<< 23-01-2014 >>

---

00:00:03	renesac	can't I index tuples with field numbers?
00:01:18	renesac	I'm trying to refactor the ParticleBench code, so unnoficial compiler tweaks are out
00:01:19	renesac	:P
00:01:30	renesac	but I will take a look
00:01:51	Araq	seems like you only need to weaken the check for tyTuple and then it should compile
00:02:07	Araq	and yes you can do tup[i] but i has to be a constant and so that doesn't help
00:02:56	renesac	I tried with a enum
00:03:05	renesac	but not constant enought it seems...
00:03:06	renesac	:P
00:03:33	Araq	it needs to be an intege
00:03:36	Araq	*integer
00:04:00	Araq	what should tup[enumValue] mean? tup[enumValue.ord] ? then just say so
00:04:01	renesac	it is strange that it prints the field name if you call it with parenthesis
00:04:11	renesac	hum
00:05:03	renesac	I get a SIGSEGV: Illegal storage access. (Attempt to read from nil?)
00:05:09	renesac	with enumvalue.ord
00:05:17	renesac	it works with a simple integer
00:05:36	Araq	yay
00:09:06	renesac	it also gives SIGSEGV when I try an array of ints...
00:09:15	renesac	const dim: array[3, int] = [0,1,2]
00:09:15	renesac	for c in dim:
00:09:15	renesac	echo min[c]
00:09:45	renesac	it works with a single const int
00:09:45	Araq	on devel?
00:10:03	renesac	I tried to compile from git
00:10:18	renesac	though it gives the same version number and date of the nimbuild one
00:10:40	renesac	or not
00:10:42	renesac	Nimrod Compiler Version 0.9.3 (2014-01-21)
00:15:02	*	Smaehtin quit (Quit: http://www.kiwiirc.com/ - A hand crafted IRC client)
00:20:33	*	brson joined #nimrod
00:27:54	renesac	I was trying to use enum as a nicer array, but it seems that the working solution is to use a simple array + enum
00:28:31	Araq	yeah
00:28:34	Araq	good night
00:28:45	renesac	good night
00:28:56	dom96	yeah, I'm away to sleep too
00:28:58	dom96	bye
00:30:09	renesac	*I was trying to use tuple
00:33:55	*	nueva quit (Quit: Bye)
00:47:25	*	LordAndrew joined #nimrod
01:07:34	*	DAddYE quit (Remote host closed the connection)
01:10:20	*	Varriount is back
01:41:53	*	vbtt_ joined #nimrod
01:42:05	vbtt_	hi
01:44:01	Varriount	Hi vbtt_
01:45:04	vbtt_	what's the primary dev platform for nimrod?
01:46:36	vbtt_	also who runs the nimrod builders?
01:47:09	vbtt_	dom96: u here?
01:49:32	LordAndrew	Varriount: Do you think it'd be worth making a pull request for this function I wrote?: https://gist.github.com/Lordovos/8552530
01:55:16	vbtt_	LordAndrew: why not a generic one that can use float or int?
01:59:48	renesac	I was going to write something similar
02:00:19	LordAndrew	Ah right, floats. Forgot about that.
02:00:26	renesac	but accepting as argument a nimrod type or subrange
02:00:27	*	capisce_ quit (Ping timeout: 260 seconds)
02:00:43	LordAndrew	I'm also wondering if I would want to just have the procedure be named random() instead of randomRange.
02:01:23	renesac	and it would return all types of values, like uint32, floats, strings, etc
02:01:29	renesac	well, that was the plan
02:01:43	renesac	and subranges, of course
02:02:21	LordAndrew	...oh. random(float) doesn't work on Windows??
02:02:35	renesac	nor in linux I think
02:02:49	renesac	the std random is very limited
02:03:03	renesac	that is why I wanted to make this super generic templated version
02:03:21	vbtt_	random(float) is documented. doesn't work?
02:03:30	LordAndrew	"This is currently not supported for windows."
02:03:31	renesac	is it?
02:03:34	renesac	hum
02:03:40	LordAndrew	Compiler throws error as well, says expected an int.
02:04:03	vbtt_	http://nimrod-lang.org/math.html#135
02:04:46	vbtt_	yeah you could just overload random().
02:05:09	vbtt_	random(max) or random(min, max). nimrod can do this type of overloading, right?
02:05:21	renesac	right
02:05:42	renesac	well, the other thing wrong with random is using global state
02:06:26	renesac	I haven't tried, but I suspect a call to randomize in a function will interfere in other unrelated functions called latter
02:07:06	vbtt_	unrelated, but what do you guys think are the top 3 distinctive features of nimrod?
02:08:20	vbtt_	just wondering since the language has grown a lot of features.
02:09:13	*	DAddYE joined #nimrod
02:09:14	renesac	I was thinking in something like random[range[0 .. 16].uint](seed=x).get()
02:09:53	renesac	though that will only work for ranges known at compile time...
02:10:12	vbtt_	renesac: the global state should always be available because some functions may not care.
02:10:25	renesac	?
02:10:25	vbtt_	also why is range better than (min, max) ?
02:10:54	vbtt_	renesac: I mean random(max) should always be allowed (using the global random generator)
02:11:01	renesac	not that it is better, but I though it would make min,max unnecessary
02:11:47	vbtt_	renesac: max and min, max are convenient.
02:11:54	renesac	vbtt_: I'm talking about the seed used by the random generator, that may be changed by unrelated functions runing at the same time
02:12:06	renesac	vbtt_: indeed, now I see it
02:12:41	renesac	but even then, I think random should also be capable of returning ranges, even if it has optional max and min arguments
02:13:33	vbtt_	renesac: why return a range? randomSample(sampleSize, min, max) is more useful imo.
02:13:53	*	DAddYE quit (Ping timeout: 272 seconds)
02:14:26	renesac	what is sampleSize?
02:14:52	vbtt_	actually randomSample(sampleSize, array or seq)
02:15:00	renesac	and my question would be, why not return a range, if you want to return int, uint64, float, etc?
02:15:03	vbtt_	select random sampleSize elements from list.
02:15:30	renesac	oh, that would be a different function
02:15:31	vbtt_	renesac: let me rephrase, what does it mean to return a random 'range'?
02:16:09	renesac	return a value of type subrange that is inside that defined subrange
02:16:50	renesac	http://build.nimrod-lang.org/docs/manual.html#subrange-types
02:16:57	renesac	<-- I'm talking about those
02:17:17	vbtt_	ah i see.
02:17:27	vbtt_	yup sure - that should work too.
02:17:44	renesac	but the limitation is that they should be known at compile time
02:17:49	vbtt_	type Color = range[0..0xFFFFF];.... random(Color)
02:17:57	renesac	so the parameters max and min have their value
02:18:06	vbtt_	it's still useful.
02:18:37	renesac	I was thinking more in random[color]()
02:18:39	*	DAddYE joined #nimrod
02:18:41	*	DAddYE quit (Client Quit)
02:19:13	renesac	but I guess overloading/implicit generics works too
02:19:26	renesac	I'm not sure you can use "when" inside the function then
02:19:41	vbtt_	either will work but random(Color) is more readable imo.
02:19:46	renesac	rigth
02:19:47	vbtt_	renesac: why cant you use when?
02:19:58	renesac	I don't know
02:20:17	renesac	probably you can indeed, I'm still starting with nimrod
02:20:51	renesac	anyway, I wanted a 64bit random value, so I was going to write my own random funcion from scratch
02:21:25	renesac	as the stdlib only provides positive integers as random values
02:21:35	renesac	(and floats, I missed that)
02:23:13	vbtt_	renesac: what's sizeof(int) in nimrod?
02:23:27	renesac	I think it is plataform dependent
02:23:33	renesac	either 32 or 64bits
02:23:43	renesac	int32 is aways 4 bytes
02:23:54	renesac	int64 aways 8bytes
02:24:14	renesac	vbtt_: about the global state again, you might have some test that depends on certain random pattern, and another function is called, that calls randomize, and your totally unrelated function fails.
02:24:50	renesac	you should be able to seed your own random number generator
02:25:11	vbtt_	renesac: code shouldn't depend on random 'patterns'. it's a bug. random by definition shouldn't have patterns (but pseudo-random happens to cos it's machine generated)
02:25:21	vbtt_	renesac: agree on 2nd point.
02:25:25	renesac	to reproduce a bug
02:25:26	renesac	for example
02:25:44	renesac	well, global state is bad
02:25:50	vbtt_	renesac: the api could be. randomizer = initRandom(seed); var x = random(randomizer, min, max);
02:28:01	renesac	I think simply 'let random1 = Random(seed); var x = random.get(max, min=2)
02:28:11	renesac	*random1.get
02:28:33	renesac	the case insensitivity of nimrod makes worse to work with classes
02:28:47	renesac	well, maybe 'let random = TRandom(seed)
02:28:54	renesac	yep
02:29:26	EXetoC	yes that's the convention
02:30:10	renesac	if you just need one random value now, you could use 'TRandom(seed).get(max)'
02:30:47	renesac	though you have to initialize separately to use in a loop
02:32:16	renesac	but if TRandom itself supported arrays, sets, strings, etc as output, it wouldn't be as necessary
02:32:55	LordAndrew	Hmm.
02:33:22	LordAndrew	For a .gitignore file in a NImrod project, what would I want to exclude? Just /nimcache?
02:34:21	renesac	I'm going to post a rant in the forums soon about how I hate some C style stdlib APIs that don't use or let you use nimrods features and nicities that C don't have
02:34:52	renesac	I'm still preparing it, and feeling the language
02:36:08	renesac	random is among them, as is Open() and Alloc()
02:36:15	renesac	/spoiler
02:39:49	EXetoC	wrapping alloc shouldn't be difficult
02:39:58	renesac	yep, it is trivial
02:40:16	EXetoC	and what about open?
02:41:30	renesac	you need to declare a variable of type TFile to feed to the function
02:41:38	renesac	you loss nimrod type inference
02:42:15	EXetoC	that has to do with how var works, no? also, check the other overloads
02:42:21	renesac	you should be able to just 'var file = Open(filename)'
02:42:44	EXetoC	you can
02:43:03	renesac	ohh
02:43:16	renesac	I didn't saw the other definitions
02:43:41	renesac	ok, that is outside my rant now
02:44:59	renesac	about alloc, you see all over the place: 'cast[ptr T](alloc0(sizeof(T)))' where it would be shorter to write alloc0[ptr T](sizeof(T))
02:45:11	renesac	yep, that is trivial, but still a C wart in nimrod
02:45:21	EXetoC	easy to fix like I say, but I'll do it tomorrow or something if you don't want to
02:45:36	renesac	yeah
02:45:37	EXetoC	assuming that sizeof works and all that
02:46:12	renesac	I'm just removing the cast, not changing the alloc0 arguments
02:46:28	vbtt_	renesac: can a template fix that?
02:46:37	Varriount	Hey guys. What's going on?
02:46:41	renesac	a simple wrapper function would fix that
02:46:41	vbtt_	why not just allocPtr(T)
02:47:36	renesac	well, most of the time is the size of the elements pointed by the pointer times the number of elements
02:47:45	EXetoC	or just overload it. alloc(T, n: int) or something?
02:48:02	renesac	yep
02:48:17	renesac	but my point is that it should be like that in the stdlib the first place
02:48:49	vbtt_	renesac: agree.
02:49:10	vbtt_	the stdlib should be convenient and consistent.
02:49:34	renesac	I also like the jemalloc experimental API style: http://www.canonware.com/download/jemalloc/jemalloc-latest/doc/jemalloc.html
02:49:38	EXetoC	there has always been a lack of man power, but ask if you want
02:49:43	renesac	instead of binary flags, optional arguments, of course
02:50:06	renesac	but that would be a more radical change
02:50:39	renesac	and even jemalloc people aren't certain that their experimental API should be keept
02:51:27	renesac	of course, instead of returning int, throw an exception and return a pointer
02:51:42	xtagon	How do you trap signals like SIGTERM in nimrod?
02:52:02	Varriount	There's a handler setting procedure in system.nim
02:52:33	xtagon	Varriount, what's the proc name?
02:52:52	renesac	actually, that experimental api is totally compatible with current nimrod, just adding optional arguments
02:53:00	xtagon	Varriount, searching for "signal" or "trap" didn't bring anything up
02:53:16	EXetoC	hook?
02:53:39	EXetoC	so, only for ctrl-c I guess
02:53:48	Varriount	Set control hook should get sigterm
02:54:20	Varriount	setControlCHook
02:54:29	Varriount	*Looks at implementation*
02:54:46	xtagon	Thanks. Is there no way to hook other signals? I don't need to, but now I'm curious :)
02:55:01	Varriount	Why do you think I'm looking at the source?
02:55:36	Varriount	Anyway, the two things I think the stdlib could use the most improvement in is unicode string handling, and asyncio
02:55:40	EXetoC	because you've watched all the cat videos
02:56:12	EXetoC	what about unicode?
02:57:12	Varriount	EXetoC: Most of the string related procedures in nimrod are untested with unicode strings
02:57:59	Varriount	Also, the manual/tutorial needs a big warning saying that the subscript operator for strings works on the width of a char type
02:57:59	renesac	well, if they affect only ASCII, they are utf-8 agnostic
02:58:21	EXetoC	got an example? many procs don't care what the string is
02:58:23	renesac	Varriount, there is a warning
02:58:58	Varriount	EXetoC: That's why I said "untested"
02:59:54	Varriount	xtagon: See this -> https://gist.github.com/Varriount/8572100
03:00:20	Varriount	That's what excpt.nim (included in system.nim) does
03:01:38	Varriount	There's also some private signal handlers procs in excpt.nim, so I would make sure you're not overriding anything if you want to play around with such things
03:03:19	xtagon	Varriount, awesome, thanks
03:03:57	xtagon	Varriount, do you know if nimrod traps INT or KILL on it's own for any reason, such as for the GC?
03:04:15	Varriount	I'm looking
03:04:26	Varriount	Yeah, it captures both.
03:05:16	Varriount	Doesn't do much with them. Just prints out stack traces (if available) and handles exiting the integrated debugger if you're debuggin
03:05:19	Varriount	*debugging
03:06:07	Varriount	xtagon: Go to ./lib/system/excpt.nim, down near the bottom
03:08:02	Varriount	I imagine that Araq doesn't have handler registering procedures for the other signals, because after getting most of those, a program has to tread very carefully to avoid termination by the OS
03:09:55	xtagon	Yes, but I can see why someone would want to catch KILL in order to free up resources or something
03:10:15	Varriount	*shrug*
03:12:00	Varriount	The code doesn't look too complicated. If you want, you could probably use sigaction to retrieve nimrod's handlers, and wrap them in your own handler. Or you could modify the handling code to allow user-set hooks.
03:12:24	Varriount	Looks like a good mini-project to me.
03:12:41	Varriount	Although, I'm focusing on strutils and unicode.nim right now.
03:13:33	xtagon	Have fun with that :)
03:14:03	Varriount	I'm just comparing what python has built-in to strings that nimrod lacks.
03:14:03	xtagon	I'm fighting a virus and not in much of a coding mood
03:14:28	Varriount	Aw. Sorry about that.
03:15:16	xtagon	Oh, one other thing I couldn't find in the docs: How do you return an exit status code?
03:15:32	xtagon	I get an error if I use return in the main code
03:19:16	Varriount	quit()
03:20:11	Varriount	xtagon: http://nimrod-lang.org/system.html#407
03:20:36	xtagon	Heh, I was searching for "exit" when I should have searched "quit" :)
03:21:09	Varriount	I really do wish that chrome had a regex or peg ctrl+find feature
03:22:00	*	vbtt_ quit (Ping timeout: 272 seconds)
03:23:40	xtagon	There's one in the Chrome web store but it has 2/5 stars...
03:24:31	xtagon	Varriount, ah, here is one that looks promising https://chrome.google.com/webstore/detail/regex-search/bcdabfmndggphffkchfdcekcokmbnkjl?hl=en
03:25:22	Varriount	Ooh, thanks!
03:25:39	*	vbtt_ joined #nimrod
03:25:48	Varriount	Regex may suck for any complicated parsing, but it does get the job done for little things.
03:26:32	xtagon	The main benefit of regex is that it's everywhere
03:27:27	Varriount	xtagon: Did you know that sublime text (and many other IDE's) user regex for highlighting and parsing languages?
03:27:32	Varriount	*use
03:28:22	xtagon	Varriount, that sounds like it would be hard for recursive grammars
03:28:22	Varriount	I help develop one of the two sublime text nimrod plugins. It's very.. uncomfortable.
03:29:24	Varriount	xtagon: It's a mixture of json and regex. You define "match rules" in regex, and then specify other things about the rules in json.
03:29:39	xtagon	The other thing about regex is that it CAN be fast, but usually isn't. Have you seen this? http://swtch.com/~rsc/regexp/regexp1.html
03:31:13	Varriount	xtagon: Do you happen to know which algorithm PCRE uses?
03:31:53	xtagon	I'm not sure
03:32:39	vbtt_	xtagon: in practice you dont really hit the pathological case.
03:33:40	xtagon	Not for small scripts, but I can see why programs like grep would get a performance boost
03:33:49	Varriount	xtagon: Does this article link to a usefull implementation?
03:34:28	LordAndrew	I really need to learn how to use regular expressions.
03:34:58	Varriount	LordAndrew: The first thing about learning to use regular expressions is when to not use regular expressions.
03:35:37	renesac	grep compiles not very complex queries to be VERY fast
03:35:45	LordAndrew	Would using a regular expression to validate, say, a dice roll notation be an appropriate use?
03:36:13	Varriount	Yes. Using it, however, to parse HTML, is not.
03:36:35	xtagon	Varriount, I don't remember the article mentioning any specific implementation
03:37:07	xtagon	Varriount, check the "History/Refernces" section, there might be something there
03:37:21	renesac	yep, up to two regular expressions, grep uses the Boyer-Moore algorithm
03:37:55	xtagon	Varriount, PCRE has two different algorithms, one is Perl-compatible and one is not (for performance) but I don't know any details on the algorithms
03:38:21	vbtt_	how do i look at the generated c code again?
03:38:38	Varriount	vbtt: Look at the nimcache directory
03:38:49	renesac	I'm starting to read a book about string matching, but haven't got very far
03:38:58	Varriount	renesac: What is it?
03:39:10	LordAndrew	What exactly is the C code in the nimcache for? Does Nimrod itself just transcompile to C?
03:39:17	Varriount	LordAndrew: Yes.
03:39:33	vbtt_	vbtt_: can't find it. i just did 'nimrod c hallo.c' and it compiled but no nimcache direction
03:39:39	vbtt_	s/direction/directory
03:39:43	vbtt_	Varriount: ^
03:39:46	renesac	"algorithms on strings, trees, and sequences" by Dan gusfield
03:39:52	Varriount	vbtt: Are you using linux or nimrod?
03:39:55	Varriount	*windows
03:40:06	xtagon	Nimrod is an OS now? nice :)
03:40:08	renesac	Varriount.
03:40:26	Varriount	renesac: Yes?
03:40:32	renesac	the name of the book
03:40:36	Varriount	I see it.
03:40:37	vbtt_	Varriount: linux x64. just bootstrapped nimrod as per the github readme.
03:41:20	vbtt_	there is a nimcache in Nimrod (doesn't contain hallo) and there's no nimcache in Nimrod/examples
03:41:39	Varriount	vbtt: Did hallo compile successfully?
03:41:46	vbtt_	Varriount: yes
03:41:57	Varriount	Where did you compile it?
03:42:40	vbtt_	cd Nimrod/examples ~/src/Nimrod/bin/nimrod c hallo.c
03:43:26	Varriount	You tried using nimrod to compile a c file?
03:43:34	vbtt_	sorry. typo.
03:43:38	vbtt_	nimrod c hallo.nim
03:43:48	vbtt_	then i have the hallo executable and it works.
03:44:24	Varriount	Well, the C code should be somewhere in the examples/nimcache directory.
03:44:32	vbtt_	there is no examples/nimcache :(
03:44:44	vbtt_	"no such file or directory"
03:44:50	xtagon	vbtt, sudo updatedb && locate nimcache
03:45:06	Varriount	vbtt: You're starting to scare me.
03:46:01	vbtt_	hmm something weird going on. locate finds it but I can't cd to it.
03:46:17	vbtt_	ah wait
03:46:20	xtagon	Varriount, I looked into a bit more just now. Thompson's algorithm was patented (which explains why it's hard to find an implementation)
03:46:21	vbtt_	examples/examples/nimcache :)
03:46:32	*	Lordovos joined #nimrod
03:46:48	Varriount	xtagon: I will never understand why people are allowed to patent an *algorithm*
03:47:02	xtagon	Varriount, or a UI pattern. Good grief.
03:47:06	Varriount	That's like patenting a recipe
03:47:18	vbtt_	yo dawg - there's an examples directory inside the examples directory
03:47:38	Varriount	:3
03:47:55	Varriount	xtagon: Any idea when the patent expires?
03:48:42	Varriount	Or who has to die to get the patent into public domain?
03:48:53	*	LordAndrew quit (Ping timeout: 272 seconds)
03:49:10	xtagon	Varriount, http://www.freepatentsonline.com/3568156.pdf
03:49:25	Lordovos	damnit
03:49:27	*	Lordovos is now known as LordAndrew
03:50:06	Varriount	Good grief, it was patented about 47 years ago.
03:50:36	Varriount	Or rather, filed 47 years ago, patented 44 years ago.
03:52:13	Varriount	Is it still valid?
03:52:55	xtagon	Varriount, IIRC design patents only last 14 years
03:53:33	renesac	"only
03:53:34	renesac	"
03:53:36	xtagon	Varriount, I just found this library which was made by the article author https://code.google.com/p/re2/
03:53:36	Varriount	xtagon: Maybe we could use this algorithm to implement a native nimrod regex engine.
03:54:14	Varriount	Ooh, and it has a BSD license.
03:54:56	xtagon	I don't see anywhere saying it's the same algorithm, just that it's written by the same author
03:55:55	xtagon	Varriount, ah, here we go: http://swtch.com/~rsc/regexp/regexp3.html
03:58:21	xtagon	It's still only shaving a few microseconds of PCRE, probably not worth implementing in nimrod
04:00:53	Varriount	Yeah. Best to follow the principle of NIH
04:02:16	Varriount	It's a shame that google discontinued code search, it was a helpful service.
04:02:46	xtagon	Yeah it was
04:03:05	xtagon	I wouldn't reimplement re2, but a wrapper might be cool
04:03:21	*	Varriount holds out c2nim
04:04:33	xtagon	Does c2nim work on C++?
04:04:46	Varriount	I think it has limited support.
04:05:05	Varriount	Only one way to find out!
04:09:36	renesac	bummer, if I have a function 'proc func(x: array[2, int]) = foo' I can't call it like 'func([1,2])'
04:11:23	Varriount	renesac: Really? what error do you get?
04:11:44	renesac	ok, maybe it is because it isn't doing implicit type conversion
04:11:53	renesac	I'm trying to fix here
04:13:17	xtagon	renesac, it works for me if I specify the proc's return type
04:14:17	renesac	ok, it was my fault
04:14:53	renesac	wasn't supplying the right number type inside the array
04:15:17	LordAndrew	Bleh, I am way too paranoid about software licenses.
04:15:50	renesac	the first thing I saw in the error message was 'got (Array constructor[' instead of 'array['
04:16:20	renesac	so I wrongly assumed a nimrod limitation :/
04:18:20	Varriount	renesac: How do you mean? What was the correct code?
04:18:51	renesac	the function was asking a array of floats, and I gave it an array of ints
04:19:06	renesac	just putting .0 on the numbers solved the issue
04:19:17	vbtt_	is there a quick way to call new(TMyObject) and initialize it with values at the same time?
04:19:35	*	ddl_smurf joined #nimrod
04:19:41	Varriount	vbtt: define 'quick'
04:20:11	Varriount	I'm assuming that TMyObject is a ref
04:20:31	vbtt_	Varriount: easy. instaead of var x = new(TMyObject); x.foo = 1; x.bar=2;..
04:20:40	renesac	var obj = TMyObject(x: bla) ?
04:20:50	vbtt_	if I don't use new, I can do var x = TMyObject(foo: 1, bar: 2)
04:21:05	Varriount	vbtt: Then don't use new.
04:21:48	vbtt_	Varriount: but then I make a TMyObject instead of a ref TMyObject, no?
04:21:56	vbtt_	is it still created on the heap?
04:22:22	renesac	type RTmyobject = ref Tmyobject
04:22:36	renesac	var obj = RTMyObject(x: bla)
04:22:50	renesac	I haven't tried, but I guess it works
04:23:22	renesac	(I think the standard is to use a P instead of a R)
04:23:39	renesac	(not sure)
04:23:51	Varriount	Actually, the standard is to not use type prefixes, although personally, I like them.
04:24:20	renesac	well, w/o case sensitivity, type prefixes help separate types from values
04:24:47	Varriount	Which is what I said to Araq, and he just grumbled that too many people complained about the prefixes.
04:25:06	Varriount	It also helps to seperate types and values from procedures.
04:25:08	renesac	the first time I saw the language I thought they were very ugly, but I'm growing to like them
04:25:16	vbtt_	yeah I dont like type prefixes. did you say the nimrod standar is to *not* use them?
04:25:21	renesac	given the style insensitivity
04:25:29	Varriount	vbtt: Yes.. sorta
04:26:53	xtagon	But all the standard libraries use them
04:27:30	Varriount	xtagon: Araq means to change the names in the future
04:27:52	vbtt_	interesting.
04:27:58	xtagon	Good to know
04:28:03	Varriount	It's too bad we don't have a larger community to draw upon, to get a general consensus.
04:28:30	renesac	the forum don't have pooling functionality right?
04:28:36	xtagon	If the prefix isn't going to be standard, then I say there should be at least first-character case sensitivity
04:28:50	xtagon	so that "version" can be a variable and "Version" can be a type
04:29:22	renesac	my vote would be: either keep the style insensitivity AND type prefixes or have full/partial style sensitiviy by default and remove the prefixes
04:29:38	xtagon	renesac, I agree
04:30:51	LordAndrew	Style insensitivity?
04:31:07	xtagon	s/style/case
04:31:13	renesac	DS_f_D == dffd
04:31:17	renesac	*dsfd
04:31:38	renesac	englobing case and _ insensitivity
04:31:48	renesac	but the main issue here is indeed the case of the first letter
04:32:02	LordAndrew	err
04:32:13	renesac	that is useful for OO style programing
04:32:15	Varriount	I bequeth to you all these three links: https://github.com/Araq/Nimrod/blob/devel/doc/apis.txt , http://forum.nimrod-lang.org/t/191/3 ,
04:32:26	Varriount	Third one coming up as soon as I can find it
04:33:17	LordAndrew	var my_var == myVar?
04:33:27	Varriount	LordAndrew: Yes
04:33:38	Varriount	Currently, though possibly not in the future
04:33:41	LordAndrew	why would that be a thing
04:34:22	Varriount	LordAndrew: So that everyone can have their own style for their own projects
04:34:33	LordAndrew	Does this extend to say:
04:34:36	LordAndrew	var t = "whatever"
04:34:37	LordAndrew	echo(T)
04:35:16	Varriount	Yes.
04:35:39	Varriount	Again, there are currently options for nimrod's compiler regarding case-sensitivity
04:35:40	renesac	but consistency is encouraged
04:35:46	Varriount	renesac: Yes
04:36:02	xtagon	Varriount, I just found another implementation: https://github.com/laurikari/tre/
04:36:11	xtagon	Varriount, this one is in C
04:36:31	Varriount	Ooh, much easier to wrap than C++
04:37:26	Varriount	DId I tell you guys about a mailing post on the cairo (2d graphics library) mailing list involving someone working for one of the standards groups?
04:37:57	xtagon	Varriount, do tell
04:38:03	LordAndrew	How does Nimrod's compiler create executables based on the code we compile?
04:38:24	Varriount	He said that cairo would be a perfect model for the standard.. except that the standard should have a C++ object oriented interface
04:39:31	Varriount	LordAndrew: The "Nimrod Compiler" is actually a transpiler. It translates nimrod code into C (or C++, or objective C, or javascript) and then tells the local compiler to compile those generated files into a binary.
04:40:27	xtagon	Whoa, I did not know C++ and Objective C were supported too
04:40:49	xtagon	Are you saying we could compile native apps for iOS/Android?
04:40:52	Varriount	Well, those backends are not quite as stable as the C backend, or so I've been lead to believe.
04:41:13	vbtt_	there's also js :)
04:41:25	LordAndrew	I see. But since the default local compiler is GCC, wouldn't the binaries depend on libgcc and glibc?
04:41:54	Varriount	LordAndrew: Yes, if the compiler is configured that way.
04:42:08	Varriount	You can pass arguments to the backend compiler if need be.
04:42:30	Varriount	xtagon: Yes, theoretically.
04:44:01	xtagon	That's awesome
04:44:43	Varriount	I think there's an example that's supposed to work on android. I don't know if it's still functional though.
04:46:21	xtagon	I'm guessing it has to be wrapped in something Java-like anyway, though
04:46:27	Varriount	xtagon: Look at the cross_calculator example.
04:47:50	xtagon	Varriount, cool
04:48:03	LordAndrew	Varriount: I see. I ask because I try to avoid accidentally using GPL licensed stuff that would in turn make my code and/or executables GPL. I'm probably worrying far too much about it though.
04:48:25	Varriount	LordAndrew: You're worrying far too much about such things.
04:50:43	xtagon	LordAndrew, libc is LGPL which lets you link it in proprietary code (in contrast to GPL)
04:52:09	Varriount	Heh, your worrying reminds me of the time I called atlassian ( the company behind bitbucket) in order to get my code removed from a repository.
04:52:52	LordAndrew	I'm totally fine with open sourcing my stuff but I prefer the MIT/BSD licenses (and not having the GPL forced upon me over hidden goblins I didn't see coming).
04:52:52	xtagon	Varriount, from your repository or someone elses?
04:53:07	Varriount	I had written a modification tracker for a minecraft server written in python, and I had placed a very specific license in it that specified that I was to be given credit
04:53:17	Varriount	(minecraft classic)
04:53:52	Varriount	The head of the dev team for that particular piece of server software took my code and incorperated it into his server, and called it his own.
04:54:05	vbtt_	can nimrod pragmas be used to tag the license onto modules and then programmers can compile with 'no-gpl' if they want?
04:54:06	Varriount	Since I was also on the dev team, I saw this.
04:54:15	Varriount	vbtt: Theoretically.
04:54:31	vbtt_	that would be nice
04:55:19	Varriount	When the lead dev wouldn't confess to what he had done, I found atlassian's phone number, and called them up one day. The person who answered the phone must have been surprised to hear a 15 year asking that a repository owner be given a warning
04:55:38	Varriount	for violating a software license.
04:56:17	Varriount	Although he took me seriously, he explained that I had to have proof, and that without proof, they couldn't / wouldn't do much
04:56:37	xtagon	What constitute's proof? Git logs?
04:57:09	Varriount	And/or an email sent with an attachment to yourself. That way you have time logs stored by a third party
04:57:16	Varriount	*That, and/or
04:57:31	Varriount	And we were using svn at the time.
04:57:58	xtagon	I wonder if that would be a useful software-as-a-service.
04:58:09	Varriount	In the end, I actually got my code removed without resorting to such drastic action.
04:58:38	Varriount	By getting the rest of the team to gang up on the leader (He was younger than I, and a lot more pig-headed)
05:00:51	Varriount	It's kinda laughable when I think about it. The modification tracker I wrote wasn't very well written, not always tracking modifications in the correct order, causing the server to freeze when loading from the sqlite database...
05:01:31	Varriount	But I was quite proud of it at the time. It was one of my first pieces of somewhat complex code.
05:01:37	LordAndrew	Hmm. Dependency Walker only show that my executables link Kernel32.dll and MSVRCT.dll.
05:02:21	Varriount	LordAndrew: Well yeah. Those are both essential libraries.
05:09:49	Varriount	xtagon: Are you able to compile that tre regex lib you linked?
05:10:08	xtagon	Varriount, I didn't try
05:10:17	xtagon	Varriount, it's on my todo list (which is long...)
05:10:45	Varriount	Odd. the configure command isn't showing in my mingw shell
05:11:54	*	filwit quit (Quit: Leaving)
05:12:12	xtagon	Varriount, you have to use autoconf on configure.ac to produce a configure script (although I'm not familiar with how)
05:12:40	Varriount	There's a configure.ac file, I don't know how to process it though.
05:12:45	xtagon	Varriount, I just say https://github.com/laurikari/tre/blob/master/configure.ac#L1
05:12:48	xtagon	saw*
05:13:12	Varriount	I'm not terribly experience on the intricacies of linux build systems.
05:13:27	xtagon	Varriount, try just running "autoconf" in that directory with no arguments
05:13:57	Varriount	Apparently I don't have autoconf
05:14:04	xtagon	Varriount, what distro?
05:14:09	Varriount	Windows.
05:14:24	xtagon	Oh right, you mentioned Mingw
05:14:30	Varriount	I have msys installed. I could have sworn that I installed autotools as well.
05:15:55	*	brson quit (Ping timeout: 272 seconds)
05:21:24	Varriount	xtagon: I've tried linux, I just can't kick the habit of using windows. I'm used to it.
05:22:59	xtagon	Varriount, nothing wrong with that. I use both regularly.
05:23:44	Varriount	Although, I do seem to find a constant between the two : No matter what OS I use, building things with autotools and the like will always be a pain for me.
05:24:01	xtagon	I hear you on that one
05:24:29	Varriount	Which is why I like nimrod. No more do I have to endure the masochistic build systems!
05:25:32	Varriount	xtagon: Have you ever managed to get GTK compiled as a 64 bit library/dll on windows?
05:25:41	xtagon	I use Windows for two reasons: 1. my job involves Windows-specific software, 2. there is better audio production software on Windows
05:25:51	xtagon	Varriount, no, that sounds like "fun"
05:26:30	Varriount	xtagon: Did you know, not only are the instructions on the GTK website nicely non-specific, GTK has circular dependancies.
05:26:50	xtagon	Varriount, my most recent "adventure" was getting LLVM compiled on 64-bit windows. It was fairly easy (in comparison to other things) and still drove me mad
05:27:10	Varriount	xtagon: I tried that. Never managed to be successful though.
05:27:29	Varriount	I can get 32 bit things compiled, 64 bit is another matter.
05:27:39	xtagon	Varriount, I was trying to get Crystal (another in-development lang) to compile on Windows. I didn't get far.
05:28:00	Varriount	Hopefully nimrod is far superior.
05:28:05	xtagon	It is
05:30:17	renesac	question: can I define a new implicit conversion?
05:30:29	Varriount	renesac: Yes. See: Converters
05:30:49	renesac	oh, cool
05:31:09	Varriount	They're a bit limited, in the fact that they operate on a global level.
05:31:34	Varriount	It would be nice to be able to apply them to only certain procedure scopes, but w/e
05:32:09	renesac	right, but they will only change the output where the compiler would give an type mismatch error anyway, right?
05:32:17	Varriount	Yeah
05:32:29	renesac	then it is ok for my usage now
05:35:41	LordAndrew	Huh, so it does look like the executables staticly link libgcc_s_dw2-1.dll internally.
05:37:21	*	brson joined #nimrod
05:39:22	Varriount	Rats. The tre regex library has only been built under windows using visual studio
05:41:23	xtagon	Bummer.
05:41:50	Varriount	It's not too bad. I have the SDK downloaded. I just need to remember how to invoke MSBuild correctly.
05:43:07	Varriount	Few people realize that you can download the compiler behind visual studio, without Visual Studio itself (and all the crud that comes with it)
05:43:34	LordAndrew	You can?
05:43:39	Varriount	Yep
05:43:54	Varriount	You just install the windows sdk downloads.
05:46:13	*	delian66 quit (Ping timeout: 248 seconds)
05:48:13	*	brson quit (Ping timeout: 272 seconds)
05:54:44	*	brson joined #nimrod
05:56:02	Varriount	xtagon: If you manage to build that library as a 64 bit, please tell me. MSBuild complains that I don't have the right Toolset, and after that, complains that I'm trying to build a 32 bit library on a 64 bit platform.
05:56:10	Varriount	I have to get to bed.
05:56:58	xtagon	Varriount, I'll let you know. I was only planning on building on Linux though
06:02:27	*	LordAndrew quit (Quit: Out.)
06:05:13	*	xtagon quit (Ping timeout: 252 seconds)
06:38:09	vbtt_	having both variants and inheritance is a bit confusing imo.
06:40:32	vbtt_	really variants can be modelled by abstract base types so not sure what the two ways of expression get us.
06:40:39	vbtt_	this should be in the faq.
07:05:17	*	aruniiird joined #nimrod
07:32:20	*	bbodi joined #nimrod
07:34:54	vbtt_	hm, how do i search github issues?
07:36:23	vbtt_	so I can't use typeclasses to parameterize other types? e.g. seq[MyTypeClass]
07:36:30	vbtt_	zahary^
07:47:56	bbodi	I think it siuld work. what is the full code and the error?
07:55:57	vbtt_	bbodi: Error: internal error: GetUniqueType
08:00:44	bbodi	https://github.com/Araq/Nimrod/search?q=GetUniqueType&ref=cmdform&type=Issues there are some issues about it
08:06:18	vbtt_	bbodi: thanks.
08:06:25	vbtt_	none of them are related to typeclasses though.
08:13:25	*	bbodi quit (Ping timeout: 248 seconds)
08:14:41	*	bbodi joined #nimrod
08:31:15	*	aruniiird quit (Ping timeout: 272 seconds)
08:32:33	*	odc joined #nimrod
08:34:02	*	odc left #nimrod (#nimrod)
08:57:08	*	brson quit (Quit: leaving)
08:57:55	*	betawaffle quit (Ping timeout: 245 seconds)
09:05:10	*	olahol_ left #nimrod (#nimrod)
09:16:37	*	darkf_ joined #nimrod
09:19:33	*	darkf quit (Disconnected by services)
09:19:34	*	darkf_ is now known as darkf
09:22:54	*	aruniiird joined #nimrod
09:37:01	*	CarpNet joined #nimrod
09:54:56	vbtt_	a bunch of basic type class functionaly seems broken. e.g. 'is' doesn't really test anything.
10:11:00	*	nueva joined #nimrod
10:18:02	bbodi	I got compilation errors related to C sources. But yesterday everything worked fine, and didn't occur any modifications since it. I think the problem is that nimrod doesn't use the gcc that was packaged with it.
10:26:10	*	bbodi quit ()
10:32:18	*	bbodi joined #nimrod
10:42:19	vbtt_	hello
10:43:06	bbodi	hi
10:43:21	vbtt_	http://vocalbit.com/posts/exploring-type-classes-in-nimrod.html
10:46:29	bbodi	it seems cool!
10:56:18	*	vbtt_ quit (Quit: Page closed)
11:27:08	NimBot	Araq/Nimrod devel cfa31d1 Araq [+16 ±12 -15]: next steps for closure iterators
11:27:08	NimBot	Araq/Nimrod devel 25f1426 Araq [+0 ±2 -0]: closure iterators almost work
11:27:08	NimBot	Araq/Nimrod devel 513f5ed Araq [+0 ±4 -0]: closure iterators work
11:27:08	NimBot	Araq/Nimrod devel fc8fa0d Araq [+0 ±3 -0]: small code cleanups
11:27:08	NimBot	1 more commits.
11:27:43	Araq	Varriount: closure iterators are now complete afaict
11:28:06	Araq	and with that I'm entering bugfixing mode ;-)
11:28:14	bbodi	Hi Araq!
11:28:21	Araq	hi bbodi
11:29:32	Araq	bbl
11:37:28	bbodi	I can't compile my sources since yesterday
11:37:51	bbodi	I get error from the GCC
11:37:53	bbodi	like
11:37:54	bbodi	c:\projects2\whitestageditor\nimcache\nimrod_system.c:11597:25: error: dereferencing pointer to incomplete
11:37:54	bbodi	type
11:37:54	bbodi	for (LOC1 = 0; LOC1 < a->Sup.len; LOC1++) {
12:01:15	bbodi	I think I found some huge bug
12:01:50	Araq	bbodi: will look at it tonirght
12:01:53	Araq	*tonight
12:02:34	bbodi	ok, I try to reproduce it in a simple way, then I will create an isse for that
12:05:31	*	easy_muffin joined #nimrod
12:09:16	nueva	looks like type classes syntax could be used also as some kind of unit tests for type
12:10:01	nueva	i mean, it's already working in this role
12:11:49	bbodi	https://github.com/Araq/Nimrod/issues/836
12:29:51	Araq	bbodi: that is not supposed to work, @[] has a gerneric seq type
12:30:14	Araq	it doesn't perform that form of type inference that you're after there
12:30:31	Araq	type the @[] and it should work
12:30:39	Araq	and yes, the compiler should complain about it
12:34:43	*	aruniiird quit (Ping timeout: 260 seconds)
12:41:48	bbodi	Yes, some compiling error messages could save a lot of time for the developers, thanks
12:43:58	*	Mordecai joined #nimrod
12:45:25	*	psquid quit (Ping timeout: 248 seconds)
12:48:55	EXetoC	sure. if the error comes from a C compiler then it's a bug
13:13:29	*	easy_muffin quit (Ping timeout: 252 seconds)
13:18:58	*	BitPuffin joined #nimrod
13:19:14	BitPuffin	ping Araq
13:36:13	EXetoC	lots of issues resolved
13:36:48	*	darkf quit (Quit: Leaving)
13:36:49	EXetoC	BitPuffin: including this https://github.com/Araq/Nimrod/issues/517
13:37:58	EXetoC	did you work around it?
13:48:57	renesac	the math.mean function requires an openarray, so that means I can't call it with slices?
13:49:50	renesac	nevermind
13:49:54	BitPuffin	EXetoC: don't remember lol
13:50:10	BitPuffin	My mind is not very organized atm
14:33:09	Varriount	vbtt: Just saw you blog post. Nice!
14:36:03	bbodi	can I invoke a method of the super type?
14:36:19	*	Mordecai is now known as psquid
14:38:10	Araq	bbodi: no, you have to make the method a proc of some unique name to invoke it, there is a feature request for foo(Base(x)) to resolve 'foo' statically but I don't think that's a good idea
14:38:50	bbodi	Araq: ok, thanks
14:39:12	Araq	hi BitPuffin
14:39:28	BitPuffin	hey Araq. Sorry I got home a bit too late yesterday in order to deal with it
14:39:32	Araq	bbodi: if you have a good idea how to do this, I'm all ears
14:39:48	BitPuffin	I'll be home much earlier today though so I'll give it a shot then :) will you be in here later?
14:40:11	*	faassen joined #nimrod
14:40:18	*	dirkk0 joined #nimrod
14:41:00	Araq	BitPuffin: what shot?
14:41:20	BitPuffin	Araq: the error I came across yesterday
14:41:26	BitPuffin	you told me to fix something in ropes
14:42:22	Araq	meh, I can do it later too
14:42:40	Araq	I'm doing a feature freeze for 0.9.4 now anyway
14:42:43	*	dirkk_ joined #nimrod
14:42:55	Araq	so, I have some time to fix bugs
14:45:33	*	dirkk0 quit (Ping timeout: 245 seconds)
14:49:00	*	dirkk__ joined #nimrod
14:50:20	BitPuffin	Araq: ah neat!
14:50:43	Araq	BitPuffin: means no improvements to the concurrency model though :P
14:50:47	BitPuffin	Araq: guess updated concurrency and vm is the main things on the roadmap for 0.9.6 ?
14:51:04	Araq	vm is in 0.9.4
14:51:10	BitPuffin	oh?
14:51:13	BitPuffin	completely done?
14:51:15	*	dirkk_ quit (Ping timeout: 245 seconds)
14:51:18	BitPuffin	well you know what I mean
14:51:23	Araq	well "completely"
14:51:25	Araq	yes
14:51:32	BitPuffin	software is never completely done
14:51:47	BitPuffin	well that's nice
14:52:05	BitPuffin	We desperately need 0.9.4 to release so no complaints here honestly :)
14:53:06	Araq	well fixing the most important bugs will take 2 weeks
14:53:14	Araq	that's my estimate
14:53:57	Araq	based on my feelings :P
14:54:43	BitPuffin	#emonimrod
14:54:51	BitPuffin	coding with feelings
14:58:58	Araq	bbl
14:59:19	nueva	#entomonimrod: dissecting bugs with feelings
14:59:42	*	AndChat|206976 joined #nimrod
15:03:43	AndChat|206976	Varriount: thanks. Any feedback on the blog post?
15:04:06	BitPuffin	what blog post
15:04:17	*	AndChat|206976 is now known as vbtt_
15:05:54	vbtt_	http://vocalbit.com/posts/exploring-type-classes-in-nimrod.html
15:06:01	vbtt_	Bbl
15:07:21	BitPuffin	vbtt_: cool! will definitely read it later!
15:11:45	*	dirkk0 joined #nimrod
15:12:55	*	dirkk__ quit (Ping timeout: 252 seconds)
15:14:11	renesac	https://gist.github.com/ReneSac/95c224ab5aab29529315 <-- how do I properly initialize a TRunningStat?
15:14:36	renesac	I must explicitly zero all fields
15:14:39	renesac	?
15:20:26	BitPuffin	lol
15:23:03	BitPuffin	that doesn't look right
15:23:34	BitPuffin	renesac: is the result the same each time?
15:24:22	renesac	actually yes
15:24:32	renesac	but it is probably because unitialized memory
15:24:46	renesac	oh, it is not
15:25:00	renesac	(big numbers all look the same)
15:25:09	BitPuffin	renesac: yeah floats seem too have issues initializing to 0
15:25:14	BitPuffin	at least completely 0
15:25:20	BitPuffin	it's usually something very close to 0
15:25:21	renesac	acc.n is an int
15:25:47	BitPuffin	yup
15:26:03	BitPuffin	oh right yeah wtf
15:29:37	vbtt_	Can anyone give feedback on correctness of the blog post?
15:29:40	vbtt_	http://vocalbit.com/posts/exploring-type-classes-in-nimrod.html
15:30:01	discoloda	type classes look usefull
15:31:12	discoloda	i cant tell you how accurate the post is, but i think its written well
15:33:16	vbtt_	Thanks discoloda
15:35:51	*	[1]Endy joined #nimrod
15:38:55	*	dirkk_ joined #nimrod
15:39:13	EXetoC	BitPuffin: that bug wasn't reported?
15:39:47	*	dirkk0 quit (Read error: Connection reset by peer)
15:42:53	BitPuffin	EXetoC: which bug
15:44:20	*	Varriount|Mobile joined #nimrod
15:45:20	Varriount|Mobile	So, my email got assaulted this morning by quite a few GitHub notifications
15:45:46	BitPuffin	lol
15:45:48	BitPuffin	:D
15:47:05	EXetoC	BitPuffin: the float init bug
16:07:37	*	Varriount|Mobile quit (Remote host closed the connection)
16:11:47	*	bbodi_ joined #nimrod
16:15:23	*	[2]Endy joined #nimrod
16:18:23	*	[1]Endy quit (Ping timeout: 272 seconds)
16:22:27	BitPuffin	EXetoC: The way I worked around it was to just loop through my vector or matrix or wtf it was and initialize it to 0
16:22:30	BitPuffin	wasted cpu cycles ftw
16:22:33	BitPuffin	but it's ze workaround
16:26:15	vbtt_	System.reset
16:26:35	EXetoC	so no
16:27:07	EXetoC	vbtt_: yes?
16:27:39	renesac	should I open a bug on my issue with TRunningStat?
16:29:44	EXetoC	BitPuffin: noInit first then?
16:29:44	*	wan quit (Quit: leaving)
16:32:11	EXetoC	vbtt_: nvm
16:32:36	EXetoC	I accidentaly called system.reset instead of the proc I had define. it's very hard to figure out what's going on in that case
16:36:56	*	LordAndrew joined #nimrod
16:37:15	*	BitPuffin quit (Ping timeout: 252 seconds)
16:39:57	*	XAMPP quit (Ping timeout: 265 seconds)
16:41:12	*	ddl_smurf quit (Quit: ddl_smurf)
16:42:40	*	ddl_smurf joined #nimrod
16:43:10	*	ddl_smurf quit (Client Quit)
17:08:43	*	dirkk_ quit (Quit: This computer has gone to sleep)
17:17:01	*	BitPuffin joined #nimrod
17:36:50	*	io2 joined #nimrod
17:48:45	*	vbtt_ quit (Ping timeout: 252 seconds)
17:48:46	Araq	renesac: the compiler intializes everything to binary 0 for you ...
17:49:39	renesac	then "n" should have been initialized too...
17:50:01	Varriount	Araq, zahary, dom96 : I'm going through issues older than one month, and testing any examples I find to see if they still exist. I've commented on the ones that I think should be closed.
17:52:39	Araq	Varriount: excellent, but do you test with devel or master?
17:52:51	Varriount	devel
17:52:57	Araq	ok
17:53:45	Araq	not sure if thats correct but merge with master is not far away anyway *cough*
17:58:53	*	xtagon joined #nimrod
17:59:20	nueva	Araq: 'nimrod --path:/tmp/example c a' doesn't found given module, though '/tmp/example/a.nim' certainly exists. bug?
18:01:23	*	BitPuffin quit (Ping timeout: 245 seconds)
18:01:26	Araq	--path is for imports
18:01:43	Araq	not if you're too lazy to give your main file properly
18:07:50	*	CarpNet quit (Quit: Leaving)
18:11:38	*	brson joined #nimrod
18:13:34	renesac	Araq: using values passed as parameters is faster than using module-global values? When used in a tight loop, for example.
18:15:16	Araq	renesac: usually, yes
18:15:41	renesac	what I do if this value is a state that the caller should not care about?
18:15:56	renesac	make a block inside the module to declare it and the function?
18:16:34	Araq	then the function is in the block too and can't be exported
18:16:45	renesac	oh
18:16:49	Araq	use a .global variable instead, which is quite like C's static
18:16:59	renesac	humm
18:17:07	renesac	I forgot this pragma
18:17:31	renesac	thanks
18:18:42	renesac	why did you name it "global" and not "static"? (I know, "global like lifetime", but it is a bit confusing ^^')
18:19:40	*	vbtt_ joined #nimrod
18:24:03	NimBot	Araq/Nimrod devel 179bfad Araq [+0 ±2 -0]: typeClasses now documented in grammar.txt
18:24:03	NimBot	Araq/Nimrod devel 581d138 Araq [+0 ±1 -0]: renderer knows about nkStaticTy and nkTypeClassTy
18:24:53	Araq	vbtt_: many thanks for your blog post, will read it tonight
18:25:24	NimBot	Araq/Nimrod master b4645a0 Araq [+0 ±1 -0]: master compiler compiles with devel
18:26:15	vbtt_	Araq: sure thing.
18:26:45	vbtt_	I'll probably continue doing minor tweaks to the post.
18:27:10	vbtt_	I've wanted something like nimrod's type classes and haven't seen it in another language. I wrote about it because I like the overall design.
18:27:15	Araq	renesac: "static" now means "at compile time"
18:27:21	*	io2 quit (Remote host closed the connection)
18:27:22	vbtt_	I'll try to do more posts over time.
18:27:27	*	DAddYE joined #nimrod
18:27:31	Araq	hi DAddYE
18:27:44	DAddYE	Hi Araq!!!!!! How are you!?
18:28:15	renesac	Araq: right, I thought it was "const" for compile time stuff
18:28:16	Araq	dunno, I never sleep more than 5 hours
18:28:33	DAddYE	Araq: kids or work?
18:28:36	Araq	renesac: we have "const", "static" and ".compileTime", ... it's a bit of a mess
18:28:46	Araq	DAddYE: kids...
18:29:07	DAddYE	heheheh
18:29:16	vbtt_	Araq: plural? I thought your only kid was nimrod?
18:29:50	Araq	vbtt_: no I have also a real life
18:30:17	vbtt_	Araq: excellent!
18:30:37	Araq	I guess ...
18:31:11	*	aftershave quit (Quit: Textual IRC Client: www.textualapp.com)
18:31:39	vbtt_	I'm always amazed by how much nimrod has, in terms of features *and* stdlib (very important) given that it's a hobby project.
18:31:40	Araq	renesac: in fact, I don't think .global has been designed. It grew into existance
18:31:46	*	aftershave joined #nimrod
18:32:55	Araq	renesac: 'sfGlobal' was a flag to mark global variables in the compiler, the step to .global was obvious then for a compiler writer
18:33:08	renesac	hum
18:33:32	Araq	later we discovered it's incredibly useful for metaprogramming
18:33:51	vbtt_	hmm I think I used master ti test type class features. perhaps thats why they didn't all work complete.
18:35:00	*	vbtt_ quit (Quit: Page closed)
18:35:07	nueva	wtf? 'nimrod c random/a.nim' puts cached file as 'nimcache/random_a.cpp', but 'nimrod c src/a.nim' puts cached file as 'nimcache/parent_a.cpp', where 'parent' is a name of parent dir of 'src'
18:35:19	*	vbtt_ joined #nimrod
18:35:28	Araq	nueva: that's the feature I was talking about
18:35:39	Araq	the compiler guesses the package name
18:35:52	Araq	as we have no notion of a package in the language
18:36:45	Araq	IMHO it should be documented how the algorithm works and then it's perfectly fine, in your opinion it's a feature from hell, I guess
18:37:38	*	io2 joined #nimrod
18:40:44	reactormonk	Araq, could we change that name generation? Or somehow control it?
18:41:23	Araq	why?
18:41:30	nueva	but why to do it for main module? so if i have my example project in '/home/user/Projects/ExampleProject/src/main.nim', but then move it to 'home/user/Projects/CoolProject/src/main.nim'. it will regenrerate *_main.cpp?
18:42:04	nueva	"Main module" is what I pass to compiler to build whole project
18:43:13	Araq	nueva: the use case was that every babel package has such nice module names as "utils" or "types" and so we need a way to distinguish modules of the same name
18:44:54	Araq	so yeah, it recompiles when you rename your project dir. wooohoo big deal
18:45:26	Araq	happens like once in a year so I already spent more time talking about it
18:48:43	reactormonk	Araq, does the compiler do that for everything or just the included libs?
18:49:14	Araq	for every module you use
18:49:18	nueva	are there so many babel packages with different intenral hierarchy?
18:49:47	nueva	i've found list of ecxceptions. "lib", "src", "source", "package", "pckg", "library", "private"
18:49:56	Araq	yup
18:50:39	nueva	so, if package writer will use something different for his internal directory name it will be his fault?
18:51:13	Araq	I can't follow you, sorry
18:51:37	Araq	packageA/src/utils.nim is common and should become packageA_utils
18:51:53	Araq	not src_utils as that wouldn't avoid collisions
18:52:26	nueva	Araq: it will not only recompile, it will leave unexpected garbage. it also makes reasoning about generator a bit harder.
18:53:04	vbtt_	really the top level package direcotry should be tagged. e.g. package.nimpkg/srcl/blah/whatever
18:53:22	nueva	packageA/myprivatedir/utils.nim and packageB/myprivatedir/utils.nim
18:53:34	vbtt_	then it's unabiguous what the package name is.
18:54:21	nueva	who should then change his 'privatedir' packager A or packager B?
18:54:59	reactormonk	nueva, so you want the compiler to be aware of packages?
18:55:04	vbtt_	and yeah, special directories should not be ignore (if you want you can create src/mypackage.nimpkg/x.nim)
18:55:09	nueva	ok, it's not my question really. my question was: should packagers just use 'src' by convention as a name of private package dir?
18:55:09	zahary	vbtt: there is a minor error in one of your examples
18:55:22	zahary	for item in f:
18:55:22	zahary	line is String
18:55:22	zahary	you meant to say "for line in f"
18:55:29	vbtt_	zahary: thanks for reviewing!
18:55:34	vbtt_	ah right - that's the one I didn't compile :)
18:55:59	vbtt_	zahary: feedback on accuracy also appreciated. also, did I miss anything important?
18:56:09	zahary	you may also mention that you could even write param types such as seq[FileLike] and Nimrod will just do the right thing
18:56:25	*	vbtt quit (Quit: Page closed)
18:57:18	vbtt_	zahary: yeah I tried that and didn't get it work. perhaps I should use devel and not master? I also found minor issues with 'is' testing in type classes.
18:57:55	zahary	hmm, I think this one should work already, but I might be wrong
18:57:55	nueva	reactormonk: I'd like to shorten this list of exceptions to just '__some_ugly_rare_name_rer34234d' and make a rule for packages to use this name only (not 'src' or something). because this list couldn't be infinite
18:58:23	Araq	nueva: well the convention is indeed a "private" directory
18:59:21	vbtt_	zahary: the compiler reported Error: internal error: GetUniqueType. when I used it as a parameter to seq[MyTypeClass]. trivial example reproduces it.
19:00:45	zahary	alright, I'll work on that soon
19:01:12	reactormonk	nueva, I suggest make a forum post and/or a github issue for that and state your point in full
19:01:16	nueva	Araq: still there is a list of conventions used by compiler. I suppose it's for legacy packages?
19:03:19	Araq	nueva: actually it's to allow for more flexibility but legacy is a reason too of course
19:03:30	Araq	as I said, I think we should document this list and then it's fine
19:03:38	*	Mordecai joined #nimrod
19:04:02	Araq	but I'm open for suggestions
19:04:14	Araq	it's not too late to change it again
19:04:33	vbtt_	zahary: also a simple 'is' test didn't work. i.e. `f(x) is Int` in the typeclass. and then define f(x:MyType): Bool. but `MyType is MyTypeClass` still returns true.
19:04:43	*	psquid quit (Ping timeout: 245 seconds)
19:04:58	Araq	I guess I'm too blame for the 'is' bugs
19:05:12	vbtt_	Araq: what do you think of my suggestion? only .nim files and sub-folders in a top level package directory that has an .npk extension.
19:05:21	vbtt_	every folder is part of the package.
19:05:37	vbtt_	i.e. no specially ignored cases (src or whatver).
19:05:48	Araq	vbtt_: so essentially we go up the tree until we find a .babel file?
19:06:11	vbtt_	that might work too.
19:06:29	vbtt_	Araq: I was suggestion you call the folder mysqllib.npk
19:06:37	vbtt_	or collections.npk
19:06:45	vbtt_	within that the folders dont have extensions.
19:07:00	vbtt_	yes you go up till you know you've found the top level folder for that package.
19:07:06	nueva	Araq: shorten list to just "private". and maybe to not so obvious name ("__private__"?)
19:07:11	Araq	folders shouldn't have extensions, that's uncommon
19:07:19	vbtt_	easy to find from the outside too (unike python where I have to see if it contains __init__
19:07:30	vbtt_	Araq: ok then just a special file in the top level folder is fine.
19:08:02	vbtt_	Araq: btw, it may be uncommon but i think it's a great idea :) and cross platform too.
19:08:51	vbtt_	my basic proposal is get rid of ignored special cases. and have an easy way to locate a source file, given an import statement.
19:08:53	reactormonk	vbtt_, .app ?
19:09:15	vbtt_	reactormonk: true! not that uncommon, haha.
19:10:20	vbtt_	i think organizational only directories like 'src' should be outside your package source top level folder, not inside.
19:12:37	Araq	vbtt_: just to be clear packageA/src/packA/module.nim produces packA_module, no need to adhere to any convention, it simply works unless you dream up unrealistic example names like "foo"
19:13:58	Araq	and what do we do with the stdlib? that has no .babel file, though perhaps it could fake one
19:14:16	nueva	btw, babel path outside of project directory by default isn't nice. it's better to follow Node.js (npm), Python (virtualenv, pip), Haskell (cabal sandbox). but I guess, it was already discussed before making a desicion...
19:14:21	vbtt_	Araq: yeah, .babel or whatever special file should be in *every* top level pkg. including stdlib.
19:14:27	reactormonk	Araq, just hardcode it to `stdlib`
19:15:10	vbtt_	Araq: how does nimrod know where to search for nim files?
19:15:24	reactormonk	vbtt_, config/nimrod.cfg
19:15:26	Araq	vbtt_: via --path and --babelPath
19:16:36	Araq	nueva: again, what do you mean? "outside of project dir"?
19:17:54	vbtt_	Araq: i'm proposing that given '--path A,B,C' is given, nimrod do only a one-level DFS for folders under A or B or C that contain the special .babel file.
19:18:18	vbtt_	no complicated search for nim files and magic package names.
19:19:03	vbtt_	s/is given//
19:19:57	Araq	nah, I leave everything as it is
19:20:20	Araq	it works quite well afaict
19:20:32	Araq	and it's telling I got not a single bug report about it
19:20:41	nueva	Araq: I mean, "config/nimrod.cfg(37, 2) Hint: added path: '/home/user/.babel/pkgs/'" by default, while it could be '/home/Projects/CoolProject/.babel/pkgs'
19:20:45	reactormonk	Araq, maybe no one is using it ^^
19:21:06	Araq	reactormonk: why do we get so many other bug reports then? :P
19:21:49	*	Mordecai is now known as psquid
19:21:51	reactormonk	right, I still need to create something like repr in JS
19:21:58	vbtt_	Araq: you'll get more colissions as the community grows.
19:22:07	*	EXetoC quit (Ping timeout: 260 seconds)
19:22:47	reactormonk	Araq, does the compiler recognize collisions?
19:22:55	vbtt_	does nimrod have a locate command? `nimrod locate foopkg`
19:23:04	nueva	Araq: bug report about what? about directory exceptions used for generated file naming? about flat scheme of naming of generated files?
19:23:10	vbtt_	in the current state of affairs, the locate command would be much appreciated.
19:23:26	reactormonk	vbtt_, I think idetools --def might help a bit
19:25:14	Varriount	Araq: For testing purposes, either we should close issues that are fixed on devel, or at least make a note that the issues are fixed on devel, and close them when devel next merges into master
19:25:16	vbtt_	fwiw, this could be fixed down the line too.
19:25:20	Kooda	Is the devel branch expected to build right now?
19:25:33	Varriount	Kooda: Yes.
19:25:36	vbtt_	ppl will just have to move their files around to match their exported hierarchy.
19:25:44	Varriount	Also, nice to see you again kooda
19:26:01	Kooda	I saw your comment on my PR, I want to test it :)
19:26:16	Kooda	Do I need to use koch to build devel too?
19:27:22	Varriount	Kooda: Of course.
19:27:43	Kooda	Hm, so I might have a problem.
19:27:53	Kooda	lib/pure/os.nim(1589, 22) Error: undeclared identifier: 'ParamStr'
19:28:15	Varriount	Kooda: Your not mergin devel into master on your local repo, are you?
19:28:31	Varriount	*merging
19:28:34	nueva	I guess, no one is using Nimrod with an external build system. but if Nimrod declares interoperability with C/C++/JS on source level, somebody will use it with external build system (like Grunt for JS or Cmake for C++)
19:28:59	Kooda	Varriount: hm… not if I didn’t mess up with my repo
19:29:26	Varriount	Kooda: It should build. I built nimrod using the devel repo this morning.
19:29:38	nueva	as all (except me) are using Nimrad as a build system, there are no bug reports about file names and stuff
19:29:45	nueva	*Nimrod, sorry
19:29:56	vbtt_	btw, where does the build farm run? what tool does it use? and can I add a builder?
19:30:14	vbtt_	iow, I didn't find an about page on build.nimrod...
19:30:20	Varriount	vbtt_: The build farm is run by different people, using nimbuild (search on github)
19:30:31	Varriount	I run the Win32 and Win64 builders
19:30:43	Kooda	Oh
19:30:51	Varriount	dom96 runs the 32 and 64 bit linux builders.
19:30:51	Kooda	Could I add NetBSD i386 and amd64?
19:31:01	Araq	vbtt_: nimbuild is currently broken and I'm having fun to get it back to work ... it's a nice multithreading issue
19:31:34	Varriount	Kooda: Yeah. You just have to get dom96 to accept the new OS config and have him set it in the hub
19:32:02	Kooda	I have a machine for it if needed
19:32:04	Varriount	Araq, vbtt_ : Well, nimbuild isn't entirely broken. It's just that it crashes when it tries to bootstrap nimrod the second time in a row.
19:32:20	Araq	nueva: ok, fair enough, but the .babel thing is an algorithm just like it currently implements an algorithm. so how exactly does it change anything?
19:32:31	Varriount	Which I bypass by running the builder in a shell loop.
19:34:16	vbtt_	some algorithms have fewer surprises.
19:34:34	vbtt_	the current violates the principle of least surprise, perhaps.
19:35:08	nueva	Araq: .babel wasn't my suggestion (though I'm not objecting it). I was suggessting shorten list of exceptions to one non-obvious, rare name and make it conventional.
19:35:08	vbtt_	anyway, i'm going to stop arguing about it cos it still sort of works.
19:35:49	Araq	Kooda: which OS are you on?
19:35:54	Kooda	NetBSD
19:36:26	*	EXetoC joined #nimrod
19:40:12	EXetoC	nueva: lib/src is common in quite a few languages aren't they? which is why I adhere to those. lib would be for libs and src for apps
19:40:20	EXetoC	I guess we just need some conventions
19:44:24	NimBot	Araq/Nimrod devel c9cdbbe Araq [+0 ±3 -0]: should compile on netbsd
19:44:29	Araq	Kooda: try again
19:44:42	nueva	EXetoC: I was speaking about predictability of generated files' names. yeah, 'src' name for source directory is common, so I'm using it in project (not in package). and I'm using external build system. so I want to know what files will be generated by Nimrod. i was suprised.
19:47:04	Araq	nueva: I'm looking at my .babel directory now
19:47:34	Araq	the opencv.babel file is in opencv-0.1.0/
19:47:36	Kooda	Araq: fixed the issue :)
19:48:13	nueva	so it's a clash between my expectations and allowing to use 'src'/'lib' in packages. right now, I'm the only who is using external build system, so I think I will not win :)
19:48:37	Araq	no on the contrary
19:48:45	Araq	you fought long enough with me
19:48:46	vbtt_	nueva: for what you're doing, nimrod should just emit the list of generated files. you shouldn't have to reproduce it's algorithm.
19:49:01	Araq	vbtt_: that's what nimrod already supports
19:49:11	vbtt_	nueva: why not use that?
19:49:27	vbtt_	nueva: that isolates you from nimrod internals and possible changes.
19:49:56	vbtt_	hmm, maybe the build system doesn't support it.
19:50:05	vbtt_	so you have to recode the logic in the other build systems
19:50:14	Araq	nueva: we also have plans to split the larger generated .c file into mutliple files
19:50:28	Araq	then there is no 1-1 mapping between modules and .c files
19:51:42	nueva	vbtt_: yeah, bu Nimrod now can output this list only when it compiles files and only to fixed file path. I need to know about generated files' paths without compilation (i.e. without producing any files).
19:52:07	*	aftersha_ joined #nimrod
19:52:41	vbtt_	that's a feature request then! bbl.
19:52:42	nueva	it was discussed earlier, PR is awaited from my side, but I want to avoid it :)
19:52:54	*	aftersha_ quit (Client Quit)
19:53:10	*	aftersha_ joined #nimrod
19:53:39	Varriount	Anyone know if the semantics for 'creating' a new closure iterator variable changed?
19:54:38	*	aftersha_ quit (Client Quit)
19:54:51	Araq	Varriount: I hope not
19:54:59	Araq	all the tests I have for them are green
19:55:13	Varriount	Araq: I'm running a test from an old issue
19:55:34	Varriount	Or rather, a code snippet provided in the issue. It concerns recursive iterators.
19:55:49	Varriount	Araq: https://github.com/Araq/Nimrod/issues/555
19:55:54	Araq	recursive iterators are not supported
19:56:33	Varriount	Araq: Are they supposed to throw an error?
19:57:18	*	aftersha_ joined #nimrod
19:57:31	Araq	yes
19:57:56	Araq	it's a nice tricky test case
19:58:05	Araq	added it to the suite, no idea why it was missing
19:58:39	Varriount	Araq: Well in the code snippet in that issue, they don't.
19:58:50	Araq	I know
19:59:00	Araq	it's an indirect recursion, I didn't check against that
19:59:55	Varriount	Although, at least it doesn't crash the compiler. ;P
20:00:00	EXetoC	what about sets in strutils for control chars, lowercase etc?
20:00:39	Araq	EXetoC: constants are usually in Uppercase except when they are not
20:01:45	nueva	Araq: splitting of .c isn't really interferes with my current approach. naming of splitted files will still be deterministic. but of course I always will be relying on unstable/fragile compiler internals in this case.
20:02:08	EXetoC	Araq: no I mean character sets for those, because I have a hard time remembering ASCII ranges such as 0-0x1F for example
20:02:40	*	vbtt_ quit (Ping timeout: 272 seconds)
20:03:14	Araq	EXetoC: we'll only get bug reports that those are not unicode aware, so please don't add it
20:03:37	EXetoC	-.-
20:03:55	nueva	and it's better to use specialized public interface
20:03:59	renesac	https://gist.github.com/ReneSac/3b56d3b3836761093436 <-- how I can get this object working like a sequence?
20:04:51	Araq	in fact ... I don't think I'll write another stdlib module anytime soon for reasons like this
20:05:29	Araq	when you do nothing you can't do anything wrong
20:06:12	dom96	When you do things right, people won't be sure you've done anything at all.
20:06:30	*	dom96 wonders who gets the reference
20:06:34	dom96	Hi btw
20:07:57	EXetoC	I haven't heard of any alternatives. I'll keep referring to asciitable.com then
20:08:44	Araq	renesac: I think you need a single `[]` that returns a 'var TPt' for this to work
20:09:41	Araq	hi dom96, there were discussions about babel and how the compiler generates C filenames
20:09:48	dom96	Holy wow at the amount of notifications on Github.
20:09:55	dom96	I'm glad I disabled the emails.
20:10:00	Araq	I would like to know your opinion
20:10:11	Araq	what? the emails can be disabled? lol
20:10:20	Varriount	I like the emails.
20:10:22	dom96	Yep. In the settings.
20:10:38	Varriount	dom96: I think I've commented on one or two issues posted by you.
20:10:41	dom96	Although you still get some. Like ones which you created or were a part of I think.
20:10:56	dom96	Yeah, that's probably why I got those emails heh.
20:11:57	dom96	Araq: Yeah, I read everything. The idea to use the .babel file to determine the top-level directory isn't a bad one.
20:12:10	*	bbodi_ quit (Ping timeout: 272 seconds)
20:12:22	Araq	the top-level directory has that annoying version number though, dom96
20:12:37	EXetoC	Araq: do you know what these people want then? only unicode-aware functions perhaps
20:13:02	Varriount	^
20:13:06	Araq	EXetoC: they want C#
20:13:19	Araq	they might pretend that they don't but they do
20:13:32	dom96	Araq: Yeah... well it's hard to keep a balance between "Nice filenames" and "unambiguous filenames"
20:13:53	dom96	It may be a good idea to lean towards "Unambiguous filenames" completely.
20:13:57	Varriount	EXetoC: Give up. Your never going to be able to convince Araq to change his mind about unicode.
20:14:06	Varriount	*unicode related stuff
20:14:44	dom96	Araq: If you're desperate you could use the .babel filename. :P
20:14:58	EXetoC	Varriount: I disagreed?
20:15:53	Varriount	It sounded like you were trying to convince Araq of something unicode-related.
20:15:54	EXetoC	it sounded to me like he wouldn't mind, but other people would coplain
20:17:08	EXetoC	nevermind that
20:18:38	dom96	Kooda: I appreciate you wanting to contribute a machine but I would prefer to get these nimbuild issues fixed first. I still think that the builders need some adjustments to be more friendly to machines which are not under my control.
20:18:39	EXetoC	I wonder what would be more acceptable. overloads for both char and TRune?
20:19:17	Araq	ah yeah, back to nimbuild debugging
20:20:04	Varriount	EXetoC: Yes.
20:20:06	Araq	Varriount: btw it took me 3 evenings to get the closure iterators to work, so don't worry. I had to change a lot
20:20:29	Varriount	Araq: I know. Believe me, your hard work is not going unappreciated.
20:21:12	Varriount	I've *seen* the lambda lifting code.
20:21:38	Araq	the lambda lifting code is actually quite clean
20:21:57	Araq	the algorithm itself drives one crazy though
20:22:13	Araq	and there are so many special cases to consider
20:23:00	Varriount	Like the one I mentioned above.
20:23:14	Araq	that returns 0 now
20:23:28	Varriount	Which is counted as "None" I assume?
20:24:02	Varriount	Anyway, I hope you don't mind me going through the issues. It's something useful that I enjoy doing.
20:24:28	Araq	are you crazy? I appreciate you going through the list of bugs!
20:25:29	Kooda	dom96: please tell me when you are ready :)
20:25:49	dom96	Kooda: I certainly will, thanks :)
20:26:13	Varriount	dom96: Thank you for putting easily testable snippets in your issues. :D
20:27:22	EXetoC	Araq: so the biggest issue is the lack of isControlChar for TRune for example? would that *and* sets be fine then? Sets are neat
20:27:36	dom96	Varriount: Thank you for appreciating them! You just put a smile on my face :)
20:27:56	Araq	EXetoC: just add the set with a comment it's not unicode aware
20:28:00	reactormonk	https://github.com/Araq/Nimrod/issues/347 gist timed out :-/
20:28:40	Araq	reactormonk: I can close it as "won't fix" if that helps you
20:29:01	reactormonk	Araq, let's see if it's still buggy
20:29:07	renesac	Araq: indeed it was just the lack of 'var' on the return variable, thanks ^^"
20:30:10	renesac	it seems I don't even need to define `[]=`
20:30:18	EXetoC	Araq: well that was simple. maybe for Whitespace/NewLines as well then, because unicode has more of both I think
20:30:23	reactormonk	Araq, yup, it is.
20:31:07	Araq	reactormonk: can you pass a function(a,b,c) to a function(a,b) in JS?
20:31:43	*	mietek quit (Ping timeout: 272 seconds)
20:31:45	reactormonk	Araq, you can pass functions all the way you want
20:31:56	*	mietek joined #nimrod
20:33:02	Araq	hmm
20:33:16	Varriount	What does "object contstructor needs an object type" mean?
20:33:22	Varriount	*constructor
20:33:29	reactormonk	Varriount, code?
20:33:44	Araq	Foo(a: 1, b: 2) # Foo should be an object?
20:34:28	Varriount	Araq, reactormonk : https://gist.github.com/Varriount/8586245
20:34:36	Varriount	The error is caused by the bottom file
20:34:56	dom96	hello mietek
20:35:05	Varriount	(I can't get the top file to compile, it seems that the kind of tuple conversion expected on line 4 is no longer supported)
20:35:07	Araq	Varriount: yes, because it's a tuple, not an object
20:35:13	dom96	oh, mietek is not new. Hello anyway.
20:35:29	Araq	and yes I know about the range in tuples problem
20:35:52	Araq	first example should compile, 2nd not
20:35:57	Varriount	Which would explain why natural() fails.
20:36:56	Araq	it's really quite simple; if A <: B then C[A] <: C[B] doesn't hold in general
20:37:07	Araq	so the compiler refuses the code
20:39:11	reactormonk	Araq, in case of a tuple it doesn't? That's a good old product type.
20:40:11	Araq	if A and B are objects they could have different sizes and then the tuple can't possible adhere to a subtype relation
20:40:28	*	aftersha_ quit (Quit: Computer has gone to sleep.)
20:40:57	reactormonk	ah sap
20:41:21	reactormonk	can't you check for sizes?
20:42:24	Araq	that's planned, yeah but it means we have 2 different subtype relations
20:42:39	Araq	to support and so it's a bit of work
20:42:56	reactormonk	and not overly urgent?
20:43:42	Araq	yeah
20:45:39	nueva	I built compiler from devel branch 6 days ago. I've just pulled new commits and 'koch boot' now fails with "lib/system/excpt.nim(83, 17) Error: undeclared identifier: 'CString'". will cleaning help?
20:46:47	*	vbtt joined #nimrod
20:47:20	Varriount	nueva: It should
20:47:49	Varriount	nueva: Also make sure that all nimcache directories are deleted.
20:47:53	Araq	nueva: something is still not case consistent ,but I wonder why "nativeStackTraces" are enabled for you
20:48:15	vbtt	btw if I declare `var x = MyObject(..)` it's goes on the stack, correct?
20:49:03	Araq	yes
20:49:04	*	[2]Endy quit (Ping timeout: 252 seconds)
20:49:10	*	Varriount will never understand why people are worrying about stack/heap allocation in a gc'ed language
20:49:15	dom96	vbtt: I'm reading your article and found some spelling mistakes: "defintion", "arbirary". Other than that so far very nice article :)
20:49:25	Varriount	*stack-vs-heap
20:50:24	nueva	Araq: because I explicitly enabled it.
20:50:26	dom96	vbtt: Another one: "evalutate"
20:50:26	Araq	Varriount: because stack allocation is 0 instructions and heap allocation is 100?
20:50:41	vbtt	dom96:thanks for the feedback! I'll spellcheck it.
20:51:00	vbtt	hah wow I was tired at night.
20:51:22	dom96	vbtt: Thank you for writing the article!
20:52:00	vbtt	Varriount:I need to know the semantics. it's not just about the gc. it's copying, lifetime, etc.
20:52:01	dom96	Araq: "var written: Int = write(f, String)". See? People already want it to be 'Int' not 'int' :P
20:52:25	vbtt	heh - not sure what it's originally, or string.
20:53:23	dom96	Araq wants the convention to be that for types (and some consts) the first letter is uppercase, except for built-in types like int, string etc.
20:53:32	nueva	'koch clean' didn't help (btw, it deleted .gitignore). there are no nimcache directory left, but it still fails with the same error. so I should just disable nativeStackTrace?
20:53:45	Varriount	nueva: Yes.
20:53:45	vbtt	I like type names capitalized. And I'm used to snake_case but I tried using camelCase in the examples. if there's a style guide i'll follow it.
20:53:52	dom96	I'd rather just make it clear that *every* type starts with a capital letter.
20:54:00	dom96	everything else does not.
20:54:02	dom96	Like in Haskell.
20:54:04	Araq	nueva: sorry I didn't catch this but "koch clean" never helps with anything
20:54:10	Araq	I never run it fwiw
20:54:31	vbtt	dom96: Python does what Araq wants. I lean towards every type being capitalized. but either is fine really.
20:54:35	Araq	you simply need to do /CString/cstring/ and then it should compile
20:55:15	Araq	dom96: uppercase Int is too annoying, I tried it
20:55:52	vbtt	will `type lowercasename` be a syntax error?
20:56:09	dom96	Araq: Ok. I can live with it.
20:56:17	vbtt	Hm, I'll change the example to lowercase Int and String, then.
20:56:19	Araq	vbtt: I wanted to do that but people dislike the idea
20:56:41	Araq	apparently consistency works best when it's not enforced ....
20:56:44	vbtt	Araq: I like the idea.
20:57:07	vbtt	Araq: that's a constraint that can always be relaxed later with no breakage. Unlike the other way..
20:57:24	Araq	vbtt: doesn't help
20:57:34	Araq	for instance we're still quite strict when it comes to comments
20:57:54	Araq	I can only guess how many people never considered nimrod because of that feature
20:57:54	vbtt	hmm ok.
20:58:38	Varriount	vbtt: Or simply write in the style you prefer, and run `nimrod pretty` over everything.
20:58:43	Varriount	:3
20:58:50	nueva	## \ and then doc comments on new line is weird, IMHO
20:58:50	vbtt	actually i originally passed on nimrod because of case/underscore insensitivity, $ and command style calls.
20:59:10	vbtt	possibly others, but now I realize I cant get everything I want (without doing any work ;)
20:59:12	Varriount	nueva: That isn't actually done anymore.
20:59:22	Araq	vbtt: there you go then
20:59:27	EXetoC	such minor reasons
20:59:40	vbtt	Varriount:ah ok. i didn't know 'pretty' works.
20:59:51	dom96	vbtt: Perhaps in the "Using Type Classes [sic]" section you could add another example where you actually define a type which satisfies the type classes you defined and shows that they can be passed to those functions you defined (``doubleBend`` and ``copyFileData``)?
20:59:56	Varriount	nueva: If you look in the libs, documentation tends to be done python style, with doc-comments underneath things
21:00:21	dom96	er, nvm my [sic] misread heh
21:00:28	vbtt	dom96:good idea. i was going to write more in there and also show a complie time error on mismatch.
21:00:39	dom96	vbtt: Yeah, great :)
21:01:11	vbtt	EXetoC:not really minor. readability counts. specially when you have to read other developer's code.
21:01:38	Varriount	Araq, dom96: Well, I've gone through about 75% of all the issues (I ignored the one's that weren't clear, lacked a test, or were feature requests)
21:01:50	vbtt	i was more happy with clay, but that language died.
21:01:59	Varriount	I've commented on the one's whose results have changed.
21:02:06	dom96	Varriount: Now fix 20% and i'll give you a hug.
21:03:05	vbtt	there's also other weird looking stuff like dangling '=' in proc. yes these things matter and ppl will pass on languages because of this. however, you can still have a compelling language if your ecosystem is solid. i.e. good, complete stdlib, nice community etc.
21:03:36	vbtt	i can't use most new languages because they dont even try to implement a stdlib or common libraries. i want to build real products.
21:04:00	vbtt	nimrod has a very good ecosystem but not compelling enough yet.
21:04:17	reactormonk	vbtt, what lib are you missing?
21:04:29	dom96	I think the reason that they don't is because the language changes too much. We try to keep backwards compatibility even though we're pre-1.0.
21:05:10	EXetoC	vbtt: right, subjectivism and all that :>
21:05:41	dom96	vbtt: Any ideas what happened to Clay? Did the author just give up/lose interest?
21:06:11	EXetoC	dom96: yeah, people should know what <1.0 implies
21:06:21	vbtt	reactormonk:just compare with pytho stdlib. not that python apis are great, but there's really a lot more than nimrod.
21:07:11	Araq	no there isn't
21:07:14	vbtt	dom96: the original author left it for another project. then 'joe' worked on it for a while and then got busy.
21:07:53	Araq	python's internet modules are extensive and battle-tested but that's about it
21:08:08	reactormonk	vbtt, make a list and post it in the wiki. And those will not go to stdlib, more likely as a babel package
21:08:22	Araq	of course if you count stuff like Python's "enum" module as lacking for nimrod, you'll get a different picture
21:08:31	nueva	Varriount: http://pastebin.com/WtDX16kE is there any other way understandabe by 'nimrod doc'?
21:09:27	Araq	also python has really good 3rd party stuff like numpy
21:10:02	Araq	but the stdlib is not that bigger
21:10:54	vbtt	forget enum. it's little things like datetime and random.sample
21:11:03	reactormonk	I mean you could port numpy to nimrod. Just rip out the python API and make a nimrod API
21:11:28	reactormonk	vbtt, time is nasty.
21:11:42	Varriount	nueva: http://pastebin.com/9U0SLngN
21:11:53	Araq	vbtt: hmm ok, I miss these details as I only go over the list of modules
21:12:39	reactormonk	Araq, what about time? Its own babel package and kick it out of the stdlib mostly?
21:12:55	Varriount	vbtt: I've heard that arrow is much better than datetime
21:13:09	Araq	reactormonk: doesn't work, but we can trim it down
21:13:29	Araq	and have a "timex" module in babel
21:13:32	vbtt	Varriount: I dont like python's datetime api, (dont know about arrow) but it's still convenient vs. nimrod where i'd have to implement it.
21:13:33	nueva	Varriount: I've tried it before asking, but it didn't work. because of outdated compiler (mine was built 6 days ago)?
21:13:52	Varriount	nueva: Huh. Odd
21:14:19	dom96	no, times should be in the stdlib
21:15:19	Varriount	nueva: Add two spaces before the doc-comment
21:15:43	vbtt	Araq:from a high level the list of stdlib in nimrod is impressive. however the impression i have is 1) somewhat lacking apis (however better functions can be added as they are used) and 2) not coherent or consistent (e.g. does the http lib use the parser lib? does it emit the standard datetime for dates, etc.
21:16:01	vbtt	Araq:still for pre 1.0, there's no other language that's close.
21:16:25	*	zahary quit (Quit: Leaving.)
21:16:33	Araq	it is quite consistent and coherent afaict
21:16:48	Varriount	^ If a bit biased toward the compiler's needs.
21:16:51	reactormonk	vbtt, write up any inconsistencies you find.
21:17:10	vbtt	ok, i'll spend more time with it. perhaps i'm missing the details.
21:17:16	dom96	I agree that the stdlib still needs some things and that there is a lot packages which could be on babel which we do not have but python and ruby do
21:17:38	dom96	This sadly means that deciding to take up some project could mean that you have to recreate some library that you need.
21:17:55	dom96	For example, for nimbuild I had to create a CIDR module (or whatever it's called)
21:18:26	Araq	Varriount: the compiler tries to avoid the stdlib
21:18:46	Araq	mostly for historical reasons though
21:19:04	Varriount	Araq: Ok then, biased towards other parts of nimrod's codebase.
21:19:05	Araq	there was no stdlib when I wrote the compiler ...
21:19:16	dom96	Anyone on Ubuntu or any other linux for that matter here have a minute to test something?
21:20:48	nueva	Varriount: pretty unintuitive. but it works now, thanks.
21:22:18	nueva	dom96: I'm on Debian. what exactly do you need to test?
21:23:01	Varriount	Hm. Which is preferable: Implementing reversed string operations from strutils as separate procs, or as a flag that can be passed to each proc?
21:23:15	vbtt	eg. theres asyncio *and* libuv.
21:23:29	Varriount	vbtt: Asyncio is a pure lib, libuv is a wrapper.
21:23:38	Araq	vbtt: if anything the stdlib is too consistent for its own good. people stumble upon TTable being value based types
21:23:59	vbtt	Varriount:my point is that perhaps one of those does not belong in the stdlib.
21:24:00	Varriount	Also, iirc, wrappers are being moved out of the stdlib (with some exceptions)
21:24:16	Araq	vbtt: libuv will become a babel package
21:24:19	vbtt	either pick libuv and then use it as the one and only event loop.
21:24:24	vbtt	Araq: ah ok.
21:24:37	vbtt	Araq:too consistent is a good thing imo.
21:24:45	Araq	quite a lot things already moved to babel
21:25:00	Araq	so now the tester needs to learn about babel ...
21:25:18	Varriount	Why?
21:25:34	Araq	so that we can test that babel packages still compile?
21:26:05	Varriount	Ah. So that was why you needed a new dir name?
21:26:35	Araq	no, I just like "testament" as a pun
21:26:39	vbtt	the parsers are all called 'parseXXX' *except* json which is just json.
21:27:51	Araq	json it not only a parser ...
21:29:54	Varriount	By the way, how much of the xml and json specs do the xml and json parsers implement? Is it enough that users should be concerned about xml and json exploits?
21:30:41	Araq	not sure what exploits you mean, but no
21:30:56	Araq	the xml parser ignores xml as much as it parses it
21:31:31	vbtt	heh
21:31:41	Araq	it's a braindead standard and when it dies I'll rejoice
21:32:40	vbtt	braindead perhaps but very commonplace.
21:33:00	vbtt	you want at least one strict xml parser.
21:33:11	Araq	we have some wrapper for that
21:33:13	vbtt	(for other people, not for me)
21:33:32	Araq	usually we have wrappers for the "proper" way and a cool module for my way
21:33:46	Araq	my way always wins in practice
21:33:51	dom96	nueva: https://github.com/dom96/ParticleBench
21:34:26	Araq	turns out nobody gives a shit about xml's namespacing feature etc. nobody uses it, nobody knows it exists
21:34:48	Araq	you should read these standards one day, it's crazy
21:35:27	dom96	oh yeah, the xmldom module seems like a waste of my time nowadays heh
21:36:22	Araq	btw nobody actually produces valid xml either, so a "strict" module doesn't help
21:36:55	vbtt	Araq:not true, there's a lot of valid xml out there. specially in-house.
21:37:11	Araq	there is more invalid xml though
21:37:17	vbtt	on the web json is fast replacing xml.
21:37:27	Araq	and it helps when you can parse that too
21:37:28	dom96	nueva: Run 'babel build' if you can with Nimrod @ master.
21:37:29	vbtt	take atom feeds e.g. or google apis
21:38:02	vbtt	Araq:sure but a switch (strict/lenient) would be nice.
21:39:08	Araq	*shrug*
21:39:36	vbtt	the peg module looks pretty slick.
21:41:17	nueva	dom96: strictly @ master? would @ devel be ok?
21:41:53	dom96	nueva: Would prefer @ master. But sure go ahead.
21:42:09	Araq	oh btw
21:42:30	*	dirkk0 joined #nimrod
21:42:38	Araq	I am *very* open for suggestions of how we should do the TParser -> Parser transition
21:42:39	*	dirkk0 quit (Client Quit)
21:43:50	vbtt	are you talking about renaming all types to remove the 'T' prefix?
21:44:04	Araq	yes
21:45:08	Araq	technically it's easy, "nimrod pretty --noTP myproject.nim" will do the job
21:45:25	nueva	babel failed to find libssl.so. library is installed. does nimrod program use predefined paths for searching libraries to use for FFI?
21:45:53	dom96	nueva: Maybe the .so has a different name
21:46:18	nueva	/usr/lib/i386-linux-gnu/libssl.so
21:46:35	nueva	ah. ok, it's my fault
21:46:51	nueva	sorry
21:47:03	vbtt	I have an elaborate idea.
21:47:51	dom96	well that suggests that something is wrong with 'devel'
21:48:00	nueva	I compiled Nimrod as an amd64 program but didn't install libssl:amd64
21:48:13	nueva	*compiled babel
21:48:16	vbtt	nimrod rename module1.TMyType MyType -o fixups.log
21:48:31	dom96	because xtagon cannot build ParticleBench on devel
21:48:34	vbtt	nimrod fixup module2,module3,module4 -i fixups.log
21:48:50	dom96	Guess I should test devel then
21:48:51	vbtt	-o is output, -i is input file
21:49:17	Araq	vbtt: no need, "nimrod pretty" can do that
21:49:18	xtagon	dom96, actually what I meant was I cannot build nimrod on devel
21:49:23	nueva	dom96: wait, I didn't proceed to testing ParticleBench yet
21:49:28	xtagon	dom96, on master I mean
21:49:34	xtagon	dom96, sorry, misread what you said.
21:49:49	Araq	master can build devel, devel can build master
21:49:59	Araq	but only since today
21:50:13	dom96	nueva: Yeah, sorry. xtagon tested it for me too.
21:50:17	dom96	I was referring to that.
21:50:57	vbtt	oh, pretty can save explicit renames specified by the user?
21:51:16	vbtt	and then apply those renames to other dependent modules later?
21:51:58	nueva	'babel install' failed with "command line(1, 1) Error: invalid command line option: '--noBabelPath'"
21:51:59	Araq	nah it works differently, you pass --pretty.overwrite and then pretty modifies the files directly
21:52:27	nueva	I mean, I'm installing the babel itself
21:52:37	Araq	welcome in the year 2014 where git tells you about the changes it made
21:52:51	EXetoC	the one and only babel
21:53:11	Araq	no need for a rename/fixup sing and dance
21:53:28	vbtt	Araq: what i'm suggesting allows piecemeal renames and also arbitrary renames ('Foo' -> 'Bar')
21:53:34	vbtt	but yeah, i said 'elaborate' :)
21:53:54	vbtt	it will be useful for software refactoring in general. not just for stdlib and not just for a one time conversion.
21:54:09	Araq	it will allow for arbitrary rename yes
21:54:24	Araq	but can we focus on the social aspect instead?
21:54:38	vbtt	what social aspect?
21:54:46	dom96	nueva: Sounds like you're running 0.9.2
21:54:47	Araq	do we release 0.9.4 and then say, for 0.9.6 run "nimrod pretty" over your entire code base?
21:55:12	dom96	Well in any case it's pretty much confirmed now that devel is faulty.
21:55:22	dom96	Araq: Your VM crashes on ParticleBench
21:55:25	vbtt	Araq: that's why i proposed fixup.log file. you just distribute this file with each release and when people upgrade they run 'nimrod fixup mymodules -i fixup.log'
21:56:04	Araq	vbtt: that's disgusting
21:56:36	nueva	dom96: Nimrod Compiler Version 0.9.3 (2014-01-17) [Linux: amd64]
21:56:44	Araq	a deprecation path is much more polite
21:57:00	dom96	nueva: babel uses the nimrod that's in your PATH. Is 0.9.2 in your PATH?
21:57:17	dom96	oh wait
21:57:22	nueva	dom96: output is from 'nimrod -v'
21:57:35	dom96	I guess Araq never merged master into devel?
21:57:43	Araq	I did
21:58:18	dom96	nueva: Git pull then
21:58:37	vbtt	why? it'll work when you make other changes to the stdlib as well. it doesn't preclude a deprecation path. it's just that you use a generic tool rather than manual or specific one.
21:58:54	vbtt	but i understand it needs extra effort cos the tool doesn't exist.
21:59:09	Araq	the tool does exist
21:59:21	Araq	I made 100K LOC case consistent with it
21:59:40	nueva	dom96: ok, but I'm right now fixing case sensitivity problems on current devel head to boot with rare options
21:59:56	vbtt	pretty is different tool from what I'm describing. lets say you decide to go through the stdlib and rename a few functions for consistency. now what? how do you get everybody to fix that up? pretty?
22:00:07	dom96	nueva: No worries. xtagon already tested it for me so it's fine.
22:00:37	Araq	vbtt: ah I see
22:01:05	xtagon	vbtt, I like your fixup idea
22:01:12	Araq	but yes, pretty will learn how to do it
22:01:51	*	brson quit (Ping timeout: 252 seconds)
22:01:53	vbtt	Araq:pretty will learn specail renames? such as 'pipe' -> 'socket' ?
22:02:05	vbtt	or 'parsecsv' -> 'csv' ?
22:02:07	Araq	that's easy to do
22:02:32	Araq	the problem is that it needs to be done in a batch
22:02:52	Araq	you can't rename it in the stdlib and then later in your project
22:03:19	vbtt	of course you can. i only need to rename it in my project when i upgrade to the new version of the stdlib.
22:03:40	*	brson joined #nimrod
22:03:46	Araq	because it's not clear if foo in your project refers to stdlib.foo
22:04:16	vbtt	what? it can be statically resolved. i'm not talking about textual replace.
22:04:17	nueva	should I rename globalError to GlobalError or vice versa?
22:04:37	Araq	but you can compile against the old stdlib and then it's obvious
22:04:40	vbtt	i'm talking about parsing the files, resolving symbols and semantic rename.
22:04:49	Araq	I know
22:05:38	Araq	you can't do a semantic rename when your code doesn't compile anymore
22:05:49	Araq	that's the problem
22:06:08	Araq	so you need to compile against the old stdlib so all names can be resolved
22:06:12	Araq	and then you can rename
22:06:49	vbtt	can you instead lookup the 'fixups.log' file to find the old symbols?
22:07:03	Araq	nueva: globalError but where do you see GlobalError?
22:07:14	nueva	Go has a similar tool http://golang.org/cmd/fix/
22:08:11	Araq	oh really? how come you think I don't know?
22:08:23	nueva	Araq: http://pastebin.com/VshmpEcV
22:08:33	vbtt	Araq:another idea. a deprecated statement. you rename TType to Type and then write depreated TType = Type. so the compiler can resolve it to the new symbol and emit a warning.
22:09:08	Araq	nueva: almost all of that is dead code
22:09:32	Araq	vbtt: we have a deprecated pragma and it also works for types
22:09:56	nueva	Araq: I don't promote Go or my extensive knowledge. just making a sidenote.
22:10:33	vbtt	actually deprecated is the wrong word. i'm suggesting obsolete but auto linked to the new version.
22:10:46	vbtt	obsolete TType = Type
22:11:01	Araq	what's the difference between obsolete and deprecated?
22:11:05	vbtt	now when you compile a dependent module, it looks up TType, find it, emits a warning, but links to Type instead.
22:11:22	vbtt	Araq:minor difference, deprecated still work.
22:11:27	Araq	I mean in english
22:11:32	nueva	Araq: not so dead. it prevents 'koch boot -d:release -d:tinyc -d:useGnuReadline -d:useFFI -d:nativeStacktrace'
22:11:57	vbtt	deprecated = going away soon. obsolete = gone already.
22:12:11	Araq	ah
22:12:25	vbtt	i think. english is not my absolute 1st language. more like 1.5th.
22:13:18	EXetoC	yes
22:13:21	vbtt	in english, actually, deprecated means discouraged.
22:13:25	Araq	nueva: are you on devel?
22:13:37	vbtt	but in programming it means what i said.
22:13:46	EXetoC	well that too, and some things stay deprecated forever
22:15:10	nueva	Araq: yes
22:16:39	Araq	nueva: you can also simply get rid of --cs:partial in nimrod.nimrod.cfg
22:20:00	nueva	internalAssert or InternalAssert? I think I got pattern and it should be internalAssert (because it's a proc)
22:20:48	nueva	or it's not a proc?
22:21:37	Araq	it's internalAssert
22:21:52	nueva	it's template. so it's still internalAssert
22:21:57	Araq	just run nimrod pretty over it with all your -d stuff
22:22:19	Araq	and use --pretty.overwrite
22:25:16	Araq	btw does -d:tinyc still work?
22:25:28	Araq	worked for me years ago
22:26:09	Araq	and then the gnu headers kept getting stranger
22:27:07	nueva	I don't use it, just compilng with it enabled :)
22:27:37	*	vbtt quit (Quit: http://www.kiwiirc.com/ - A hand crafted IRC client)
22:27:45	Araq	btw the idea to *share* builtin header files between compilers is ... bizzare
22:29:10	nueva	oh, well... error: no such file or directory: 'compiler/nimcache/libtcc.o'. is it tinyc related?
22:32:48	nueva	yeah, it's tinyc-related
22:52:07	*	BitPuffin joined #nimrod
23:04:16	Araq	good night
23:12:25	*	darkf joined #nimrod
23:17:12	nueva	dom96: ParicleBench building on devel: "SIGSEGV: Illegal storage access. (Attempt to read from nil?)"
23:17:24	dom96	nueva: Yeah, same results here.
23:17:30	dom96	Appreciate you testing though
23:19:42	nueva	dom96: full log http://pastebin.com/YYHykB17 (ust for the record)
23:22:13	*	vbtt joined #nimrod
23:30:57	*	ddl_smurf joined #nimrod
23:30:59	dom96	Good night
23:35:01	*	io2 quit (Ping timeout: 272 seconds)
23:43:22	Varriount	Meep
23:44:26	vbtt	beep
23:44:41	renesac	Error: execution of an external program failed; rerun with --parallelBuild:1 to see the error message
23:44:42	renesac	FAILURE: Execution failed with exit code 256
23:44:56	renesac	<-- this is running with --parallelBuild:1
23:46:01	vbtt	try parallelBuild:0
23:46:17	renesac	same failure
23:48:37	renesac	I'm trying to compile ParticleBench with the latest nimrod master
23:51:57	renesac	ok, changed a var to a const and now it compiles
23:51:58	renesac	\o/
23:52:26	renesac	(well, uncomented a line too