<< 07-12-2013 >>

---

00:00:53	*	Demos joined #nimrod
00:03:31	BitPuffin	yeah I think I'm gonna do that
00:03:33	BitPuffin	night!
00:03:46	*	BitPuffin quit (Quit: WeeChat 0.4.2)
00:04:10	*	Icefoz quit (Quit: leaving)
00:05:06	*	Icefoz joined #nimrod
00:05:34	*	Icefoz left #nimrod (#nimrod)
00:06:01	*	Icefoz joined #nimrod
00:10:48	*	Amrykid quit (Excess Flood)
00:11:18	*	Amrykid joined #nimrod
00:17:40	*	ics quit (Ping timeout: 246 seconds)
00:20:36	*	ics joined #nimrod
00:31:48	*	Hannibal_Smith quit (Quit: Sto andando via)
00:32:22	*	zielmicha quit (Ping timeout: 246 seconds)
00:49:22	*	ics quit (Quit: My MacBook Pro has gone to sleep. ZZZzzz…)
01:06:27	*	boydgreenfield quit (Quit: boydgreenfield)
01:50:33	*	DAddYE quit (Remote host closed the connection)
01:51:07	*	DAddYE joined #nimrod
01:55:19	*	DAddYE quit (Ping timeout: 246 seconds)
02:03:09	*	webskipper quit (Ping timeout: 246 seconds)
02:03:43	*	brson quit (Quit: leaving)
02:06:05	OrionPKM	varriount you here?
02:12:39	*	Demos quit (Quit: Textual IRC Client: www.textualapp.com)
02:36:06	Varriount	OrionPKM, yeah
02:36:26	Varriount	I was playing a fullscreen game, sorry.
02:36:39	OrionPKM	np
02:37:30	OrionPKM	never mind, I was going to ask you something, but it doesn't matter anymore
02:37:34	Varriount	Oh?
02:41:10	OrionPKM	yeah. at the gym now though, so can't talk
02:44:03	OrionPKM	I'm working on babel integration when I get back tho
03:50:42	*	EXetoC quit (Quit: WeeChat 0.4.2)
03:53:07	*	DAddYE joined #nimrod
03:57:26	*	DAddYE quit (Ping timeout: 240 seconds)
04:16:31	*	fundamental quit (Read error: Operation timed out)
04:32:18	*	fundamental joined #nimrod
04:51:09	*	mflamer joined #nimrod
04:54:20	*	DAddYE joined #nimrod
04:58:51	*	ics joined #nimrod
04:59:50	*	DAddYE quit (Ping timeout: 240 seconds)
05:21:00	Varriount	Araq, how would one do nimrod/python style indentation in the peg parser? Is it possible?
05:21:10	Varriount	*with the peg parser module
05:56:37	*	DAddYE joined #nimrod
06:00:38	*	DAddYE quit (Ping timeout: 240 seconds)
06:00:50	*	ics quit (Ping timeout: 246 seconds)
06:02:26	*	ics joined #nimrod
06:57:43	*	DAddYE joined #nimrod
07:03:02	*	DAddYE quit (Ping timeout: 240 seconds)
07:18:49	*	hoverbear quit (Quit: Hibernating, be back soon.)
07:35:39	*	boydgreenfield joined #nimrod
07:59:55	*	DAddYE joined #nimrod
08:04:14	*	DAddYE quit (Ping timeout: 240 seconds)
08:15:07	*	webskipper joined #nimrod
08:16:40	*	dymk quit (Ping timeout: 245 seconds)
08:21:51	*	dymk joined #nimrod
08:37:21	*	_dymk joined #nimrod
08:37:45	*	dymk quit (Ping timeout: 245 seconds)
08:46:58	*	boydgreenfield quit (Quit: boydgreenfield)
09:00:48	*	DAddYE joined #nimrod
09:00:58	*	webskipper quit (Ping timeout: 240 seconds)
09:05:11	*	DAddYE quit (Ping timeout: 252 seconds)
09:19:13	*	silven joined #nimrod
09:27:14	*	shodan45 quit (Quit: Konversation terminated!)
09:49:04	*	io2 joined #nimrod
10:02:01	*	DAddYE joined #nimrod
10:06:20	*	DAddYE quit (Ping timeout: 240 seconds)
10:42:23	*	_dymk quit (Quit: ZNC Bouncer Quitting)
10:43:59	*	dymk joined #nimrod
10:44:38	*	mflamer quit (Ping timeout: 240 seconds)
11:00:13	*	PortablePuffin joined #nimrod
11:03:20	*	DAddYE joined #nimrod
11:05:50	*	BitPuffin joined #nimrod
11:06:03	BitPuffin	Araq: But openarrays are for parameters!
11:06:05	BitPuffin	anyways
11:07:26	*	DAddYE quit (Ping timeout: 240 seconds)
11:07:28	BitPuffin	Araq: I think I'm gonna need to send pretty big quantities of data through channels and that's apparently slow. Is there a good option to this though? maybe having some global seq that one can write and one can read and you send a message to read and push it in to that stack and pop it after reading maybe?
11:11:18	BitPuffin	guess it would be the reverse of push and pop
11:11:22	BitPuffin	but you get the idea
11:12:44	fowl	hi BitPuffin
11:12:51	fowl	hows life
11:13:07	BitPuffin	fowl: hey fow
11:13:10	BitPuffin	l
11:13:14	BitPuffin	stressful, and yours? :)
11:17:48	BitPuffin	Araq: well I guess it isn't like video files I'm gonna be passing arounds. But say 250 000 vertices? so with float32's that's 4*4*250000 bytes
11:18:14	BitPuffin	3906.5kb
11:18:52	BitPuffin	~3.8 MB
11:18:59	BitPuffin	Is that too much to pass through a channel?
11:19:06	BitPuffin	dom96: opinion plox
11:19:25	BitPuffin	and fowl if you know this stuff
11:27:27	fowl	what are you doing
11:27:33	fowl	with so much data
11:31:24	BitPuffin	fowl: terrain
11:32:36	BitPuffin	that's only one vertex per meter, I'm experimenting with interpolating via the geometry shader
11:32:57	BitPuffin	ofc geometry shaders aren't the fastest things so that's why it is experimenting
11:33:04	BitPuffin	but that allows me to send less data to the gpu
11:36:39	fowl	how are channels "apparently slow" did you do actually try it
11:50:18	BitPuffin	fowl: not yet, but there are warnings on the site and dom96 freaked out
11:52:17	fowl	BitPuffin, do you have libjit installed
11:52:52	BitPuffin	fowl: no
11:52:55	BitPuffin	or well let me check
11:53:23	BitPuffin	don't think so no
11:54:40	BitPuffin	nope
11:56:04	fowl	o
11:57:17	PortablePuffin	fowl: Why do you ask?
11:58:03	fowl	im getting a segfault running an example from the tutorial (in nimrod)
11:58:18	fowl	it compiles and runs fine from c
11:58:34	PortablePuffin	hm. That's weird
11:59:03	PortablePuffin	fowl: look at the generated c source I guess
12:04:52	*	DAddYE joined #nimrod
12:07:38	BitPuffin	who the hell is PortablePuffin
12:08:01	PortablePuffin	BitPuffin: yo moma
12:08:06	BitPuffin	'kay
12:08:18	PortablePuffin	:
12:08:18	BitPuffin	D
12:09:02	*	DAddYE quit (Ping timeout: 240 seconds)
12:11:49	fowl	the generated c is fine
12:12:25	BitPuffin	fowl: gdb it? :P
12:12:31	fowl	i dunno how :(
12:17:14	BitPuffin	fowl: this one is a decent introduction http://youtu.be/sCtY--xRUyI
12:24:41	BitPuffin	NOOB CAN'T EVEN CODE FACTORIAL
12:26:18	*	Hannibal_Smith joined #nimrod
12:28:59	fowl	well
12:29:13	fowl	that doesnt help
12:29:23	fowl	a function thats imported is segfaulting
12:32:06	*	Kristina_ is now known as kristina
12:32:29	BitPuffin	fowl: well you can also use gdb to dissassemble
12:32:39	BitPuffin	fowl: disassemble*
12:32:58	BitPuffin	fowl: so you could compare the asm of the c and nimrod versions
12:33:06	BitPuffin	maybe that can give you some hints on what's going on
12:33:59	fowl	but the c version works
12:34:20	BitPuffin	fowl: yeah, that's the point, you can compare the C version that works with the nimrod version that doesn't
12:34:28	BitPuffin	you are drunk today aren't you
12:35:30	fowl	no :(
12:35:36	fowl	i ran out of money, no beer
12:37:49	BitPuffin	no beer is probably a good thing
12:39:23	BitPuffin	http://cgdb.github.io/ is also really cool
12:46:28	Araq	hi PortablePuffin welcome
12:46:39	Araq	BitPuffin: just send a 'ptr' then
12:46:43	BitPuffin	Araq: that's just me on my phone but thanks :P
12:47:01	Araq	no, he's your mum
12:47:08	BitPuffin	oh yes
12:47:09	BitPuffin	true
12:47:11	BitPuffin	I am my own mother
12:47:28	Kooda	D:
12:47:35	fowl	BitPuffin, its no use unless i recompile libjit with debugging symbols
12:47:58	BitPuffin	Araq: okay maybe a ptr could work. I guess ref is forbidden because the GC only works on local threads?
12:48:19	BitPuffin	fowl: well libjit obviously works so there is some difference in your code
12:48:30	fowl	nope there is none
12:48:34	Araq	'ref' is not forbidden but causes a copy of the whole thing
12:48:57	BitPuffin	ah yeah true. I remember it saying that there is a deep copy
12:49:03	Araq	and yes ... if the data structure contains cycles stuff crashes
12:49:20	BitPuffin	Araq: do you think sending ca 3.8mb is that slow though?
12:49:35	Araq	I think it's stupid, yes
12:49:48	BitPuffin	alright
12:49:56	BitPuffin	well after its creation the data will be static anyway
12:50:00	BitPuffin	so what could go wrong :D
12:50:47	BitPuffin	in fact it doesn't even have to break gc? Because all the message will trigger is data being sent to the gpu. So I could just get a ptr with addr no?
12:51:42	Araq	"level" is a palindrome
12:53:45	BitPuffin	haha correct
12:54:15	fowl	radar
12:54:18	Araq	well the thing is: you're on your own when use 'ptr'
12:54:39	Araq	you need to ensure the data lives long enough in the thread that owns the data
12:54:45	BitPuffin	Araq: so the gc stops trying the moment I do addr x?
12:55:15	Araq	I guess you can say that. The GC doesn't trace addr x
12:55:22	*	Hannibal_Smith quit (Quit: Sto andando via)
12:55:26	BitPuffin	no but does it keep tracing x?
12:55:32	BitPuffin	if x is ref
12:55:51	Araq	well it traces 'x' on the heap that 'x' was created
12:56:29	BitPuffin	hmm, but I guess it's not a good idea to let the gc keep tracking x at that point
12:56:39	Araq	when that thread dies, the memory is freed no matter if other threads access that data
12:56:40	BitPuffin	since it could delete x while the receiver of addr is using it
12:57:01	BitPuffin	could be less of a problem though
12:57:27	Araq	you can also use allocShared
12:57:37	Araq	which seems to be easier to use from what I can tell
12:57:39	BitPuffin	since that thread with the data will only die when the program dies. And that data will probably be kept in that thread and not freed until you walk too far away from that terrain section
12:58:19	BitPuffin	and by then it's obviously already on the gpu
12:58:27	BitPuffin	being rendered and stuff
12:58:38	BitPuffin	or actually done rendering for the moment
13:00:22	BitPuffin	Araq: hmm, well if I send a ptr through a channel doesn't that more or less have the same effect as allocShared?
13:01:04	BitPuffin	or an addr or whatever
13:01:42	BitPuffin	I guess I could even make it nicely GC'd by creating a finalizer that sends a message to the opengl thread to delete the buffer
13:01:50	Araq	no, it's completely different
13:02:26	Araq	sending a ptr sends the data but the ownership is still bound to the creating thread
13:02:50	Araq	allocShared creates something that has no owner
13:02:57	BitPuffin	right
13:03:09	BitPuffin	but that doesn't affect the performance I assume?
13:03:32	Araq	sue it does, allocShared uses a lock
13:03:35	Araq	*sure
13:03:45	BitPuffin	but that's no good
13:04:18	Araq	well according to you you only allocate once
13:04:26	BitPuffin	yeah
13:04:41	Araq	so taking a lock for that allocation is irrelevant
13:04:42	BitPuffin	guess it wouldn't ever need to lock if everyone is only reading
13:05:36	Araq	well just to be clear the code does: aquire(heapLock); result = allocateSomehow(); release(heapLock)
13:05:59	BitPuffin	right
13:06:01	Araq	you don't have to do anything with that heap lock and it only affects the allocation itself
13:06:09	*	DAddYE joined #nimrod
13:06:18	BitPuffin	Araq: and reading causes no lock?
13:07:02	Araq	what do you mean? myPointer[] surely doesn't lock
13:07:15	BitPuffin	yeah that
13:07:17	BitPuffin	cool
13:07:21	BitPuffin	well hrm
13:08:00	BitPuffin	I guess to get the data from thread to thread (even if nobody is owning, just the thread that is really controlling it) I should still send it through a message
13:09:01	Araq	you can do that, yes
13:09:20	BitPuffin	Araq: does actors have the same kind of inefficiency? Like if I want to generate 100 trees, is it fine to use 100 threads for that?
13:09:50	BitPuffin	I'll probably be using both but for different purposes
13:10:43	*	DAddYE quit (Ping timeout: 250 seconds)
13:10:53	Araq	actors have some shitty thread abstraction, so you create a thread pool of N that work on the tasks
13:11:36	Araq	so it would use N threads to work on 100 trees
13:11:54	BitPuffin	yeah sure my computer sure as hell don't have 100 cores or whatever
13:12:21	BitPuffin	so it doesn't make that much sense to use 100 threads
13:12:35	BitPuffin	or oh it doesn't adjust automatically?
13:12:51	Araq	nope
13:14:07	BitPuffin	well, I guess one could get that with some system call
13:14:33	Araq	so why do I say it's a "shitty" abstraction? good question. It's because there is thread-local storage but not task-local storage.
13:15:09	BitPuffin	mm gotcha
13:15:23	BitPuffin	you mean that multiple actors might be able to access the same memory
13:15:34	Araq	and people spend lots of work to make thread local storage as efficient as possible. That's hard to get with task-local storage.
13:15:53	BitPuffin	goroutines xD
13:16:03	*	BitPuffin prepares to get slapped
13:16:15	*	Araq slaps BitPuffin
13:16:19	BitPuffin	ouch
13:18:58	BitPuffin	I wonder if I should even spend the time to make this game so heavily threaded
13:19:02	BitPuffin	I mean it's kind of cool
13:19:14	BitPuffin	but really it's not a heavy mega super game
13:19:28	Araq	don't thread it then
13:19:34	fowl	lets replace error messages with haikus
13:19:38	BitPuffin	ofc some things should be threaded
13:19:46	BitPuffin	like generating trees etc
13:19:48	Araq	especially if you have a deadline to meet
13:20:00	BitPuffin	but maybe sending messages etc to a rendering thread is overkill
13:20:01	BitPuffin	I do
13:20:26	Araq	I'd rather optimize after profiling and still keep everything single threaded
13:20:41	Araq	though admittedly somethings are easier when you use threads
13:20:47	BitPuffin	how come?
13:21:15	Araq	because multithreading in a domain where global state abounds simply is no fun
13:21:29	Araq	and games are all about global state
13:21:42	BitPuffin	yeah
13:21:48	BitPuffin	I mean you can thread some things easily
13:22:02	BitPuffin	like loading assets etc
13:22:40	fowl	hehe
13:22:42	fowl	"ass"ets
13:23:16	BitPuffin	but for some reason it can get really difficult to thread things like physics simulation and AI etc. Which is weird considering how those things surely aren't single threaded in real life :P
13:23:19	fowl	libjit, if anyone cares: https://gist.github.com/fowlmouth/7841229
13:23:43	BitPuffin	fowl: make something you can sell instead
13:23:51	fowl	BitPuffin, are you saying string theory is not threads?
13:24:55	fowl	BitPuffin, im not good at selling things
13:25:04	fowl	or making things
13:25:32	BitPuffin	fowl: sure you are. Just make some lame android/ios game or something and sell it for a buck
13:25:50	fowl	but i hate java and dont own the mac
13:26:18	BitPuffin	fowl: you don't need java and you don't need mac for android
13:26:38	BitPuffin	fowl: use the ndk
13:26:46	BitPuffin	set yourself a deadline of like 1 month
13:26:46	fowl	im pretty sure you need at least some java for android and i was talking about mac for iso
13:27:30	BitPuffin	fowl: well maybe a line or two I dunno. Not something that you need to write yourself, doesn't gradha already have samples for it on github? steal it
13:27:41	BitPuffin	https://github.com/gradha/nimrod-on-android
13:28:31	Araq	fowl: semicolons? you'll burn in hell for this one ... ;-)
13:28:47	fowl	there is like no nimrod in Nimrod/examples/cross_calculator/android
13:28:58	BitPuffin	you'll figure it out
13:29:07	BitPuffin	stop the excuses and get to it :)
13:29:39	BitPuffin	I have to go
13:29:42	BitPuffin	but I'll be portable
13:29:55	fowl	Araq, its copied from a c example
13:30:03	Araq	fowl: I know
13:30:06	PortablePuffin	awwwwww yeeeeeee
13:31:16	Araq	PortablePuffin: your son surely is straining sometimes
13:34:07	*	BitPuffin quit (Ping timeout: 250 seconds)
13:35:05	PortablePuffin	Araq: Yes, I'm gonna give him a real beating later on
13:39:18	PortablePuffin	fowl: seriously though do some project like that. It can only cause good things!
13:40:32	fowl	i dont want to write java
13:40:41	fowl	not now, not never
13:41:31	Araq	see you later guys
13:41:36	PortablePuffin	fowl: *sigh* you only need to bootstrap with java
13:41:36	fowl	cya
13:42:09	fowl	PortablePuffin, ive talked about this at length with grADHa, you have to write 99% java
13:42:33	PortablePuffin	fowl: what! how come
13:43:34	PortablePuffin	doesn't sound right
13:44:06	PortablePuffin	compile nimrod to java lol
13:44:38	PortablePuffin	fowl: jruby?
13:47:09	fowl	i thought about that but ive been in ruby channels long enough to see enough complains about jruby to know that it suxx
13:48:09	PortablePuffin	haha
13:48:18	PortablePuffin	well maybe that's changed
13:54:42	*	PortablePuffin quit (Ping timeout: 246 seconds)
14:02:15	*	PortablePuffin joined #nimrod
14:04:34	*	dyu joined #nimrod
14:06:09	*	PortablePuffin quit (Read error: Connection reset by peer)
14:07:16	*	io2 quit ()
14:07:43	*	DAddYE joined #nimrod
14:11:23	*	radsoc joined #nimrod
14:11:50	*	DAddYE quit (Ping timeout: 240 seconds)
14:13:50	radsoc	Hi, has anybody tried 'eval'? I get: Error: undeclared identifier: 'eval'. eval is part of system, no?
14:27:33	Kooda	Hm, I have something odd happening at runtime
14:28:20	Kooda	I have a type which is an object of 4 uint32, when I try to fill it I get “(invalid data!)” in all the fields, what does it mean?
14:36:05	*	dyu quit (Ping timeout: 250 seconds)
14:40:04	fowl	Kooda, repr() derps on uints, it always shows (invalid data!)
14:40:15	Kooda	Ah…
14:40:49	fowl	did you report the bug you found yesterday
14:41:13	Kooda	No, I think it already exists
14:42:00	Kooda	Well, there’s a lot of integers related PR
14:45:07	Kooda	fowl: also, why is the rational behind discouraging the use of unsigned integers? :o
14:45:11	Kooda	what is*
14:45:33	Kooda	I *have* to work with them, and it’s pretty painful
14:45:55	Kooda	(I have to use pointers and unsigned integers of various sizes ^^')
14:49:37	fowl	hmm import unsigned
14:49:46	Kooda	Yes yes, but why? :Þ
14:50:11	Kooda	Also, I use ntohl(), which is for int32 in nimrod but uint32 in C… That’s confusing
14:50:35	Kooda	Both should exist, imo.
14:55:24	fowl	nimrod has ntohl?
14:57:45	Kooda	In sockets
14:58:30	fowl	radsoc, eval is disabled
14:59:50	radsoc	thanks fowl! Do you know why? I have to use a macro?
15:00:41	OrionPKM	jesus, -23 C out
15:01:21	fowl	radsoc, what are you trying to do?
15:03:36	Kooda	fowl: $ works for pretty printing my structure, where repr() fail
15:03:42	Kooda	That’s odd ^^'
15:06:27	radsoc	fowl, I need to duplicate a template call for different parameter values.
15:06:44	*	webskipper joined #nimrod
15:07:54	radsoc	If I use a variable I get: Error: constant expression expected
15:08:34	*	DAddYE joined #nimrod
15:10:23	fowl	Kooda, I dont see an issue for the repr problem or the uint problem you had from last night
15:10:59	fowl	radsoc, you mean calling a template with different parameters? you cant do this?
15:11:48	Kooda	fowl: ok, I’ll report them then :
15:11:49	Kooda	:)
15:12:41	*	PortablePuffin joined #nimrod
15:13:48	*	DAddYE quit (Ping timeout: 246 seconds)
15:13:59	radsoc	Yes, but I don't want to type: myTemplate(1) myTemplate(2) myTemplate(3) and so on. I'd like to have: for i in 1.100: eval("myTemplate(" & $i &")")
15:14:48	fowl	mytemplate(i) doesnt work?
15:20:21	radsoc	No, I get: "Info: instantiation from here" and "Error: constant expression expected" because somewhere in my template code I have array[0..i, char]. I want to duplicate the code inside my template so I need multiple const, not a var.
15:21:26	radsoc	does it make sense?
15:21:41	fowl	radsoc, arrays are fixed sized
15:21:41	*	PortablePuffin quit (Read error: Connection reset by peer)
15:21:48	*	PortablePuffin joined #nimrod
15:21:58	fowl	you cannot say in this iteration i want the array to hold 2, the next one i want it to hold 3
15:22:00	*	EXetoC joined #nimrod
15:22:37	fowl	you can replace it with a seq (use newseq[char](i+1) to get the same effect)
15:22:39	radsoc	No, I want the code to be duplicated at compile time
15:25:13	radsoc	I can't use a seq in my configuration. We had a conversation about that with Araq yesterday.
15:26:02	dom96	Perhaps you should use a macro to generate those template calls.
15:26:42	dom96	You should be able to use 'emit' in a macro to do what you want.
15:27:44	fowl	well if you {.unroll.} the loop then you should be able to do this, but i just tested it and it doesnt work
15:28:03	radsoc	dom96: I tried emit but I also get Error: undeclared identifier: 'eval'
15:29:06	fowl	i will write something for you.. one sec
15:30:53	dom96	radsoc: how did that happen? Are you emitting 'eval'?
15:31:13	Araq	I removed system.eval as I couldn't see the point of it
15:31:21	Araq	we have 'static' for it now
15:31:34	Araq	and "eval" is a confusing name
15:32:09	*	joelmo joined #nimrod
15:33:33	Araq	hi joelmo welcome
15:33:46	fowl	radsoc, try this: https://gist.github.com/fowlmouth/aa06e723494708c8d21b
15:33:57	joelmo	hi Araq, thank you :)
15:34:02	Araq	radsoc: you surely could use a sequence but I can't see that being faster than a string
15:34:04	radsoc	dom96: no I didn't emit an 'eval' but I guess emit calls eval
15:34:30	Araq	good point, I'll remove emit too then
15:35:53	dom96	Araq: er, perhaps you should deprecate it?
15:36:09	Araq	Kooda: unsigned arithmetic is brain dead, http://critical.eschertech.com/2010/04/07/danger-unsigned-types-used-here/
15:37:30	Araq	dom96: in general I agree with you, but in this case I find both eval and emit too annoying to provide a deprecation path
15:37:48	Araq	I could move 'eval' to macros.nim I guess
15:38:13	radsoc	fowl: thanks, I'll try to understand your gist!
15:38:44	fowl	radsoc, you can add at the end of the macro "echo(repr(result))" to see the resulting code
15:39:22	dom96	Araq: Fair enough
15:39:50	*	PortablePuffin quit (Ping timeout: 240 seconds)
15:39:58	Araq	fowl: the 'unroll' pragma does nothing currently
15:40:08	Araq	and the docs say so
15:41:26	fowl	oh
15:41:45	OrionPKM	you shouldnt be too afraid of breaking code :P
15:41:51	OrionPKM	it's < 1.0 still
15:42:16	*	XAMPP_ quit (*.net *.split)
15:42:17	*	noam quit (*.net *.split)
15:42:17	*	reactormonk quit (*.net *.split)
15:43:13	radsoc	Araq: eval syntax is a lot simpler than macro (from a user point of view)
15:43:18	*	PortablePuffin joined #nimrod
15:43:38	*	PortablePuffin2 joined #nimrod
15:43:38	*	PortablePuffin quit (Read error: Connection reset by peer)
15:44:09	radsoc	moving 'eval' to macros.nim would be great
15:44:11	Araq	oh really? how come you used it wrong then? :P
15:44:25	Araq	(eval takes a block and not a string)
15:48:22	*	reactormonk joined #nimrod
15:48:42	*	BitPuffin joined #nimrod
15:49:21	BitPuffin	hay ho! hay ho!
15:50:00	BitPuffin	Araq: eval takes a brock and not a sling
15:51:32	radsoc	I tried emit first (with a string) and didn't change when I tried with eval. But BTW none of them works!
15:51:54	BitPuffin	radsoc: try using admit instead
15:53:47	radsoc	Araq: I don't think anybody needs both (emit & eval)
15:54:20	Araq	template emit*(e: expr[string]): stmt =
15:54:21	Araq	macro payload: stmt {.gensym.} =
15:54:23	Araq	result = e.parseStmt
15:54:24	Araq	payload()
15:54:33	Araq	now patch your macros.nim, make a pull request and be happy
15:57:16	radsoc	Araq: I'd like to but I'm not a github user. (actually I'm not even a real developer)
15:57:29	fowl	that's ok, we're not even real people
15:57:51	fowl	we're figures of your imagination
15:58:22	radsoc	I guess so!
15:58:26	Araq	well then wait until somebody did it; I'm working on a different branch and switching branches annoys me :-)
15:59:45	BitPuffin	Araq: yeah that annoys me too
16:00:51	BitPuffin	in fact git annoys me
16:00:59	BitPuffin	so I just use hg :D
16:07:43	fowl	git is scary
16:08:01	*	XAMPP_ joined #nimrod
16:08:01	*	noam joined #nimrod
16:08:24	Araq	radsoc: did you try a TCritBitTree or whatever I called it?
16:08:50	*	dirkk0 joined #nimrod
16:09:56	BitPuffin	in mercurial you do things the way you'd expect to do it
16:10:02	BitPuffin	and things are called what you'd expect
16:10:09	*	webskipper quit (Ping timeout: 246 seconds)
16:10:13	BitPuffin	like subrepository instead of submodule
16:10:45	fowl	do they automatically download with the repository when you clone it
16:10:51	*	DAddYE joined #nimrod
16:11:03	fowl	or do you have to do something like git submodule init
16:11:30	Zuchto	Hi, I'm trying to figure out how to statically link a library to my nimrod application. Is there any documentation on this?
16:11:48	BitPuffin	fowl: not sure
16:12:33	fowl	Zuchto, you have {.passl: "-lyourlib".} and use {.header: "myheader.h".} on things you {.importc.}
16:12:59	BitPuffin	"Notably, the 'pull' command is by default not recursive. This is because Mercurial won't know which subrepos are required until an update to a specific changeset is requested. The update will pull the requested subrepositories and changesets on demand. To get pull and update in one step, use 'pull --update'. "
16:13:02	*	zielmicha joined #nimrod
16:13:03	fowl	Zuchto, heres an example https://gist.github.com/fowlmouth/7841229#file-jit-nim-L19
16:13:36	Zuchto	fowl: ah, ok. Thanks!
16:13:44	radsoc	Araq: I tried to copy paste your code but I get: macros.nim(226, 8) Warning: unknown magic 'NGenSym' might crash the compiler [UnknownMagic] / macros.nim(225, 11) Error: implementation of 'macros.genSym(kind: TNimrodSymKind, ident: string): PNimrodNode' expected
16:14:24	BitPuffin	fowl: there is an important difference there though, hg update is a common command to changing to different change sets so it would pull the subrepos you need
16:14:57	BitPuffin	fowl: update has the aliases up, checkout, c so you know what it is
16:15:02	*	DAddYE quit (Ping timeout: 240 seconds)
16:15:18	Zuchto	fowl: is the LinkStatically defined by some compiletime switch?
16:15:48	BitPuffin	fowl: oh wait this is with pulling
16:15:54	BitPuffin	that doesn't answer your question lol
16:16:51	BitPuffin	fowl: meh I'll just test it for you
16:17:24	fowl	Zuchto, in this example linking statically is enabled by passing -d:linkstatically
16:17:48	Araq	radsoc: alright, will have a look later
16:20:25	Zuchto	fowl: ok, thanks
16:21:06	radsoc	Araq: thanks, no problem
16:23:36	BitPuffin	fowl: it clones them by default
16:24:05	BitPuffin	fowl: and you can have git and svn subrepos
16:26:11	BitPuffin	fowl: also you don't have to add each file before every commit
16:26:37	fowl	usually i dont add every file in a commit
16:26:59	BitPuffin	well I mean you don't need to do hg add .
16:27:35	BitPuffin	ofc if you just want to commit some files you can
16:27:54	BitPuffin	hg commit src/graphics/*.nim
16:28:13	BitPuffin	hg commit src/io/network.nim src/main.nim
16:28:18	fowl	oh
16:28:32	fowl	how do you handle wchar_t*
16:28:45	BitPuffin	oh wait for the glob syntax you use -I before
16:28:50	BitPuffin	or --include
16:28:52	*	zielmicha1 joined #nimrod
16:29:25	BitPuffin	but you don't need -I for a dir
16:29:34	*	boydgreenfield joined #nimrod
16:29:47	Araq	zielmicha1: I'd argue for stdout for stack traces ...
16:30:22	Araq	"gah, can't read the error messages"
16:30:29	*	zielmicha quit (Ping timeout: 250 seconds)
16:30:30	Araq	./foo >output.txt
16:30:49	Araq	"wtf? doesn't work?"
16:31:14	Araq	"oh right, 'stderr' is the 'correct' way of doing it. thanks for that, gcc"
16:31:31	Araq	"how do I redirect stderr again? no idea, let me google it"
16:31:41	BitPuffin	Araq: which year were you born?
16:31:48	fowl	2&>1 or something like that
16:32:25	Araq	BitPuffin: that's none of your business
16:33:48	Araq	input vs output makes some sense, input vs output vs error stream is stupid IMHO
16:33:52	BitPuffin	Araq: too bad, I thought it would be nice to have a slogan for you like "Araq, slapping developers since <insert year>"
16:34:26	Araq	why errors? what if I have warnings too?
16:36:02	zielmicha1	Araq: then redirects are useless for everything else other than manual inspection
16:36:12	*	zielmicha1 is now known as zielmicha
16:36:43	Araq	zielmicha: they already are useless, most programs have something like --out:filename
16:36:44	zielmicha	Araq: and you know there is "2>&1" and even "&>" in bash
16:37:14	Araq	programs dumbing their results on stdout are not common anymore anyway
16:37:21	Araq	*dumping
16:37:41	zielmicha	so you know any language than prints stacktraces to stdout?
16:38:37	Araq	well I know GCC is annoying with its approach to "correctness"
16:39:18	zielmicha	well, GCC is a compiler, not a runtime
16:39:37	Araq	stack traces are arguably not really part of the runtime
16:40:15	Araq	however I can live with a patch that uses stderr but allows me to override with -d:useStdoutForStackTraces or something
16:40:38	zielmicha	ok, I will write one
16:42:28	zielmicha	and "stackTraceNewLine*: string ## undocumented feature; it is replaced by ``<br>``" is a really bad idea
16:42:59	zielmicha	if user can controll error message than he may write <script>doevilthings</script> to html if he manages to cause error
16:43:04	zielmicha	unless I'm missing something
16:43:12	Araq	it's really nice for cgi apps
16:43:35	Araq	which are so uncommon these days that we could remove it, I guess
16:43:39	zielmicha	better solution would be "var printBeforeStacktrace"
16:43:49	zielmicha	and use printBeforeStackTrace = "<plaintext>"
16:43:51	Araq	that misses the point though
16:43:59	Araq	oh wait
16:44:01	Araq	hmm
16:44:48	Zuchto	So, when dynamically linking to a library using the {.dynlib.} pragma it doesn't show up when i then run ldd on the executable. Could someone explain this to me?
16:44:54	Araq	well I dunno, I can't see how a user can affect stackTraceNewLine
16:45:14	Araq	Zuchto: nimrod uses dlsym to load symbols
16:47:10	Zuchto	Araq: ok, is there any way to have nimrod actually add the library as a dynamically linked dependency then?
16:47:23	Araq	Zuchto: there are lots of ways
16:47:54	Araq	you can use --dynlibOverride:foo and --passL:libfoo.a instead
16:48:23	Araq	you can use --verbosity:2 at compile time to see the dependencies
16:48:45	Araq	ok ... that's about it I guess ;-)
16:49:09	fowl	wchar_t* in nimrod, how?
16:49:34	Araq	zielmicha: so users can crack stackTraceNewline but not printBeforeStackTrace? how does that work?
16:49:55	zielmicha	Araq: <plaintext> is special, it causes browser to stop interpreting HTML
16:50:03	zielmicha	you can't close it
16:50:22	zielmicha	thought this is non-standard
16:50:27	Araq	zielmicha: so? I can make it <script>evil</script> instead according to you?
16:51:27	zielmicha	I mean, if you use: raise newException(MyError, "invalid user input: $1" % [somethingControlledByUser])
16:51:40	Zuchto	Araq: hmm, I think I was unclear about what I was actually asking for :) If I have libfoo.so as a dynamic dependency, is there any way to have this be listed when I do `ldd executable`?
16:52:08	zielmicha	and user provides somethingControlledByUser="<script>evil</script>" than it will go to stdout with stacktrace
16:52:27	zielmicha	and browser will interpret it (I think, I haven't read cgi module source)
16:52:28	Araq	Zuchto: I don't think so, but as I said you can link in a conventional way too
16:52:48	fowl	stacktracefilter: proc(x: string): string
16:52:49	Zuchto	Araq: ok, thanks for taking your time :)
16:52:59	fowl	set it to htmlentities() in cgi.nim
16:53:20	zielmicha	Zuchto: maybe {passl: "-lmylib"}?
16:53:23	Araq	fowl: that's what I fear too
16:53:46	Zuchto	zielmicha: but that would just statically link the library, yes?
16:54:05	zielmicha	no (assuming you don't pass -static to compiker)
16:54:11	Araq	zielmicha: alright fair point, but that doesn't argue against stackTraceNewLine at all
16:54:28	Araq	but against exception messages being not escaped
16:54:56	zielmicha	yes, but than we can have fowl's stacktracefilter than will do htmlspecialchars(foo) & "<br>"
16:55:35	zielmicha	and do the same without need for stackTraceNewLine
16:55:50	zielmicha	and I think this is not enough
16:55:52	Zuchto	zielmicha: ah, thanks!
16:55:59	Araq	ok, I'm sold
16:56:22	zielmicha	Python's cgitb.py has complicated code than closes headers, <script> tags, html comments
16:56:41	zielmicha	(i.e. what happens if exception happens when you are echoing <script> tag?)
16:57:35	fowl	well you could just replace \L with <br/> in the filter function, or wrap the trace in <pre> or w/e
17:00:23	zielmicha	(https://github.com/python-git/python/blob/master/Lib/cgitb.py#L30)
17:01:42	*	fundamental2 joined #nimrod
17:03:42	Araq	well I dunno. a filter function looks good but requires more discussions/thought
17:03:48	Araq	but bbl
17:05:07	*	fundamental quit (Ping timeout: 260 seconds)
17:06:13	*	q66_ joined #nimrod
17:06:18	*	hoverbear joined #nimrod
17:06:39	*	q66 quit (Disconnected by services)
17:06:41	*	q66_ is now known as q66
17:08:45	*	mflamer joined #nimrod
17:12:02	*	DAddYE joined #nimrod
17:16:14	*	DAddYE quit (Ping timeout: 240 seconds)
17:26:48	*	PortablePuffin2 quit (Ping timeout: 246 seconds)
17:30:55	radsoc	fowl: the code you gave me works great. thanks a lot.
17:37:13	Zuchto	Ok, yet another question about linking. I can statically link a library using {.passl: "/usr/lib/libfoo.a".} and that works great... but it means I have to actually specify that the library is in /usr/lib... can I somehow tell the compiler to search for the file where it expects to find it?
17:37:29	Zuchto	similar to how the "-lfoo" flag works
17:37:47	zielmicha	I think you can't, without -static (and then just use -lfoo).
17:37:48	fowl	idk, i was under the impression that -lfoo was static linking
17:37:59	fowl	<- drunk
17:38:04	*	PortablePuffin joined #nimrod
17:38:06	zielmicha	but probably there is a script like ldconfig which can find it
17:38:09	Zuchto	zielmicha: yeah, only problem with -static is that it statically links _everything_
17:38:34	Zuchto	fowl: no, it just adds it as a dynamic dependency (which is statically listed in the executable :))
17:43:45	Zuchto	oh, figured it out! by passing "-Wl,-static -lfoo -Wl,-Bdynamic" to the compiler it works as it should... at least in gcc :)
17:58:52	Amrykid	fowl, I saw your PR
17:59:04	Amrykid	merged it
18:00:38	*	ics quit (Ping timeout: 250 seconds)
18:05:35	*	ics joined #nimrod
18:13:14	*	DAddYE joined #nimrod
18:17:26	*	DAddYE quit (Ping timeout: 240 seconds)
18:18:41	fowl	yay
18:18:46	fowl	i contributed^^;
18:20:50	EXetoC	BitPuffin: I don't know when you added the warning to TStream, but pointer should be the same thing as void*: both are pointers to untyped memory
18:21:33	EXetoC	I'm trying to port one of the examples now. some error is being raised though
18:21:57	*	Hannibal_Smith joined #nimrod
18:22:45	BitPuffin	EXetoC: I haven't added a warning :o
18:22:59	BitPuffin	or do you mean a comment?
18:23:28	*	boydgreenfield quit (Quit: boydgreenfield)
18:23:39	EXetoC	BitPuffin: ok warning comment
18:24:44	BitPuffin	EXetoC: uh, I can't remember what the reasoning I came up with was
18:24:45	Araq	thinking about it some more ... we need an "exceptionMessageFilter" instead
18:25:06	BitPuffin	EXetoC: I know that there is one place that takes a *void though and 2 or so that takes **void
18:25:06	Araq	it doesn't make sense to filter stack traces the same way as user level error messages
18:25:15	BitPuffin	so TStream = pointer
18:25:20	BitPuffin	used where it's a *void
18:25:26	BitPuffin	and PStream = ptr TStream
18:25:31	BitPuffin	used where it's a **void
18:25:33	BitPuffin	or something like that
18:26:29	BitPuffin	EXetoC: would be awesome if you figured out the error and patched it if there is a problem with the binding
18:26:53	EXetoC	yes, so PStream only serves as documentation. still useful in other words
18:27:08	BitPuffin	I'm currently working on graphics for my engine since you guys are only horny for screenshots and not soundshots
18:27:21	*	mflamer quit (Ping timeout: 246 seconds)
18:27:24	BitPuffin	EXetoC: Yeah and sprinkling some typing on it or something like that
18:27:26	fowl	protip: dont use type X = pointer because type X functions match for other generic pointers
18:27:33	fowl	use "ptr object"
18:27:36	EXetoC	I don't think there's anything wrong with the related types in the binding, but I will try the C example
18:28:03	EXetoC	fowl: when the equivalent in C is void*?
18:28:24	EXetoC	which it is, but I'll see if it makes sense to introduce a dummy type
18:28:38	Araq	but ... generating stack traces and making them part of the generated HTML is horrible really. I guess CGI needs a stderr stream lol
18:29:37	Araq	oh well who uses cgi in production these days anyway
18:29:39	fowl	EXetoC, at least use distinct pointer, the downside to thise is you have to borrow functions like isNil whereas ptr object just works
18:31:01	Araq	fowl: people should pay you for giving "nimrod - best practices"
18:32:04	fowl	i know right
18:32:18	fowl	mail me the change you found under your couch, you werent using it
18:48:33	fowl	Zuchto, thanks
18:50:27	*	radsoc quit (Ping timeout: 250 seconds)
18:50:58	Araq	fowl: there is system.WideCString
18:51:27	Araq	and newWideCString("foo")
18:52:11	fowl	i dont see wideCstring in system
18:52:37	Araq	well there is system/widestrs.nim and it's included
18:52:43	Araq	but maybe only on windows :P
18:53:44	Zuchto	fowl: likewise
18:58:43	*	webskipper joined #nimrod
19:07:03	zielmicha	maybe something like Python's sys.excepthook would make sense, instead of filters
19:07:18	zielmicha	sys.excepthook is a function that is called when interpreter would print stacktrace
19:08:31	zielmicha	the cgi one could, for example "<pre>"getStacktraceToString().htmlspecialchars
19:08:44	zielmicha	and than echo it
19:09:22	zielmicha	it's better than message filter in that it could for example generate html by inspecting frames manually (if it's supported)
19:10:26	fowl	so the default exception hook would be stderr.write(backtrace())?
19:13:06	zielmicha	yes
19:13:42	zielmicha	(other possible use case: error reporting to central error reporting service like Sentry)
19:14:12	Araq	actually we already have a raiseHook for that
19:14:28	*	DAddYE joined #nimrod
19:16:55	*	PortablePuffin quit (Ping timeout: 246 seconds)
19:18:40	*	DAddYE quit (Ping timeout: 246 seconds)
19:19:02	Araq	and btw hooks suck for verification
19:20:46	zielmicha	localRaiseHook and globalRaiseHook are not the same
19:21:30	zielmicha	they are called even if there are pending try..except, if I understand code correctly
19:23:07	*	viteqqq joined #nimrod
19:23:10	Araq	hmm true
19:23:15	Araq	hi viteqqq welcome
19:24:47	viteqqq	hi there
19:32:06	*	ics quit (Quit: My MacBook Pro has gone to sleep. ZZZzzz…)
19:33:37	OrionPKM	araq are there any examples of using CaaS TCP server?
19:33:57	Araq	there are lots of tests
19:34:24	OrionPKM	I see the tests folder, but none of them seem to actually show connecting to the service, maybe i'll look more closely
19:38:11	fowl	http://adamralph.com/2013/12/06/ndc-diary-day-3/?1
19:38:14	fowl	new stuff in c#
19:41:34	Hannibal_Smith	I'm a C# programmer, and I can't find anything that would make my code better
19:42:41	*	DAddYE joined #nimrod
19:42:44	*	DAddYE quit (Remote host closed the connection)
19:42:50	*	DAddYE joined #nimrod
19:43:19	fowl	now your one line functions don't need braces
19:43:44	Hannibal_Smith	And create more style inconsistency
19:45:34	Hannibal_Smith	Previous version of the language, at least added some good addiction for "daily use" code
19:45:46	Hannibal_Smith	Like Linq, async, dynamic...
19:45:56	Hannibal_Smith	Now I have nothing
19:46:06	fowl	you still have those things :D
19:46:14	Hannibal_Smith	Probably there is more that MS are keeping secrets
19:46:37	Hannibal_Smith	fowl, after using C++...I really like generic programming
19:47:11	Hannibal_Smith	fowl, probably now the think I would like is a jit like the OracleVM one
19:47:22	Hannibal_Smith	Not more language features
19:48:02	Hannibal_Smith	*the thing
19:50:16	Hannibal_Smith	Ah fowl and probably, some people migrated from C# to F#
19:51:06	Hannibal_Smith	Probably the same people who in the past would use some of these new features
19:51:34	Hannibal_Smith	But now get directly a better language
19:51:52	Hannibal_Smith	And not a "blob" that is C#
20:19:40	*	PortablePuffin joined #nimrod
20:21:45	Zuchto	Question: If you want to pass a structure as through a function in a C library that will populate the structure with information (in other word, the C function doesn't store a pointer to the structure) would there be any case where the structure (in nimrod) still has to be a ptr rather than a ref?
20:26:25	EXetoC	PortablePuffin: ok I hear some horrible noise now
20:26:38	EXetoC	should sound like a saw wave, but at least it's working
20:27:55	*	Hannibal_Smith quit (Quit: Sto andando via)
20:46:25	Araq	Zuchto: when you can 'new' it, you can make it a 'ref'
20:49:35	Zuchto	Araq: Yes, but I guess what I'm asking for is if there are any sneaky edge-cases where you would think you could use a ref but you really should use a ptr... but I guess it's a bit too wide question for it to be meaningful.
20:49:56	OrionPKM	araq couldnt find anything about testing the caas tcp server
20:50:03	OrionPKM	just stdin
20:51:51	Araq	how do you start a new terminal? clicking on your shortcut icon in the task bar was the old way
20:52:15	Araq	now I have to type in "cmd" in the search field to get a fresh terminal. Thank you Microsoft.
20:52:51	Araq	And thank you too Apple for showing the world how to ruin a task bar.
20:53:09	Araq	of cource everybody needs to copy Apple's braindead and utterly broken approach.
20:53:35	OrionPKM	i'm hitting it from python
20:54:10	Araq	OrionPKM: now that I got a new terminal, I could grep for it ;-)
20:54:21	Araq	have a look at compiler/service.nim
20:54:44	EXetoC	Zuchto: can't think of any reasons why it'd cause any problems in that case
20:54:50	EXetoC	anyway, maybe you could wrap the pointer
20:56:02	OrionPKM	araq hmm, ok
20:56:02	Araq	Zuchto: it really depends on who allocates the memory. If the C library allocates, you can't use a 'ref'
20:56:44	Araq	good libraries provide both Foo* newFoo() and void initFoo(Foo* f)
20:57:10	Araq	unfortunately there are barely any good libraries around
20:57:13	Zuchto	then i really wish more libraries where good :P
20:57:37	Zuchto	ok, thanks for answering my very vague question :)
20:58:23	Araq	hey, it's C everything can talk to C. No need for good libraries when you're building on a fundament of ignorance.
20:58:58	OrionPKM	araq huh.. it crashes when I run from cmd line vs running it from a subproc in ST
20:59:49	Araq	OrionPKM: well the people who could develop it further are busy patching Nimrod to become more like haskell. ;-)
21:00:02	OrionPKM	lol
21:03:04	*	brson joined #nimrod
21:05:40	*	brson_ joined #nimrod
21:08:47	*	oszo joined #nimrod
21:08:53	Araq	hi oszo welcome
21:09:10	OrionPKM	araq, getting this: 'N_NOINLINE' # define N_NOINLINE(rettype, name) rettype __attribute__((noinline)) name
21:09:14	OrionPKM	runtime error
21:09:26	*	brson quit (Ping timeout: 240 seconds)
21:09:38	*	oszo quit (Client Quit)
21:09:55	Araq	OrionPKM: idetools is supposed to run the C compiler
21:09:59	Araq	*is not
21:10:24	OrionPKM	strange then
21:10:34	OrionPKM	gcc.exe: error: E:\Development\IRCFamiliar\trunk\src\nimcache\trunk_.o: No such file or directory
21:10:38	OrionPKM	thats happening at runtime
21:10:38	*	viteqqq_ joined #nimrod
21:12:39	*	viteqqq quit (Read error: Connection reset by peer)
21:12:57	Araq	EXetoC: fyi http://critical.eschertech.com/2010/05/26/specification-with-ghost-functions/
21:13:20	Araq	this has 'pre' and 'post' conditions and they are verified, not sugar for 'assert'
21:15:03	Zuchto	so the overload resolution doesn't look at expected return type or if the procedure is exported?
21:16:17	Zuchto	oh wait... the problem isn't related to export
21:16:20	Zuchto	my mistake
21:16:24	*	PortablePuffin quit (Ping timeout: 246 seconds)
21:27:24	*	ics joined #nimrod
21:47:28	*	PortablePuffin joined #nimrod
21:51:01	zielmicha	Araq: what about merging https://github.com/Araq/Nimrod/pull/702 (quoteIfContainsWhite)? Does it need improvements? Or should I squash commits?
21:52:49	Araq	make the 'let safeUnixChars' a 'const' please and it's perfectly fine
21:56:58	zielmicha	done
21:59:54	NimBot	Araq/Nimrod master 5dd5c78 Michał Zieliński [+0 ±2 -0]: Make quoteIfContainsWhite quote argument, so it can be safely passed to shell.... 6 more lines
21:59:54	NimBot	Araq/Nimrod master 887466d Andreas Rumpf [+0 ±2 -0]: Merge pull request #702 from zielmicha/master... 2 more lines
22:00:27	Araq	thanks for nagging btw, it's bad to not merge pull requests in a timely fashion
22:01:16	*	mflamer joined #nimrod
22:01:41	Araq	mflamer: you have a new job
22:01:47	Araq	congrats
22:02:07	Araq	people really want idetools to work ... ;-)
22:02:31	Araq	and you're definitely up for the task I think
22:03:42	Araq	zielmicha: instaed of let output = when defined(useStdoutForStackTraces): stdout else: stderr
22:03:53	Araq	I'd prefer:
22:04:18	Araq	when defined(weReallyHateStdErr):
22:04:29	Araq	template stdmsg*: expr = stdout
22:04:30	Araq	else:
22:04:35	Araq	template stdmsg*: expr = stderr
22:04:53	Araq	for system.nim :-)
22:05:21	zielmicha	maybe you just need to patch kernel to pipe fd 2 to fd 1?
22:05:22	EXetoC	https://gist.github.com/EXetoC/528d88cfa5bc2ca69b19 this crashes on line 21. cb is passed to opendefaultstream as you can see. it works when 'output' isn't a var
22:06:07	Araq	zielmicha: ultimately indeed I'll write my own OS :P
22:06:14	EXetoC	actually, maybe I don't need to modify 'output'. I need to check the C source again
22:06:54	EXetoC	it's incremented, right? "*out++ = data->left_phase"
22:08:30	Araq	EXetoC: I never checked codegen for a pointer to (a pointer to?) a function
22:09:09	Araq	iirc (*fn)() is the same as fn() in C and it wents downhill from there (who said C is good at pointers?)
22:16:59	*	PortablePuffin quit (Ping timeout: 246 seconds)
22:18:37	Zuchto	Is it possible to dereference a variable that I've declared as a ref? My code is essentially: "var a: ref TFoo ; new(a) ; some_c_function(a) ; result = a" and I want the returning value to be an actual object, not a reference.
22:23:03	Araq	some_c_function(a[])
22:23:11	EXetoC	Araq: no, output is "void *outputBuffer", but it doesn't matter because what I was trying to do makes no sense
22:24:00	Araq	apperently fn() is the same as (*fn)() but when you need a pointer to that you have no choice but to use (**fn)()
22:24:39	*	Araq needs an Ansi C standard under his pillow
22:26:13	zielmicha	how does compiler looks for modules? Is there a way to add directory to search path?
22:26:29	EXetoC	I don't think we're talking about the same thing, but it doesn't matter anymore
22:26:48	EXetoC	zielmicha: see the help output and nimrod.cfg
22:26:57	Zuchto	Araq: if a is of type [ref TFoo] then what is the type of a[]?
22:26:59	EXetoC	/etc/nimrod.cfg in linux
22:27:10	Araq	TFoo?
22:27:16	EXetoC	you can also create project-specific configs
22:27:36	zielmicha	No way to do this at compile time?
22:27:42	zielmicha	from macros, for example
22:27:47	Zuchto	Araq: ok, and I need to pass a ref TFoo into the c-function but I want to return a TFoo from the procedure, is this possible?
22:28:29	Araq	only with some inefficiency
22:28:38	Zuchto	I see
22:28:57	Araq	or with hacks ;-)
22:29:30	Araq	sounds like you want to make the C function take a ptr TFoo though
22:29:30	Zuchto	i guess the easiest way would be to use create a new object and copy all the values from the ref TFoo?
22:29:44	Zuchto	Araq: how so?
22:30:14	Araq	Zuchto: then you can simply use 'addr' of your TFoo value
22:31:16	Zuchto	hmm... ok
22:31:34	fowl	myref[].addr
22:31:42	fowl	&(*myref)
22:31:54	Araq	fowl: yeah but that produces a 'ptr'
22:32:10	*	PortablePuffin joined #nimrod
22:32:16	Zuchto	the unsafeness of addr wouldn't be a problem since the reference wouldn't be kept outside of the function anyways, right?
22:32:23	Araq	right
22:33:11	*	ics quit (Quit: My MacBook Pro has gone to sleep. ZZZzzz…)
22:37:35	Araq	EXetoC: if proc `+=`*[T:float32|float](x: var T, y: type(x)) provides an implicit conversion of the 2nd argument but (var T, T) does not that's a bug
22:40:03	*	mflamer quit (Ping timeout: 246 seconds)
22:41:29	EXetoC	I suck at this
22:42:39	Araq	zielmicha: what's the status of https://github.com/Araq/Nimrod/issues/701 ?
22:43:45	zielmicha	with new quoteIfWhite it's no longer security issue, but the flag is still rather unfortunate
22:44:06	EXetoC	I think I tried to shorten the code while fixing the bug, but there's no way to get the un-qualified type, so I guess it can't be done
22:44:32	Araq	well I copied it over from FPC, zielmicha
22:44:44	zielmicha	I plan to someday add poUsePath flag and maybe make code clearer
22:45:00	Araq	that would be nice, osproc is my nightmare
22:45:18	zielmicha	Ah, I mentally copy everything from Python :)
22:45:29	Araq	whenever I change something there I break things for some obscure OS
22:47:11	Araq	took me a good deal amount of time to figure out how to use posix_spawn
22:48:28	Araq	was happy when I got it to work. I learned later that posix_spawn is not supported on Posix compatible OSes
22:49:25	OrionPKM	blarg im getting those errors with stdin as well
22:52:03	EXetoC	Araq: So, only the second parameter should be generic, right? and then I need to add additional overloads
22:52:41	*	zielmicha quit (Ping timeout: 250 seconds)
22:53:37	Araq	EXetoC: well I think (var T, T) should work and provide the implicit conversion for T
22:53:57	Araq	especially when 'type(x)' already seems to provide it
22:54:05	Araq	makes no sense they behave differently
22:55:06	*	dirkk0 quit (Quit: This computer has gone to sleep)
22:58:49	*	mflamer joined #nimrod
22:59:11	EXetoC	Araq: what I said makes no sense, and I guess you're referring to a compiler bug
22:59:49	Araq	lol yes
23:03:51	*	mflamer quit (Ping timeout: 246 seconds)
23:17:31	*	PortablePuffin quit (Ping timeout: 260 seconds)
23:20:32	*	BitPuffin quit (Quit: WeeChat 0.4.2)
23:25:26	EXetoC	Araq: should other procs me merged when this works? might as well just separate these functions otherwise, and not have to deal with this 'type' bug in this case
23:27:40	Araq	dunno. there is an important edge case which I forgot.
23:28:13	Araq	all I remember is that it's really important to consider "this"
23:28:46	Araq	where "this" is what I forgot :P
23:30:54	EXetoC	ok, well I'll give it a go once this is fixed
23:31:20	Araq	check if it already is please
23:31:37	Araq	I remember a change in the compiler affecting this
23:36:18	EXetoC	if it is a compiler bug that you were referring to, doesn't this imply that you just tested it?
23:36:54	EXetoC	do integers need those two sets of compiler magics, compared to the ones for all float types, that end with F64?
23:37:44	*	PortablePuffin joined #nimrod
23:38:47	Araq	gah, don't ask me now
23:38:56	Araq	I'm working
23:40:51	*	mflamer joined #nimrod
23:47:57	*	mflamer quit (Ping timeout: 246 seconds)
23:49:07	EXetoC	ok, cya later
23:53:48	*	mflamer joined #nimrod
23:56:19	*	PortablePuffin quit (Ping timeout: 260 seconds)