<< 16-01-2016 >>

---

00:15:15	*	jaco60 quit (Ping timeout: 260 seconds)
00:16:21	*	Trustable quit (Remote host closed the connection)
00:37:29	*	kas_ quit (Ping timeout: 255 seconds)
00:51:31	*	yglukhov quit (Remote host closed the connection)
01:23:59	*	awsteele joined #nim
01:24:07	*	awsteele quit (Client Quit)
02:10:15	*	Jachy quit (Ping timeout: 240 seconds)
02:11:25	*	Jachy joined #nim
02:58:27	*	BitPuffin|osx joined #nim
03:20:13	*	brson quit (Quit: leaving)
04:32:52	*	nande quit (Remote host closed the connection)
05:21:22	*	darkf joined #nim
06:14:07	*	minmax joined #nim
06:33:56	*	minmax left #nim (#nim)
06:39:21	*	brson joined #nim
07:18:28	*	brson quit (Quit: leaving)
07:23:22	*	brson joined #nim
07:34:03	*	yglukhov joined #nim
07:42:06	*	desophos quit (Read error: Connection reset by peer)
07:59:06	ldlework	I've tried to replicate the sigsev in a minimal example for a few hours
07:59:09	ldlework	can't do it
08:05:39	*	nsf joined #nim
08:06:26	Araq_	yeah, the VM is register-based.
08:07:04	Araq_	when you change the code slightly, the register allocator uses different registers and this makes things non-reproducible
08:07:26	Araq_	a definite downside of its design.
08:08:15	Araq_	but I told you I don't need minimal, only self-sufficient
08:39:16	*	yglukhov quit (Read error: Connection reset by peer)
08:39:48	*	yglukhov joined #nim
09:28:29	*	kas_ joined #nim
09:29:08	*	avsej quit (Quit: Quit)
09:30:29	*	avsej joined #nim
09:30:36	*	avsej quit (Changing host)
09:30:36	*	avsej joined #nim
09:44:51	*	Demon_Fox quit (Quit: Leaving)
10:09:29	*	brson quit (Ping timeout: 246 seconds)
10:42:10	*	exebook quit (Quit: Leaving)
11:12:39	*	Matthias247 joined #nim
11:24:35	*	jaco60 joined #nim
11:26:34	*	mountaingoat joined #nim
11:26:41	mountaingoat	how is nim support for arm?
11:42:28	dom96	mountaingoat: It's great
11:42:48	dom96	Nim compiles to C, so as long as your C compiler supports arm it will work well.
11:44:10	mountaingoat	cool
11:44:23	mountaingoat	i presume the C compiler picks up the hardware floating point support than too
11:44:26	mountaingoat	then*
11:50:59	dom96	possibly :)
12:03:11	*	kas_ quit (Ping timeout: 264 seconds)
12:06:09	Araq_	I'm merging no-echo-gotcha into devel
12:07:30	*	kas_ joined #nim
12:07:39	def-	Araq_: cool, didn't find any regressions with it
12:07:52	Araq_	yay :-)
12:08:12	Araq_	I did find a few tests that had to be adapted
12:08:20	Araq_	but the code always was questionable
12:08:43	Araq_	and it's good that 'a+1' works and 'a + 1' and 'a +1' not anymore. IMO.
12:09:04	Araq_	I mean both 'a+1' and 'a + 1' continue to work.
12:10:13	Araq_	dom96: we can actually make the "async GC" way more efficient than I thought.
12:10:41	Araq_	since stack roots are not possible we can go after a copying collector.
12:25:51	dom96	cool
12:31:27	*	vendethiel- joined #nim
12:31:58	*	vendethiel quit (Ping timeout: 260 seconds)
12:39:51	Araq_	dom96: one problem is that capturing a 'ref' in a closure keeps the pointer semantics
12:40:44	Araq_	but if we transfer the whole heap to a worker thread that means we get copy semantics for it
12:41:15	Araq_	this will be a problem for users.
12:42:05	Araq_	I think we can statically detect it though.
12:42:47	dom96	That would be a PITA
12:43:17	dom96	Pretty sure that is a really common occurrence
12:43:48	Araq_	I dunno. does 'asyncCheck foo(data)' work?
12:46:13	dom96	you mean the fact that you're passing `data`? yes
12:47:47	Araq_	well we can allow it
12:48:04	Araq_	but if that feature is used the task won't be migrated to the thread pool
13:01:04	*	Trustable joined #nim
13:01:13	dom96	practically every async proc takes parameters though
13:04:47	Araq_	ok then we need to allow it and change the semantics under the hood
13:05:02	Araq_	it's an ownership transfer, so quite a few things continue to work
13:11:07	dom96	not sure how that can possibly work
13:33:04	*	wuehlmaus joined #nim
13:33:35	wuehlmaus	hi, all, i have a very strange problem: when i do
13:33:47	wuehlmaus	``import posix'' and try to compile that i get
13:33:48	wuehlmaus	lib/nim/posix/posix.nim(32, 19) Error: undeclared identifier: 'Time'
13:34:00	wuehlmaus	it's that early :(
13:34:21	wuehlmaus	nim 0.12 on archlinux
13:34:36	wuehlmaus	git hash: 30cc353831a9727cfc6b4e7c379d9920b7041059
13:40:44	dom96	hello wuehlmaus
13:41:37	wuehlmaus	hi, dom96
13:42:07	dom96	that's really strange
13:42:26	dom96	How did you install Nim?
13:42:55	wuehlmaus	via pacman
13:43:08	dom96	ahh, could be an AUR issue.
13:43:34	dom96	I would suggest you try bootstrapping the compiler manually
13:43:47	dom96	and instead of installing it, just add the path of nim/bin to your PATH
13:43:50	wuehlmaus	hmm, nim is not in AUR
13:44:00	wuehlmaus	it's in community
13:44:05	dom96	oh
13:44:11	dom96	could still be a packaging issue though
13:44:16	wuehlmaus	yeah
13:44:22	wuehlmaus	i try to get the git version
13:45:42	dom96	please let me know if that works for you
13:46:31	wuehlmaus	of course i will do :)
13:46:58	wuehlmaus	the wonderful thing about nim is that it can be compiled in a few minutes on my netbook
13:47:01	dom96	make sure to remove the pacman package first
13:47:06	wuehlmaus	try that with swift or rust :)
13:47:09	dom96	yeah :)
13:47:31	dom96	if this fixes the problem then I will create a packaging issue on github
13:47:42	dom96	I recall other users having similar issues with installation
13:59:24	wuehlmaus	hmm, sadly no difference :(
13:59:57	wuehlmaus	i try to restart my computer "have you tried to turn it on and off again" :)
14:03:32	wuehlmaus	didn't help :(
14:06:18	dom96	right, can you tell me how you bootstrapped and installed Nim?
14:06:37	dom96	wuehlmaus: also, did you bootstrap the devel branch or the master branch?
14:06:46	wuehlmaus	./bootstrap.sh
14:07:08	wuehlmaus	and then ./install.sh /usr/local/bin
14:08:05	wuehlmaus	git clone https://github.com/nim-lang/Nim.git
14:08:12	wuehlmaus	that's what i used
14:30:39	dom96	wuehlmaus: hehe
14:30:52	dom96	wuehlmaus: Like I said, "instead of installing it, just add the path of nim/bin to your PATH"
14:30:57	dom96	Don't use install.sh
14:31:12	dom96	That is the root cause of this AFAIK
14:32:02	wuehlmaus	seems not to be a solution for me
14:32:18	dom96	what do you mean?
14:32:32	wuehlmaus	i still get that error
14:33:07	dom96	did you add Nim to your PATH instead of using install.sh?
14:33:42	wuehlmaus	~/Nim/bin/nim c ~/p/nim/problem.nim
14:33:51	wuehlmaus	it still gives the error :(
14:34:21	dom96	gist the output of `cat ~/p/nim/problem.nim` please
14:35:02	dom96	also please try `PATH=$PATH:/home/yourusername/Nim/bin nim c ~/p/nim/problem.nim`
14:35:04	dom96	it may just work
14:36:17	wuehlmaus	problem.nim has just ``import posix''
14:40:05	dom96	can you gist the full output of `PATH=$PATH:/home/yourusername/Nim/bin nim c ~/p/nim/problem.nim` and `nim -v`?
14:41:21	wuehlmaus	afk
15:34:27	wuehlmaus	back
15:35:37	def-	dom96: if you want to make sure that nim version is used you should prepend to the path, not append
15:35:59	dom96	true
15:36:10	def-	(but no idea if that's the issue)
15:36:30	dom96	wuehlmaus: run `PATH=/home/yourusername/Nim/bin:$PATH nim c ~/p/nim/problem.nim`
15:36:34	dom96	and gist the output please
15:36:36	dom96	and also `nim -v`.
15:44:39	*	yglukhov quit (Remote host closed the connection)
15:46:32	wuehlmaus	Nim Compiler Version 0.12.1 (2016-01-16) [Linux: amd64]
15:46:47	wuehlmaus	the rest is absolutely identical to what i wrote before
15:49:37	wuehlmaus	and it doesn't change one bit if i use another shell
15:51:01	dom96	Could you please just paste the full output of the compilation process? There are other messages besides the error which I would like to see.
15:53:08	wuehlmaus	https://gist.github.com/pfefferminz132/d928660110575c10ff4e
16:00:41	wuehlmaus	it must be something in the problem.nim folder
16:00:50	wuehlmaus	it compiles in another directory
16:05:56	dom96	Do you have any other files in that directory?
16:06:17	wuehlmaus	82 :)
16:06:33	dom96	lol
16:06:45	dom96	I guess you have a `times.nim` file in there? :P
16:06:51	dom96	I just reproduced your issue
16:07:04	wuehlmaus	yes i have :)
16:07:18	dom96	I'm going to report this as an issue heh
16:07:23	wuehlmaus	:)
16:07:30	wuehlmaus	i am happy that we found it!
16:07:54	wuehlmaus	the reason i had that file was that i wanted to generate an html file from times.nim to have docu
16:09:10	dom96	heh, me too.
16:09:25	dom96	At least now you have the latest version of Nim to play around with :)
16:09:26	federico3	uh? Is 0.12.1 out?
16:09:30	dom96	https://github.com/nim-lang/Nim/issues/3725
16:09:43	dom96	federico3: nope?
16:17:02	wuehlmaus	dom96: exactly, i love it. and kudos to Araq_ for giving us such a wonderful language that even on my relatively weak system can compile without problems
16:17:21	dom96	:)
16:17:23	wuehlmaus	btw i have compiled crystal and it quit because i had not enough memory
16:17:32	wuehlmaus	nim is wonderfully light
16:17:48	wuehlmaus	that's a huge plus
16:17:50	dom96	that's nice to hear
16:18:08	wuehlmaus	i love it! :)
16:19:12	*	Jachy quit (Quit: Leaving.)
16:23:28	federico3	:(
16:24:27	Xe	\o/ got my main nim project behind HTTP/2, Nim 0.12 and Let's Encrypt!
16:25:17	dom96	Xe: cool, what's your project?
16:25:45	Xe	dom96: a simple little JSON api for My Little Pony: Friendship is Magic episode air dates, etc
16:26:40	dom96	ooh, does it support other TV shows?
16:26:46	*	thotypous quit (Ping timeout: 240 seconds)
16:27:13	Xe	no
16:27:23	Xe	well
16:27:25	Xe	it can
16:27:31	Xe	but it has no disambiguation
16:27:39	Xe	https://github.com/Xe/PonyAPI/blob/master/fim.list
16:27:43	Xe	this is the list of its episodes
16:28:49	Xe	but yeah
16:28:53	Xe	if you can get the info for it all
16:28:58	Xe	can adapt it for other shows
16:29:04	*	strcmp1 quit (Quit: cycling backwards like charlie)
16:29:11	Xe	it's not designed for it atm though
16:29:17	*	strcmp1 joined #nim
16:29:17	Xe	never was a need, etc
16:30:05	*	strcmp1 quit (Client Quit)
16:30:11	dom96	hrm, still awesome. Nice job!
16:30:18	*	strcmp1 joined #nim
16:30:21	dom96	was Let's Encrypt tough to set up?
16:31:31	*	strcmp1 quit (Client Quit)
16:31:40	Xe	nope
16:31:42	*	strcmp1 joined #nim
16:31:47	Xe	https://caddyserver.com
16:31:50	Xe	this automates it for you
16:31:54	*	dom96 should set it up for the forum
16:32:12	Xe	yeah
16:33:06	def-	dom96: Let's Encrypt doesn't work on WinXP though
16:33:22	Xe	def-: oh god the horror of not supporting a dead os
16:34:01	dom96	well, if anything I will make https optional
16:34:15	dom96	at least for the forseeable future
16:34:22	*	yglukhov joined #nim
16:34:35	Xe	caddy kinda enforces https though so might not be your best option :P
16:34:54	def-	dom96: Google automatically checks if https is available and all google search links to the forum will then be https, and this can't be disabled anymore
16:35:29	dom96	def-: :\
16:35:32	def-	Xe: I understand that WinXP isn't supported anymore, but in some cases people still use it and you don't want to exclude them, so it might be good to know
16:40:15	dom96	bbl
16:43:58	*	yglukhov quit (Remote host closed the connection)
17:16:41	*	thotypous joined #nim
17:36:39	*	vikaton joined #nim
17:44:26	*	yglukhov joined #nim
17:49:46	*	yglukhov quit (Ping timeout: 240 seconds)
18:02:13	*	Jesin quit (Quit: Leaving)
18:07:25	*	tankfeeder joined #nim
18:09:16	*	tankfeeder left #nim (#nim)
18:13:47	*	desophos joined #nim
18:28:11	*	brson joined #nim
18:36:30	*	yglukhov joined #nim
18:40:50	*	yglukhov quit (Ping timeout: 246 seconds)
18:41:18	*	Salewski joined #nim
18:42:24	*	cyraxjoe_ joined #nim
18:44:45	*	cyraxjoe quit (Ping timeout: 260 seconds)
18:47:54	Salewski	There is sequtils: proc filter*[T](seq1: seq[T], pred: proc(item: T): bool {.closure.}): seq[T]. Seems not to work when seq1 is a plain array, as expected. Is there a proc for arrays available? And why is seq1 not an open array, should work for seqs and array if I understand correctly? And yes, I think I can make my own filter for array.
18:49:38	Araq_	but what would it return?
18:49:43	Araq_	a seq?
18:50:12	Salewski	Yes, a seq I assume.
18:50:22	Araq_	hrm, try to patch it please
18:50:34	Araq_	s/seq/openarray
18:50:42	Araq_	should work
18:50:44	Salewski	Will try, thanks.
18:51:02	*	Salewski quit ()
18:58:10	*	Jesin joined #nim
18:58:41	*	Matthias247 quit (Read error: Connection reset by peer)
19:01:48	*	vendethiel- quit (Ping timeout: 272 seconds)
19:20:04	*	vendethiel joined #nim
19:25:12	*	onionhammer quit (Quit: WeeChat 1.0.1)
19:37:33	*	pregressive joined #nim
19:38:27	*	onionhammer joined #nim
19:42:58	*	vendethiel quit (Ping timeout: 272 seconds)
19:45:25	*	vendethiel joined #nim
19:48:21	*	kniteli joined #nim
20:14:11	*	awsteele joined #nim
20:15:36	*	awsteele quit (Client Quit)
20:23:06	*	brson quit (Quit: leaving)
20:27:34	*	yglukhov joined #nim
20:29:19	*	yglukhov quit (Remote host closed the connection)
20:41:17	*	darkf quit (Quit: Leaving)
20:44:50	*	irrequietus joined #nim
20:48:37	*	irrequietus quit (Client Quit)
20:52:30	*	yglukhov joined #nim
20:56:38	*	yglukhov quit (Ping timeout: 246 seconds)
21:00:51	*	Jesin quit (Remote host closed the connection)
21:06:32	wuehlmaus	hmmm, i use tnim to play around with nim.
21:06:56	wuehlmaus	import nre and it crashes on me wit the git nim version
21:07:14	Araq_	pretty sure the tests are green for nre
21:19:38	*	yglukhov joined #nim
21:21:47	*	pregressive quit (Remote host closed the connection)
21:24:22	*	yglukhov quit (Ping timeout: 265 seconds)
21:35:47	*	yglukhov joined #nim
21:42:26	dom96	Araq_: Why the change from `#[ #]` to `#[ ]` for multi-line comments?
21:43:12	Araq_	#[ #] is ugly
21:43:22	Araq_	#[ ]# would be the alternative
21:43:41	Araq_	but I am beginning to doubt I want the feature at all.
21:43:56	Araq_	Updating every highlighter out there ...
21:45:24	dom96	Hrm, I suppose #[ ]# will work just fine anyway, the last # will just be optional.
21:45:36	dom96	You don't have to update those highlighters.
21:45:41	dom96	Just the Nim one :P
21:46:05	yglukhov	#[ ]# Should not be optional to let commenting [] operators
21:47:09	yglukhov	Araq: why not nested?
21:48:36	yglukhov	#[ + ]# nest as you wish. don't waste time deciding what start/end tokens to use based on commented content.
21:49:01	Araq_	sorry, but it doesn't work this way.
21:49:11	Araq_	let foo = "#["
21:49:27	Araq_	# damn, you always need to take the content into account.
21:50:00	Araq_	allowing for nesting mitigates the problem, but doesn't solve it.
21:51:44	*	OderWat joined #nim
21:51:56	OderWat	hi @all
21:51:59	yglukhov	imho, nesting #[ ]# will work in 99% sane cases.
21:52:40	OderWat	I fell like using # for something else than "comment to end of line" is not so nice.. I did not think about that before but seeing it just hurts me now :(
21:52:45	yglukhov	and its much more handy in those 99%. the other 1%. well, put a space between false-ending-token and thats it
21:53:16	Araq_	well you can always do this:
21:53:28	Araq_	let foo = "#[" # ]#
21:53:35	Araq_	;-)
21:53:38	OderWat	maybe that impression changes when syntax highlighter actually do highlight it correctly ..
21:54:19	Araq_	OderWat: I know that's the case for me. I updated my highlighters already.
21:54:42	OderWat	whats the case? thats ok or that it is disturbing?
21:54:54	yglukhov	but that is an obviously unambiguos example, no? since string literal starts first, the multiline comment never starts
21:55:15	dom96	wait, is `let foo = "#["` not going to comment out the rest of the file (until the matching ])?
21:55:28	dom96	hrm, well I guess that makes sense.
21:55:45	OderWat	no it is a en end of line comment: # ]# ...
21:55:51	*	Demon_Fox joined #nim
21:56:02	OderWat	no it is an end of line comment: # ]# ...
21:56:24	yglukhov	i really cant see a. single. problem. with. nesting. come on!
21:56:52	dom96	yeah, me neither.
21:57:29	OderWat	not as long as the end is ]# ... araq showed me that the end is ] ... which makes proc(@[1,2#[,3]],4] a problem already
21:58:16	yglukhov	of course. single-char terminator is too ambiguous. but ]# is pretty unique already
21:58:36	dom96	indeed, that would be a good case for making ]# mandatory
21:59:23	OderWat	i only wished that its not # ... i think proc(@[1,2~[,3]]~,4]) would be nicer.. not mixin "end of line comment" # with the new ones..
22:00:33	yglukhov	OderWat I guess other symbols are allowed as operators now. so only # has a power of comment =)
22:01:41	OderWat	we are not 1.0 and i don't think it would break the world to remove the single ~ from being available. buuuuut... maybe
22:02:20	Araq_	pretty sure pegs already use ~
22:02:25	OderWat	one could also use [:blabla:]
22:02:47	OderWat	[:this is multiline\nand here it stops:]
22:03:04	dom96	or #| |#
22:03:05	Araq_	yglukhov: ok, so let's allow nesting. I don't really care.
22:03:16	OderWat	no thats containing # again dom
22:03:52	Araq_	OderWat: it's totally fine that that multi line comment starts with a # too.
22:04:02	dom96	What's the problem with the #?
22:04:08	yglukhov	OderWat well, i cant see why you dont like #[. i on the countrary like it very much that both comments and multiline comments have something common in tokens
22:04:10	OderWat	i feel var foo = "hello" #[test]# and here code again?
22:04:47	yglukhov	code again - yes
22:05:27	OderWat	after all the time # was end of line comment and is in other languages.. dunno.. don't like that
22:05:33	Araq_	so ... we use #[ ]# and allow nesting. and ##[ ]## for doc comments
22:05:37	dom96	As far as I can see there is only one other language which uses #[ ]# http://rosettacode.org/wiki/Comments
22:05:41	dom96	(That's Lily)
22:06:03	dom96	And it doesn't allow nesting
22:06:50	Araq_	what's done in other languages never bothers me.
22:07:18	dom96	I know. But I find it interesting and think others might as well.
22:07:34	Araq_	everything that is not like in Ruby is wrong anyway...
22:07:40	yglukhov	if the only argument - "in other languages" - its not enough. there should be some practical explanation.
22:08:06	Araq_	how about we deprecated #[ for years now as an argument?
22:08:31	Araq_	we *know* it won't break any code out there.
22:08:54	OderWat	hmm i tend to agree that #[ and ]# .. is ok... hum..
22:09:40	Araq_	yeah, I will change it to #[]# and ##[]##
22:09:40	OderWat	but not #[ something ] ...
22:09:54	Araq_	and they can nest.
22:10:24	Araq_	#[[ ]]# are not supported. or should they?
22:10:24	OderWat	nesting by using #[[ nested #[here]# ]]# .. right?
22:10:36	Araq_	no, nesting as yglukhov wants it.
22:10:42	OderWat	or nested by #[ test #[ test ]# ]# ..
22:10:46	Araq_	yes.
22:11:11	OderWat	ok with me.. i love that :)
22:11:28	OderWat	but I do not need to write highlighters :)
22:11:29	Araq_	dom96: just fyi #| would break Nim's parser.nim file ...
22:11:40	yglukhov	omg i cant express my happiness! :D
22:12:13	Araq_	hrm... seems like a bad compromise ... everybody is happy.
22:12:18	OderWat	i can give you my paypal address :)
22:12:20	Araq_	I have done something wrong.
22:12:55	OderWat	haha :) you did a very good job with the LL already. why not doing more good stuff :)
22:13:38	yglukhov	+1, ll is really nice, thanks Araq =)
22:13:39	Araq_	I'm making async multi-core ready ...
22:14:43	OderWat	omg in the end people don't want to use c# or go anymore.
22:14:44	yglukhov	do i understand correctly that all closures created in one proc will share the same context?
22:14:54	dom96	It's time for everyone to start making lists of things they want to be changed before 1.0
22:15:07	dom96	Because after 1.0 changing things will not be so easy
22:15:32	Araq_	yglukhov: erm yes.
22:15:33	OderWat	i want curly braces ... *giggles*
22:15:54	dom96	OderWat: How about.... LOLCODE syntax?
22:16:01	OderWat	that template from the changes "captureEnv" would help?
22:17:29	OderWat	I think that I really see some advantage in end (block) decorations and a "strict" mode in the compiler for that. but then i don't care. I never hat a big problem in python because of indentation problems.
22:17:42	OderWat	^had
22:19:13	dom96	There was a time when Araq wanted to make Nim support curly brackets, indentation and 'end'. Giving you the ability to choose between them.
22:19:35	dom96	Personally I think I value consistency more nowadays.
22:19:47	Araq_	yeah but nowadays I think the IDE should do that and we need to get away from text based tools
22:20:03	Araq_	it's 2016 ffs
22:20:17	OderWat	and some people write an ide which didn't do this in the past
22:20:19	dom96	Would you prefer to drag boxes around?
22:20:39	dom96	Text is still the most efficient way to write software.
22:20:45	Araq_	there is some nice middleground and you know it.
22:20:56	Araq_	so don't act like you don't, I hate that.
22:20:59	OderWat	i think that you need to copy/paste code effortlessly which is why I think end markers could be nice and even help the ide to indent it correctly
22:21:59	Araq_	nimedit still gets it wrong most of the time but the problem is actually easy enough to solve without end markers
22:22:02	*	pregressive joined #nim
22:22:09	OderWat	i am not sure if "evaluation of indent and matching it on paste" is good enough
22:22:42	Araq_	IMO it definitely is.
22:23:12	OderWat	I did not write code which has to do that. I thought it can work but I am not sure.
22:24:26	OderWat	after all you need good indent/"outdent" routines anyway with nim and for real with any language as the curly or end marker will not help until you do a full code format on every insert and that sucks for other reasons (partial code in other places)
22:25:55	*	yglukhov quit (Remote host closed the connection)
22:26:55	dom96	btw syntax highlighters need to be updated anyway
22:26:59	dom96	look at this crap https://github.com/yglukhov/nimx/blob/f15fcce7f4d4223997ee546643ce90bd5dd688fd/nimx/linkage_details.nim#L68
22:27:55	*	yglukhov joined #nim
22:31:03	OderWat	I found that the ACE code editor does feature >120 languages but no Nim :(
22:31:26	OderWat	Afaik GitHub uses this for code editing somewhere too.
22:32:03	OderWat	see .. https://github.com/blog/905-edit-like-an-ace
22:32:11	*	yglukhov quit (Ping timeout: 246 seconds)
22:32:47	Araq_	I use 4 different editors but no ACE :P
22:35:38	OderWat	well they said 2011 they do .. but that can have changed of course. ace still says they use it.. and i think that adding Nim is probably not that hard... https://github.com/ajaxorg/ace/tree/master/lib/ace/mode .. I of course have another motivation for wanting that :)
22:47:36	*	jakesyl joined #nim
23:02:48	*	apotheon joined #nim
23:03:11	apotheon	Hey, all.
23:03:23	dom96	hey apotheon
23:03:28	OderWat	why was '' (double single quote) not being considered for multi line comments? I just thought thats also an option which does not break existing code imho?
23:04:22	OderWat	var test = [1,''2,''3] # end of line comment
23:04:37	OderWat	(besides looking like " .. well)
23:05:26	apotheon	dom96: Do you know whether Nim supports a way to "guarantee" (at least as well as C, though it's not exactly a guarantee per se sometimes) clearing a particular segment of RAM without using inlined C or a C library to do it?
23:06:18	*	pregressive quit (Remote host closed the connection)
23:06:34	dom96	apotheon: no idea sorry
23:06:45	apotheon	I guess I'll ask Araq_ . . .
23:07:16	apotheon	Araq_: Can you tell me whether Nim supports a way to "guarantee" (at least as well as C, though it's not exactly a guarantee per se sometimes) clearing a particular segment of RAM without using inlined C or a C library to do it?
23:07:58	apotheon	Araq_: Also, keeping RAM from being written to swap, while I'm asking about memory management.
23:09:23	OderWat	guess you mean for storing passwords/auth data @apotheon?
23:09:52	apotheon	Yeah, sensitive data management is the use case.
23:10:37	apotheon	Encryption keys and data temporarily unencrypted on the system are more directly the kind of thing than passwords.
23:11:21	apotheon	. . . and I guess I don't mean *storing* so much as *in use*, so that (for instance) with a long-running process unencrypted data won't get left in RAM beyond when it's needed, and won't get swapped to disk.
23:15:10	OderWat	i think this is kinda unsolveable as there is hibernation in the game. so far the only way to store such data "safe" is using a smartcard like device. I think as soon as the data is (even temporary) in memory you trust the system (kernel) for it
23:15:59	OderWat	same goes for DMA like in those apple vulnerabilities through the thunderbold port.
23:16:58	*	joelmo joined #nim
23:19:56	dom96	yay https://twitter.com/d0m96/status/688500909052145664
23:26:39	OderWat	cool dom.. this may replace https://github.com/julienXX/terminal-notifier for me :)
23:27:34	apotheon	OderWat: Considering most OSes don't seem able to do suspend-to-disk reliably, that doesn't seem like a huge problem, and that shouldn't be a real issue for the kind of software I'm likely to write in the next year anyway.
23:28:20	apotheon	. . . and I try to be very careful about DMA stuff. I don't buy $2.5K monitors or play graphics-intensive games, anyway.
23:28:21	dom96	OderWat: It's a pity that I need to overwrite Nim's main() function and use Cocoa's NSApplication event loop thing.
23:29:31	apotheon	OderWat: If I can just get the level of confidence I can get from C, that'd be good enough for me, for now. The big problem is that I don't want to have *less* assurance of memory safety than C provides.
23:29:42	apotheon	. . . but I'd like to be able to use something other than C, sometimes.
23:29:46	OderWat	on linux you can just `swapoff -a` and nothing will get swapped to disk anyway... erasing the data in ram could be easy if you allocated it yourself such that it is never relocated.
23:30:32	apotheon	OderWat: Yeah, but I realized what I said about displays and games was irrelevant to my needs for the same reason `swapoff -a` is irrelevant; my software will presumably have users ot her than me.
23:30:39	apotheon	err, s/ot her/other/
23:33:17	OderWat	what type of memory safety you are talking about? afaik you can simply allocate some ram in Nim too?
23:35:21	OderWat	so you can wipe it.. if that gets swapped to the harddrive .. well still not that easy to find. i am not sure how and if you can lock that ram to not be swapable. but if somebody has the right to "read" all your applications memory thats something you can't stop imho and then there comes the smartcard or an external server which decrypts the data outside of the system.
23:37:12	apotheon	Preventing swap is probably the big problem.
23:37:24	apotheon	(on a per-process basis, as opposed to per-system or something like that)
23:37:56	apotheon	In theory, in C you can prevent allocated memory from being swapped. In practice, it sometimes depends on the OS, I think.
23:39:11	apotheon	Some languages' implementations are obviously particularly deficient in this area, using garbage collection that offers no guarantee of clearing RAM at any given point and stuff like that.
23:39:34	OderWat	hmm there is https://msdn.microsoft.com/en-us/library/aa366895%28VS.85%29.aspx ..
23:39:59	OderWat	which locks a memory region from getting swapped out
23:40:08	apotheon	Yeah, C++ can do it, too -- but I'm not particularly interested in using C++ for this stuff.
23:40:53	OderWat	Nim can do "anything" C or C++ can.. worst case you emit some code for it to the c compiler.
23:41:33	OderWat	that windows thing is not about c++ but about windows itself
23:41:37	apotheon	Yeah, but you might notice I asked about whether it can be done in Nim, rather than in some inlined C or something like that. One reason to use something like Nim or Rust is to avoid having to use C.
23:41:51	apotheon	Oh, I didn't really look at it very closely.
23:42:04	apotheon	C++ can do much the same thing C can in this respect, though, as I recall.
23:42:25	apotheon	(on Unixy systems, which is my primary target platform)
23:42:45	OderWat	http://linux.die.net/man/2/mlock ..
23:42:47	dom96	I don't think there are any Nim libraries which do this for you, so you would need to write one yourself.
23:42:52	apotheon	mlock is the thing, indeed
23:42:58	dom96	For that you would need to wrap C functions
23:43:09	OderWat	so just use it :)
23:43:19	apotheon	Yeah, I guess so.
23:43:25	dom96	Even if one existed, it would still use external C functions to do its job.
23:43:39	apotheon	dom96: As long as I wouldn't have to write the C, it'd make me happy.
23:44:37	apotheon	Almost any generally available, widely used language can make use of some C stuff, of course, but (for instance) that doesn't mean you can write a program in Python or Ruby and use C to manage the memory used by the rest of the program.
23:44:51	apotheon	I get the impression you can in Nim(?).
23:44:54	OderWat	int mlock(const void *addr, size_t len);
23:44:54	OderWat	int munlock(const void *addr, size_t len); .. should be "importc" able.. and are pretty easy to use with self allocated memory I guess (so the gc does not relocate it. but that may be safe for not changing it's size anyway.. dunno).
23:45:40	apotheon	Well . . . I've been at this coffee shop long enough. I need to buy some ingredients for dinner and go collapse at home.
23:45:45	OderWat	Nim compiles to C and has probably the best FF interface of all languages available
23:46:00	apotheon	Being ill sucks. Thanks for the info. I'll leave my IRC client connected in my absence.
23:46:34	apotheon	later
23:49:54	*	jakesyl quit (Ping timeout: 256 seconds)