<< 11-04-2015 >>

---

00:00:06	BlaXpirit	renesac, sure .get gives u exceptions
00:00:22	BlaXpirit	and yes, these are are also my favorite Maybe names
00:00:46	renesac	well, but then the code ends up littered with .gets
00:00:53	renesac	if people starts using maybes everywhere
00:01:06	renesac	I don't know
00:01:18	flaviu	Doing headers with RST is a PITA.
00:01:32	BlaXpirit	3 first words were unnecessary
00:01:41	renesac	lol
00:01:43	flaviu	I can't write the header until I've rendered it, since the underline needs to be the same length as the header.
00:01:53	BlaXpirit	o.o
00:02:07	BlaXpirit	are you doing nre?
00:02:16	renesac	when the nim forum will accept markdown instead of rst?
00:02:20	renesac	:P
00:02:27	flaviu	BlaXpirit: I'm calling it "docweave".
00:02:32	*	saml_ joined #nim
00:02:50	BlaXpirit	so you're making some trickery
00:03:07	BlaXpirit	flaviu, but you'd still need to do rst without trickery if it is to be in stdlib
00:03:17	BlaXpirit	there will be no readme
00:03:36	flaviu	BlaXpirit: I don't really care about this module getting into the stdlib.
00:03:43	flaviu	It's just a utility.
00:03:49	BlaXpirit	oh come on
00:04:23	flaviu	Actually, I'd be actively against this being put in the stdlib.
00:04:31	BlaXpirit	-__-
00:05:07	BlaXpirit	no reason
00:05:15	renesac	what module?
00:05:53	flaviu	BlaXpirit: It doesn't belong.
00:05:58	flaviu	https://github.com/flaviut/docweave
00:06:40	*	bcinman quit (Quit: My Mac has gone to sleep. ZZZzzz…)
00:06:50	BlaXpirit	flaviu, it will be the dawn of stdlib
00:07:02	flaviu	This module does not belong.
00:07:07	*	bcinman joined #nim
00:07:08	BlaXpirit	well i'm talking about nre
00:07:20	flaviu	Oh, sorry for the confusion then.
00:07:26	BlaXpirit	and i assumed you wanted to make "docweave" to do tricks with "nre"s docs
00:07:46	flaviu	Nah, I just want it to extract nre's docstrings into a readme.
00:08:16	BlaXpirit	ok
00:08:34	renesac	no comments or documentation, I'm not really sure what it does from a first glance
00:08:44	BlaXpirit	> extract nre's docstrings into a readme
00:08:45	renesac	an alternative for html doc gen?
00:09:02	BlaXpirit	i assume it's like rst doc gen
00:09:32	BlaXpirit	generate a rst file
00:10:06	renesac	hum
00:10:43	flaviu	BlaXpirit: Yeah.
00:11:26	renesac	BlaXpirit: any reason why your proposed Maybe don't have the `or` operator?
00:11:50	renesac	and I'm not really sure if I'm a fan of the "?" and "?=" operators
00:12:03	BlaXpirit	renesac, the reason is i don't get it
00:12:28	BlaXpirit	if you are knowledgable of these things, do give feedback
00:12:38	BlaXpirit	for now i'm just gonna make a pull request with the base version
00:12:41	renesac	well, I programmed some in Lua
00:13:14	renesac	and there the "or" on non booleans would test for nil and return the first non nil object
00:13:17	renesac	or nil
00:13:40	BlaXpirit	this is not Nim's style
00:13:45	renesac	that seems exactly what that `or` operator is for
00:13:56	renesac	what is nim's style?
00:14:11	BlaXpirit	`or` operator returns a boolean or the result of a bitwise operation
00:14:34	renesac	the `or` operator only returns a boolean if comparing two booleans
00:14:49	renesac	it is overloaded for integers, to do something completely different
00:14:58	BlaXpirit	like i said
00:15:08	renesac	it can also be overloaded to Maybe[T]
00:15:23	dom96	it's also overloaded for Future[T]
00:15:26	renesac	to do something closer to it's logical meaning
00:16:15	renesac	the only problem with all those overloads is if you make a mistake somewhere, it can pass quietly because the other type is also acceptable there
00:16:35	renesac	I'm not sure how much of a problem it is in real life though
00:16:50	BlaXpirit	m or x is not worth it because nim has if m: m else: x
00:16:51	*	bcinman quit (Quit: My Mac has gone to sleep. ZZZzzz…)
00:16:54	BlaXpirit	or something like that
00:17:08	flaviu	Instead of or, it's possible to do myOption.get(foo)
00:17:16	flaviu	If myOption is None, that returns foo.
00:17:25	BlaXpirit	it's slightly different but yes, i have that in
00:18:34	renesac	flaviu: it isn't as clear what "foo" means in this expression by just looking at it w/o knowing the name of the parameter
00:18:56	BlaXpirit	it's anything
00:19:03	renesac	I would guess you were getting a "foo" field in myOption
00:19:08	BlaXpirit	m.get(foo) is pretty much m or foo
00:19:18	renesac	yeah
00:19:26	BlaXpirit	it's a kind of a pythonism
00:19:38	flaviu	Yes, inspired by python.
00:19:52	renesac	it is?
00:20:01	dom96	OT: I love macros.quote.
00:20:23	BlaXpirit	https://docs.python.org/dev/library/stdtypes.html#dict.get
00:21:02	renesac	hum
00:21:29	renesac	I still don't like the look of it
00:22:02	renesac	of course, it would be annoying if dic["abc"] returned an option I guess
00:22:03	BlaXpirit	it's just so usual to me from python that i didn't consider that it's weird
00:22:32	BlaXpirit	other names welcome
00:22:38	renesac	`or`
00:22:46	BlaXpirit	nope :|
00:23:16	BlaXpirit	although...... actually it kinda makes sense
00:23:46	BlaXpirit	if m1 or m2: will still work
00:24:16	*	Matthias247 quit (Read error: Connection reset by peer)
00:24:29	*	davidhq quit (Quit: My MacBook Pro has gone to sleep. ZZZzzz…)
00:24:55	BlaXpirit	should just steal these https://github.com/fowlmouth/nimlibs/blob/master/fowltek/maybe_t.nim#L31-L35
00:25:04	*	jholland quit (Quit: Connection closed for inactivity)
00:25:24	BlaXpirit	and by L31-L35 I mean L24-L35
00:25:25	renesac	I don't understand that one
00:25:43	renesac	(it's my turn now)
00:25:54	BlaXpirit	just read the code, it's easy
00:26:33	renesac	oh
00:26:52	renesac	instead of doing my.x in a normal object
00:27:03	renesac	do my.?x if that object is wrapped by maybe?
00:27:47	BlaXpirit	.? is just . for 'maybe', yes
00:28:00	BlaXpirit	but it returns a maybe and not just the value
00:28:48	renesac	oh, it returns a maybe?
00:29:38	renesac	I guess it makes sense
00:34:45	renesac	and yeah, "none" will certaninly generate a lot of name clashes
00:35:03	renesac	when importing this module
00:35:20	BlaXpirit	nothing
00:36:01	renesac	much longer, but probably not very used exacly for this reason
00:36:07	renesac	so a possibility
00:37:15	*	user7181 quit (Quit: leaving)
00:37:27	renesac	though there is the overload resolution
00:37:41	renesac	I'm not sure if a variable with the same name would be usable
00:37:55	renesac	or if it would shadow it
00:39:54	BlaXpirit	shadow
00:42:28	*	vendethiel quit (Ping timeout: 272 seconds)
00:44:40	flaviu	Lol, that code has some really catastrophic backtracking.
00:44:57	BlaXpirit	which
00:45:29	*	irrequietus quit ()
00:45:45	flaviu	Mydocweave
00:45:55	flaviu	s/Mydocweave/docweave/
00:50:52	Varriount	Yay, finally removed Bittorrent in favor of something less... horrid.
00:52:58	flaviu	Varriount: I'd be happy to discuss the merits of various torrent clients in #nim-offtopic.
01:02:29	*	wb quit (Ping timeout: 276 seconds)
01:09:27	*	chop1n joined #nim
01:10:20	*	chop1n left #nim ("Textual IRC Client: www.textualapp.com")
01:27:22	*	vendethiel joined #nim
01:34:54	*	HakanD___ joined #nim
01:36:03	*	ChrisMAN joined #nim
01:39:12	*	HakanD___ quit (Ping timeout: 250 seconds)
01:43:20	BlaXpirit	how to type `()` inside `` ``
01:43:29	BlaXpirit	in doc comment
01:43:45	BlaXpirit	```()``` doesn't work, neither does `` `()` ``
01:47:43	*	BitPuffin quit (Ping timeout: 252 seconds)
01:49:41	*	flaviu quit (Read error: Connection reset by peer)
01:49:41	*	Mimbus quit (Read error: Connection reset by peer)
01:50:31	def-	BlaXpirit: I wondered about this as well. Haven't found a solution yet
01:54:04	*	flaviu joined #nim
02:02:30	BlaXpirit	https://github.com/Araq/Nim/pull/2515 "Add optionals module (Maybe[T])"
02:06:27	*	BlaXpirit quit (Quit: Quit Konversation)
02:10:56	*	ChrisMAN quit (Ping timeout: 264 seconds)
02:12:26	*	a5i joined #nim
02:26:04	*	pregressive joined #nim
02:30:20	*	pregressive quit (Ping timeout: 250 seconds)
02:40:43	*	pregressive joined #nim
02:42:58	*	gsingh93 joined #nim
02:45:07	*	jholland joined #nim
02:45:34	*	vendethiel quit (Ping timeout: 250 seconds)
02:53:55	*	darkf joined #nim
02:55:43	*	TEttinger joined #nim
03:38:11	*	ajpocus joined #nim
03:38:27	*	gsingh93 quit (Quit: WeeChat 1.1.1)
03:39:22	*	gsingh93 joined #nim
03:41:43	*	ajpocus quit (Quit: Leaving)
03:52:19	reactormonk	any way to get the return value of an executed process AND the input/output streams?
03:52:42	reactormonk	ah, just go with waitForExit
03:52:48	reactormonk	or peekExitCode. neat.
03:59:55	*	mpthrapp_ joined #nim
04:02:40	*	mpthrapp quit (Ping timeout: 255 seconds)
04:04:44	*	bcinman joined #nim
04:14:46	*	endragor joined #nim
04:30:22	*	endragor quit (Remote host closed the connection)
04:33:15	*	a5i quit (Quit: Connection closed for inactivity)
04:53:50	*	dashed joined #nim
04:58:45	*	wb joined #nim
05:21:12	*	dtscode is now known as dtscokeacola
05:46:07	*	saml_ quit (Remote host closed the connection)
05:59:19	*	pregressive quit (Remote host closed the connection)
06:04:46	*	endragor joined #nim
06:30:22	*	dtscokeacola is now known as dtscode
06:34:36	*	dtscode is now known as charmander
06:42:09	*	Demon_Fox joined #nim
06:49:24	*	bcinman quit (Quit: My Mac has gone to sleep. ZZZzzz…)
06:58:42	*	EXetoC quit (Remote host closed the connection)
07:07:29	*	charmander is now known as dtscode
07:08:22	novist	I get scratch.nim(19, 43) Error: cannot evaluate at compile time: ctor_counter when my code is imported from other module. any idea why that could be? executing module itself works fine.. https://paste2box.com/iQY-cQ#/P6EiojsvWIv6czhlvRk8O6QlFFuVnvMWOedq7UhypTU/scratch.nim
07:12:08	Varriount	novist: Likely a bug or vm limitation
07:12:17	*	akiradeveloper joined #nim
07:12:44	*	BlaXpirit joined #nim
07:14:49	*	gsingh93 quit (Ping timeout: 245 seconds)
07:15:45	*	bcinman joined #nim
07:15:58	*	bcinman quit (Client Quit)
07:17:04	novist	hmm i must have bad karma, tripping on all these bugs hehe
07:25:24	*	MagusOTB joined #nim
07:27:58	fowl	novist, try var ctor_counter{.compileTime.} = 0
07:28:56	novist	hey that worked, thanks!
07:29:07	novist	although.. whats the point of "static:" then?
07:29:25	novist	it works even without static block
07:29:41	fowl	to execute code at compile time
07:29:55	fowl	it doesnt necessarily live beyond the life time of the module though
07:29:58	Varriount	novist: Could you submit a PR to the pragma part of the manual adding that the compileTime pragma can be used on variables?
07:30:14	Varriount	https://github.com/Araq/Nim/blob/devel/doc/manual/pragmas.txt
07:30:34	novist	sure thing
07:30:56	fowl	whats up Varriount
07:31:16	fowl	working on anything interesting?
07:31:41	Varriount	fowl: I don't know. Do Calculus tests counts as 'interesting'?
07:32:10	fowl	not to me :p
07:34:21	*	dashed quit (Quit: Connection closed for inactivity)
07:42:39	*	vendethiel joined #nim
07:56:53	*	akiradeveloper quit (Ping timeout: 250 seconds)
07:58:06	*	baby_rage_nospac joined #nim
08:01:34	*	Arrrrrrr joined #nim
08:03:04	Varriount	Hello Arrrrrrr
08:03:12	baby_rage_nospac	I'm asian and that's racist
08:03:25	Varriount	?
08:03:42	Arrrrrrr	Hello Varriount. Hello Nimrods.
08:03:42	baby_rage_nospac	we don't spell "l" as "r"
08:04:11	Arrrrrrr	Im very sorry baby_rage_nospac
08:04:16	*	Arrrrrrr is now known as Arrrrrrrrrrrrrrr
08:04:36	baby_rage_nospac	now I'm 200% triggered now
08:05:13	Varriount	Um... ok?
08:05:31	fowl	do not continue this conversation
08:05:43	baby_rage_nospac	:^)
08:06:20	*	baby_rage_nospac quit (Quit: Leaving)
08:07:12	fowl	obvious troll is obvious
08:07:13	Araq	aha single letters can be racist too now. interesting.
08:07:40	*	vendethiel quit (Ping timeout: 244 seconds)
08:08:55	Arrrrrrrrrrrrrrr	How does nim resolves the concurrency problem? http://blog.rust-lang.org/2015/04/10/Fearless-Concurrency.html
08:08:57	BlaXpirit	some modules in nim standard library do tests via isMainModule. are they included in koch test?
08:09:20	Varriount	BlaXpirit: Yes.
08:09:26	BlaXpirit	Varriount, but how?
08:09:38	Varriount	BlaXpirit: The tester compiles and runs the module.
08:09:51	BlaXpirit	but why does it run it?
08:10:01	Varriount	BlaXpirit: To test things?
08:10:03	BlaXpirit	does it just run all of stdlib?
08:10:15	Varriount	BlaXpirit: Yes.
08:10:15	BlaXpirit	even if most modules have no tests?
08:10:19	BlaXpirit	ok..
08:10:22	Varriount	BlaXpirit: Yes.
08:10:28	BlaXpirit	so i suppose i'll test my new module in ismainmodule
08:11:37	Araq	Arrrrrrrrrrrrrrr: thread local heaps + a *bunch* of other features like locking levels to prevent deadlocks statically, annotations to enforce you do your locking consistently, message passing etc
08:12:20	Varriount	Araq: I don't think I've seen any other language that has locking levels. It's quite an itneresting feature.
08:12:56	Arrrrrrrrrrrrrrr	You should write a post explaining this, people would like to learn more (i guess).
08:13:02	Araq	BlaXpirit: the tester tests the module at least compiles and if it has isMainModule it is also *run*
08:13:17	BlaXpirit	is it in any category?
08:13:35	Araq	yes, category 'lib'
08:13:46	Varriount	Araq: I don't think that the tester actually tries to detect if a module has an 'isMainModule' in it.
08:13:54	Varriount	It just runs the exe
08:13:58	BlaXpirit	oh ok. lib+stdlib for stdlib then
08:14:22	Araq	Varriount: but I did program it this way, let me take a look
08:14:59	BlaXpirit	I see exactly as many executables as sources
08:15:35	Araq	if contents.contains("when isMainModule"):
08:15:36	Araq	testSpec r, makeTest(test, options, cat, actionRun)
08:15:38	Araq	else:
08:15:39	Araq	testNoSpec r, makeTest(test, options, cat, actionCompile)
08:15:51	Araq	maybe I got the logic wrong :P
08:16:15	Araq	also it's apparently case sensitive so if you write ismainmodule it won't work :P
08:16:19	Varriount	Well, I stand corrected.
08:16:41	BlaXpirit	so weird, I put an when isMainModule: assert false and it passed
08:17:14	Araq	BlaXpirit: now that *sucks*
08:17:26	BlaXpirit	i'm probably doing something wrong
08:20:31	BlaXpirit	the module is being tested for a smaller amount of time than the rest
08:20:38	fowl	Araq, do you use generics with macros.getType at all? i want to change them quite a bit
08:20:55	Araq	fowl: go head
08:21:08	Araq	we don't use them in getType for now
08:21:10	fowl	cool
08:23:07	BlaXpirit	reNimcCrash when I put syntactically wrong code, even in isMainModule, but assert(1 == 2) just passes
08:27:10	Varriount	BlaXpirit: Well duh. Didn't you hear? Math is a lie.
08:27:27	Varriount	1 == 2, and 3/0 = Fish
08:28:20	BlaXpirit	dude
08:28:29	BlaXpirit	does this mean lib testing actually doesn't work?
08:28:39	BlaXpirit	i added assert false to other module and it passed
08:28:45	BlaXpirit	can someone try this too?
08:29:06	BlaXpirit	add assert false to actors.nim or something and koch test c lib
08:31:02	fowl	BlaXpirit, try do_assert before you run through the town screaming that the british are coming
08:31:07	BlaXpirit	i did
08:31:16	fowl	carry on then
08:31:33	fowl	lol
08:31:46	TEttinger	it's kinda amazing to me how active this channel is (maybe because development is so active and not afraid to break things once to improve them later :P )
08:33:07	BlaXpirit	hellooo
08:33:58	Arrrrrrrrrrrrrrr	Is nice because your questions receive a fast answer.
08:34:09	Arrrrrrrrrrrrrrr	I have learned a lot in 2 days.
08:36:08	novist	just noticed [0 .. ^2] thing
08:36:12	novist	man slices are total mess..
08:36:22	BlaXpirit	were
08:36:26	*	Jehan_ joined #nim
08:37:20	novist	i really dont like special cases... it doesnt look self explanatory at all. usual characteristic of special cases
08:37:35	*	Sembei joined #nim
08:37:37	BlaXpirit	dont use it
08:38:08	novist	why ranges and slices had to be smashed into the one syntax at all
08:38:32	Arrrrrrrrrrrrrrr	what does ^2 in that context?
08:38:44	novist	like [0:-2] in python
08:38:47	TEttinger	there's quite a few languages coming close to or just getting into a "mature and usable" state now. Julia is nice, but last I checked couldn't compile a standalone executable (which makes it kinda pointless for games). Rust I kinda doubt is as close to ready as advertised, a friend of mine uses the rust nightly and things break constantly.
08:39:15	Jehan_	Julia really has a different target audience.
08:39:36	Jehan_	Mathematicians and scientists in general.
08:39:54	BlaXpirit	which is a pity
08:40:00	Arrrrrrrrrrrrrrr	what does [0:-2] in python?
08:40:29	TEttinger	Nim's older than those two, I think, and it also performs very well... I think the imperative non-heavily-OOP model is a pretty good fit for a lot more software than people think if they've been conditioned into thinking with design patterns and best practices
08:40:31	BlaXpirit	a[0 .. ^2] is exactly the same as a[0 .. a.len - 2]
08:40:32	novist	but seriously why array/string slicing could not be like [0:-2]? then .. could have stayed inclusive and ones that dong like could implement range(start, stop, step) that produces non-inclusive range. world would be happy, not weird
08:40:46	Araq	novist: negative indexing in Python is a *special* case, moreso than Nim's ^
08:40:54	BlaXpirit	novist, you don't understand it. it's all been discussed.
08:40:57	novist	but it sure does not look like special case
08:41:17	Araq	proof: the implementation in Python uses an 'if' to implement it :P
08:41:17	novist	yeah i have seen discussion stop two weeks ago..
08:42:19	Araq	it's called "coming to a conclusion"
08:42:23	BlaXpirit	so can anyone confirm or deny that nim's standard library is not being tested at all?
08:42:45	Araq	BlaXpirit: will look into it, but I'm afraid you're right
08:42:59	BlaXpirit	it takes 1 minute to check, srsly
08:42:59	fowl	BlaXpirit, tonight is the first i've heard of those tests supposed to be running
08:43:06	Araq	good thing we also have the stdlib directory
08:43:13	BlaXpirit	that one works
08:44:15	Araq	novist: in Nim foo[-1] never worked, now foo[^1] does. a definite improvement and no runtime overhead, in contrast to -1
08:44:27	Araq	but I won't repeat more of the arguments here
08:44:36	novist	i just dont understand why there has to be introduced special less obvious syntax
08:44:54	BlaXpirit	"less obvious to a Python programmer"
08:44:59	novist	everyone knows what -1 is. now what is ^1? nothing xor 1?
08:45:02	BlaXpirit	you're not making sense
08:45:15	BlaXpirit	no, not everyone knows what -1 "is"
08:45:23	BlaXpirit	i will stop now.
08:45:51	novist	its pretty obvious its "negative" right? and if it is used as index of indexable object then its negative index. no?
08:45:55	Triplefox	the thing that triggered the most discussion (at least for me) was that inclusive range behavior plus negative indexes created some non-obvious bug situations
08:46:26	novist	Triplefox: dont forget pervasive -1 with inclusive ranges
08:46:47	novist	for i in 0..ID_LENGTH-1: .. /facepalm
08:47:26	Araq	that too has been discussed to death
08:47:30	*	Sembei quit (Max SendQ exceeded)
08:47:45	Araq	for c in 'A'..'Z': # not so stupid to be consistent, is it?
08:47:53	novist	no its great
08:48:23	fowl	i do wish for ... too but only because i come from ruby
08:48:24	novist	but its not used more often than 0..a-1
08:49:16	Arrrrrrrrrrrrrrr	is in ruby x...y == x..y-1 in nim?
08:49:37	fowl	yea
08:49:53	Arrrrrrrrrrrrrrr	It's a good idea.
08:50:12	novist	anyone see a conflict in that? shorter .. produces longer range, while longer ... produces shorter range. odd
08:50:45	novist	although i think nothing prevents defining custom ... operator right?
08:50:55	Arrrrrrrrrrrrrrr	I would not care.
08:51:11	fowl	lol
08:51:12	Arrrrrrrrrrrrrrr	Is a good observation tho.,
08:51:18	BlaXpirit	novist, yes, after negative indices were removed, nothing prevents it
08:51:31	*	HakanD___ joined #nim
08:52:01	novist	anyways for slicing in making myself proc slice(...). less weirdness ^_^
08:53:55	Arrrrrrrrrrrrrrr	Is there a way to making procs/types be autoincluded in every module like ints, without hacking the source code?
08:54:45	BlaXpirit	Arrrrrrrrrrrrrrr, http://blaxpirit.com/blog/4/make-a-nim-module-available-to-all-your-projects.html
08:55:01	Arrrrrrrrrrrrrrr	Ah, you have a blog. I'll have a look a tit
08:56:26	novist	BlaXpirit: i dont think its what he meant
08:56:28	Arrrrrrrrrrrrrrr	But does it save you to import the code?
08:56:33	BlaXpirit	no
08:57:06	novist	i was too thinking that a way to define common pragmas/imports in main project file and have them effective through entire project would be useful
08:57:30	BlaXpirit	include would work
08:57:32	Jehan_	You can use --import or --include to add a module that's automatically imported/included.
08:57:54	novist	oh was not aware of cmd line switches, awesome then, thanks Jehan_
08:58:53	Araq	yup, --import and --include are so often overlooked it's crazy
09:00:00	*	vendethiel joined #nim
09:00:28	*	pregressive joined #nim
09:02:28	Jehan_	There are a lot of other interesting command line switches that are often overlooked, too. :)
09:03:00	Arrrrrrrrrrrrrrr	Your random is fantastic BlaXpirit, have you seen this? http://www.pcg-random.org/ in theory looks even better than mersenne.
09:03:28	BlaXpirit	author sucks
09:03:36	BlaXpirit	promised to at least 3 people to change license
09:03:40	Arrrrrrrrrrrrrrr	lol
09:03:50	Arrrrrrrrrrrrrrr	Maybe he doesn't have the time
09:03:53	BlaXpirit	asked for a pull request to change license, got 3 of them for each project
09:03:59	BlaXpirit	completely ignored
09:04:23	BlaXpirit	also it's a new thing without any references from serious sources
09:04:23	Arrrrrrrrrrrrrrr	What's the problem with apache 2 license?
09:04:29	BlaXpirit	read it and see
09:04:56	*	pregressive quit (Ping timeout: 264 seconds)
09:05:53	Arrrrrrrrrrrrrrr	tl;dr: ?
09:06:07	Arrrrrrrrrrrrrrr	I saw it in some java projects like libGDX
09:06:42	Jehan_	Araq: Are there any plans to get rodfiles working again in the future (I realize that it's not going to be a pre-1.0 thing regardless)?
09:06:49	*	Sembei joined #nim
09:09:00	fowl	why isnt string just "distinct seq[char]"
09:09:38	HakanD___	thanks all for help
09:09:43	HakanD___	for yesterday
09:09:45	HakanD___	bugs fixed (:
09:09:49	*	Sembei quit (Max SendQ exceeded)
09:10:19	fowl	cool HakanD___
09:11:39	Jehan_	fowl: Because it's magic? I mean, literally. :)
09:12:00	*	Sembei joined #nim
09:14:40	*	vendethiel- joined #nim
09:14:44	fowl	Jehan_, i dont follow
09:14:50	*	vendethiel quit (Ping timeout: 272 seconds)
09:15:00	fowl	im asking 'why is it the way it is' and you're saying 'it is the way it is'
09:15:39	fowl	every string function casts it to pgenericseq
09:17:10	Jehan_	fowl: That's the code generator.
09:17:25	Jehan_	If you look at the compiler, you'll note that there's plenty of code specific for tyString.
09:18:10	*	Sembei quit (Max SendQ exceeded)
09:19:45	*	Sembei joined #nim
09:20:32	*	milosn quit (Ping timeout: 272 seconds)
09:21:26	BlaXpirit	nim-lang.org down?
09:21:44	BlaXpirit	no. apparently it's just me.
09:21:55	TEttinger	BlaXpirit, I'm also unclear on apache license stuff. MIT is problematic if someone commits something that violates a third party's patent, which BSD licenses specifically address
09:22:42	*	Sembei quit (Max SendQ exceeded)
09:25:22	TEttinger	http://xorshift.di.unimi.it/splitmix64.c is a good PRNG that fits most of the bill that PCG does
09:25:35	*	akiradeveloper joined #nim
09:25:49	TEttinger	it's SplitMix64 in the table on http://xorshift.di.unimi.it/
09:26:01	BlaXpirit	was that added recently?
09:26:26	TEttinger	yes
09:26:58	TEttinger	it's derived from the same reference implementation that Java 8 uses for its latest RNG
09:27:18	BlaXpirit	well it's extremely easy to add if u want it in your project
09:27:44	BlaXpirit	and for now i'll just say, this is the first time i see it
09:27:50	TEttinger	yep
09:28:04	TEttinger	I'm using it in a java lib I'm working on/with
09:28:21	TEttinger	(not splittablerandom, since that would need users to use java 8)
09:28:46	TEttinger	there's an advantage in that it can use ANY long as a seed
09:29:24	TEttinger	XS+128, which is otherwise excellent, needs the seed to be not "everywhere 0" after a certain number of iterations
09:29:42	TEttinger	so they recommend you use an avalanche function from murmurhash3
09:30:09	TEttinger	(both are public domain, but it's kinda a hassle)
09:30:12	BlaXpirit	TEttinger, i have this
09:30:55	BlaXpirit	https://github.com/BlaXpirit/nim-random/blob/1170147/src/random/xorshift.nim#L65
09:31:58	TEttinger	looks like that will work yeah
09:32:51	TEttinger	it's such an amazing field isn't it? this research into random numbers and how to generate them
09:33:30	TEttinger	one month, the fastest uses 1024 bits of state, the next, it's 128, and the next it's kinda 64 if you're using GCC
09:33:34	TEttinger	*clang
09:34:59	TEttinger	if I understood the math behind quasirandom number generators or low-discrepancy sequences, I'd try to implement the faure sequence
09:35:17	*	Sembei joined #nim
09:35:31	BlaXpirit	k
09:36:45	*	Sembei quit (Read error: Connection reset by peer)
09:37:16	TEttinger	heh
09:37:18	TEttinger	https://en.wikipedia.org/wiki/Low-discrepancy_sequence
09:37:43	*	akiradeveloper quit ()
09:42:10	*	Sembei joined #nim
09:43:53	*	user7181 joined #nim
09:44:57	HakanD___	Nim attracts lots of python programmers, especially those who are looking for statically typed and/or faster languages. Working on https://github.com/Araq/Nim/wiki/Nim-for-Python-Programmers , and putting on the docs might help with adoption
09:45:34	*	Sembei quit (Max SendQ exceeded)
09:45:48	*	HakanD___ is now known as HakanD
09:48:53	Arrrrrrrrrrrrrrr	I noticed there aren't too many contributors, so i suppose they are very busy.
09:49:00	*	Arrrrrrrrrrrrrrr is now known as Arrrrr
09:50:19	*	BitPuffin joined #nim
09:59:45	*	Sembei joined #nim
10:04:22	*	davidhq joined #nim
10:16:20	*	Sembei quit (Read error: Connection reset by peer)
10:16:58	*	MyMind joined #nim
10:22:22	Arrrrr	{.inheritable.} is long and ugly, can it be changed to simply {.open.} ?
10:24:07	*	rr joined #nim
10:24:30	*	rr is now known as Guest17376
10:25:08	*	Guest17376 quit (Client Quit)
10:33:43	*	johnsoft quit (Ping timeout: 245 seconds)
10:33:50	*	johnsoft joined #nim
10:38:18	Araq	Arrrrr: it's long and ugly because you should inherit from RootObj instead
10:40:16	Araq	Jehan_: rodfiles are planned for 2.0 but I'm also playing with other ideas how to implement caching
10:40:27	Jehan_	Araq: *nod*
10:40:41	Arrrrr	Ah thanks, this is (a bit) better than using {.inheritable.}
10:40:52	Jehan_	As long as there's a long term plan for huge projects. :)
10:41:29	Araq	it's easy enough to get them back into the shape where it once was
10:41:31	Jehan_	Note that it's generally advisable to use inheritance only with ref object types, not object types.
10:41:46	Arrrrr	Why?
10:41:51	Jehan_	I'm not sure if the latter still crash, but the semantics are problematic, anyway.
10:42:38	Jehan_	Because assignment to a super type has to strip off the extraneous attributes and cast down to the supertype.
10:42:54	Jehan_	Meaning that you won't get any actual polymorphism out of it.
10:43:38	Jehan_	It works only (kinda sorta) with procedure parameters that are passed (explicitly or implicitly) by reference.
10:43:46	*	Trustable joined #nim
10:44:43	Arrrrr	and why doesn't it happen with refs ?
10:45:04	Araq	because every ref is the same size, it's a pointer
10:46:14	Arrrrr	Ok ... then why it has to strip off the differences instead of, i dont know, hide them.
10:47:48	Jehan_	Where should the extra storage come from?
10:48:01	Jehan_	E.g. when you're copying a 2-word object into a 1-word object?
10:48:09	*	milosn joined #nim
10:49:07	Jehan_	If the entire hierarchy were known, you could do it, but not with open inheritance (where you don't know what subtypes may be added later on).
10:49:36	Arrrrr	Ok, i get it. Does this happen in c++ ?
10:49:50	Arrrrr	It seems (for me at least) counterintuitive
10:50:22	Jehan_	It's not even possible in C++.
10:50:22	Arrrrr	although i understand what you mean.
10:50:34	Jehan_	if you have class B : public A { … }
10:50:42	Araq	no? I thought C++ has the same problem
10:50:51	Jehan_	Then you cannot assign a B to A. Only a B* to an A*.
10:51:04	Jehan_	Araq: yeah, that's what I was saying.
10:51:11	Araq	aye that's what I wanted to do too and at some point
10:51:53	Araq	I mean, crashing at runtime when the types are not exactly the same
10:52:16	*	MyMind quit (Ping timeout: 240 seconds)
10:52:35	Jehan_	The problem right now is that Nim doesn't actually cast to the supertype upon assignment.
10:53:16	Arrrrr	Would not be possible to make it working like with refs?
10:53:41	Araq	actually what could be done is to transfer the object type field on assignment if the sizes are the same and if not, raise an exception
10:53:57	Jehan_	Araq: Hmm, that'd work, too.
10:54:04	Araq	that can be useful for a Ruby wrapper
10:54:39	Jehan_	It would be nice for this case to be able to reserve memory for subtypes in the supertype.
10:55:04	Araq	you can do that on your own though
10:55:17	Araq	with mild hacks
10:55:32	Jehan_	Hmm, how?
10:55:46	*	milosn quit (Read error: Connection reset by peer)
10:56:34	*	milosn joined #nim
10:56:48	Araq	add some fields to the supertype on your own
10:57:08	Arrrrr	could not the size be checked at runtime when assigning object values? like "This A class has more weight than a normal A class, so put more space" or something.
10:57:14	Araq	and use them in the subtypes, via 'cast' perhaps
10:57:36	Jehan_	Well, the problem is that this doesn't work so well with ref fields.
10:57:46	Araq	true
10:58:21	Araq	but internally we have the very same problems then with the GC tracing etc
10:58:48	Araq	seems like too much work for a niche feature
10:58:58	Jehan_	Do we? The GC would know that the rest of the object is essentially empty.
10:59:33	Arrrrr	Maybe is niche, but people with little idea about values vs refs dont expect that.
10:59:37	*	milosn quit (Read error: Connection reset by peer)
10:59:52	Arrrrr	You can work around it but well.
11:00:30	Araq	Arrrrr: Nim is getting better at handling it, but it's still not optimized for stupidity ;-)
11:01:39	*	milosn joined #nim
11:05:39	Arrrrr	I would say that many will land to nim expecting an easy programing language but fast, so they will be used to how assigning subtypes to supertypes works with refs types, but will not expect it with values. I'm not saying it should be changed because of this fact, but i would expect most being surprised because of this fact.
11:06:25	HakanD	value types/ref types are common source of confusion for people coming from dynamically typed languages
11:07:42	Arrrrr	Yeah, but one thing is not knowing something, so you cannot make false assumptions, and be confused with something because of its likeness with a previous and very used feature.
11:07:48	*	untitaker quit (Ping timeout: 250 seconds)
11:08:05	HakanD	compiler is giving "Error: unhandled exception: false [AssertionError]"
11:08:07	HakanD	is it a bug?
11:08:24	HakanD	i mean, in the compiler?
11:10:00	flaviu	TEttinger: Have you read the PCG paper? It's great!
11:11:14	HakanD	nvm, false alarm
11:11:50	flaviu	HakanD: If it comes from the compiler process, then yes, it is a bug.
11:12:30	HakanD	thanks
11:12:35	HakanD	it was coming from my code tho -_-
11:13:44	*	untitaker joined #nim
11:17:19	Araq	Arrrrr: I am not aware of any official description of Nim that uses the word "easy"
11:18:28	Araq	I'm much more concerned about "bug prone for people who programmed in Nim for a couple of days or weeks"
11:19:31	Arrrrr	Ok, whatever. I'm just saying it makes sense, is not a matter of stupidity or intelligence, but cohesion. Maybe is niche because noone can actually do it.
11:19:36	*	banister quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
11:22:11	Triplefox	probably the most pressing thing for "nim newbies" is a summary of the safe/productive subset
11:23:56	Araq	BlaXpirit: can you fix the tester bug?
11:24:29	*	Matthias247 joined #nim
11:26:06	Triplefox	today's languages tend to have so many features that you never quite know which features are designed for everyday tasks and which are covering a really difficult edge case and shouldn't be thrown in lightly...it's no wonder people end up thinking Go is okay
11:27:53	*	Strikecarl joined #nim
11:28:04	Araq	Triplefox: you are very welcome to improve our tutorial ;-)
11:28:41	Triplefox	i'll think about it the next time i work in nim...however as it is, i'm writing C :p
11:29:04	flaviu	Is there some way to tell Nim to use a different executable extension?
11:29:26	*	milosn quit (Read error: Connection reset by peer)
11:29:27	Araq	Triplefox: now that's just *sad* :-(
11:29:30	Arrrrr	But would you tell me not having the full power of object inheritance in value objects is just a matter of being a newbe? I know that in a month or 2 or three i will be able to overcome it, but that only means having to write more code or doing it in a counterintuitive way, no that i have become more clever. However i can understand there is no time/manpower to employ it in something people gave up for decades.
11:29:31	dom96	flaviu: likely only with --out:foo.bin
11:29:32	def-	flaviu: -o:file.exe
11:29:51	flaviu	yeah, but I want to put it in a nim.cfg file.
11:29:57	pigmej	Arrrrr: isn't it that you're trying ot use too much java inside nim ?
11:30:16	*	milosn joined #nim
11:31:15	Arrrrr	Well, tell me how making that assumption is wrong or useless.
11:31:24	dom96	flaviu: you can put that in a nim.cfg file
11:31:35	Arrrrr	Maybe is not a problem in java, doesnt mean is only usefull in java.
11:31:52	flaviu	dom96: Sure I can. But then every executable is called foo.bin.
11:32:07	dom96	flaviu: put it in a project-specific nim.cfg then
11:32:10	Araq	Arrrrr: expecting the "full power of inheritance" in value objects makes you a newbie.
11:32:18	Triplefox	easy test: does the rule of least power apply to your pet feature
11:32:24	pigmej	Arrrrr: well, the thing is inheritance is not the best for quite a lot of cases
11:32:26	Arrrrr	Why
11:32:29	Triplefox	if it's not "least powerful" you probably don't need it
11:32:30	flaviu	dom96: There are multiple executables in my project.
11:32:31	Triplefox	yet
11:32:48	Arrrrr	If using inheritance in value objects is wrong, then dont allow it.
11:32:56	dom96	flaviu: you can have an executable-specific cfg file then
11:33:24	pigmej	Arrrrr: genrally, when you're learning new language you shouldn't try to use it as another language
11:33:27	flaviu	dom96: And I can also figure out a way to get autotools to write my nim.cfg files!
11:33:37	pigmej	so if you're learning Nim, then stop thinking 'in java I would...'
11:33:47	flaviu	Arrrrr: for what it's worth, I agree.
11:33:49	pigmej	it will probably have terrible consequences
11:33:55	Araq	well *I* can do it ;-) it's not simply always wrong.
11:33:58	dom96	flaviu: what you want isn't implemented, so i'm suggesting alternatives.
11:34:07	pigmej	Araq: ;D
11:35:04	pigmej	Arrrrr: for example composition combined with awesome nim ucfs could give you way better / readable code (even If I'm Nim newbie too)
11:36:14	Arrrrr	I gree with you that composition is always prefered. But i dont see then why we call bicycle something that only has one wheel.
11:36:45	pigmej	"[...] Nim's support for object oriented programming (OOP) is minimalistic [...]"
11:36:47	pigmej	from docs
11:36:47	*	milosn quit (Ping timeout: 256 seconds)
11:37:11	Triplefox	there's a basic architecture pattern i've started applying across all the languages i'm using today, and it requires essentially no language features, although some can eliminate boilerplate
11:37:11	BlaXpirit	Araq, I don't think i'm the right person to fix testing. it will be much faster for someone who is familiar with it
11:37:49	def-	BlaXpirit: i think i know already why it fails
11:38:01	def-	BlaXpirit: I'm just not sure how to fix it
11:38:05	Arrrrr	but pigmej i can still use a lot of stuff with refs types, and the only reason you are giving me to dont use them in values is "only a noob would do that", which for me is nonsense
11:39:02	Arrrrr	I know for example why i would prefer using values instead of refs, and is not because of a language limitation, but a cpu/cache limitation.
11:39:56	Arrrrr	I dont see why limit that on nim (of course, besides you want to put effort in other things, which is different from saying is good as it is)
11:40:31	Araq	well I already said it's not good and gave a couple of alternatives we should try
11:42:33	BlaXpirit	can non-immediate templates access variables in scope?
11:44:05	BlaXpirit	nvm, i just need mixin
11:47:39	*	HakanD_ joined #nim
11:48:20	Araq	hrm our JS tests are quite broken, right?
11:49:04	BlaXpirit	you fixed testing and found broken tests, huh?
11:49:10	def-	Araq: the JS backend has problems, yes
11:49:23	Araq	no I fixed yet another JS codegen bug
11:49:33	Araq	and didn't run the tests before but after my patch
11:49:50	Araq	and of course now I wonder whether I broke something or if was broken before
11:50:01	def-	the js tests never succeeded
11:50:17	def-	at least many of them
11:50:34	def-	taddr, tunittests, texcpt1 (all with and without -d:release)
11:50:48	*	HakanD quit (Ping timeout: 256 seconds)
11:51:15	*	jefus quit (Quit: WeeChat 0.4.2)
11:52:07	*	jefus joined #nim
11:54:50	Araq	taddr works for me
11:55:08	Araq	unittest cannot work and I'm disabling it
11:55:15	Araq	looking at the exception stuff now
11:55:42	Araq	tactiontable fails for me
11:55:59	*	milosn joined #nim
11:57:35	*	user7181 quit (Quit: leaving)
12:11:18	*	BlaXpirit quit (Read error: Connection reset by peer)
12:11:54	*	BlaXpirit joined #nim
12:12:36	Arrrrr	I go to go, thanks for explaining to me how value objects work in the stack, nimroders.
12:12:39	*	Arrrrr quit (Quit: Page closed)
12:13:00	Araq	"for example asynchttpyserver just starts and http server and lets it run", def-
12:13:07	Araq	that's a very good point
12:13:19	*	Strikecarl quit (Quit: http://www.kiwiirc.com/ - A hand crafted IRC client)
12:13:23	Araq	we need some other tag in the source that is checked for
12:13:49	Araq	like #TESTRUN
12:14:07	Araq	bbl
12:15:12	BlaXpirit	Araq, what about when defined(test):
12:15:46	BlaXpirit	testing in isMainModule never was a good idea anyway
12:17:38	fowl	cursed from the start?
12:48:26	*	Kingsquee quit (Quit: Konversation terminated!)
12:58:21	*	milosn quit (Ping timeout: 264 seconds)
13:14:01	*	milosn joined #nim
13:23:45	*	HakanD_ quit (Read error: Connection reset by peer)
13:27:53	novist	is it somehow possible to override proc that is defined in stdlib? for example proc echo defined in system.nim. Can i make my own custom echo which is preferred than what is defined in system.nim in my project?
13:29:40	dom96	you can make one but you will need to prefix it with the module name to disambiguate it
13:30:21	novist	i see.. that wouldnt work for stuff like operators though. okie..
13:31:13	*	TEttinger quit (Ping timeout: 245 seconds)
13:34:13	*	dhasenan quit (Remote host closed the connection)
13:34:33	*	dhasenan joined #nim
13:35:19	dhasenan	Hrm. Can they turn you into a republic while you're at war?
13:35:56	dhasenan	Apparently you can.
13:39:51	dhasenan	Yay, wrong channel.
13:48:24	*	ob_ joined #nim
13:53:25	*	Matthias247 quit (Quit: Matthias247)
13:58:27	*	milosn quit (Ping timeout: 256 seconds)
13:59:31	fowl	novist, dom96 that is not a problem if its from system
13:59:59	flaviu	I'm clearly using the wrong tool here :P
13:59:59	flaviu	It takes 2 seconds to parse a standard nim file with my regexes.
14:00:12	fowl	of course your echo won't flow back through modules that use system.echo
14:00:46	fowl	flaviu, parsing nim syntax or searching for something?
14:01:12	flaviu	Searching for something that happens to involve parsing a subset of nim syntax.
14:02:00	flaviu	https://github.com/flaviut/docweave/blob/master/src/docweave.nim#L9-L43
14:02:39	dom96	fowl: interesting
14:02:52	*	Matthias247 joined #nim
14:09:39	*	MagusOTB quit (Remote host closed the connection)
14:10:37	*	xcombelle joined #nim
14:14:37	fowl	dom96, seems to be that way and i just tested defining my own echo
14:26:59	*	johnsoft quit (Ping timeout: 246 seconds)
14:27:08	*	gmpreussner joined #nim
14:27:28	*	HakanD joined #nim
14:29:07	*	Matthias247 quit (Read error: Connection reset by peer)
14:32:29	xcombelle	hi
14:34:57	def-	hi xcombelle
14:35:36	xcombelle	hi def-
14:41:31	repax	BlaXpirit: Stunning implementation of Maybe[T]. :)
14:43:17	*	jefus quit (Quit: WeeChat 1.1.1)
14:56:22	*	johnsoft joined #nim
15:02:37	flaviu	When using fsanitize with nim, `fun:markstackandregisters_[:digit:]*` should be added to the blacklist.
15:12:53	BlaXpirit	2 overloaded procs: one accepts typedesc, other accepts concept - doesn't work, because apparently typedesc accepts everything
15:13:12	BlaXpirit	1 proc with `when val is typedesc` causes a crash
15:13:22	BlaXpirit	1 proc with `when compiles(typetraits.name(val))` works
15:13:39	*	darkf quit (Quit: Leaving)
15:19:19	BlaXpirit	gee, showstoppers cropping up like crazy
15:19:52	flaviu	I'm trying to test and make sure I don't leak memory, how would I do that? I'm using GC_fullCollect(), but my tooling still says I'm leaking memory.
15:20:12	*	milosn joined #nim
15:22:32	Jehan_	flaviu: Difficult to say without context. Do you mean that there's memory that should be freed by the GC but isn't? Or that there's foreign memory that's not being managed properly?
15:23:19	flaviu	foreign memory, but I'm not sure if there is a problem or not.
15:23:38	federico3	I'm tempted to suggest not using async on Nim-for-Python-Programmers
15:25:15	Jehan_	federico3: In general, I recommend against using concurrency unless you inarguably need it.
15:25:42	Jehan_	You're incurring a lot of cost, and you have to make sure the benefits are worth it.
15:26:15	federico3	well, sometimes you really want it
15:27:00	*	HakanD quit (Quit: Be back later ...)
15:27:08	Jehan_	federico3: Of course.
15:27:41	Jehan_	But for a lot of applications it's essentially pointless.
15:28:00	Jehan_	Well, not so much pointless, but really not worth the trouble.
15:28:23	dom96	federico3: why?
15:28:43	federico3	speaking of which, any recommended message passing library/design/method for Nim?
15:29:02	flaviu	one GC_fullCollect at the end shows 21 leaks, two GC_fullcollect shows only 16 leaks.
15:29:08	flaviu	Not sure what's going on.
15:30:29	federico3	dom96: 1) it seems to be at a relatively early stage 2) Mixing Python-like OO design with asyncs leads to a lot of memory safety errors that confuse newcomers coming from Python
15:30:49	BlaXpirit	isn't async singlethreaded
15:31:02	federico3	yep
15:34:09	dom96	what memory safety errors are you referring to?
15:36:01	federico3	dom96: well, for example when an async proc is called on an object instance and it tries to change values in internal attributes or attributes on other instances
15:38:20	*	jefus joined #nim
15:40:32	dom96	federico3: can you give an example?
15:41:21	*	jefus quit (Client Quit)
15:42:42	*	jefus joined #nim
15:43:57	*	mpthrapp joined #nim
15:46:28	*	mpthrapp_ quit (Ping timeout: 255 seconds)
15:53:21	federico3	hm, I'll have to recreate one.
16:02:11	*	banister joined #nim
16:17:33	*	pregressive joined #nim
16:23:15	Araq	BlaXpirit: 'when defined(testing)' is fine too, but I don't consider isMainModule as wrong
16:26:45	flaviu	Araq: What about longer test suites? At some point, a test suite it big enough that it needs to be split out.
16:27:21	Araq	really? I thought a test suite always can run in under 5 minutes :P
16:27:57	Araq	flaviu: but to be serious:
16:28:06	flaviu	There can be multiple files and still run in under 5 minutes.
16:28:08	Araq	we already have a nice lib vs stdlib split in the tester
16:28:26	*	clynamen quit (Ping timeout: 246 seconds)
16:29:31	*	banister quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
16:32:59	Araq	dom96: please review https://github.com/Araq/Nim/pull/2513
16:38:16	*	bcinman joined #nim
16:39:09	dom96	done
16:45:48	Araq	can we merge https://github.com/Araq/Nim/pull/2244 now?
16:46:05	Araq	if async is unstable anyway, we might as well make it faster
16:50:51	*	BitPuffin quit (Ping timeout: 276 seconds)
16:56:36	dom96	and make it even less stable in the process?
16:56:41	dom96	let's not do that
16:57:11	Araq	actually fewer closure iterators and allocations could also make it more stable as a side-effect
16:57:24	*	BitPuffin joined #nim
16:57:29	dom96	This doesn't just make the implementation unsafe
16:57:33	dom96	it makes the API unsafe!
16:57:42	Araq	why?
16:58:20	dom96	https://github.com/Araq/Nim/pull/2244/files#diff-3a2daa5a0e408529216e7d6e0900bc29R422
16:58:39	dom96	Async and pointers in params don't mix.
16:59:43	Araq	why does this have to be exported?
17:02:18	dom96	I guess it could be moved to asynchttpserver
17:03:23	dom96	This isn't the way this should be done though.
17:03:32	dom96	We need to come up with a way to speed up async whole.
17:05:53	Araq	bbs
17:21:27	*	yymoto2 joined #nim
17:32:34	*	BitPuffin quit (Remote host closed the connection)
17:33:09	*	BitPuffin joined #nim
17:39:18	*	Senketsu quit (Quit: Leaving)
17:40:47	*	gimireh joined #nim
17:41:44	gmpreussner	what's the best way to manage a project with a lot of packages? would package dependencies go into each package's nim.cfg?
17:41:58	gmpreussner	i don't want to go through nimble for my personal stuff fyi
17:44:21	Jehan_	gmpreussner: Use Peru to keep them in sync, nimble build to build them (locally), and add the paths to your config. At least that would be my suggestion for now.
17:44:26	BlaXpirit	gmpreussner, nimble can install all dependencies specified in a .nimble file, why avoid it
17:45:20	Jehan_	BlaXpirit: The point is that you don't want a package update in a global directory to break your stuff.
17:45:31	BlaXpirit	oh i'm a dummy
17:45:41	Jehan_	One of the big shortcomings of Nimble is that you can have only one installation.
17:45:59	pigmej	Jehan_: what's Peru?
17:46:09	BlaXpirit	it's a bad sign that people aren't crying everywhere about this
17:46:22	Jehan_	pigmej: https://github.com/buildinspace/peru
17:46:54	Jehan_	BlaXpirit: Meh, I (mostly) just don't use Nimble. Except to look up where packages are.
17:46:54	BlaXpirit	sure looks cool as heck
17:47:40	Araq	er I think Nimble can handle just fine it's just that nobody knows about it
17:47:52	pigmej	Jehan_: ee, weird
17:48:06	BlaXpirit	Araq, i wouldn't be surprised
17:48:21	Jehan_	Araq: If there's a way to specify a different installation path, it's not documented anywhere.
17:49:16	flaviu	BlaXpirit: I use it in Nimlets to do dependency management.
17:49:17	pigmej	for me the biggest problem with nimble is versioning
17:49:30	flaviu	https://github.com/nimlets/nimlets
17:50:20	pigmej	correct me if I'm wrong, but isn't nimble replacing global package versions "just when some package needs it" ?
17:50:37	pigmej	and there is no way to have multiple versions of the same package, isn't it?
17:51:00	BlaXpirit	something a bit like that
17:51:18	pigmej	just like pythons pip without virtualenv
17:51:59	BlaXpirit	pigmej, actually it keeps all old versions of packages
17:52:24	BlaXpirit	problem is only the biggest version is ever used
17:52:36	pigmej	ah so exactly like in python pip ;)
17:52:52	BlaXpirit	i don't think pip even can keep old version
17:53:04	pigmej	BlaXpirit: and probably 'biggest' means lastest on disk when sorted by name?
17:53:13	BlaXpirit	i dont know
17:53:15	pigmej	BlaXpirit: it can, it will in fact keep it
17:53:21	pigmej	sometimies by accident sometimes not
17:53:26	pigmej	who know, pip is broken as hell
17:53:36	BlaXpirit	i think you are mistaken here.
17:53:45	pigmej	about pip ?
17:54:09	*	gsingh93 joined #nim
17:54:38	pigmej	there is even thingy called yolk which will list multiple packages
17:54:59	pigmej	and it will show you which version is used (but sometimes it will fail too)
17:55:10	Jehan_	Well, dealing with multiple versions isn't trivial.
17:55:16	Jehan_	Unless you do something like virtualenv.
17:55:41	*	yymoto2 quit (Quit: Lost terminal)
17:55:41	pigmej	virtualenv is also quite hmm, hacky :)
17:55:48	*	gsingh93 quit (Client Quit)
17:56:14	pigmej	generally, default python package management / installing shouln't be used as example as "good package management"
17:56:55	*	gsingh93 joined #nim
17:59:00	Jehan_	What virtualenv does well is creating a standalone version of a piece of software that won't get screwed up by messing around with the rest of the system.
17:59:18	Jehan_	That alone is enormously valuable.
17:59:23	pigmej	I like how rebar works
17:59:24	*	shodan45 quit (Quit: Konversation terminated!)
18:00:10	pigmej	Jehan_: well, in theory you're right, but it can crash after system upgrades or some other upgrades (some dynamic loaded libs are copied as they are)
18:00:18	pigmej	and soo on
18:00:47	Jehan_	pigmej: Yes, I know. Nothing's perfect (though --always-copy deals with some of the worst stuff).
18:01:15	pigmej	Jehan_: sure, I just wnated to say that pythonic 'separtion' is not really good
18:01:20	pigmej	it works, but that's mostly all :)
18:01:34	Jehan_	It is better than what most other systems offer.
18:01:57	pigmej	Jehan_: rebar, hex from erlang world are hmm, wayt better
18:02:16	federico3	Jehan_: which speaks volumes about the quality of language-specific packaging
18:02:18	pigmej	for python conda does pretty cool work, but it's non standard thingy,
18:02:19	Jehan_	Erlang isn't "most other systems".
18:02:23	pigmej	nix is cool also for python things
18:02:27	Jehan_	Erlang isn't even remotely mainstream.
18:02:40	pigmej	probably nix + nim would allow multiple versions
18:02:40	pigmej	etc
18:02:53	Jehan_	And I remember the last time I had to clean up after uninstalling RVM, for example.
18:02:53	pigmej	Jehan_: huh, erlang world is quite big
18:03:09	pigmej	rebar3 and hex are quite cool, really
18:03:16	Jehan_	pigmej: That's different from being mainstream. And I wish Erlang were more mainstream, I'm not dissing it.
18:03:19	pigmej	before that it was mostly crap :)
18:04:10	pigmej	Jehan_: yeah, but not beeing mainstream doesn't mean that it does not have cool tools ;)
18:04:29	dom96	Why do people seem to think that Nimble cannot handle multiple versions of the same packages being installed at the same time?
18:04:34	Jehan_	pigmej: Yeah, I know. In fact, most cool tools are somewhat obscure.
18:04:58	pigmej	dom96: probably because docs are not good enough ?
18:05:33	BlaXpirit	lol maybe Nimble can handle, then Nim can't
18:05:58	pigmej	BlaXpirit: rotfl :D
18:06:09	pigmej	but the result is the same :)
18:06:42	pigmej	hmm, just checked nix with nim and it will work pretty well, so that will be a solution too
18:06:50	dom96	BlaXpirit: That is correct, Nimble resolves the dependencies not Nim.
18:07:35	Jehan_	dom96: How does Nim know which path the correct version is on?
18:07:50	Jehan_	Nevermind, how do I even specify a version in Nim?
18:08:06	dom96	Jehan_: Nimble passes the paths explicitly to Nim
18:08:20	dom96	You specify the version in a .nimble file.
18:08:33	Jehan_	I'm calling Nim, not Nimble, to build my main program.
18:08:59	dom96	Well, sorry, but you will have to call Nimble to build your main program not Nim.
18:09:01	Jehan_	And there's no way around that, because I have to do other things as part of the build.
18:09:02	BlaXpirit	oooh i get it. you're not supposed to call nim directly
18:09:20	BlaXpirit	that can work
18:09:28	Jehan_	Okay, then Nimble is still not for me. Aside from the lack of virtualenv support.
18:09:40	pigmej	Jehan_: check nix ;)
18:09:41	gmpreussner	i have to admit that, now that i'm starting to work on larger projects, i'm not at all clear on how to organize them and manage dependencies. neither Nim nor the Nimble docs are very helpful in this regard.
18:09:43	BlaXpirit	you don't need virtualenv
18:09:55	BlaXpirit	Nimble manages dependencies well
18:10:06	dom96	Jehan_: All we need is `nimble c file.nim`.
18:10:25	BlaXpirit	each package is allowed to want different versions, no problem
18:10:25	dom96	After that is implemented I don't see any issues.
18:10:31	Jehan_	BlaXpirit: There's a lot more to virtualenv than just managing dependencies.
18:10:41	BlaXpirit	like a ton of hacks?
18:10:49	pigmej	BlaXpirit: check nix ;)
18:11:01	pigmej	sorry for repeating, but nix is really cool if you want to do 'separation'
18:11:09	Jehan_	Being able to create a standalone distribution, for example.
18:11:22	BlaXpirit	it doesn't do that so well actually
18:11:24	gmpreussner	can i use Nimble with packages that only exist on the local file system?
18:11:30	BlaXpirit	yes
18:11:44	BlaXpirit	`nimble install` in the folder
18:12:09	gmpreussner	will that create a copy into nimbleDir, or will it fetch files from the folder when needed?
18:12:10	dom96	gmpreussner: How can I improve the docs? I think that the docs are already pretty detailed.
18:12:17	BlaXpirit	gmpreussner, copy
18:12:25	dom96	Perhaps the issue is that people are not reading developers.markdown?
18:12:31	dom96	https://github.com/nim-lang/nimble/blob/master/developers.markdown
18:12:59	Jehan_	nimble c file.nim does not work?
18:13:06	Araq	IMHO the problem is that Nimble has no real website. I never read these github pages
18:13:13	BlaXpirit	>_>
18:13:23	Jehan_	Scratch that, nimble doesn't even work with the devel branch at the moment.
18:13:50	dom96	Jehan_: It hasn't been implemented yet.
18:13:58	gmpreussner	dom96: sorry, i didn't mean to bitch. what i'm saying is that there seem to be no guidelines on how to organize large projects. i'm thinking of frameworks and programs that may consist of dozens or hundreds of modules.
18:13:58	Jehan_	Ah. Well, that doesn't help me, then.
18:14:14	gmpreussner	and by modules i mean packages
18:14:21	pigmej	dom96: you say docs are detailed
18:14:27	gmpreussner	sorry, i always mix up those two, because we refer to packages as 'modules' at work :)
18:14:29	pigmej	and you said that nimble can manage multiple versions
18:14:39	pigmej	even knowing this I don't see this in docs
18:15:06	dom96	gmpreussner: None of us have projects of that magnitude so coming up with guidelines for them hasn't been a priority as of yet.
18:15:18	gmpreussner	dom96: yeah, understood
18:15:23	Araq	gmpreussner: so where can I find this information for C++? I don't think this criticsm is fair.
18:15:47	dom96	pigmej: fair enough
18:16:02	pigmej	dom96: I'm not saying that those docs are bad
18:16:07	gmpreussner	well, it's not criticism. it's just an obstacle i ran into today. i haven't put much thought into this, and now that i have to, i feel stuck :)
18:16:13	pigmej	I think that they have too high entry level
18:16:26	pigmej	so for someone new, it's a problem
18:16:50	gmpreussner	Araq: with C++, the IDEs usually make it easy
18:17:18	gmpreussner	for example, in Visual Studio, organizing a solution into projects is the natural worklow, and VS takes care of the dependencies
18:17:32	gmpreussner	or you can at least tweak them in the project settings
18:18:00	gmpreussner	i found the workflow for C# particularly easy. i don't have to worry about anything there. the IDE figures out what needs to be imported
18:18:22	Jehan_	How does Nimble know which version of Nim to call, by the way?
18:18:44	dom96	Jehan_: It just calls whatever is in your PATH.
18:19:01	Araq	that's not the same at all. so you don't want docs of how to organize stuff, you want IDE support for Nim.
18:19:18	*	bcinman quit (Quit: My Mac has gone to sleep. ZZZzzz…)
18:19:19	Araq	which is of course a very valid desire ;-)
18:19:34	pigmej	Araq: I will try to improve nim support in emacs if so :)
18:19:47	*	bcinman joined #nim
18:19:53	dom96	pigmej: It would be awesome if somebody could help out with docs, perhaps you would be willing to do so?
18:20:18	pigmej	dom96: I'm probably too new in nim world
18:20:24	gmpreussner	Araq: with Nim, i don't know what the idiomatic workflow is, not even in principle. is it nim.cfg? is it nimble? how do i deal with sub-directories? how do i handle module name collisions across different packages? how do i make sure that the main program compiles, but that individual modules are compilable as well? what about separate test harnesses? etc. etc.
18:20:42	dom96	pigmej: That's a good thing, you know the questions newcomers ask.
18:20:58	Jehan_	Right now … no offense, but I'll just keep not using Nimble for the time being. I don't think what I get out of it is worth the problem it's causing me.
18:21:05	dom96	pigmej: You can document your journey to understanding how Nimble works :)
18:21:13	pigmej	dom96: np
18:21:22	Jehan_	For what it's worth, I think it's essential that Nimble works with existing build tools and doesn't try to take over the build process.
18:21:26	pigmej	I can remember my questions / thoughts ;-)
18:21:55	BlaXpirit	i think Nimble is very good
18:22:02	Araq	:O
18:22:30	Araq	BlaXpirit: I don't believe you.
18:22:58	dom96	Araq: hah
18:23:27	Jehan_	For better or worse, using external libraries with autoconf/make/etc. is simply unavoidable for a lot of stuff.
18:23:48	dom96	Jehan_: is there any chance you could give me a detailed description of your project so that I could perhaps help you understand how I would use Nimble to build it?
18:24:09	Jehan_	dom96: My point is that I don't want to use Nimble to build it.
18:24:24	dom96	is that because you prefer to use make to build your project?
18:24:34	flaviu	It should be possible to hack around with staticExec to run make.
18:24:36	Jehan_	I want Nimble to simply put packages where I can find them.
18:24:50	Jehan_	dom96: Not so much prefer but need.
18:25:07	dom96	Jehan_: It already does that.
18:25:27	dom96	They will always be in ~/.nimble/pkgs/pkgName-ver
18:25:44	Jehan_	dom96: Yeah, that's another problem.
18:25:59	Jehan_	How does Nimble deal with packages that need to run autoconf, by the way?
18:26:36	Jehan_	Simply put: I need to give my code, as a standalone distribution, to other people.
18:26:38	repax	~/.nimble is a weird place for non-config files.
18:26:51	Jehan_	Those people don't want stuff in their home directory that doesn't belong there.
18:27:21	Jehan_	The distribution absolutely needs to be standalone.
18:27:37	Jehan_	Not to mention that there may be multiple versions of the code.
18:27:53	Araq	that's not nimble's business
18:28:04	Araq	nimble is not for end users of software
18:28:10	Araq	it's for Nim developers
18:28:22	flaviu	Is there some way to have Nim use standard malloc?
18:28:48	dom96	Jehan_: Nimble already has a config file in which you can override where the "Nimble dir" is (~/.nimble). Perhaps if this became a cmd line argument it would be helpful to you?
18:28:48	Jehan_	Araq: yeah. But given that I already need to maintain my own package tree then, Nimble is essentially useless for me.
18:28:59	dom96	That said, I do agree with Araq.
18:29:18	BlaXpirit	dom96, that would definitely be best
18:29:29	BlaXpirit	> cmd line argument
18:29:40	Jehan_	dom96: When I say "no stuff in the home directory of other people", that means exactly that.
18:29:40	Araq	flaviu: no and that wouldn't help you anyway to detect leaks.
18:29:53	Jehan_	Not "except for a nimble config file".
18:30:05	flaviu	Araq: I'm not looking to locate leaks.
18:30:20	flaviu	I'm looking for buffer overflows.
18:30:26	dom96	Jehan_: Yeah, which is why I suggest implementing a cmd line argument in Nimble to replace the nimble config file.
18:30:32	Jehan_	What's the problem with giving Nimble a --dir option to tell it where to look anyway?
18:30:54	Araq	flaviu: lol.
18:30:54	Jehan_	Why the obsession with putting stuff in $HOME?
18:31:03	dom96	That's pretty much my suggestion
18:31:22	Jehan_	Fine. That's all I've been asking for. Sigh.
18:31:39	Araq	$HOME is cool, it's a fucking mess already and doesn't require sudo. :P
18:31:54	federico3	...where else?
18:32:11	flaviu	$XDG_CONFIG_HOME is better.
18:32:20	Jehan_	Araq: It's not $HOME vs. /usr/lib, it's $HOME vs. $HOME/some/local/development/dir
18:32:21	Araq	I have ~/projects for this reason, so ~ can be full of garbage
18:32:36	dom96	Jehan_: My goal from the very beginning with Nimble was to keep it simple, which is why I prefer to hear the user's full story before simply implementing every feature they ask for.
18:32:48	federico3	flaviu: sure, if it's available (but it's still under $HOME)
18:33:05	repax	home is for "your" files. nimble-packages can always be downloaded anew
18:33:27	Jehan_	dom96: That's okay. As I said, I'm doing fine without Nimble so far.
18:33:58	Jehan_	I'm only telling you that it's virtually useless for me right now in that it only creates extra work and doesn't give me any benefits.
18:35:20	dom96	alright.
18:35:52	*	federico3 stating the obvious: nimble could check for an optional $NIMBLE_HOME
18:35:54	Jehan_	You don't need to feel forced to do anything about it, but if you want feedback, that's why I'm not using Nimble.
18:36:30	gmpreussner	dom96: i put together a gist with some quick ideas here: https://gist.github.com/gmpreussner/df7705a7c2c3c57f0e62 - tell me if it's unreasonable or if i'm crazy :)
18:36:30	*	MagusOTB joined #nim
18:36:31	Jehan_	What I've been telling you is the long answer to why versioning doesn't do what I need, that's all.
18:37:06	Jehan_	SInce everybody seemed to try to tell me that I don't need what I know I need.
18:37:06	MagusOTB	the default constructor for a type zero initializes everything, right?
18:37:12	def-	MagusOTB: right
18:37:34	repax	I'm not using nimble because I want all dependencies in subdirectories of the project so that I can control exactly what version I'm using and when I'm updating any one depencency. I'm sure Nimble is a useful tool though
18:38:45	*	xcombelle quit (Ping timeout: 248 seconds)
18:38:51	Araq	I do use nimble but I want to know how I can make it "git clone" so that I can contribute easily to a nimble package
18:39:04	dom96	gmpreussner: thank you
18:39:05	BlaXpirit	repax, or just specify exact versions in .nimble file and have nimble do it for u
18:39:13	dom96	repax: ^^
18:39:37	dom96	Araq: you can't :P
18:39:56	Araq	btw instead of version nailing you can also do feature detection via system.declared and system.compiles :P
18:40:34	repax	BlaXpirit: I'm probably just not sufficiently familiar then. Good to know it can handle it
18:40:53	BlaXpirit	i wasn't either, until just nopw
18:41:50	Jehan_	Looking at the documentation right now, there doesn't seem a way to tell Nimble to use specific command line options for Nim?
18:42:05	BlaXpirit	isn't that nim.cfg
18:43:09	dom96	indeed. You should be putting them in your nim.cfg file.
18:43:35	dom96	My planned "nimble c file.nim" feature will support passing command line options to Nim though.
18:44:47	BlaXpirit	the design is so awesome
18:45:14	Jehan_	What I was thinking of how to deal with debug vs. release builds.
18:45:19	Jehan_	… was how ...
18:46:11	Jehan_	Given that right now "nimble build" seems to be the way how you'd go about building something locally.
18:46:46	dom96	"nimble build" was in fact designed for end-users of your package so it will always build in release mode
18:47:02	repax	Good to know
18:47:08	Jehan_	Umm. That's another thing. I very rarely want to build in release mode.
18:47:26	Jehan_	I almost always want to at least have checks enabled that guarantee memory safety.
18:48:14	dom96	That's an option we should introduce for the .nimble file.
18:48:24	Jehan_	Release is (1) for code that absolutely needs every last bit of performance or (2) benchmark games, IMHO.
18:48:44	Jehan_	But a lot of code doesn't fall under (1) or (2) and is better off with not having buffer overflows.
18:49:09	repax	I usually have different shell scripts for doing debug / release / test builds / or just "nim check"s
18:49:34	Varriount	Jehan_: You do know that alternate 'modes' can be defined in the config file, right?
18:49:52	Jehan_	Varriount: Yeah, that's what I'm doing.
18:50:00	Jehan_	The question is how to tell Nimble which one to pick.
18:50:12	Jehan_	In short, I have -d:memsafe for this mode.
18:52:16	gmpreussner	Araq: wrt better IDE support, i don't think that needs to be the first step. the things i mentioned for visual studio ultimately boil down to making the right settings in some configuration files. i'd be totally fine configuring those files by hand now.
18:52:50	Jehan_	Another fun problem: one of my projects has Nim code that generates other Nim code (from a plain text database).
18:52:52	gmpreussner	if there is some workflow with, say, nim.cfg that allows me to manage large projects, i'd be good to go :)
18:53:49	gmpreussner	i should qualify: as long as i don't end up hard-coding a bunch of absolute paths :)
18:54:12	repax	A config file can never be sufficiently flexible
18:54:23	gmpreussner	yeah, i think something like nimble is needed
18:55:13	gmpreussner	maybe i'll just write my own build script that discovers available packages and generates appropriate command lines
18:55:14	*	Arrrrr joined #nim
18:55:16	repax	In the end you would have to implement a scripting language to perform the setup of a build (like all the build systems out there)
18:55:39	*	xcombelle_ joined #nim
18:55:57	Jehan_	repax: Yeah. Which is why essentially all I need is "nimble path".
18:56:30	Arrrrr	(Hello Nimrod. I want to know how to do this: 'for y in 0..(HEIGHT-1 / 2)', it does not compile)
18:56:35	*	xcombelle_ is now known as xcombelle
18:56:58	repax	Jehan_: how about your nim to nim build. I'm sure it can be built by nimble somehow. It just seems overwhelming to me
18:57:32	BlaXpirit	Arrrrr, / returns float
18:57:37	Jehan_	repax: It's not even something fancy, it mostly just generates a case statement. But it does require an extra step.
18:57:49	repax	Perhaps I'm thinking too much in imperative terms
18:57:53	Arrrrr	Yeah, that's what the compiler yells to me, i dont know how to make the conversion
18:58:09	BlaXpirit	you probably meant 0..(HEIGHT-1) div 2
18:58:35	Jehan_	Arrrrr: x/y gives a floating point result if x and y are ints.
18:58:36	Arrrrr	wow you were right. Thanks!
18:58:50	flaviu	Arrrrr: *Nim
18:59:00	Arrrrr	So, div gives an entire result?
18:59:13	Jehan_	x div y produces an int.
18:59:25	BlaXpirit	floor div
19:00:33	Arrrrr	Ok, i'll remember that. Thanks
19:01:05	*	bjz quit (Quit: Textual IRC Client: www.textualapp.com)
19:01:19	MagusOTB	what's the deal with 'result' in the docs?
19:01:33	repax	MagusOTB: can you elaborate?
19:01:35	MagusOTB	is it some kind of magic, or is using that to refer to the return value a convention?
19:01:41	dom96	gmpreussner: replied
19:01:48	BlaXpirit	MagusOTB, i'm sure it's mentioned somewhere
19:02:16	BlaXpirit	maybe this http://nim-lang.org/manual.html#return-statement
19:02:49	MagusOTB	ok so result is magic
19:02:57	MagusOTB	it's zero initialized by the compiler?
19:02:57	gmpreussner	dom96: awesome, thanks
19:03:01	def-	MagusOTB: yes
19:03:01	dom96	it's also a convention to use 'result' instead of defining your own var
19:03:03	MagusOTB	what if the result is a compound type?
19:03:22	Jehan_	MagusOTB: Every procedure has an implicit variable result that holds the return value. It is initialized as though reset(result) had been called.
19:03:27	Varriount	MagusOTB: If it's an object, then all it's fields are initialized to zero.
19:03:42	Varriount	MagusOTB: If it's a reference, the reference is 0 (a null reference)
19:03:48	MagusOTB	so the same thing you get by calling the proc that comes with the typename
19:03:53	dom96	gmpreussner: np. Hope that helps somewhat, I know it's not ideal but that is how I would do it. Let me know if you come up with a better solution or if you have any questions.
19:04:50	gmpreussner	oki
19:06:08	gmpreussner	dom96: i think your suggestions will work fine for now. it is kind of what i concluded, except i didn't know about 'nimble install' for local files
19:13:02	*	bjz joined #nim
19:17:58	repax	One might also want to build object files and temporaries on a scratch disc or ram disc instead of among source files
19:18:38	Araq	repax: --nimcache is already configurable
19:18:44	repax	Great
19:18:46	BlaXpirit	:o
19:23:24	*	brson joined #nim
19:24:48	pigmej	what does nimsuggest expect on projectfile ?
19:26:56	*	xcombelle quit (Ping timeout: 240 seconds)
19:28:32	gmpreussner	Araq: for the --out parameter, can we make it generate the target directory if it doesn't exist?
19:28:33	Arrrrr	Does nim have a switch to test if a file has any problem without having to compile it?
19:28:47	*	ChrisMAN joined #nim
19:28:48	repax	pigmej: That's a good question. Nimsuggest should perhaps have access to as much information as nimble has (in order to find all dependencies)
19:28:50	def-	Arrrrr: nim check
19:28:54	gmpreussner	Araq: for the --nimcache directory it seems to work that way already
19:29:18	Arrrrr	Fantastic , thanks def-
19:29:31	pigmej	repax: are you aware of some documentation for nimsuggest?
19:29:41	Araq	gmpreussner PRs welcome :P
19:30:03	Araq	pigmej: there is no documentation. ask me
19:30:06	repax	pigmej: No. I think it is more like the compiler - i.e. having limited knowledge of the project
19:30:52	pigmej	Araq: what should I know before I will start integration of it with emacs company mode
19:31:12	pigmej	Araq: I think about one nimsuggest process per project
19:31:21	pigmej	but what more should I know about nimsuggest itself before :)
19:31:31	pigmej	like what the heck is that projectfile.nim
19:31:54	Araq	it usually helps to be completely ignorant when one starts a new project.
19:32:08	repax	I find it hard not to be ;)
19:32:22	pigmej	Araq: you mean ?
19:32:31	Araq	too much knowledge keeps one from doing it
19:32:58	pigmej	ok
19:33:08	pigmej	fine, just wondering what's that projectfile.nim :)
19:33:14	*	gimireh quit (Quit: Leaving)
19:33:32	pigmej	aporia uses current file as projectfile argument for example
19:33:45	Araq	pretty sure it doesn't
19:33:52	*	banister joined #nim
19:33:55	*	banister quit (Max SendQ exceeded)
19:34:06	pigmej	it does ;) at least for me
19:34:07	repax	The most pressing issue with nimsuggest as I see it has to do with its knowledge of include dirs and project-wide defined symbols
19:34:29	Araq	the projectfile is the file that is the main file of your project
19:34:30	pigmej	Araq: or maybe it's first file where ctrl+space was hit
19:35:09	Araq	repax: NimStudio works around that
19:35:29	pigmej	Araq: yeah, aporia uses first file where completion was called
19:36:05	pigmej	(it even copies it to /tmp/aporia)
19:37:36	BlaXpirit	please help me fix this code. https://bpaste.net/show/19c223b03522
19:37:45	BlaXpirit	as far as i'm aware, this is a bug, but i dunno
19:37:48	pigmej	Araq: what are sug|con|def|use ?
19:38:00	repax	So, perhaps NimStudio is a useful source for pigmej?
19:38:17	pigmej	sug => suggestion, con ?, def ?, use?
19:39:07	Araq	sug # dot was pressed
19:39:20	Araq	con # ( was pressed (show parameters and types)
19:39:27	Araq	def # goto definition
19:39:35	Araq	use # list usages
19:39:51	pigmej	cool
19:40:00	pigmej	and the difference between file and dirtyfile ?
19:40:01	*	xcombelle joined #nim
19:40:13	pigmej	dirtyfile is something like 'may be broken/ under construction file' ?
19:40:23	pigmej	kind of ignore syntax errors etc?
19:40:40	Araq	dirtyfile is the temporary file
19:40:51	Araq	you need to save the file for nimsuggest
19:40:53	gmpreussner	dom96: i got this working as suggested. i think 'nimble c <file>' is going to be critical though - otherwise i can build either all or nothing :)
19:41:05	Araq	but might not want that
19:41:09	*	dewdrop quit (Read error: Connection reset by peer)
19:41:25	Araq	so the editor uses a temporary file
19:41:38	gmpreussner	dom96: in particular, there is no convenient way to build the tests associated with a package right now
19:41:59	Araq	and you tell nimsuggest "foo.nim" is actually in "/tmp/foo.nim"
19:42:24	pigmej	Araq: ah cool
19:42:25	*	cjbest joined #nim
19:43:20	Araq	pigmej: also nimsuggest can compile unrelated modules, so you don't really have to create one process per project
19:43:30	Araq	but it's true that "per project" works better
19:43:53	BlaXpirit	Araq, please advise on this very short piece of code. https://bpaste.net/show/19c223b03522
19:44:21	dom96	gmpreussner: cool. I'll see about implementing that ASAP.
19:44:34	flaviu	Wow, the performance of findAll in nre is terrible!
19:45:15	*	devdri joined #nim
19:45:49	Araq	BlaXpirit: known bug.
19:45:52	Araq	hi devdri welcome
19:45:55	BlaXpirit	what can i do?
19:46:04	*	bcinman quit (Quit: My Mac has gone to sleep. ZZZzzz…)
19:46:18	Araq	I never understood why 'distinct G[T]' is required
19:46:22	*	bcinman joined #nim
19:46:23	*	devdri quit (Client Quit)
19:46:43	BlaXpirit	because i oversimplified the example
19:46:46	*	pregressive quit (Remote host closed the connection)
19:46:48	BlaXpirit	in reality it's actually needed
19:47:06	def-	BlaXpirit: that's the bug i think: https://github.com/Araq/Nim/issues/1385
19:47:38	BlaXpirit	:s
19:50:08	Arrrrr	omg, it's been a hard hour but at last i managed to get only one char from window's input without having to press enter http://pastebin.com/rEsqQrTT i feel less stupid now
19:50:51	Arrrrr	I'm trying to program a roguelike just for experimentation purposes
19:51:01	federico3	can I list all procs that can act on a given type? The equivalent of dir() ?
19:51:37	def-	Arrrrr: i think we're missing just that for the terminal module, where it's only defined for non-windows
19:51:56	*	cjbest quit (Quit: My Mac has gone to sleep. ZZZzzz…)
19:52:12	Arrrrr	i didnt found it inside windows module, but found a way to program it from c++
19:53:12	pigmej	hmm, Araq i'm getting werid errors from nimsuggest
19:53:17	pigmej	for example Error: cannot open '/home/pigmej/Apps/Nim/compiler/lib/system.nim
19:54:38	gmpreussner	dom96: is --NimblePath more of an internal feature used by nimble, or is there a way to make Nim read this from the environment? putting it in a nim.cfg is probably not good (absolute path)
19:55:20	dom96	gmpreussner: it's so that Nim knows where to find Nimble packages
19:55:37	gmpreussner	right, but i suppose it's only passed by nimble right now?
19:55:48	dom96	gmpreussner: nope, it's in Nim's cfg file.
19:56:10	dom96	It's what allows you to use installed Nimble packages even when compiling with 'nim c'
19:56:13	gmpreussner	can it be added as a global environment variable, so i don't have to have an absolute path in nim.cfg?
19:56:37	gmpreussner	i guess i could throw it into the global nim.cfg for now
19:57:15	Araq	pigmej: add 'nim' to your PATH
19:57:35	Araq	pigmej: put 'nimsuggest' into the dir where 'nim' is
19:57:42	dom96	gmpreussner: nope
19:57:52	federico3	how can I access "ftell"?
19:58:10	Araq	federico3: no, but we know everybody's life depends on this feature
19:58:37	pigmej	Araq: hmm, in fact they are both symlinked
19:58:42	federico3	hum?
19:59:10	pigmej	Araq: where nim binary or where nim sources ?
19:59:21	federico3	Araq: were you talking to me?
19:59:28	Araq	federico3: system.getFilePos
19:59:29	def-	federico3: getFilePos?
20:00:28	federico3	thanks
20:00:41	Araq	federico3: yes, "can I list all procs that can act on a given type?"
20:00:46	pigmej	ok it seems in the same place where binary is located
20:00:51	Araq	I replied
20:00:56	Araq	to that question
20:00:58	*	endragor quit (Ping timeout: 256 seconds)
20:01:19	federico3	ah - well, I find python's help() and dir() very useful to explore the language
20:01:46	pigmej	Araq: now I will need to know what to do with: skVar, skForVar, and so on
20:01:47	pigmej	what are those ?
20:02:02	pigmej	skVar => seems as a local variable, skForVar same but for for loop
20:02:18	Araq	well yes
20:02:30	pigmej	what more is there ?
20:02:43	*	HakanD joined #nim
20:03:40	gmpreussner	dom96: is --NimblePath the path to nimble, the pkgs directory, or to individual packages within pkgs? i can't get it to work :|
20:03:52	dom96	the pkgs directory
20:04:07	dom96	i.e. ~/.nimble/pkgs
20:05:03	Araq	pigmej: procs, converters, iterators, macros, templates, lets, types, consts, methods
20:05:17	pigmej	Araq: ok so mostly 'everything' ;-)
20:05:18	pigmej	cool
20:05:41	*	Arrrrr quit (Quit: Page closed)
20:07:24	renesac	Araq: the manual says "If the operator ends with ``=`` and its first character is none of
20:07:24	renesac	``<``, ``>``, ``!``, ``=``, ``~``, ``?``, it is an *assignment operator* which
20:07:24	renesac	has the lowest precedence."
20:07:33	Araq	pigmej: skField. that might be all.
20:07:39	renesac	and right after it, there is the table
20:07:56	renesac	where the precedence level 0 (lowest) has the arrow like operators
20:08:10	renesac	with the assignment operators above that
20:08:26	Araq	yea yeah yeah, fix it. the table is correct
20:08:27	renesac	this is very confusing
20:08:35	Araq	the text needs an update
20:08:53	renesac	ok
20:10:34	renesac	and I think the paragraph about arrow operators should be moved to the precedence section
20:10:41	renesac	as it isn't right associative anymore
20:10:48	renesac	it only changes precedence
20:10:54	gmpreussner	Araq: i find --out rather useless. wouldn't something like --outPath be better?
20:11:46	gmpreussner	--out seems only useful if you want to rename the build product
20:11:59	gmpreussner	most of the time i'm just test compiling modules, however
20:13:28	Araq	gmpreussner: meh I guess I agree
20:13:52	Araq	pigmej: https://github.com/Araq/Nim/blob/devel/doc/idetools.txt#L201
20:14:00	Araq	still somewhat up to date
20:14:30	repax	gmpreussner: And it's not --nimcache:dir/ that you're looking for?
20:15:27	pigmej	Araq: cool.
20:15:39	pigmej	i will be there later, I have to go off for hour or two
20:15:55	pigmej	but well nimsuggest seems fine
20:17:23	Araq	bbl
20:20:25	gmpreussner	repax: no, the nimcache is separate. i'm talking about the binaries that are created. they usually end up alongside their corresponding .nim modules
20:22:01	*	xcombelle quit (Ping timeout: 252 seconds)
20:22:17	gmpreussner	dom96: import packageName/module does not work for me in conjuction with --NimblePath. i can do "import module", and it finds the right one (by accident?), but i cannot qualify the package name
20:22:35	gmpreussner	Error: cannot find 'packageName/module'
20:22:50	dom96	where did you point --nimblepath to?
20:23:11	gmpreussner	--NimblePath:"~/.nimble/pkgs"
20:23:26	gmpreussner	the files live in "~/.nimble/pkgs/package-0.1/"
20:23:49	gmpreussner	0.1 is the version i specified in the .nimble file
20:25:07	dom96	What does the directory structure in package-0.1 look like?
20:25:39	gmpreussner	it's flattened out. just a bunch of files, including module.nim
20:25:58	gmpreussner	there's also a .nimble and a .json. that's it
20:26:01	Jehan_	Just import module, then.
20:26:17	Jehan_	If you want to import foo.module or foo/module, you'll have to put it in a foo subdirectory.
20:26:29	dom96	yeah, you need to put it in a packageName subdir
20:27:05	gmpreussner	oh ok, so packageName refers to a subdirectory WITHIN the package's srcDir
20:27:13	Jehan_	Yeah.
20:27:27	Jehan_	The Nimble package directories are simply added to the search path.
20:27:36	gmpreussner	understood
20:28:13	Jehan_	Which is why I have a feature request that foo:module is resolved so that it looks for foo.nimble, then module in that directory. :)
20:28:32	gmpreussner	yeah, that might be better. packageName subdir seems rather ugly
20:28:36	Jehan_	To avoid conflicts and stuff (plus, smarter lookup of local modules).
20:29:04	Jehan_	Right now, you can (for example) override posix.nim by accident, which of course breaks everything.
20:29:23	gmpreussner	couldn't we just always infer the 'scope' from the package name?
20:29:34	dom96	Jehan_: I like that idea.
20:29:45	gmpreussner	so if i have packageName.nimble, then "import packageName/module" should just work?
20:29:58	dom96	Would require tons of code changes though.
20:30:01	gmpreussner	regardless of where srcDir is within packageName
20:30:19	Jehan_	dom96: What idea?
20:30:32	dom96	Jehan_: foo:module
20:31:16	Jehan_	dom96: Yeah, not sure how much work it would be.
20:31:17	gmpreussner	for example, i have /packageName/packageName.nimble, /packageName/src/moduleName.nim, srcDir="src". then "import packageName:moduleName" should just work.
20:31:43	Jehan_	But most importantly, module lookup should start with modules local to the current package.
20:32:17	Jehan_	It only requires two packages to have a foo.nim module to get some really bad problems.
20:32:42	Jehan_	Especially if they just happen to be slightly different versions that were copied at different times from the same source.
20:34:31	*	xcombelle joined #nim
20:35:42	dom96	gmpreussner: Nimble c is now ready for action.
20:36:05	dom96	only tested briefly so let me know if you find any bugs.
20:36:11	gmpreussner	Jehan_: yep, i have this problem right now. complex.nim in a nimble package colliding with the one in stdlib, and no way to disambiguate
20:36:17	gmpreussner	dom96: ok, will do right now
20:36:52	dom96	gmpreussner: hrm, that's a serious problem :\
20:37:19	Jehan_	gmpreussner: Yeah, I had similar issues myself more than once.
20:38:52	Jehan_	These days, I mostly stuff things in a subdirectory (but of course that doesn't help with packages).
20:39:18	Jehan_	Also, if I were ever to use one of the same directories as the stlib, I might run into issues again.
20:39:51	Jehan_	Anyhow, good night. :)
20:39:59	*	Jehan_ quit (Quit: Leaving)
20:42:32	*	dtscode is now known as dtscocain
20:43:05	federico3	oh - there is no sound support
20:44:01	*	xcombelle quit (Ping timeout: 252 seconds)
20:54:36	gmpreussner	dom96: gonna pm you with some stuff
20:57:00	onionhammer	cool s
20:57:02	*	bcinman quit (Quit: My Mac has gone to sleep. ZZZzzz…)
20:57:04	reactormonk	how can I set environment variables for startProcess?
20:57:18	onionhammer	so nimble is getting a path command arg? dom96
20:58:12	dom96	I suppose. It won't be convenient though.
20:58:22	onionhammer	why?
20:58:24	dom96	It won't be what you asked for.
20:58:48	dom96	It will be what you can already override in the nimble config file
20:58:52	dom96	i.e. the nimble dir
20:58:58	onionhammer	nimble install name .
20:59:24	onionhammer	. meaning here
20:59:33	onionhammer	or nim l
20:59:48	onionhammer	oops, phone
21:00:06	fowl	araq would say that being able to override posix.nim or system.nim is a feature
21:00:22	fowl	:p
21:07:02	federico3	uh oh, there is no syslog
21:08:05	*	gsingh93 quit (Ping timeout: 246 seconds)
21:11:31	*	vendethiel joined #nim
21:11:40	*	HakanD quit (Quit: Be back later ...)
21:12:40	*	vendethiel- quit (Ping timeout: 265 seconds)
21:12:45	*	HakanD joined #nim
21:13:58	*	Kingsquee joined #nim
21:15:12	*	bcinman joined #nim
21:23:53	BlaXpirit	repax, renesac, fowl, flaviu: I updated optionals module. https://github.com/Araq/Nim/pull/2515
21:24:10	BlaXpirit	(tell me if you don't care about this so i don't bother you anymore)
21:24:57	*	a5i joined #nim
21:25:03	flaviu	I care, but my opinion on that implementation is unchanged.
21:25:18	BlaXpirit	flaviu, on what implementation
21:25:29	flaviu	Of the optionals module.
21:25:37	flaviu	Araq, mind taking a look at https://gist.github.com/flaviut/234c784502a81376dd37 ?
21:25:42	BlaXpirit	what in particular?
21:25:56	BlaXpirit	look at my last comment there
21:26:43	fowl	BlaXpirit, or should be a template
21:26:53	BlaXpirit	why?
21:27:15	fowl	to avoid executing the default param until its needed
21:27:26	flaviu	BlaXpirit: That's going in the opposite direction, IMO.
21:27:28	BlaXpirit	ah right, right
21:27:48	BlaXpirit	flaviu, i'm not sure you understand
21:28:02	BlaXpirit	the only bad direction i've taken here is the complexity of the module.
21:28:08	fowl	i dont understand why you need just[t] and none[t]
21:28:18	flaviu	BlaXpirit: I do understand, I just don't like it.
21:28:41	BlaXpirit	fowl, let me try to explain
21:28:55	BlaXpirit	there is that `?` operator which returns one of the two implementations, right?
21:29:10	BlaXpirit	the procedures `just` and `nothing` relied on it
21:29:34	BlaXpirit	which means that you cannot choose a different implementation of maybe for your type
21:30:04	flaviu	The module API should be no more than Option[T], `?=`, `[]`, `or`, rawGet, isSome, isNone. Everything else is unnecessary and should not exist.
21:31:02	BlaXpirit	oh but consider this too. a procedure that accepts a Maybe can also accept a zero-overhead Just
21:31:11	BlaXpirit	it is literally zero overhead
21:32:17	flaviu	Why would that be useful?
21:32:26	BlaXpirit	> it is literally zero overhead
21:32:28	*	dtscocain is now known as dtscode
21:32:59	flaviu	I understand its advantages, but why is it useful?
21:33:42	BlaXpirit	flaviu, let's drop the thought that 2 different implementations of Maybe are not useful. people thought they were.
21:33:50	*	banister joined #nim
21:33:54	*	banister quit (Max SendQ exceeded)
21:34:19	flaviu	BlaXpirit: I don't understand how Just and Nothing are useful.
21:34:47	BlaXpirit	they are required if you want to opt out of nil-based implementation
21:35:34	flaviu	But they are concrete types? Why even use an Just[T] when a T does the exact same thing.
21:35:40	BlaXpirit	they are required because you don't know what underlying implementation of Maybe you'll be using when you're instantiating them
21:36:11	BlaXpirit	later they can be converted to a concrete Maybe type
21:36:28	fowl	BlaXpirit, so Just[T] will be one of the maybe kinds?
21:36:36	BlaXpirit	well, yes, yes, it is
21:36:40	BlaXpirit	it's all in the docs
21:36:56	BlaXpirit	now, why is Just[T] needed: because there are converters from Just to Maybes
21:37:57	flaviu	Why not have just() return a Maybe directly?
21:38:12	BlaXpirit	i've covered that, although explanation was very poor
21:38:19	*	banister joined #nim
21:38:23	*	banister quit (Max SendQ exceeded)
21:38:58	BlaXpirit	you wouldn't be able to opt out of nil-based implementation
21:39:11	BlaXpirit	you don't know what underlying implementation of Maybe you'll be using when you're instantiating a Just
21:40:11	flaviu	I see.
21:40:13	fowl	flaviu, isNil(x) will never work, isNil returns bool
21:40:39	BlaXpirit	?
21:40:44	flaviu	fowl: How do concepts work, I don't understand.
21:41:12	flaviu	I assumed that each expression is wrapped in a when compiles(...), but I don't really understand it.
21:41:25	BlaXpirit	how i'm trying to understand it: each part must compile, and, if it returns an expression, it must be boolean true
21:41:51	BlaXpirit	isNil(x) returns an expression which is not even boolean, so yeah
21:42:05	BlaXpirit	(but this is all conjecture)
21:42:13	BlaXpirit	wait, it is lol
21:42:20	flaviu	Aha!
21:42:25	flaviu	That was the problem, works now.
21:42:33	fowl	flaviu, if theres a return type isNil(x) is bool
21:43:08	fowl	i can't get the "x = nil" or anything like that ie "var y = x; y = nil" to work
21:43:13	flaviu	fowl: I can't parse your sentence, but "isNil(x) is bool" works fine.
21:44:40	fowl	flaviu, i meant if it returns something you need "is T"
21:45:15	flaviu	fowl: Ok, I understand! Thanks!
21:46:26	flaviu	https://gist.github.com/cf4b6f97d60a5c6db888 works great!
21:46:38	fowl	as to how they work, i believe the code inside the concept is semantically checked
21:47:01	fowl	i stumbled on the section in the compiler but it didnt make much sense
21:48:43	BlaXpirit	please write some of your comments under the pull request :p
21:48:56	*	EXetoC joined #nim
21:49:25	flaviu	BlaXpirit: Working on it, actually.
21:49:33	BlaXpirit	thx
22:00:33	flaviu	Error: type mismatch: got (typedesc[ref int], nil) *sigh*
22:01:09	BlaXpirit	you don't know how much fun i had today
22:01:16	BlaXpirit	both literally and sarcastically
22:06:18	*	banister joined #nim
22:06:22	*	banister quit (Max SendQ exceeded)
22:08:34	flaviu	Is there some way to say "Anything but this type"?
22:08:58	BlaXpirit	flaviu, maybe implement 2nd proc to capture that type
22:10:57	*	bcinman quit (Quit: My Mac has gone to sleep. ZZZzzz…)
22:11:22	HakanD	gn
22:11:25	*	HakanD quit (Quit: Be back later ...)
22:11:58	*	renesac left #nim ("Once you know what it is you want to be true, instinct is a very useful device for enabling you to know that it is")
22:12:22	*	renesac joined #nim
22:13:16	flaviu	BlaXpirit: I don't understand what you mean.
22:13:54	*	renesac_ joined #nim
22:14:22	*	renesac_ left #nim (#nim)
22:15:10	*	bcinman joined #nim
22:19:44	*	pregressive joined #nim
22:25:07	*	brson quit (Quit: leaving)
22:30:42	*	HakanD joined #nim
22:35:09	*	HakanD quit (Ping timeout: 244 seconds)
22:37:55	flaviu	Ah, I see.
22:45:34	*	pregressive quit (Remote host closed the connection)
22:46:56	*	Matthias247 joined #nim
22:47:00	*	bcinman quit (Ping timeout: 256 seconds)
22:49:43	*	bcinman joined #nim
23:08:41	*	user7181 joined #nim
23:12:51	pigmej	sadly my elisp skills are not enough to make nimsuggest + emacs working (and I don't have probably enough time to improve my elisp now ;/).
23:26:33	*	Matthias247 quit (Read error: Connection reset by peer)
23:29:29	*	bcinman quit (Quit: My Mac has gone to sleep. ZZZzzz…)
23:30:16	*	bcinman joined #nim
23:40:52	*	bcinman quit (Quit: My Mac has gone to sleep. ZZZzzz…)
23:41:01	reactormonk	pigmej, wanna team up? Got time in... 18h from now?
23:42:06	pigmej	reactormonk: if I will have something usefull then, ok ;)
23:43:58	*	BlaXpirit quit (Quit: Quit Konversation)
23:46:02	*	davidhq quit (Quit: My MacBook Pro has gone to sleep. ZZZzzz…)