzz…)
08:09:36	*	awesomo4000 left #nim (#nim)
08:15:13	*	Arrrrrrr joined #nim
08:15:21	Arrrrrrr	Hello Nimrod.
08:15:45	coffeepot	Good morning :)
08:16:08	Arrrrrrr	How is your game going on coffeepot ?
08:16:50	coffeepot	right now, slowly - haven't had time to do anything on it for the past few days
08:17:02	coffeepot	working on physics atm tho
08:17:35	Arrrrrrr	I made this http://i.imgur.com/h2V0o7c.gif
08:17:43	coffeepot	how's your stuff going? What are your thoughts of nim so far?
08:18:17	coffeepot	awesome :)
08:19:17	Arrrrrrr	Good, i get a lot of help here. For now i'm just trying to settle down everything that i learned. But there's so much.
08:21:43	*	Ven quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
08:22:27	Arrrrrrr	Btw, is 'generic' and 'constraint' the 'interface' of nim?
08:25:01	coffeepot	yeah I'm still really just getting my feet wet in nim - still a lot to learn for me, especially macros and things
08:25:27	coffeepot	as for your question, I don't know. Hoepefully someone else with more knowledge can answer
08:25:48	HakanD_	Arrrrrrr: No, no, hell no
08:25:52	HakanD_	Don't go that way (:
08:26:39	Arrrrrrr	Hah, so?
08:26:43	HakanD_	They look like interfaces, but sadly they don't work that way, they are a different thing all together, at least in it's curren state
08:26:57	Arrrrrrr	will they evolve then?
08:27:07	HakanD_	No idea.
08:27:32	HakanD_	Maybe, if enough people bugs Araq :P
08:27:45	coffeepot	what's the intended purpose of constraints?
08:27:55	coffeepot	to constrain generics?
08:28:11	coffeepot	for params, etc?
08:28:22	HakanD_	constraints for generics
08:28:35	coffeepot	ah
08:28:43	*	bjz joined #nim
08:29:23	HakanD_	They are awesome, and opens of lots of interesting options. I really wish they also worked like Go's interfaces tho.
08:30:22	coffeepot	i'm sure someone is working on adapting constraints to be interfaces, I think there was a discussion about it the other day
08:30:37	HakanD_	some topics if you are interested: http://forum.nim-lang.org/t/673 http://forum.nim-lang.org/t/282 http://forum.nim-lang.org/t/754/1 http://forum.nim-lang.org/t/278/1
08:31:37	*	Ven joined #nim
08:31:51	HakanD_	there is something fowlmouth posted here: https://gist.github.com/fowlmouth/ac51baef7956af1ebdf8
08:32:20	Arrrrrrr	Thanks HakanD_
08:32:26	coffeepot	hakenD_ that's what I was thinking of :)
08:36:15	*	bjz quit (Ping timeout: 250 seconds)
08:37:27	coffeepot	what Araq mentions in one of your links is pretty interesting - using closures. I guess that'd be much more adaptable than interfaces
08:38:58	*	milosn quit (Ping timeout: 255 seconds)
08:41:22	HakanD_	As lots of people are looking for interfaces in Nim, I'm sure some proper way to do them will emerge, either baked in to the language or in userland
08:41:52	HakanD_	bbl
08:41:54	*	HakanD_ quit (Quit: Be back later ...)
08:42:11	coffeepot	yeah I was looking at the closures equivalent and thinking that someone will macro-ise this sooner or later
08:42:42	coffeepot	one of the reasons why I chose nim out of the newer languages was because it's so extensible
08:42:55	coffeepot	some degree of future proofing
08:44:11	coffeepot	tbh tho nim kinda ruins me for other languages for simple things like templating. Kind of like super-inlining really, saves bugs, increases run time speed and reduces reading complexity
08:45:42	coffeepot	as an aside from game programming I'm looking at a native sql library
08:46:45	coffeepot	since I do a lot of sql server at work. Actually that reminds me, must look at nim's rtti capabilities to see how easy it'd be to create a simple ORM framework
08:49:15	reactormonk	coffeepot, what would you like the API to look like?
08:49:32	coffeepot	well you know, that's what i was wondering
08:50:25	coffeepot	i wonder if it'd be possible to attach some kind of trait to an object's fieldname that represents the fieldname in the db and uses that access things
08:50:36	coffeepot	it depends how flexible and fast the rtti is
08:50:41	reactormonk	rtti?
08:50:47	coffeepot	run time type information
08:50:52	reactormonk	none.
08:50:53	coffeepot	reflection
08:51:15	reactormonk	nim doesn't have a VM, so you can't do reflection.
08:52:10	coffeepot	ideally i'd like to be able to define an object and tag each field with it's db fieldname, perhaps tag the object as a table, and be able to query the db by making a seq of clauses
08:52:24	coffeepot	not even static reflection?
08:52:31	reactormonk	that's macro.
08:52:49	coffeepot	i thought it might involve some macro wizardry :)
08:53:41	coffeepot	queryseq.addclause(myobject.id, clauseEquals, 17) or something
08:54:21	coffeepot	and queryseq.addclause(myobject.description, clauseLike, "%test%")
08:54:31	reactormonk	I'd prefer something like select[Foo].where(bar == 2))
08:54:45	coffeepot	yeah that's nice
08:55:32	coffeepot	how would you do multiple clauses though?
08:55:38	coffeepot	in that syntax
08:55:50	BlaXpirit	chain more onto it
08:55:58	reactormonk	select[Foo].where(bar == 2).group_by("baz")
08:55:59	coffeepot	fair enough :)
08:57:19	reactormonk	just make sure to handle TaintedString correctly
08:57:23	coffeepot	how would you pass "bar == 2" as a parameter and have it be equate to the field bar - pass as an expression?
08:57:45	reactormonk	pretty much
08:58:00	coffeepot	yeah I remember seeing this http://nim-lang.org/manual.html#avoiding-sql-injection-attacks
08:58:00	emilsp	aren't there predicates in nim ?
08:58:11	reactormonk	emilsp, what's a predicate?
08:58:17	*	endragor joined #nim
08:58:21	*	endragor quit (Remote host closed the connection)
08:59:06	*	endragor joined #nim
09:00:36	Arrrrrrr	(x: T) -> bool
09:00:38	Arrrrrrr	i believe
09:01:22	coffeepot	I guess with expression parameters you don't need reflection, and there's no speed penalty. However not sure if that would work if you wanted to dynamically query at runtime, would it? As in, if a user gets to select the field and data to search on
09:02:47	*	bjz joined #nim
09:04:02	reactormonk	you can't go for dynamic queries anyway on an ORM
09:04:07	*	Demon_Fox quit (Quit: Leaving)
09:04:15	coffeepot	why not?
09:04:38	reactormonk	depends on your ORM I guess. but when the select return type is typed, you're kinda set
09:05:39	coffeepot	i wrote a simple orm that does dynamic user queries in delphi for an app at work, I'm sure nim can do it better
09:05:54	BlaXpirit	.eval macro test(e: varargs[expr]): expr = e; discard test(a)
09:05:55	reactormonk	yup, but you probably can't type them in an ORM way
09:05:56	Mimbus	BlaXpirit: eval.nim(4, 13) Error: undeclared identifier: 'a'
09:05:59	BlaXpirit	why?
09:07:15	BlaXpirit	well that's actually a bad example
09:07:48	BlaXpirit	.eval import macros; macro test(e: expr): expr = bindSym"true"; echo test(a)
09:07:50	Mimbus	BlaXpirit: true
09:07:55	BlaXpirit	.eval import macros; macro test(e: varargs[expr]): expr = bindSym"true"; echo test(a)
09:07:56	Mimbus	BlaXpirit: eval.nim(5, 10) Error: undeclared identifier: 'a'
09:09:19	*	akiradeveloper joined #nim
09:11:11	coffeepot	In delphi I just parsed the object fields at start up (using traits on the fields to associate the object fields with the DB fieldnames) and allowed users to select from this list, then I just construct the sql myself
09:17:51	*	HakanD_ joined #nim
09:22:36	*	vendethiel joined #nim
09:26:29	Arrrrrrr	I read the links about type classes and i have a light idea of what they are, but still can't see what do people when they want interfaces. Real scenarios. I'm so used to them i cannot think without them
09:27:27	reactormonk	Arrrrrrr, a typeclass is basically a definition of an interface, except you don't attach the implementation to the object.
09:28:45	HakanD_	Arrrrrrr: tuple of closures work, but they require a slightly different type of thinking
09:29:37	HakanD_	https://gist.github.com/fowlmouth/ac51baef7956af1ebdf8 this might give some ideas
09:31:29	HakanD_	it took me a few days to grok nim way of thinking regarding to interfaces.
09:32:00	HakanD_	still not sure if i really understood totally tho. still hacking on stuff.
09:32:06	Arrrrrrr	But does it reduce the number of lines one needs to produce interfaces?
09:32:22	Arrrrrrr	or it is more like "better than nothing"
09:32:41	*	Ven quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
09:33:05	HakanD_	not an expert, but it looks to me is the way forward regarding interfaces
09:33:14	HakanD_	I still haven't dived into macros and stuff
09:33:44	Arrrrrrr	I haven't seen anythig about concepts
09:34:40	reactormonk	I think we need some more concepts in the stdlib
09:35:20	BlaXpirit	yes
09:35:34	BlaXpirit	down with openarray
09:35:40	coffeepot	haha
09:35:40	HakanD_	btw, 'MyConcept: generic x' is now 'MyConcept: concept x'
09:36:43	Arrrrrrr	Ah, ok ok
09:36:48	Arrrrrrr	Now i see it
09:36:55	fowl	the idea is that we can get an interface type from a concept that defines functions like draw(obj); shouldDelete(obj) is bool; ...
09:37:33	Arrrrrrr	So, does your example work out of the box with 0.10.3 ?
09:38:08	BlaXpirit	fowl, why is such a type needed?
09:38:24	fowl	Arrrrrrr, it does
09:40:11	fowl	BlaXpirit, i dunno, kind of cool to show that it can be implemented without a compiler extension, working at compile time and run time to provide polymorphism
09:40:37	coffeepot	do concepts have any overhead?
09:40:47	*	milosn joined #nim
09:41:45	BlaXpirit	i bet they do have a considerable compile time overhead
09:42:08	fowl	coffeepot, if you have a large generic library and it gets instantiated a couple times your exe would be bigger
09:42:31	coffeepot	compile time or file size i'm not bothered about, just wondering if i could refactor my gaming stuff to use them
09:43:21	fowl	they would come in handy in places where you would use multiple inheritance in c++
09:43:24	*	milosn quit (Read error: Connection reset by peer)
09:44:04	fowl	in fact you can use generics in place of multiple inheritance in cpp wrappers
09:44:11	coffeepot	the example you give fowl is actually the one i was thinking of: drawable (also has physics etc)
09:44:17	*	vendethiel quit (Ping timeout: 250 seconds)
09:45:50	*	milosn joined #nim
09:47:07	Arrrrrrr	I can only pray for a simpler way to introduce interfaces in the future.
09:47:41	BlaXpirit	this is the simplest way
09:48:03	HakanD_	fowl: the geninterface macro still doesn't work on its own tho
09:48:05	Arrrrrrr	I will pray harder
09:48:12	HakanD_	(:
09:48:51	fowl	HakanD_, i know, when i print out the code and paste into the file it works, need to examine the ast
09:54:36	*	akiradeveloper quit (Read error: Connection reset by peer)
10:01:59	*	irrequietus joined #nim
10:02:47	BlaXpirit	I have tried to implement this bottom example here https://github.com/flaviut/nre/issues/11#issuecomment-92144985 but have run into a problem that you can't pass unexisting identifiers to varargs macro, although without varargs/commas it works.
10:02:55	BlaXpirit	.eval import macros; macro test(e: varargs[expr]): expr = bindSym"true"; echo test(a)
10:02:56	Mimbus	BlaXpirit: eval.nim(5, 10) Error: undeclared identifier: 'a'
10:03:05	BlaXpirit	fowl, maybe you know something about this?
10:03:06	*	milosn quit (Write error: Connection reset by peer)
10:07:20	*	milosn joined #nim
10:08:06	fowl	BlaXpirit, i think its a bug because expr is supposed to be untyped
10:08:29	fowl	so it doesnt need to be semantic'd
10:08:51	BlaXpirit	how would you name such a bug?
10:09:11	fowl	you can just make the macro immediate
10:09:29	BlaXpirit	no, actually i can't
10:09:52	fowl	i dunno
10:09:54	BlaXpirit	actually is macro `{}`*(pattern: Regex, options: varargs[expr]): expr =
10:09:57	fowl	naming is half the battle
10:17:22	*	milosn quit (Read error: Connection reset by peer)
10:17:41	*	milosn joined #nim
10:21:54	*	milosn quit (Ping timeout: 244 seconds)
10:30:01	*	milosn joined #nim
10:32:15	*	BitPuffin joined #nim
10:40:29	*	banister quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
10:48:23	*	Ven joined #nim
10:49:19	*	Ven quit (Client Quit)
10:51:19	*	milosn quit (Read error: Connection reset by peer)
10:51:42	*	milosn joined #nim
10:56:08	*	coffeepot quit (Quit: http://www.kiwiirc.com/ - A hand crafted IRC client)
10:57:17	*	coffeepot joined #nim
10:58:29	*	milosn quit (Read error: Connection reset by peer)
10:58:55	Arrrrrrr	Right now, are contributors working into introducing a 'major' feature? or just bugs fixing?
10:59:01	*	milosn joined #nim
10:59:36	BlaXpirit	right now is the worst time for major features
11:04:33	Arrrrrrr	And is something planned for the future?
11:04:46	Arrrrrrr	*anything
11:05:27	*	milosn quit (Read error: Connection reset by peer)
11:05:59	dtscode	BlaXpirit, why is that?
11:06:26	BlaXpirit	next release is sorely needed
11:06:31	*	bogen joined #nim
11:06:45	*	milosn joined #nim
11:06:55	BlaXpirit	a low percentage of my code compiles in 0.10.2
11:08:27	*	untitaker quit (Ping timeout: 276 seconds)
11:13:08	BlaXpirit	btw maybe 0.11.0 is warranted
11:14:15	*	untitaker joined #nim
11:18:49	*	milosn quit (Read error: Connection reset by peer)
11:20:44	coffeepot	yeah current version on website is well old now, in nim terms
11:21:23	HakanD_	well, new release also means updating the docs
11:22:21	*	nulpunkt left #nim (#nim)
11:35:26	*	Ven joined #nim
11:36:50	*	saml_ joined #nim
11:43:25	*	milosn joined #nim
11:52:32	Arrrrrrr	"in nim terms" haha
11:54:17	*	milosn quit (Read error: Connection reset by peer)
11:55:22	*	milosn joined #nim
12:02:37	*	milosn quit (Read error: Connection reset by peer)
12:03:01	*	milosn joined #nim
12:03:58	Arrrrrrr	It is possible to breakthrough in case statements?
12:04:18	BlaXpirit	Arrrrrrr, i remember something like that
12:04:45	BlaXpirit	might have been from a different language :p
12:05:00	Arrrrrrr	Hah, no wonder
12:05:02	def-	go has an explicit fallthrough
12:05:30	BlaXpirit	wow, nim doesn't
12:05:30	def-	in nim you can list multiple cases, but i don't think we have a fallthrough
12:05:39	def-	case 2, 5, 10..20:
12:05:45	Arrrrrrr	I see.
12:05:59	Arrrrrrr	That is quite usefull, but not the same.
12:06:39	*	milosn quit (Read error: Connection reset by peer)
12:08:02	*	milosn joined #nim
12:08:09	coffeepot	had too look up what case fallthrough was :D why would you even want this?
12:08:18	coffeepot	or rather why would people want this
12:08:46	*	BitPuffin quit (Ping timeout: 272 seconds)
12:09:47	coffeepot	aside from the duff engine thing that more or less sounds like a hack
12:09:51	BlaXpirit	http://programmers.stackexchange.com/questions/116221/appropriate-uses-of-fall-through-switch-statements
12:10:01	dom96	duff's device mainly benefits from it
12:10:27	coffeepot	thanks BlaXpirit that explains it quite well :)
12:11:14	Arrrrrrr	could be done with an "and": http://pastebin.com/VEiuTg7C
12:11:30	BlaXpirit	naaaaah
12:11:33	coffeepot	ok so i only read about duffs device today but sounds like it wouldn't provide much performance with modern hardware
12:11:38	BlaXpirit	case 2..10 {.fallthrough.}
12:11:47	BlaXpirit	much more in the spirit of the language
12:12:14	Arrrrrrr	but that's longer.
12:12:23	dom96	yeah, it's a pity that most languages optimise their syntax for the edge case here.
12:12:25	Arrrrrrr	and "and" is there.
12:12:27	BlaXpirit	well it should be as long as reasonably possible
12:12:34	BlaXpirit	Arrrrrrr, you don't get it :|
12:12:59	Arrrrrrr	sorry
12:13:17	BlaXpirit	{.intentionalFallthough.} is pretty good too
12:13:30	BlaXpirit	well ok it's not
12:14:13	Arrrrrrr	c# has goto if i recall well
12:14:29	BlaXpirit	that can't be right
12:14:43	*	milosn quit (Read error: Connection reset by peer)
12:14:46	BlaXpirit	oh it does :|
12:15:17	Arrrrrrr	Hah, "goto case x"
12:15:18	*	milosn joined #nim
12:15:40	BlaXpirit	is actually pretty good
12:15:58	coffeepot	i'm now in two minds about duffs device - on the one hand it is really slick, on the other hand it makes me recoil in buggy terror
12:16:20	BlaXpirit	geez, i'll have to google this
12:16:55	*	saml_ quit (Quit: Leaving)
12:26:20	*	milosn quit (Read error: Connection reset by peer)
12:26:43	*	milosn joined #nim
12:33:20	Arrrrrrr	BlaXpirit: in which languages have you programmed before?
12:35:25	*	vendethiel joined #nim
12:38:13	*	milosn quit (Read error: Connection reset by peer)
12:40:44	*	Arrrrrrr quit (Quit: Page closed)
12:45:13	coffeepot	is nim's approach to GC unique in the programming world? Seems most languages that use GC use the boehm type. Also: is nims GC kind of like what swift calls ARC?
12:49:33	*	milosn joined #nim
12:52:55	emilsp	isn't sphinxs GC more like C++ shard_ptr approach?
12:53:19	emilsp	tfu, not sphinx's but nim's
12:53:56	coffeepot	i didn't think it was, as the gc runs thread local and you need to put extra work in to share GC'd objects across threads.
12:54:37	coffeepot	as i understand it shared_ptr has more overhead as it has to be shared between threads?
12:54:49	emilsp	yes
12:55:11	*	milosn quit (Read error: Connection reset by peer)
12:55:19	coffeepot	the reason i ask is because i was reading this http://sebastiansylvan.com/2015/04/13/why-most-high-level-languages-are-slow/
12:55:36	coffeepot	and thought that his comments might not apply to nim's GC
12:55:54	*	milosn joined #nim
12:57:03	coffeepot	i assume he hasn't heard of nim since he comments "I’m relatively convinced that you could write a GC more suitable for high performance and low latency applications (e.g. an incremental GC where you spend a fixed amount of time per frame doing collection)" which AIUI is what you can do with nim's GC
12:58:02	coffeepot	anyway, it's a common comment for some people "GC = slow" without considering that there's different types of GC - though I'm sure the writer knows more about this than I do
12:59:37	*	vendethiel quit (Ping timeout: 248 seconds)
13:01:33	*	Ven quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
13:04:02	*	vendethiel joined #nim
13:07:42	emilsp	to be honest, nims memory model doesn't make allocating stuff cheap either, does it ?
13:08:26	Varriount	emilsp: Not as cheap as, say, Java's allocator
13:08:29	coffeepot	emilsp, i'd be interested to know this
13:09:08	Varriount	Java (and other languages) can move objects around in memory
13:09:20	Varriount	Which enables them to use a slab allocator
13:09:52	Varriount	The cost comes in the form of GC and runtime complexity.
13:10:16	emilsp	yes, indeed. How would one deal with the cost of allocations in nim ? Construct used stuff as far top as possible ?
13:11:31	Varriount	emilsp: Huh? What do you mean?
13:12:18	emilsp	Well, do stupid stuff such as use a massive object to hold intermediate strings in ?
13:12:43	Varriount	emilsp: Well, try to reuse memory.
13:13:10	emilsp	yes
13:13:26	Varriount	emilsp: You'll have to ask Araq about the particulars of memory allocation.
13:13:44	emilsp	I wish I had more time on my hands to do something awesome in nim
13:14:24	Varriount	I know very little concrete information. I know that it uses memory maps, it's primarily refcount based, and it hasn't been much of a problem
13:16:27	Varriount	emilsp: Also, though Java might have faster allocation times, it's still very memory hungry.
13:17:19	emilsp	Varriount, java is memory hungry and slow until you warm it up. But don't worry, I'd never use Java if I don't have to
13:17:48	*	Varriount looks at the 1.5 GB of ram taken up by IntelliJ Idea
13:18:34	*	milosn quit (Ping timeout: 245 seconds)
13:18:47	emilsp	>not using vim
13:18:55	Varriount	emilsp: And (although this a completely uninformed guess) I'd wager that the cost of allocations isn't too much higher than that of standard malloc in C/C++
13:19:52	*	Ven joined #nim
13:20:16	emilsp	Varriount, I'd guess there should be no oberhead when compared to c/c++
13:24:40	*	vendethiel quit (Ping timeout: 256 seconds)
13:33:50	*	vendethiel joined #nim
13:33:59	coffeepot	i've been very impressed by the speed of nim without having to do anything particularly optimise-y
13:35:14	coffeepot	to be honest I'm always amazed by how fast java is for an 'interpreted' language
13:35:51	repax	Java is commonly jitted
13:36:01	coffeepot	but you know, there's the joke about "enterprise fizzbuzz" that seems to apply to most java
13:36:38	coffeepot	i find it impressive how effective on the fly jitting is
13:37:37	*	milosn joined #nim
13:38:03	repax	I don't dare guess how many hundreds of man-years have gone into the development of Java
13:38:23	coffeepot	yeah it's a pretty huge language
13:39:15	repax	I think nim has found a sweet niche, though
13:39:19	coffeepot	definitely
13:39:34	*	BitPuffin joined #nim
13:40:27	coffeepot	i was looking for ages for a new language and before nim it was "fast, powerful, sane - choose two"
13:40:40	HakanD_	^
13:40:52	coffeepot	i think nim *finally* provides all three
13:41:01	repax	Indeed
13:41:29	repax	Also, productive
13:41:33	coffeepot	yeah
13:41:54	coffeepot	i've heard it said many times with regard to nim and I agree: "nim makes me love programming again"
13:42:14	repax	:)
13:43:01	a5i	Crystal and Nim are the only 2 languages that provide those traits as of now
13:43:37	*	ChrisMAN joined #nim
13:43:48	coffeepot	i remember the first time I tried to write something with python at work and it was fairly productive, then I realised in order to distribute it to collegues they would have to carry around a python interpretter o_O Eventually got it compiles into one file but it was just so disappointing
13:44:10	coffeepot	crystal is definitely on my radar but it's a bit of a case of nim got there first and is further along
13:44:41	a5i	Yeah
13:44:43	repax	Haven't had the time yet to look at crystal
13:45:12	repax	Anything in particular it does better than nim?
13:45:14	coffeepot	also I really like Araq's focus on efficiency and elegance
13:46:06	coffeepot	crystal: "Ruby-inspired syntax" that's actually a turn off fo rme
13:46:08	coffeepot	for me
13:46:50	a5i	Nim and Crystal are roughly the same, one compiles to C and the other to LLVM
13:47:54	a5i	They both have amazing native speed, syntactic sugar and low-level functionalites
13:48:00	a5i	like C bindings
13:48:04	repax	I used to like C++ for its goal of zero-cost abstractions. In the end any large piece of software is a heap of abstractions. Writing in c++ is not productive though
13:48:43	repax	Nim might someday go straight to llvm
13:49:36	a5i	I'm myself love Ruby syntax, the better thing is that they fix some of Ruby's issues
13:49:53	coffeepot	repax: totally agree. in C++ I feel drunk with power at the freedom, but then fall down drunk with freedom
13:49:54	repax	I don't see that as important right now though because the C and C++ back ends are both powerful and portable
13:50:45	coffeepot	my time with nim has left me wondering why do we need all these pointless symbols everywhere in most languages
13:52:33	*	detox joined #nim
13:53:46	a5i	detox is an irc bot written in crystal, it has an eval function I would use to compare behaviors between Mimbus and detox
13:54:06	a5i	!eval p "test"
13:54:07	detox	"test"
13:54:32	a5i	.eval "tesT"
13:54:34	Mimbus	a5i: eval.nim(3, 0) Error: value of type 'string' has to be discarded
13:54:41	a5i	.eval echo "tesT"
13:54:42	Mimbus	a5i: tesT
13:55:09	BlaXpirit	a5i pls
13:55:28	BlaXpirit	not cool
13:55:31	a5i	D:
13:56:01	BlaXpirit	even Mimbus is not really welcome here, and this...
13:56:05	emilsp	repax, but crystal isn't even typed
13:56:30	repax	I like typed languages though
13:56:45	a5i	emilsp: What do you mean it isnt typed ?
13:57:07	emilsp	>Never have to specify the type of a variable or method argument.
13:57:20	emilsp	this is why I am starting to get sick of python
13:57:34	a5i	You don't have to, but you can
13:57:36	coffeepot	i was trying to find more info on crystals type system myself...
13:57:54	coffeepot	i have developed a real aversion to dynamic typing
13:58:12	coffeepot	thanks to python
13:58:24	a5i	!eval x = 35.to_i64; p x.is_a?(Int64)
13:58:25	detox	true
13:58:33	a5i	Types ^
13:58:37	repax	It's too easy to mess things up with dyn typing
13:59:01	a5i	It's type inferencing, you can declare it if you want to
13:59:42	*	bcinman joined #nim
14:00:09	coffeepot	yeah i remember the first time i started db access in python and was like "wait wtf is this returning, how does any of this work?" and the answer seems to be "you just hope the caller sends you the right bindings"
14:00:10	repax	But worse than that are languages with automatic var declarations, i.e. no declarations
14:00:31	*	clynamen_ joined #nim
14:00:43	coffeepot	repax: like old BASIC? :D
14:00:49	a5i	!eval p StringIO{1,2,3}
14:00:50	detox	#<StringIO:0xCACFC0 @buffer=Pointer(UInt8)@CADFA0, @bytesize=3, @capacity=64, @pos=0>
14:01:05	repax	Yeah, or just php
14:01:05	*	bcinman quit (Client Quit)
14:01:17	BlaXpirit	a5i, seriously, please stop
14:01:37	coffeepot	thankfully never done any php. Was going to learn a few years back. Feel like I've saved myself some sanity!
14:01:42	BlaXpirit	(i'm not any kind of authority here, but whatever)
14:01:53	*	Jehan_ joined #nim
14:01:57	*	bcinman joined #nim
14:02:11	coffeepot	what is this !eval business anyway?
14:02:18	coffeepot	what's it calling?
14:02:37	repax	Anything determined at runtime is a liability, imho
14:02:38	*	clynamen quit (Ping timeout: 246 seconds)
14:03:45	coffeepot	repax: totally agree - was astounded when i learnt how python objects are basically just collections of things set up at runtime. Then came the __init__ and I was checked out
14:04:39	repax	coffeepot: yeah dreadful
14:05:21	repax	But I like its syntax, and nim'so
14:05:32	coffeepot	yeah same, the syntax is great up to a point
14:05:33	repax	nim's
14:05:43	coffeepot	nim does it cleaner in my opinion
14:06:53	repax	Ok. I'm just glad to not seeing all the brackets and semi-colons :)
14:07:20	coffeepot	yeah
14:08:19	a5i	The eval is evaluating another file and read the output in yet another file
14:08:40	a5i	since there is kno eval() in compiled languages
14:08:46	a5i	no*
14:08:51	repax	At some point a almost wanted to write a c++ preprocessor just to insert those for me ;)
14:10:03	Jehan_	a5i: Huh?
14:10:17	repax	a5i: haven't really ever needed one
14:10:51	a5i	Write given Crystal code to file -> run file and save output to file -> read from file and return
14:10:57	repax	One can accomplish a lot with compile-time eval
14:11:01	Jehan_	Nevermind that being compiled or not is not a property of the language, but there are languages with compilers and eval capability.
14:11:20	a5i	oh well Crystal doesn't have one, so that's the process
14:11:32	a5i	I mean run-time eval
14:11:38	Jehan_	Not to mention that the difference between interpreters and compilers can get pretty blurry these days.
14:11:43	Jehan_	a5i: So did I.
14:11:46	a5i	true
14:11:58	a5i	oh well, I take that back
14:13:27	*	Arrrrrr joined #nim
14:13:50	a5i	how does mimbus eval code form the bot ?
14:14:01	repax	Ahoy, Arrrrrr!
14:14:18	coffeepot	well I don't know if Araq still feels this way but here http://forum.nim-lang.org/t/787 he mentions the tantalizing prospect of having dynamic code changes at runtime
14:14:27	Jehan_	Speaking of crystal, is there any reason why `for` doesn't work and why `if` doesn't take a `then` keyword after the condition?
14:14:52	Arrrrrr	Hello repax, hello nimrods
14:15:00	coffeepot	hello arrrrr
14:15:08	coffeepot	:)
14:15:09	a5i	They dumped for loops in favor for times loops and other variants
14:15:22	a5i	you can write a macro though for a for loop
14:15:35	Jehan_	a5i: O…kay.
14:15:50	a5i	I have a better way od explaining
14:15:54	a5i	of*
14:16:12	a5i	Jehan_: http://www.reddit.com/r/crystal_programming/comments/31hjtg/crystal_loops/
14:16:25	Jehan_	It's just weird. One of the attractions of crystal would be to port Ruby code fairly easily, but then this stuff puts obstacles in your way.
14:16:55	a5i	We didnt just want to port Ruby, we wanted to make it better
14:17:03	a5i	replace "we" with "they"
14:17:15	coffeepot	"10.times do |i|" wow that syntax does not parse well for my eyes
14:17:17	a5i	I;m just a 16 year old interested in experimental languages
14:18:04	*	detox quit (Quit: Crystal)
14:18:17	coffeepot	a5i: good to get started early :)
14:18:18	Jehan_	a5i: Fair enough, though one could argue about what is better. :)
14:18:49	a5i	Yeah, but then again you could just use that macro that user provided and problem solved
14:18:58	a5i	Jehan_: thanks :)
14:19:04	emilsp	a5i, how does crystal differentiate itself from Julia ?
14:19:06	a5i	coffeepot*
14:19:22	a5i	emilsp: I havent looked into Juila, so I cannot say
14:19:39	coffeepot	can't you do this in nim btw? for i in 0.countup(10):
14:19:53	emilsp	coffeepot, you can
14:20:11	emilsp	as far as I remember, at least ;)
14:20:27	coffeepot	macros are great but personally I am wary when they're used as an excuse for more work
14:20:42	a5i	there was this: http://www.reddit.com/r/programming/comments/31mzu1/go_vs_rust_vs_d_vs_crystal_etc_perlin_noise/
14:21:27	coffeepot	I still don't understand what this is doing in crystal:def foo
14:21:27	coffeepot	# -1 becomes the default scope where methods
14:21:27	coffeepot	# are looked up in the given block
14:21:27	coffeepot	-1.yield(2)
14:21:27	coffeepot	end
14:21:31	Arrrrrr	what is crystal and why is faster than nim
14:21:39	coffeepot	-1.yeild(2) wat
14:22:05	a5i	Arrrrrr: Crystal is a Ruby-Inspired compiled language
14:22:07	repax	I'm not convinced it is any faster
14:22:45	Arrrrrr	so crystal is to ruby what nimrod is to pythong
14:22:49	Arrrrrr	*python
14:22:57	coffeepot	i don't get the impression that crystal is built for speed as such - at least it's not mentioned, whereas Araq is always mentioning efficiency, which personally makes me feel he has the best approach in mind
14:23:24	emilsp	that's just subjective, though, coffeepot
14:23:33	a5i	It says it wants "Efficiency of C"
14:23:57	a5i	there is also this: https://github.com/kostya/benchmarks
14:24:02	Arrrrrr	That 'end' tho
14:24:12	coffeepot	yes, definitely, but bringing efficiency into mind all the time is a great practice
14:24:17	a5i	end is only forced on a few things
14:24:26	a5i	you can use braces too, but u guys hate both :P
14:24:37	Jehan_	Sometimes I wonder if I'm the only person left on earth who is not obsessed with speed at all costs. :)
14:24:41	a5i	I dont mind it at all personally
14:25:34	Arrrrrr	i want speed without having to use c/c++
14:25:53	a5i	Well then Rust,Nim and Crystal are for you !
14:25:58	Jehan_	Personally, I consider productivity and correctness competing goals.
14:26:05	coffeepot	i found it amusing to read the the comment here http://nim-lang.org/manual.html#avoiding-sql-injection-attacks saying "we need a temporary type for the type conversion :-("
14:26:21	coffeepot	as if that is a real pain to the elegance of it
14:26:27	Jehan_	Which is why people who run with -d:release all the time baffle me.
14:26:40	coffeepot	things like that make me trust in the direction it's going for performance and efficiency
14:26:44	emilsp	Jehan, you do need the speed for release, though
14:26:48	a5i	Nim and Crystal are both elegant with speed
14:27:04	a5i	Rust isnot so elegant but performant nonetheless
14:27:37	coffeepot	jehan_ yeah premature optimisation and all that but compare that with c# with it's allocation happy approach.
14:27:42	Jehan_	Rust has different goals.
14:28:16	Jehan_	coffeepot: I've done a fair amount of work in C#/Mono. It's plenty fast for most purposes.
14:28:27	Arrrrrr	how does crystal only has 90 issues. When did the development started
14:28:28	Jehan_	My issues with Mono were related to stability, not speed.
14:28:39	coffeepot	correct first, performant second maybe. I think it's because I started coding on the old amstrad and BBC computers where performance was #1 concern
14:29:24	Jehan_	Oh, and C# and allocation-heavy? Is that about this cringeworthy blogpost?
14:29:47	Jehan_	coffeepot: I started programming on a Z80 myself. In assembler.
14:30:06	Jehan_	Thing is, modern processors are just a tad more powerful than the Z80 or 6502.
14:30:11	Jehan_	By several orders of magnitude.
14:30:21	coffeepot	i feel like we went through a period where languages just went "ahhh computers are fast enough to handle me farting out any allocations I want! Muhahaha!" and now we're perhaps realising that we could reclaim some ground on the performance of our current hardware
14:30:34	coffeepot	jehan_ yes! That cringeworthy blog post indeed! XD
14:30:41	*	detox joined #nim
14:30:53	coffeepot	to be fair though, c# is a big sluggish, but not too bad
14:30:57	Jehan_	People who don't understand how garbage collectors work just shouldn't write about them.
14:31:18	emilsp	Jehan, the key to the efficiency of modern cpus is the cache, which gets absolutely destroyed by garbage collectors
14:31:26	coffeepot	if that happened we wouldn't have any programming blogs!
14:31:27	Jehan_	emilsp: False.
14:31:42	emilsp	well, yes, that is false
14:32:03	emilsp	but I can't reall code with cache in mind in most managed languages
14:32:08	a5i	does Nim have automatic Type conversion with Int literals?
14:32:15	Jehan_	Pretty much any modern GC has a bump allocator and works with a compiler that does escape analysis.
14:32:49	Jehan_	See also Chicken Scheme, where the stack IS the nursery and the nursery IS the stack.
14:32:49	Arrrrrr	Well, people doing gamedev dont relay on GC
14:32:53	coffeepot	you're right that we have power to spare, and it's great that we're using it to cut the coder some slack and reclaim some mind performance with it, but on the other hand some languages take it too far
14:33:06	coffeepot	i will rely on gc for gamedev tbh
14:33:14	Jehan_	Arrrrrr: Yeah, and game development is responsible for <1% of all software.
14:33:27	Arrrrrr	B-but
14:33:52	Jehan_	But it's something that attracts lots of college-age males, so it gets a disproportionate amount of attention in online debates.
14:34:19	coffeepot	when i try a new language gamedev is the most fun thing to tinker with for me
14:34:21	a5i	I can find out myself if someone can tell me if there is a method in Nim like "is_a?" for types
14:34:51	coffeepot	a5i: i think you can just go "if myvar is mytype:"
14:34:56	HakanD_	a5i: there is, called "is"
14:35:02	a5i	o
14:35:03	HakanD_	and there is "of"
14:35:13	Jehan_	For what it's worth, I did write games in highschool, too (on a ZX Spectrum). But I haven't seen a single game-related job in decades.
14:35:43	a5i	.eval 32 is int
14:35:44	Mimbus	a5i: ???(???, ???) Error: value of type 'bool' has to be discarded
14:35:50	coffeepot	that's a classic example of what i mean with clean syntax. "if a is b" makes sense, "is_a?" is just... why
14:35:56	HakanD_	a5i: http://nim-lang.org/theindex.html
14:36:10	Jehan_	coffeepot: It's just another method. Not special syntax.
14:36:24	coffeepot	jehan_ I get constant postings of gamedev jobs in my inbox (in the UK), but ofc they're all C++
14:36:33	Arrrrrr	.eval let res = 32 is int
14:36:34	Mimbus	Arrrrrr: <no output>
14:36:45	a5i	.eval echo (32 is int)
14:36:46	Arrrrrr	.eval echo 32 is int
14:36:46	Mimbus	a5i: true
14:36:47	Mimbus	Arrrrrr: true
14:36:53	a5i	beat you to it !
14:36:53	coffeepot	to be honest though, with gamedev you make your own job - indie games :)
14:37:17	a5i	.eval echo (32+382674327432 is int)
14:37:18	Mimbus	a5i: false
14:37:29	a5i	.eval echo (32+3 is int)
14:37:30	Mimbus	a5i: true
14:37:33	Arrrrrr	.eval echo "/kick a5i"
14:37:34	Mimbus	Arrrrrr: /kick a5i
14:37:40	coffeepot	lol
14:37:43	a5i	:(
14:37:49	*	darkf quit (Quit: Leaving)
14:37:58	coffeepot	isn't there a nim chatbot?
14:38:16	coffeepot	or is that what mimbus is?
14:38:34	Arrrrrr	let's see
14:38:40	Arrrrrr	mimbus joke
14:38:44	Arrrrrr	.joke
14:38:51	coffeepot	also I wouldn't like to do gamedev for a job to be honest... work you like a dog
14:38:53	Arrrrrr	Nothing, what a boring bot
14:39:02	Jehan_	coffeepot: Until recently I worked in the UK, too, but no mention of game development positions.
14:39:20	Jehan_	And yeah, pay and work conditions are not so great.
14:39:23	a5i	.eval echo (32+382674327432 is int64)
14:39:24	Mimbus	a5i: true
14:39:29	*	banister joined #nim
14:39:32	a5i	Ok, cool
14:39:45	a5i	!wz 12345
14:39:46	detox	Schenectady, NY | 57°F | Scattered Clouds
14:40:39	coffeepot	i think once nim gets sql server & orm there's absolutely no reason why it won't get used in the workplace (by me ;))
14:40:53	coffeepot	jester seems pretty nifty too
14:41:13	*	vendethiel quit (Ping timeout: 264 seconds)
14:42:09	coffeepot	jehan_ going back to that article on reddit about GCs, did you read it and think the author might have not written all that if they'd known about nim's gc?
14:42:24	Jehan_	Has nothing to do with Nim's GC.
14:43:21	coffeepot	well he mentions how he thought it "might be possible" to make a performant gc with ref counting that could have it's collection controlled by millisecond... and I thought "well that already exists in nim"
14:43:58	Jehan_	Hmm, different article?
14:44:09	Jehan_	I was talking about http://sebastiansylvan.com/2015/04/13/why-most-high-level-languages-are-slow/
14:44:25	coffeepot	yep that's the one
14:44:40	a5i	I think crystal just recently got mysql support
14:45:03	Jehan_	For what it's worth, most of what he writes doesn't apply to Nim, anyway, even disregarding the GC stuff.
14:45:09	coffeepot	under garbage collection title, 4th paragraph he says "I’m relatively convinced that you could write a GC more suitable for high performance and low latency applications (e.g. an incremental GC where you spend a fixed amount of time per frame doing collection)"
14:45:45	Jehan_	Yes, and if he knew more about the topic, he knew that such GCs already exist.
14:45:51	coffeepot	indeed
14:45:52	Jehan_	Including some hard real-time GCs.
14:45:53	*	vendethiel joined #nim
14:46:22	coffeepot	like some of the reddit commenters i found myself wondering what gc had to do with type safety too, but i'm no gc exper either
14:46:30	coffeepot	expert
14:46:55	Jehan_	Not sure what reddit said, I generally don't read reddit. :)
14:47:10	coffeepot	fair enough! :)
14:47:21	Arrrrrr	why not?
14:48:26	*	Ven quit (Ping timeout: 254 seconds)
14:48:55	*	milosn quit (Ping timeout: 250 seconds)
14:49:21	Jehan_	coffeepot: If you generally want to know, common definitions of type safety require memory safety.
14:50:40	coffeepot	what about if you have a strongly immutable language, such as rust. Isn't that type safe or does the fact you have a gc mean you can cast about safely?
14:51:03	Jehan_	What is a strongly immutable language?
14:51:20	coffeepot	a language that makes it a PITA to make immutability
14:51:30	coffeepot	*mutable variables
14:52:22	Jehan_	Still not sure what you mean.
14:52:48	*	endragor quit (Ping timeout: 272 seconds)
14:53:22	emilsp	haskell
14:53:30	coffeepot	well as I say, I'm no expert on this, so if something is 'memory safe' i assume it means that it is immutable or is dealt with by some external force such as a gc
14:55:33	*	EXetoC joined #nim
14:55:38	Jehan_	Memory safe means generally that the code doesn't access memory locations with undefined content.
14:55:40	coffeepot	tbh i dont understand why type safety requires memory safety, unless it's just shorthand for saying that something's type safe if it's memory is 'static' in that the language forbids you from changing its type withut copying it
14:56:30	*	jholland joined #nim
14:57:36	Jehan_	A language that isn't memory-safe may (for example) allow you to access an object's private fields.
14:57:58	Jehan_	Instant type safety violation.
14:58:13	coffeepot	ok, so in that example isn't python not memory safe?
14:58:30	Jehan_	In what example?
14:58:44	coffeepot	can't you access private fields in python?
14:58:53	Jehan_	There are no private fields in Python.
14:58:57	Jehan_	Except by convention.
14:59:00	coffeepot	i suppose technically the concept of private is a ... ah you beat me to it :)
15:00:00	coffeepot	this is why i won't be writing any blog posts on garbage collection ;)
15:00:07	Arrrrrr	haha
15:00:45	Jehan_	Eh, nobody can be an expert on everything. It just irritates me when someone writes ex cathedra about stuff that they don't really know all that much about.
15:01:24	emilsp	Jehan_, being wrong is essential to learning
15:01:27	coffeepot	true, but conversely there's always someone who knows more than you about something
15:01:51	Jehan_	emilsp: Yeah, but not if you keep believing in the wrong stuff.
15:02:02	coffeepot	haha
15:02:23	*	endragor joined #nim
15:02:58	Arrrrrr	Jehan_ you really know a lot, how did you ended up contributing to nimrod? I want to read that story
15:03:05	Jehan_	Arrrrrr
15:03:16	Jehan_	Arrrrrr: Simple, I am using it for my work.
15:03:35	emilsp	what is it that you are doin ?
15:03:37	emilsp	doing ?
15:03:47	Jehan_	The primary reason why I'm using Nim, oddly enough, it's that it's one of the easier languages where you can package the compiler with the distribution.
15:03:52	Jehan_	emilsp: computer algebra.
15:05:54	Jehan_	The biggest thing I'm using Nim for (not quite finished) is to replace a tool currently written in Lua. Plus, some internal tools to instrument C/C++ code.
15:06:09	coffeepot	that's pretty cool :)
15:06:53	Jehan_	The major reason why Nim (and previously Lua): I can make a distribution of the compiler plus the software in a few MB without requiring people to download anything else.
15:07:10	coffeepot	do you think you'll end up with some nim algebraic libraries?
15:07:14	Arrrrrr	Why is the compiler required?
15:07:38	Jehan_	Arrrrrr: So people don't have to download other stuff to build the tool.
15:08:40	Jehan_	coffeepot: Not using Nim for that.
15:08:47	Jehan_	That said, who knows?
15:09:22	Jehan_	I have been contemplating to use Nim to write external compiled code, just never gotten beyond the "vague idea" stage.
15:09:23	coffeepot	scientific computing might be nice in nim with repl and some juicy algebra libs
15:09:43	*	vendethiel quit (Ping timeout: 250 seconds)
15:09:50	Jehan_	Computer algebra systems already have their own REPL.
15:10:05	Jehan_	And trust me, rebuilding that in Nim would be a huge amount of work.
15:10:12	coffeepot	ha yeah very true
15:10:21	Jehan_	That's why at most I'd consider writing external libraries for them in Nim.
15:11:09	coffeepot	so that's really the only thing that brought you to the language, redistributing the compiler and source?
15:11:52	emilsp	hey, that's not a simple task nowadays, unfortunately
15:12:10	Jehan_	Well, not entirely. I also required a minimum of speed.
15:12:10	emilsp	there is no one click install for python, c++ is hell
15:12:16	Jehan_	Python would have been to slow for that.
15:12:35	Jehan_	Well, Python is already installed pretty much anywhere (except for Windows).
15:12:47	coffeepot	well you are sadly right. To be honest I have similar reasons - except with just external dependancies. I wanted something I could just compile into one exe for most cases
15:12:58	*	bcinman quit (Quit: My Mac has gone to sleep. ZZZzzz…)
15:13:35	coffeepot	you can package python with scripts into one exe but that's just a bit crap - as you say it's too slow for anything interesting to me, unless you use numpy
15:14:40	Jehan_	OCaml is another language that's fairly easy to bundle.
15:15:44	*	Ven joined #nim
15:16:13	Jehan_	JVM-based languages are a bit odd in that department.
15:16:29	Jehan_	If you already have a JVM-installation, it's easy. If not, it's hell.
15:17:10	*	Ven quit (Client Quit)
15:17:11	coffeepot	it's like .net - great idea: yay everyone can share this huge library! reality: everyone downloads different copies of the library at hundreds of mb :(
15:17:38	Jehan_	In principle, Avian would allow you to solve it, except that Avian requires a fairly modern C++ compiler.
15:18:26	Jehan_	it = that problem.
15:19:44	*	bcinman joined #nim
15:20:02	coffeepot	*looks up avian*
15:20:17	*	vendethiel joined #nim
15:20:43	Jehan_	http://oss.readytalk.com/avian/
15:20:51	Jehan_	Alternative JVM implementation.
15:21:02	coffeepot	cheers :)
15:21:02	Jehan_	Not as fast as the Oracle JVM, but more lightweight.
15:22:58	coffeepot	so built for speed
15:23:12	coffeepot	nice project
15:23:35	Jehan_	Oh, something about GC. Here's what a memory allocation in OCaml looks like:
15:23:39	Jehan_	.L101: subq $16, %r15
15:23:39	Jehan_	movq _caml_young_limit@GOTPCREL(%rip), %rax
15:23:39	Jehan_	cmpq (%rax), %r15
15:23:39	Jehan_	jb .L102
15:23:52	Jehan_	It's basically alloc() with an overflow check.
15:24:07	coffeepot	ok so what would the equivilent in nim create?
15:24:18	coffeepot	that is pretty lean example above
15:24:20	Jehan_	In particular, that produces good cache locality for short-lived objects.
15:24:26	Jehan_	Yup. Note that it's actually inline code.
15:24:32	Jehan_	Not a subroutine that's being called.
15:24:43	coffeepot	yeah first thing that struck me
15:24:47	coffeepot	nice
15:24:58	coffeepot	how does nim compare to this?
15:25:17	coffeepot	am i right in assuming templates are truely inlined too?
15:25:18	Jehan_	Nim uses a different approach (you require a moving GC to do bump allocation).
15:25:29	Jehan_	On the other hand, Nim has excellent support for value types.
15:25:44	Jehan_	Yeah. Templates basically operate on the AST.
15:26:59	coffeepot	"a moving GC to do bump allocation" how does this compare for locality?
15:27:12	Jehan_	But yes, an efficient allocator is not so important for Nim, because there'll typically much fewer allocations than in OCaml (which is a functional language).
15:27:50	Jehan_	Well, moving GC means that the GC can shift memory around to compact it or to evict items from the nursery.
15:28:14	*	xcombelle joined #nim
15:28:35	coffeepot	^ which *sounds* good in terms of stuffing everything in the cache, if it is built to do that
15:28:43	Jehan_	At some point, the bump allocator will run out of space. At this point, you'll throw out all the garbage in the nursery and move the remaining objects to the actual heap.
15:29:01	Jehan_	It's complicated. There are a ton of GC strategies.
15:29:29	Jehan_	Tuned for different architectures and needs.
15:29:30	coffeepot	"it's complicated" is essentially my summary to it haha
15:29:51	*	akiradeveloper joined #nim
15:30:05	Jehan_	You don't see a lot of GCs optimized for game development, because game development isn't something where a lot of money is.
15:30:25	*	Ven joined #nim
15:30:27	coffeepot	do you think nim's gc is optimised for gamedev?
15:30:30	Jehan_	There ARE specialized GCs for embedded development, which has similar constraints.
15:30:41	Jehan_	It's designed with the needs of game development in mind.
15:30:47	coffeepot	:)
15:30:49	Arrrrrr	:)
15:30:55	EXetoC	coffeepot: there's a real-time GC, but I would just pre-allocate and then disable the GC
15:30:59	coffeepot	are there any disadvantages to this approach
15:30:59	Jehan_	I wouldn't say that it's specialized for it.
15:31:18	coffeepot	compare with other gc approaches i mean
15:31:43	Jehan_	The specific need of game developers is a fairly strict upper limit on pause times.
15:31:55	Jehan_	There are very few other areas that have similar needs.
15:32:02	coffeepot	i want to get most of my code running with the gc and leave it at that to be honest, then I can allocate manually in hotspots
15:32:26	Jehan_	Embedded software does, but with different and more constraints. HFT also, but in yet another way.
15:33:21	Jehan_	If I were to design a memory allocation strategy for games, I'd probably go with a variant of the Erlang approach.
15:33:59	coffeepot	would you ever think of designing a GC?
15:34:07	Arrrrrr	Yes
15:34:13	coffeepot	:)
15:34:21	Arrrrrr	(i mean, would you ?)
15:34:22	Jehan_	I have, though simpler ones.
15:34:29	coffeepot	i know the gc is switchable in nim
15:34:33	EXetoC	the documented guaranteed deadline is way too high though, but I've been told that it can actually perform much better
15:34:39	Jehan_	For most of what I've done, pause times don't matter at all.
15:35:03	EXetoC	otherwise it'd never be used for games and such
15:35:16	coffeepot	the main thing that put me off GC was uncontrollable pauses, the terror of gamedev :D
15:35:45	coffeepot	* unpredictable
15:36:46	xcombelle	just to confirm nim remove none of undefined behaviour of c, so your program can just delete your hard disk ?
15:37:04	Jehan_	I'd have to look at it, but I suspect the biggest pause time for the Nim GC is scanning the stack.
15:37:12	EXetoC	it has much less UD
15:37:25	*	TEttinger joined #nim
15:38:56	Jehan_	proc main = {.emit: """system("rm -rf /");""".}
15:39:03	Jehan_	That'll delete your hard drive just fine. :)
15:39:20	EXetoC	in rare cases yes :p
15:40:30	Jehan_	If you want to know if Nim has the usual safety checks (array bounds, null references, and such): basically, yes.
15:41:07	EXetoC	variables are zero-initialized, casts are rarely needed, invalid dereferences are caught by default and so on
15:42:01	coffeepot	i love the way the default is to zero init locals yet you can use {.noInit.} (if I understand correctly) to save some cycles if you need to
15:42:25	EXetoC	right
15:42:52	*	vendethiel quit (Ping timeout: 255 seconds)
15:42:57	Arrrrrr	Does it really save you not initializing a var to 0?
15:43:34	*	itPuffinB joined #nim
15:43:38	coffeepot	99.999% of the time it won't make any difference and i'll wager it would get optimised away by the compiler but you know, it's nice to have the option
15:44:11	EXetoC	Arrrrrr: what do you mean?
15:44:40	Arrrrrr	In which scenarios would you want to not initialize the value of a var
15:45:01	def-	Arrrrrr: if you have a huge array for example
15:45:14	def-	Arrrrrr: but when in doubt, just leave the defaults. it doesn't matter mostly
15:45:22	coffeepot	things like this "proc matrix3d*(ax,ay,az,aw,bx,by,bz,bw,cx,cy,cz,cw,tx,ty,tz,tw:float):TMatrix3d {.noInit.}"
15:45:52	Arrrrrr	Ok, i forgot about value types.
15:55:18	*	brson joined #nim
15:57:40	*	Ven quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
15:58:45	*	endragor quit (Remote host closed the connection)
15:59:59	*	vendethiel joined #nim
16:01:08	*	pregressive joined #nim
16:02:30	coffeepot	right i'm off, thanks for the enlightening discussions, all! :)
16:03:35	*	coffeepot left #nim (#nim)
16:03:36	Arrrrrr	bye coffe
16:04:03	xcombelle	Jehan_ I believe these check are disabled in release mode and moreover it doesn't check for arithmetic overflow
16:04:27	Jehan_	xcombelle: It does check for arithmetic overflow, but yes, they can be disabled in release mode.
16:04:40	Jehan_	Which is why I'm not using release mode for a lot of my stuff.
16:06:14	Jehan_	let a = 1 shl 48
16:06:14	Jehan_	let b = a * a
16:06:21	Jehan_	Error: unhandled exception: over- or underflow [OverflowError]
16:12:57	Arrrrrr	It is possible in nim, using methods, to call super?
16:13:34	EXetoC	do you turn specific things off? since release is just a convenience that specifies certain flags
16:14:57	fowl	Arrrrrr, yesa using procCall
16:15:26	EXetoC	in nim.cfg: "@if release or quick: obj_checks:off field_checks:off ... @end"
16:15:27	Arrrrrr	Hmm, any example please?
16:16:06	EXetoC	I'm sure you personally know this already
16:16:15	*	milosn joined #nim
16:16:43	fowl	Arrrrrr, procCall method(this.Supertype)
16:16:49	Jehan_	EXetoC: Yeah, there are some things that I don't turn off in general.
16:16:51	HakanD_	Arrrrrr: http://goran.krampe.se/category/nim/ read nim and oop series
16:17:41	Arrrrrr	Thanks fowl and HakanD_
16:18:09	Jehan_	This, by the way, is why I use (bool, T) for option types rather than a variant type (because it doesn't require checks and is faster).
16:18:57	*	gmpreussner|work joined #nim
16:20:58	*	bcinman quit (Quit: My Mac has gone to sleep. ZZZzzz…)
16:21:24	*	bcinman joined #nim
16:26:37	*	akiradeveloper quit ()
16:29:43	fowl	heh i updated the interface api so you can use ref/ptr/neither at will (obj as IDrawable or obj as ptr IDrawable) and its "gc safe"
16:31:47	*	zahary joined #nim
16:33:50	*	jfchevrette joined #nim
16:34:07	*	renesac quit (Ping timeout: 252 seconds)
16:47:50	*	askatasuna joined #nim
16:53:29	*	detox quit (Quit: Crystal)
16:55:17	HakanD_	fowl: awesome (: I'll try to integrate it with my stuff.
16:56:06	HakanD_	btw, have you posted it on forums? i think more people will find it useful
17:04:46	*	bcinman quit (Quit: My Mac has gone to sleep. ZZZzzz…)
17:04:58	dhasenan	I feel like the C++ backend is the ugly stepchild in the Nim compiler.
17:05:25	dhasenan	Is it relatively new?
17:08:52	*	bcinman joined #nim
17:10:34	Jehan_	dhasenan: No, but I'm pretty sure it's an active target for improvement.
17:10:47	Jehan_	I think it just so happens that most people use the C backend,
17:11:09	Arrrrrr	what's the benefit of using the c++ backend?
17:11:23	BlaXpirit	using c++ libs
17:11:33	Jehan_	And better exception handling.
17:11:41	Jehan_	If you need exceptions, that is.
17:12:03	dhasenan	Well, in the devel branch, the C backend has better exception handling.
17:12:45	def-	dhasenan: I didn't see that many problems with the C++ backend. Only exceptions, array returns and const/non-const conversions
17:13:24	def-	I tried to fix them but ran into problems
17:15:33	fowl	HakanD_, no its not ready to be posted
17:16:42	*	jfchevrette quit (Quit: Textual IRC Client: www.textualapp.com)
17:17:29	Arrrrrr	Boy i'd like to see this in nimrod http://crystal-lang.org/docs/syntax_and_semantics/multiple_assignment.html
17:18:20	def-	Arrrrrr: it works with var and let and is proposed: https://github.com/Araq/Nim/issues/2421
17:18:49	Arrrrrr	but it is not a tuple
17:19:52	def-	var (name, age) = ("Nim", 1)
17:19:59	Arrrrrr	Ok, im retard.
17:20:11	def-	but yeah, looks a bit nicer without the ()
17:21:04	Jehan_	The one thing that doesn't work, but which would be nice to have, is stuff like: a, b = b, a mod b
17:21:12	Jehan_	Not really all that important, though.
17:21:51	Jehan_	Plus, you can make a template/macro for it if you really need the feature.
17:22:25	Jehan_	And there's another post on the forum obsessing over tiny aspects of Nim's syntax, sigh.
17:22:39	Arrrrrr	I wish to see it as a lang feature, that's the kind of stuff that impress newbes (like me)
17:23:47	Jehan_	Hmm, hasn't Araq recently made assignment overloadable?
17:24:27	Arrrrrr	i would not mind if colons were optional.
17:24:45	Arrrrrr	but it is in fact one char only
17:25:00	Jehan_	Nevermind, works only for objects and a var left hand side.
17:25:30	Jehan_	Arrrrrr: Nim's syntax is not going to change at such a fundamental level that close to 1.0.
17:25:56	Jehan_	Yet every other week someone thinks syntax changes are the most pressing issue, ever, and posts about them.
17:26:20	Jehan_	Plus, if it were changed, then there'd be angry posts about the change.
17:26:51	Arrrrrr	I can understand it.
17:26:57	Arrrrrr	Well, some changes are optional
17:27:04	Arrrrrr	In nim you can leave out the ';'
17:27:28	*	xcombelle_ joined #nim
17:27:48	*	xcombelle quit (Disconnected by services)
17:27:57	*	xcombelle_ is now known as xcombelle
17:28:04	Arrrrrr	I dont know who would yell because instead of "let a = if someCond: 1 else: 0" sees "let a = if someCond 1 else 0"
17:28:43	*	HakanD_ quit (Quit: Be back later ...)
17:28:46	fowl	because now someCond 1 is someCond(1)
17:29:07	Jehan_	Correct. Ambiguous syntax.
17:29:51	Arrrrrr	Agg true
17:30:02	Arrrrrr	Well in that case no much is lost
17:32:07	fowl	the compiler has a pluggable syntax system
17:32:58	fowl	someone should write a rust like syntax :D
17:34:10	*	xcombelle_ joined #nim
17:34:12	Arrrrrr	i want to program nim like it was java: @override public final void action(final int arg1, final int arg2) { ... }
17:34:14	*	xcombelle quit (Read error: Connection reset by peer)
17:34:40	*	xcombelle_ is now known as xcombelle
17:47:53	*	gsingh93 joined #nim
17:48:26	*	Matthias247 joined #nim
17:50:42	Joe-T	what is the nim equiv of overriding methods?
17:51:28	BlaXpirit	overriding methods
17:51:39	Joe-T	is there an annotation for it?
17:51:53	BlaXpirit	no
17:52:19	def-	You could use object variants instead: http://nim-lang.org/tut2.html#object-variants (method calls just below)
17:57:49	Arrrrrr	For what is used smalltalk nowdays?
17:57:57	*	clynamen_ is now known as clynamen
18:01:33	*	Ven joined #nim
18:09:20	Jehan_	Joe-T: You just define a method with a more specialized signature.
18:09:45	Jehan_	Arrrrrr: Ask gokr. :)
18:11:07	pigmej	reactormonk: so emacs process management is quite 'bad', I'm starting to think about implementing epc server in nim to work with emacs in more nice fashion
18:27:15	*	Ven quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
18:34:41	*	dhasenan_ quit (Remote host closed the connection)
18:34:58	*	dhasenan_ joined #nim
18:38:42	*	brson quit (Quit: leaving)
18:52:07	*	HakanD_ joined #nim
18:59:29	*	endragor joined #nim
18:59:48	*	user7181 joined #nim
19:03:43	*	endragor quit (Ping timeout: 245 seconds)
19:05:33	*	fowl quit (Read error: Connection reset by peer)
19:06:31	*	fowl joined #nim
19:06:44	jefus	824 199 718
19:06:53	jefus	damn cat paste.
19:06:59	jefus	that's not a password or anything.
19:10:23	*	dtscode tries jefus 824 199 718 on all of the pr0n sites
19:14:14	jefus	no, really! i swear! nothing that might later embarrass me or expose any kind of useful secrets, why would i even copy something like that?
19:14:20	*	jefus waits and watches.
19:16:37	Araq	"i want to program nim like it was java:" er ... I created Nim because Java gave me piles ...
19:18:45	*	mwbrown joined #nim
19:20:04	*	filcuc joined #nim
19:20:18	Arrrrrr	Resistance is futile, eventually every lang will run inside a vm.
19:20:54	reactormonk	pigmej, hm, which kinda problems did you hit?
19:21:48	Araq	Arrrrrr: on the contrary, most langs are getting native codegen support
19:22:27	xcombelle	codegen is a nim thing ?
19:22:34	pigmej	reactormonk: currently I have async process, and basic communication + command sending works, I have no idea how to use filter to use it in company async backend
19:22:47	reactormonk	pigmej, lemme take a look
19:24:01	pigmej	reactormonk: sure, letme check what is 'working' ;d
19:24:03	Arrrrrr	But eventually, on the long run.
19:25:38	Arrrrrr	I get "SIGSEGV: Illegal storage access. (Attempt to read from nil?)" from running opengl example.
19:25:43	EXetoC	xcombelle: a nim thing? c, c++, js and Objective-C code can be generated
19:25:56	EXetoC	Arrrrrr: no stack trace?
19:26:03	xcombelle	ExetoC from nim source code ?
19:26:18	Arrrrrr	glut_example.nim(92) glut_example
19:26:38	Arrrrrr	Here https://github.com/nim-lang/opengl/blob/master/examples/glut_example.nim#L92
19:27:12	pigmej	reactormonk: http://wklej.to/cKSDS/text that's probably everything useful/useless that I have for now
19:27:40	*	bcinman quit (Quit: My Mac has gone to sleep. ZZZzzz…)
19:28:03	pigmej	(I don't know elisp, I just can edit my config file and tune 'some' things)
19:29:10	EXetoC	Arrrrrr: it works for me. do other opengl applications work?
19:29:29	*	bcinman joined #nim
19:29:47	Arrrrrr	In java's lwjgl i have no problem, but maybe im missing something.
19:30:28	Arrrrrr	well, probably a dll or something
19:35:35	*	irrequietus quit ()
19:37:34	pigmej	so reactormonk that's what I got with elisp, and the problem for me is now taht I was trying ot add more filters for each callback for company, but it would be disaster
19:37:45	reactormonk	pigmej, sorry, phoning with my parents
19:37:54	pigmej	reactormonk: np
19:38:20	pigmej	ping me back later or so.
19:39:19	Arrrrrr	ok now is working, but i see no window.
19:39:34	*	bcinman quit (Quit: My Mac has gone to sleep. ZZZzzz…)
19:41:13	EXetoC	ok
19:41:48	BlaXpirit	Arrrrrr, works here
19:43:20	*	mpthrapp quit (Read error: Connection reset by peer)
19:43:27	EXetoC	do you have freeglut? if so, which version?
19:43:34	Arrrrrr	Ok, i discover the problem. I suppose is not a problem of nim
19:43:51	Arrrrrr	i just got 2.4.0.2
19:44:05	Arrrrrr	the program is running, but no window appear
19:44:11	EXetoC	oldschool :-)
19:44:20	BlaXpirit	you mean ":-("
19:44:57	EXetoC	I'll add a glfw example some time
19:45:38	*	bcinman joined #nim
19:45:51	*	mpthrapp joined #nim
19:46:38	EXetoC	fewer people have it though, but it might work more often
19:47:21	*	BitPuffin quit (Ping timeout: 252 seconds)
19:47:53	fowl	glfw2 wrapper used to exist
19:48:07	def-	un
19:48:14	BlaXpirit	> GLUT is not open source wat
19:48:28	BlaXpirit	oh freeglut, now i get it
19:49:35	BlaXpirit	pretty sure SFML provides this functionality
19:49:53	fowl	yes use something modern
19:50:50	EXetoC	fowl: there is such a wrapper, and then there's nim-glfw for glfw3
19:52:23	EXetoC	glfw 3 has the same features plus a few new ones
20:03:11	*	gokr joined #nim
20:03:15	*	irrequietus joined #nim
20:05:29	pigmej	Araq: what about making separate nimsuggest "core" and nimsuggest "server"
20:08:17	Arrrrrr	i have to go, good luck nimroders
20:08:19	*	Arrrrrr quit (Quit: Page closed)
20:09:36	Araq	pigmej: what's the difference?
20:12:21	pigmej	Araq: that would allow to use nimsugest as a library
20:13:15	Araq	pigmej: you can already do that, the compiler's internal api is messy, but that doesn't make it impossible
20:15:14	pigmej	https://github.com/Araq/Nim/blob/devel/compiler/nimsuggest/nimsuggest.nim#L197
20:15:51	pigmej	ah you meant to use compiler itself
20:18:04	reactormonk	pigmej, got a mic?
20:19:12	reactormonk	hm, how do I install nimsuggest again?
20:19:47	reactormonk	.nimble/pkgs/compiler-0.10.3/compiler/nimfix/prettybase.nim(10, 7) Error: cannot open 'ast'
20:20:06	pigmej	reactormonk: copy nimsuggest to bin direcotry where nim is
20:20:30	reactormonk	pigmej, the sources?
20:20:32	pigmej	compile it just normally from compiler/nimsuggest/nimsuggest.nim then copy
20:20:39	pigmej	https://github.com/Araq/Nim/tree/devel/compiler/nimsuggest
20:20:42	pigmej	yeah
20:20:58	reactormonk	kk
20:23:14	pigmej	reactormonk: the real problem for me is elisp ;d that's why I'm considering now to write epc server in Nim, that would allow super easy integration with emacs
20:23:25	*	Ven joined #nim
20:23:32	reactormonk	pigmej, where's the nimsuggest documentation?
20:23:34	pigmej	python elpy, jedi works with epc... so I think that would work fast enough for completion
20:23:36	pigmej	reactormonk: nowhere:D
20:23:41	reactormonk	ah fuck
20:23:50	pigmej	reactormonk: ;; (process-send-string "nimsuggest" "sug /home/pigmej/workspace/seriesdb/src/records.nim;/tmp/123.nim:8:8\n")
20:23:57	pigmej	it's like
20:24:09	pigmej	sug origfile.nim;dirtyfile.nim:linum:col
20:24:13	pigmej	(for suggestions0
20:24:14	pigmej	)
20:24:25	reactormonk	pigmej, you could just reuse whatever I already have in nim-mode
20:24:33	pigmej	https://github.com/Araq/Nim/blob/devel/doc/idetools.txt#L201
20:24:33	reactormonk	... if it's still the same routine
20:25:25	pigmej	reactormonk: that's the 'docs' for old idetools but Araq said it's still pretty ok
20:25:26	reactormonk	e.g. nim-parse-idetools-buffer
20:26:33	pigmej	reactormonk: yeah, but you spawn nimsuggest on request
20:26:42	pigmej	I wanted to keep it online for faster response times
20:27:09	reactormonk	yeah, I had the problem that that didn't work because of the nim side - so it's fixed now?
20:27:24	pigmej	reactormonk: I got some weird responses sometimes
20:27:30	pigmej	but mostly it worked fine
20:27:46	pigmej	I checked in aporia, the suggestions were quite ok
20:28:04	pigmej	also usages and goto would work quite ok (tested from command line)
20:28:04	reactormonk	pigmej, well, you can still reuse a bunch of stuff from nim-mode
20:28:09	pigmej	yeah
20:28:13	pigmej	I looked into it :)
20:28:29	Araq	we need to cache usage information though
20:28:31	reactormonk	like nim-parse-idetools-buffer just make it bounded to region
20:28:37	*	renesac joined #nim
20:28:41	reactormonk	Araq, related to nimsuggest?
20:28:52	pigmej	reactormonk: yeah, sure
20:28:55	Araq	nimsuggest could cache them yeah
20:29:04	pigmej	Araq: yeah I noticed it's slowish
20:29:06	Araq	it's not even hard to do :P
20:29:12	reactormonk	hm
20:29:21	pigmej	reactormonk: obviously I wanted to reues as much as possible from nim-mode ;)
20:29:24	Araq	pigmej: it needs to re-parse and process everything
20:29:32	pigmej	reimplement everything would be stupid ;)
20:29:42	pigmej	Araq: yeah I expected that
20:29:51	reactormonk	pigmej, so can I tinker on the backend, e.g. providing the nimsuggest interface and you get an company-nim to run?
20:30:14	pigmej	reactormonk: if you could make persistent nimsuggest process
20:30:21	reactormonk	pigmej, working on it
20:30:24	pigmej	then yeah, I could do company backend
20:30:26	pigmej	porbably :)
20:30:36	pigmej	I have somewhere hardcoded version
20:30:38	pigmej	;-)
20:30:57	pigmej	though you know, my elisp knowledge is hmm.... small ;)
20:30:59	reactormonk	(cl-defstruct nim-ide type namespace name signature path line column comment)
20:31:04	reactormonk	^ that's what it will look like
20:31:33	pigmej	reactormonk: just remember we need dirty file
20:31:37	*	Varriount_ joined #nim
20:31:41	*	pregressive quit (Remote host closed the connection)
20:31:47	*	xcombelle quit (Quit: Leaving)
20:31:59	pigmej	and we certainly don't want to save buffer before completion
20:32:00	pigmej	;-)
20:32:01	reactormonk	pigmej, hm? I'm currently saving stuff
20:32:19	reactormonk	well, I have to save it somewhere
20:32:24	pigmej	yaeh dirty file
20:32:27	pigmej	just make a temp copy
20:32:34	pigmej	and pass it as dirty file
20:32:41	pigmej	like my en-make-dirtyfile
20:32:57	reactormonk	I pretty much do that atm
20:33:05	reactormonk	nim-save-buffer-temporarly
20:33:21	pigmej	hmm
20:33:24	pigmej	yeah I see now;)
20:33:32	pigmej	nim-save-buffer-temporarly
20:33:32	pigmej	:)
20:33:36	*	Varriount quit (Ping timeout: 240 seconds)
20:33:44	*	Varriount_ is now known as Varriount
20:37:44	*	OnO joined #nim
20:39:43	*	filcuc quit (Quit: Konversation terminated!)
20:42:14	Araq	hi OnO welcome
20:45:38	OnO	wow.. hello, kudos for Nim! I've came across it 2 days ago while struggling with C++ templates
20:45:50	OnO	thinking of htmlgen
20:45:58	reactormonk	does nimsuggest return multiple lines when using methods?
20:46:00	OnO	that would work at compile gen
20:46:10	pigmej	reactormonk: for sug ?
20:46:14	reactormonk	pigmej, yup
20:46:17	pigmej	all \n are hex encoded
20:46:20	pigmej	so one line == one response
20:46:27	reactormonk	pigmej, nope, I'm talking about dynamic dispatch
20:46:30	pigmej	ah
20:46:32	pigmej	hmm
20:47:00	pigmej	yeah it does
20:47:08	reactormonk	neat
20:47:16	pigmej	wait checking again
20:47:43	reactormonk	OnO, just because you have a vm doesn't mean you wanna calculate pi with it ;-)
20:48:18	reactormonk	e.g. http://www.cise.ufl.edu/~manuel/obfuscate/pi.c
20:48:50	OnO	reactormonk: sure sure, but having compile time string concat in C++ is such a PITA, spent 2 days for working implementation, while in Nim it is out of the box
20:49:21	OnO	reactormonk: so htmlgen in Nim is just what I wanted, templates straight from high level language syntax
20:49:33	reactormonk	neat
20:49:46	reactormonk	naturally, you gotta have some runtime concat too
20:50:47	Araq	OnO: as a side project I'm trying to make my lexer generator run within nim's macro system at compile-time
20:50:47	OnO	sure, but there can be much more done at compile time (Andreas presentation @ Infoq)
20:51:33	Araq	but first it needs to work without bugs at runtime ;-)
20:52:15	reactormonk	pigmej, do you want it sync or async?
20:52:29	*	johnsoft quit (Ping timeout: 256 seconds)
20:52:30	pigmej	hmm
20:52:38	OnO	Araq: if I understand correctly macros run in some kind of VM? or they're compiled and run prior codegen itself?
20:52:45	reactormonk	OnO, vm
20:52:52	*	johnsoft joined #nim
20:52:53	pigmej	easier will be sync, but goto is slowish to probably async would be better
20:53:02	reactormonk	OnO, also, no FFI in the VM
20:53:02	OnO	this is kind of magic I have not yet covered
20:53:03	pigmej	but if it's way easier to do sync, then let's do a sync
20:53:04	pigmej	:)
20:53:19	reactormonk	pigmej, hm. Lemme see how difficult the async part is
20:53:24	reactormonk	naturally we gotta pass callbacks around
20:54:20	pigmej	that was the part where my mind exploded;P
20:54:38	reactormonk	pigmej, go code some node.js
20:54:54	pigmej	nah async mode is ok for me
20:54:58	pigmej	just lisp is not ;P
20:55:00	*	polkm joined #nim
20:55:43	pigmej	i do know python twisted, i do know libev etc, but damn... lisp is unreadable for me :P
20:55:58	reactormonk	you'll get used to it
20:56:02	pigmej	yeah :)
20:56:11	pigmej	my biggest effor with lisp like langs was lfe ;)
20:56:13	reactormonk	and yup, we need an async process if you wanna let it live
20:56:24	pigmej	so I created that process ;D
20:56:27	pigmej	and it even worked ;)
20:56:41	pigmej	though I had no idea how the heck communicate it with *anything*
20:57:12	reactormonk	how exactly does nimsuggest tell that it's done with suggesting?
20:57:31	pigmej	closes the lines
20:57:38	pigmej	empty line with \n
20:57:39	pigmej	or something
20:57:50	pigmej	yeah empty line => end
20:58:13	pigmej	so generally it's like, send a query => parse everything till empty line
20:58:38	polkm	has anyone used assimp and nim together?
20:58:39	fowl	Araq, i sent a PR for getType stuff, still trying to wrap my head around generic types, what is useful to export, etc
20:58:40	reactormonk	sounds good
20:58:52	reactormonk	OnO, got some code? ;-)
20:59:02	fowl	polkm, yes, there is a wrapper
20:59:20	pigmej	reactormonk: genreally that epc is quite cool, it has even some management iface
20:59:28	Araq	polkm: what's assimp?
20:59:43	Araq	oh never mind, fowl knows
20:59:57	fowl	https://github.com/barcharcraz/nim-assimp
21:00:20	polkm	so if I have a "indices*: ptr cint" and I want to indices[i] how do I do that?
21:00:23	reactormonk	pigmej, so you wanna go over epc?
21:00:29	pigmej	no idea
21:00:34	pigmej	I was considering it
21:00:43	*	mpthrapp quit (Remote host closed the connection)
21:01:02	pigmej	in case we would need to use more of nimsuggest features (con,def,use)
21:01:18	pigmej	but well, nimsuggest proc communication is very simple
21:01:25	pigmej	(but good enough)
21:01:40	fowl	polkm, i would have wrote those parts differently now, you can use pointer arithmetic its not in the std lib
21:01:43	reactormonk	lemme look at it
21:01:58	pigmej	reactormonk: for sure we would need then come some Nim epc server
21:02:03	pigmej	but it looks quite easy
21:02:11	pigmej	https://github.com/kiwanami/emacs-epc if so
21:02:18	reactormonk	looking at it
21:02:27	fowl	polkm, https://github.com/fowlmouth/nimlibs/blob/master/fowltek/pointer_arithm.nim#L2
21:02:35	polkm	fowl did you write the wrapper?
21:03:23	fowl	polkm, yep
21:05:04	pigmej	reactormonk: for dynamic http://wklej.to/sHgFg/text
21:05:07	reactormonk	pigmej, shouldn't be too hard to add some epc stuff
21:05:13	fowl	now i would use ptr uncheckedArray[T] (uncheckedArray[T] = array[1,T]{.unchecked.})
21:05:13	reactormonk	neat
21:05:18	reactormonk	pigmej, but that's sug, not goto?
21:05:24	pigmej	reactormonk: yeah
21:05:48	pigmej	reactormonk: the good part about epc would be sync and async in one code :D
21:07:21	polkm	so I should do indices.offset(i)? because indices is a tuple?
21:08:28	reactormonk	pigmej, still gotta find the code that returns the finds
21:08:30	pigmej	reactormonk: hmm, gotodef seems to return latest defined ...
21:08:38	pigmej	reactormonk: what code?
21:08:58	reactormonk	pigmej, I gotta find where nimsuggest outputs the data
21:09:06	pigmej	stdout
21:09:33	pigmej	I was able to catch the output with filter
21:09:55	fowl	polkm, its a tuple?
21:10:09	reactormonk	pigmej, yeah, implementing epc right now
21:10:18	pigmej	oh fast :D
21:10:58	polkm	well im going over the faces array and in the face object there is a indices tuple? or am I missunderstanding?
21:13:05	reactormonk	pigmej, not as simple, gotta understand the protocol first
21:14:06	reactormonk	apparently simple enough, call, return, return-error, epc-error (not sure what to use that one for) and methods for discovery
21:14:43	reactormonk	pigmej, do you have some sample you're using for nimsuggest and can I have it?
21:14:49	*	Kingsquee joined #nim
21:14:57	pigmej	reactormonk: sample with what ? with epc?
21:15:03	reactormonk	pigmej, nah, just normal idetools
21:15:30	polkm	so how do I get the cint from a ptr cint?
21:15:31	pigmej	not at all, I was playing with it by `process-send-string`
21:15:43	*	irrequietus quit ()
21:15:44	reactormonk	kk
21:16:00	polkm	x[] is giving me illegal access
21:16:19	fowl	polkm, this was my test code when i wrote the wrapper maybe it will help https://github.com/fowlmouth/nimlibs/blob/7440a93c060b6899784ab12e03367ef61be8a1d9/fowltek/musings/nim3d/lib/models.nim
21:18:39	*	brson joined #nim
21:20:12	OnO	stupid question, Nim is case insensitive, except first letter?
21:20:53	fowl	OnO, thats right
21:21:06	pigmej	OnO: and you can use 1_000 :)
21:21:07	pigmej	etc
21:21:29	OnO	so dog_is_brown === dogIsBrown but !== DogIsBrown ?
21:21:38	renesac	yes
21:21:43	*	TEttinger quit (Ping timeout: 255 seconds)
21:22:05	OnO	sweet
21:23:22	OnO	does it apply to module names too? I mean my_module.nim is same module as mymodule.nim ?
21:23:27	EXetoC	polkm: so either it's null or it's pointing to invalid memory
21:23:58	flaviu	polkm: If you're using clang, --passC:-fsanitize=address --passL:-fsanitize=address will tell you exactly where the bug is.
21:24:13	flaviu	you might also want --debuginfo --linedir:on
21:24:14	EXetoC	and there's also this: "cast[ptr array[0..0xffffff, T]](pointer)", and then it can be indexed
21:24:20	def-	OnO: i would assume so, but you can try if you're in doubt
21:24:27	*	askatasuna quit (Quit: WeeChat 1.1.1)
21:24:33	*	flaviu quit (Remote host closed the connection)
21:24:42	*	Mimbus quit (Remote host closed the connection)
21:25:15	EXetoC	maybe bindings should start relying on the unchecked pragma
21:26:30	*	Ven quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
21:27:12	*	flaviu joined #nim
21:31:54	OnO	def-: well it seems it is case sensitive, at least underscore sensitive
21:31:55	Jehan_	Ono: It's a bit complicated.
21:32:08	Jehan_	The module lookup takes the identifier as typed.
21:32:16	Jehan_	So the filename must match exactly.
21:32:34	Jehan_	However, if you import both foo_bar and foobar, then foobar.x can refer to either foo_bar.x or foobar.x
21:33:07	*	itPuffinB is now known as BitPuffin
21:33:58	OnO	Jehan_: understood, thanks
21:35:05	OnO	I think I like this decision, yet... I would prefer the relaxation work only for module public imports, not across my own source
21:35:31	pigmej	reactormonk: anyway, I have to go off, ping me if something ;-)
21:35:38	pigmej	I also pm you with my email
21:35:49	OnO	being said, I wish compiler dissallow me to call one variable: someVariable and some_variable in other place
21:35:53	*	pregressive joined #nim
21:36:09	OnO	that could (possibly) lead to some strange errors
21:36:41	OnO	or maybe at least issue me a warning
21:37:44	*	mwbrown quit (Ping timeout: 245 seconds)
21:40:26	OnO	Jehan_: also regarding your answer on forum, can you please tell if colon is necessary for syntax disambiguation?
21:41:45	OnO	I don't really mind colon, but wish to understand better reason of its existence :>
21:42:06	Jehan_	OnO: Simple example: "if foo 1 + 1" could be parsed as either "if foo(1): 1" or "if foo: 1+1"
21:42:27	Jehan_	The primary reason it exists is the same as in Python. As a visual marker.
21:42:38	Jehan_	Same reason we have punctuation in sentences, basically.
21:42:42	OnO	oh, sure this is clear for one liner
21:42:57	OnO	but if I use indent for block, is it really necessary?
21:43:15	polkm	so fowl, in your tests you wrote "face.indices[iii] " but when I do that I get "expression has no address"
21:43:44	Araq	OnO: there are ways to do without it
21:43:48	*	HakanD_ quit (Quit: Be back later ...)
21:44:20	*	Trustable quit (Quit: Leaving)
21:44:54	fowl	polkm, face here is a pointer (from above: let face = mesh.faces.offset(ii))
21:45:17	OnO	Araq: so it could be optional, right? and Nim could let complainers to skip it
21:47:42	Jehan_	OnO: Yes, but then you'd have multiple dialects.
21:48:08	*	OnO_ joined #nim
21:49:21	polkm	fowl, I coppied that line too, but am still getting the error, my code: http://pastebin.com/8HeBieW6
21:49:25	reactormonk	for an s-expression serializer/deserializer, take the json module as example?
21:49:59	Jehan_	I'm planning to have an FAQ for my own language that says that any requests for syntax changes have to be accompanied by a peer-reviewed study. :)
21:50:49	*	OnO quit (Ping timeout: 264 seconds)
21:52:21	Araq	OnO_: yeah, but there is always be a lighthouse, there is always a man, there is always a city
21:52:22	*	bcinman quit (Quit: My Mac has gone to sleep. ZZZzzz…)
21:53:38	OnO_	okay, fair enough, as I said, I don't mind colon, even I'd say I like it as it provides punctuation
21:53:57	Jehan_	Ono: The guy also wants "then" as a keyword in lieu of the colon. Then someone will want "do" for while, and someone else prefers "then" for while instead (because consistency).
21:54:30	Jehan_	At some point, someone has to make a decision. It's easier to learn a common language rather than having N different dialects.
21:54:37	OnO_	Jehan_: if deferred RC is the one is default, why I see major speedup using M&S GC in my little benchmark?
21:54:58	Araq	because throughput != responsiveness
21:55:45	OnO_	so M&S GC would cause regular hiccups or sthing?
21:56:17	Araq	exactly
21:56:26	Jehan_	OnO_: Because you're dealing with a small heap, presumably.
21:56:45	Jehan_	Or a heap that gets collected in its entirety.
21:57:01	Jehan_	The bigger the "unchanging" part of a heap, the worse M&S gets.
21:57:10	pigmej	cu guys!
21:57:23	OnO_	cya
21:57:24	Jehan_	Because you have to do tracing work for it each time even though you aren't collecting it.
21:58:31	Araq	that said, M&S could do bulk deallocations more easily (but doesn't)
21:58:49	OnO_	okie, but still I don't understand why RC suck so much, are there any autorelease pools or sthing like in ObjC
21:59:46	OnO_	so we can defer deallocation once the response is completely over for this web example?
21:59:53	Jehan_	DRC doesn't suck.
22:00:14	Jehan_	But it has some overhead for small heaps that M&S doesn't have, such as a write barrier.
22:00:36	Jehan_	On the plus side, it generally has less memory overhead than M&S. Or should. :)
22:01:40	Jehan_	Basically, each time you write a pointer to the heap (or a global variable), the write barrier needs to update reference counts. That's a cost that the M&S collector doesn't have.
22:03:28	Jehan_	Oh, M&S can also be slower for some simple programs for other reasons.
22:03:44	Araq	a pointer *in* the heap to the heap to be more precise
22:04:21	Araq	but async transforms stack allocations into heap allocations
22:04:25	Jehan_	Araq: May have been unclear, but I meant writing a pointer to the heap.
22:04:37	Araq	so the pointers actually are in heap and not on the stack
22:04:50	Araq	and the write barrier gets expensive
22:04:53	Araq	for async
22:05:50	Jehan_	Incidentally, if you build the Boehm GC for parallel mark and sweep, you can sometimes outperform both the built-in M&S collector and deferred reference counting.
22:06:01	Jehan_	Pauses are still going to be long for multi-GB heaps.
22:06:10	*	bcinman joined #nim
22:06:20	Jehan_	But it's useful for certain kinds of batch processing of data.
22:07:05	Araq	when we decided on this async implementation we had 2 ideas of how to do it
22:07:06	OnO_	Jehan_: I just wonder if multi GB heaps will be a case for web service
22:07:36	Jehan_	OnO_: It's what the Azul GC optimizes for, for example.
22:07:45	OnO_	Jehan_: I don't see why you should handle more than N x CPU cores, where N is not so high constant
22:08:06	Araq	1. do a continuation passing transformation in the compiler and don't touch the GC
22:08:15	Jehan_	Depends on what you are doing in your backend.
22:08:16	OnO_	just let other request wait till I finish what I am doing now
22:08:34	Araq	2. implement a stack switching primitive and tell the GC a thread can have different stacks
22:08:58	*	transfuturist joined #nim
22:08:59	Araq	we picked (1). in retrospect (2) is likely to be much faster ... ;-)
22:09:23	transfuturist	does the dynlib pragma look up symbols at compile time?
22:09:39	Araq	transfuturist: no it does not
22:10:06	OnO_	Jehan_: true, that can be many requests that will stall on IO in some services.. and can grow stack
22:10:15	transfuturist	so you can unload and reload the library, call the proc, and it'll use the new library?
22:10:22	*	polkm quit (Ping timeout: 246 seconds)
22:10:31	Araq	transfuturist: yep
22:10:39	transfuturist	not what i'm getting...
22:10:43	Araq	well with a little codegen patch, it can
22:10:48	Jehan_	Araq: Well, stack switching might be painful with non-C backends.
22:11:15	Jehan_	Depending on how C++ compilers implement exceptions, for example.
22:12:44	transfuturist	Araq: but the pragma has to be given a non-nil LibHandle?
22:13:11	Jehan_	Also, it's worth keeping in mind that Stackoverflow/Stackexchange runs on some pretty modest hardware and is written in C#.
22:13:13	Araq	transfuturist: the pragma doesn't get any libhandles from your code
22:13:43	transfuturist	i was using symAddr/checkedSymAddr like in the opengl package
22:13:54	OnO_	or have thread-groups with single GC and GIL?
22:14:23	Jehan_	There are very few things that you need hyperoptimized code for.
22:14:50	transfuturist	{.pragma: ogl, dynlib: glGetProc(oglHandle, "0").}
22:15:05	transfuturist	where glGetProc boils down to symAddr
22:16:01	Jehan_	Speaking of which, with the open-sourcing of Microsoft's .NET implementation, that may become an interesting alternative backend for Nim.
22:17:02	Araq	transfuturist: the openGL package uses undocumented DLL loading features and is nothing you should do in your code :P
22:17:07	transfuturist	>>
22:17:25	OnO_	Jehan_: honestly I think LLVM would be nicer, at least if you want then to run code on GPU or sthing using LLVM numerous backends
22:17:59	Araq	not that much is lost in the C interim step. Nim -> C -> LLVM (clang)
22:18:07	Araq	vs Nim -> LLVM
22:18:28	Araq	the benefits of Nim -> LLVM are overrated
22:18:53	transfuturist	but does that mean that since the dynlib pragma uses a library path, it loads the library each time the proc is called?
22:19:14	Araq	transfuturist: no, it's loaded once on program startup for you
22:19:20	OnO_	yeah I know, the only one I can imagine is better native debugging, or maybe handling multiple return values
22:19:25	Araq	and even the proc calls are resolved at startup
22:19:35	transfuturist	but, uh... then i can't load it or unload it at runtime
22:19:40	Araq	calling works by calling a global function pointer
22:19:41	flaviu	Araq: Proper exceptions and devirtualization are lost.
22:20:14	Araq	flaviu: we do devirt on our own, proper exceptions work with the C++ backend
22:20:24	renesac	Araq: talking abou the GC/allocator, would something like this be useful for Nim: http://stackoverflow.com/questions/23044153/sized-deallocation-feature-in-memory-management-in-c1y
22:20:25	renesac	?
22:20:43	flaviu	Araq: does Nim do speculative devirtualization?
22:20:45	renesac	the dealloc funciton don't seem to take a size currently
22:21:03	transfuturist	how could hot swapping be implemented, in that case?
22:21:22	OnO_	Araq: but LLVM would definitely make easier to have interactive REPL
22:21:28	Araq	transfuturist: by a codegen patch. not hard.
22:21:43	renesac	OnO_: people are making do with TinyC
22:22:00	Araq	OnO_: how so? the problems largely remain the same
22:22:19	renesac	*TCC
22:22:38	flaviu	http://blogs.msdn.com/b/oldnewthing/archive/2011/09/21/10214405.aspx is interesting.
22:22:42	Araq	flaviu: no. but then speculative devirt can also be done for function pointers, c++ methods have no advantage afaict
22:22:57	Araq	renesac: no benefit.
22:23:00	renesac	and yeah, "nim i" was implemented with the nim VM, but couldn't handle imports
22:23:14	renesac	or FFI
22:23:16	flaviu	Araq: But instead of Nim having it's own VM, it can use the known-correct and fast LLVM VM.
22:23:48	Araq	flaviu: that's only possible when you have no idea how we use our VM :P
22:25:20	Araq	how does LLVM's VM builds Nim's ASTs for consumption of the Nim compiler? that's the primary feature of Nim's VM. and of course little details like that are always happily ignored when you propose yet again the "tried and proven solutions"
22:25:47	EXetoC	you and your details :p
22:26:07	transfuturist	Araq: could you give me a couple more keywords? where is dlsym used?
22:26:27	transfuturist	like, cgen?
22:26:40	flaviu	Araq: Bytes are bytes. The compiler could be run as a library in the VM.
22:28:28	flaviu	Of course, the LLVM VM does have other issues, like performance.
22:28:45	flaviu	IIRC Julia has it's own interpreter because LLVM is too slow.
22:29:10	transfuturist	>known-correct and fast LLVM
22:29:10	OnO_	you mean LLVM codegen being to slow?
22:29:14	transfuturist	>too slow
22:29:19	Araq	transfuturist: compiler/cgen.nim
22:29:28	OnO_	because LLVM has no VM ;)
22:29:35	EXetoC	transfuturist: yeah, some people will want to do selective loading and querying
22:29:47	gokr	I would also guess Julia doesn't use LLVM since (again a little guess) its dynamically typed and LLVM doesn't seem too targeted at that.
22:29:57	EXetoC	actually, you could just omit that call then I think
22:30:00	renesac	Araq: what was the paper that you based nim's allocator?
22:30:09	OnO_	gokr: no it is not dynamically typed
22:30:32	renesac	I remember you linking it to me, but I can't find anymore
22:30:34	Araq	flaviu: ah I see, so *if* Julia implements its own interpreter that's because of valid reasons, but if Nim does it, it's all wrong because you happen to know Nim's creator.
22:30:39	OnO_	it is just functions are templates by default that get concretized before generating call instruction
22:30:56	transfuturist	kek
22:31:03	gokr	OnO_: Care to elaborate?
22:31:10	flaviu	Araq: Perhaps Julia's decision is wrong, I don't know enough.
22:31:24	Araq	your double standards are so annoying that I'm about to /ignore you.
22:31:59	OnO_	gokr: if you define method with no types it behaves like a template, it does not emit any code, until it encounters a call with solid type
22:32:25	OnO_	gokr: but in the end you have several "versions" for a method based on given parameters type
22:32:38	OnO_	gokr: so it works exactly like in Nim
22:32:45	Varriount	flaviu: On the subject of that oldnewthing article, Yeah, quite interesting.
22:33:04	Varriount	It's one of the few things windows has that linux didn't
22:33:08	gokr	OnO_: http://stackoverflow.com/questions/28078089/is-julia-dynamically-typed
22:33:19	gokr	I would argue its dynamically typed.
22:35:03	gokr	Or just read here: http://docs.julialang.org/en/release-0.3/manual/types/
22:35:59	*	johnsoft quit (Ping timeout: 265 seconds)
22:36:56	Jehan_	If there's genuine interest, I can ask my officemate tomorrow, he's pretty deeply invested in Julia.
22:37:38	flaviu	Varriount: Just wait for kpatch :)
22:37:50	gokr	Jehan_: Its not hard to read the page I pointed at.
22:37:57	gokr	Its dynamically typed, period.
22:38:13	Jehan_	gokr: Yeah, but from what I recall, that's not really the story of the implementation.
22:38:29	Jehan_	Which I think is what people were talking about.
22:38:30	OnO_	gokr: well... okay, you are right Julia as language is dynamically types, but the execution model uses static typing
22:38:34	Varriount	flaviu: There's a reason I used the word "didn't"
22:39:03	OnO_	gokr: I mean machine code gets concretized for particular arguements
22:39:37	*	johnsoft joined #nim
22:39:44	gokr	OnO_: But you know, that's what a JIT does.
22:39:45	OnO_	gokr: so... in many places Julia tend to use static typing, to avoid dynamic paths
22:40:35	OnO_	gokr: not only JIT, in many places and many Julia libraries you prefer naming types or your method arguments explicitly
22:40:47	transfuturist	was the entire point of the dynlib package to provide support to opengl and similar?
22:41:34	OnO_	gokr: I think Methods section describe it better http://julia.readthedocs.org/en/latest/manual/methods/
22:43:11	OnO_	so Julia's function are dynamic, where Julia methods are statically typed materializations of these dynamic functions
22:44:19	OnO_	what is most sexy about Julia is that you can actyally dump machine code of such methods
22:45:43	gokr	My guess was that the "type specific" methods was simply a way to help the JIT to be able to generate more efficient code.
22:46:30	gokr	Their manualy clearly states only values have types and there is no meaningful concept of a "compile-time type".
22:47:08	gmpreussner|work	Varriount: i will update the PR when i get home tonight. thanks for taking a look!
22:47:18	OnO_	but "compile-time" refers to time of compiling script, not the time when final machine code is generated
22:47:44	gokr	But anyway, Eliot Miranda who implements the Cog VM for Smalltalk has looked at LLVM and also came to the conclusion that its not really fitting for making a JIT for dynamically typed languages.
22:47:55	OnO_	http://www.evanmiller.org/why-im-betting-on-julia.html
22:48:27	Varriount	gmpreussner: What version of Sublime Text do you use?
22:48:30	OnO_	quite similar to LuaJIT with its tracing compiler
22:48:51	OnO_	erm... tracing JIT compiler
22:48:55	*	bjz quit (Quit: My MacBook Pro has gone to sleep. ZZZzzz…)
22:49:46	flaviu	Huh, julia might be quite nice for learning assembler.
22:51:08	gokr	OnO_: Humhum, Julia does use LLVM. I thought someone said it didn't?
22:52:49	OnO_	gokr: I am bit sleepy, but I think someone said that it has it's own interpreter which at some points generates LLVM then machine code
22:52:56	gokr	flaviu: Ah, it was you.
22:53:04	OnO_	but it still needs interpreter to handle dynamic types
22:53:07	renesac	flaviu: I think I might make a feature request for the nim repl
22:53:09	renesac	XD
22:53:40	onionhammer	Varriount what was it you mentioned the other day about syntax
22:54:11	gokr	OnO_: As I have said all along - its a JIT. That uses LLVM for the final assembly generation.
22:54:23	renesac	it shouldn't be too difficult to do an objdump on the generated executable for a new function
22:54:59	*	fowl quit (Quit: Leaving)
22:55:04	OnO_	gokr: yup
22:55:17	*	vendethiel quit (Ping timeout: 248 seconds)
22:56:25	OnO_	renesac: you can do it with most compilers with -c -S -o -, would be easier than translating back from binary
22:56:43	renesac	indeed
22:58:29	OnO_	okay guys, I am going sleep, that was great discussion, still lot of new things to learn about Nim... see you tommorow
22:58:46	gokr	cya
22:58:57	*	OnO_ quit (Quit: Textual IRC Client: www.textualapp.com)
22:59:05	Varriount	onionhammer: ST3 is getting new syntax definition format
22:59:48	Varriount	onionhammer: http://www.sublimetext.com/docs/3/syntax.html
23:00:10	Jehan_	Varriount: I saw that, too. Can't be worse than the TextMate format. :)
23:00:22	Jehan_	And I'm saying this is as someone who really likes TextMate.
23:00:54	Varriount	Jehan_: Well, the context mechanism should really help cut down on repetitive regexes
23:01:32	Jehan_	Varriount: I may have to give Sublime Text a closer look after all.
23:02:45	*	Mimbus joined #nim
23:03:30	Varriount	Though, a built-in parsing feature might be better. :/
23:09:42	*	buMPnet quit (Read error: Connection reset by peer)
23:10:41	*	buMPnet joined #nim
23:12:03	*	Matthias247 quit (Read error: Connection reset by peer)
23:28:29	gmpreussner|work	Jehan_: sublime with NimLime is awesome
23:28:53	gmpreussner|work	i use it on all platforms now
23:29:32	Jehan_	gmpreussner|work: I believe you, but I'm still heavily invested in Vim.
23:29:44	BlaXpirit	I'm heavily invested in Kate.
23:30:01	Jehan_	Oddly enough, one of the major reasons is that on OS X that's one of the ways I can have focus-follows-mouse. :)
23:30:07	Varriount	Araq: Have you tried Arak? (The drink)
23:31:45	Araq	hrm ... I don't think so
23:34:01	gmpreussner|work	BlaXpirit: i stopped using Kate, because their GDB plug-in does not work with Nim (because of the way it communicates with GDB - the author is aware)
23:34:10	gmpreussner|work	even then, sublime is a better editor than kate, imho
23:34:20	BlaXpirit	why does it have to be GDB o.o
23:34:30	BlaXpirit	i mean, why is Kate even supposed to support it
23:34:53	BlaXpirit	i'm probably missing out.
23:35:32	gmpreussner|work	i like to step through my programs within the editor :)
23:35:49	BlaXpirit	yup, i'm missing out
23:36:08	gmpreussner|work	to be fair, i haven't tried it with sublime yet either, but i looked at their code, and i think it will work
23:36:33	Jehan_	Hmm, does NimLine work with ST2 or do you have to get ST3?
23:36:44	Jehan_	NimLime*
23:37:23	BlaXpirit	Jehan_, it can't not work with 2
23:37:43	gmpreussner|work	i'm using 2.0.2
23:37:49	BlaXpirit	pretty sure it's much older than any alpha version of sublime 3
23:37:58	Jehan_	I'm honestly not at all familiar with the ST ecosystem, and I know some stuff requires ST3.
23:51:06	*	TEttinger joined #nim
23:54:39	*	fungos joined #nim
23:55:42	transfuturist	Araq: is there more documentation on the compiler source besides http://nim-lang.org/intern.html ?