<< 27-05-2014 >>

---

00:03:21	*	EXetoC quit (Quit: WeeChat 0.4.3)
00:17:49	*	flaviu1 quit (Ping timeout: 240 seconds)
00:32:50	*	q66 quit (Quit: Leaving)
01:06:54	*	nande_ joined #nimrod
01:12:54	*	hoverbear joined #nimrod
01:13:29	*	flaviu1 joined #nimrod
01:13:30	flaviu1	dom96: Sorry, couldn't resist :P
01:17:33	Skrylar	hrmm
01:17:40	Skrylar	Sublime 3 still isn't out? o_O
01:19:05	caioariede	hi guys! can someone give me some light? I'm getting "out of memory" error here: https://gist.github.com/caioariede/9986786099eef53e093a#file-gistfile1-nim-L72
01:19:35	caioariede	I think I'm missing something but I can't figure out what I have to do
01:21:30	flaviu1	caioariede: No idea, but try dumping the object using repr
01:22:08	flaviu1	echo(repr(tpl)) will print out the object graph for tpl
01:24:06	caioariede	repr entered a infinite recursion loop :o
01:27:48	flaviu1	Yeah, it does that sometimes. I don't know about the layout of your thing, but line 68 looks suspicious.
01:27:55	*	def- joined #nimrod
01:29:08	flaviu1	since in the other cases you do something like getInt or getString, but in that case you just do .len without any getSeq. I think that should just crash though...
01:31:21	caioariede	it's not the case, the error only happens when I pass ctx to another proc
01:31:43	caioariede	it works otherwise: https://gist.github.com/caioariede/9986786099eef53e093a#file-gistfile1-nim-L84
01:32:12	*	Jesin joined #nimrod
01:32:31	caioariede	it seems to get collected by the gc or something like this
01:34:30	flaviu1	GC sounds reasonable. "TAny keeps a non-traced pointer to its wrapped value and must not live longer than its wrapped value"
01:36:43	fowl	well tany is not meant to be passed around like that
01:38:03	caioariede	what I could do to keep it alive?
01:38:22	flaviu1	caioariede: What is your end goal?
01:40:27	caioariede	I need a key-value dictionary where key is a string and values are TAny, and pass it through procs without losing them
01:42:54	fowl	caioariede, TAny is like a view on an object, thats why it takes a var T to instnatiate it, and why the TAny cant live longer than the object it views
01:44:53	fowl	caioariede, if all you need is a small subset of data types i would just use PJsonNode from the json module
01:45:00	flaviu1	I don't know how Araq'd do it, but the way I'd do it as a variant type.
01:45:23	flaviu1	fowl: He also wants to store regular object refrences
02:12:00	*	freezerburnv quit (Quit: freezerburnv)
02:32:56	fowl	not sure
02:48:17	*	bjz joined #nimrod
02:55:49	*	bjz quit (Ping timeout: 252 seconds)
03:09:19	*	xenagi joined #nimrod
03:11:05	*	def-_ joined #nimrod
03:14:14	*	def- quit (Ping timeout: 252 seconds)
03:22:20	*	flaviu1 quit (Ping timeout: 255 seconds)
03:31:30	*	xenagi quit (Read error: Connection reset by peer)
03:40:04	Demos	Varriount: ping
03:47:47	*	flaviu1 joined #nimrod
03:48:03	*	flaviu1 quit (Remote host closed the connection)
04:16:29	*	OrionPK joined #nimrod
04:34:11	*	caioariede quit (Ping timeout: 252 seconds)
04:51:31	*	xtagon quit (Ping timeout: 276 seconds)
05:04:56	*	New-Guy joined #nimrod
05:05:06	New-Guy	Howdy fellas.
05:05:30	New-Guy	Sorry about all the questions I've had lately. Some of which easily answered.
05:06:31	New-Guy	But got another question - I tried making a type T_Flag_Incomplete = tuple[symbol: expr, value: expr], but the compiler complains that expr isn't a concrete type.
05:07:27	New-Guy	I can just use tuple[symbol: expr, value: expr] in my templates, but I wanted to keep it clean, so does the compiler support a static type like that?
05:07:49	New-Guy	Again, I tried looking around and a few different ideas, but couldn't find anything.
05:07:52	New-Guy	Thanks in advance.
05:15:02	*	Demos quit (Read error: Connection reset by peer)
05:17:09	fowl	expr doesnt exist at runtime
05:23:00	New-Guy	fowl: And no compile-time only types?
05:34:09	fowl	New-Guy, PNimrodNode
05:35:43	fowl	New-Guy, templates are substitution-based, like #defines, it doesnt make sense to store an expression in the context of a template, you want macros
05:36:15	New-Guy	Too simply to warrant a macro.
05:36:42	Skrylar	New-Guy: compile-time only types are implicit :P
05:36:53	Skrylar	If you only use the type in macros, it gets cut out by dead code elimination AFAIK
05:37:05	New-Guy	Yeah, no problem.
05:38:00	fowl	New-Guy, its not simple if you're doing anything more than substitution
05:38:19	New-Guy	All of the templates making use of the types are one line long. Simply substitution.
05:41:45	New-Guy	Like I said, I don't need to use types, it was just cleaner that way. Thanks for the explanation, though.
05:46:16	*	hoverbear quit ()
05:47:00	New-Guy	So how are you guys doin'? Don't see much casual chat around here.
05:47:34	Skrylar	right now is largely sleeptime
05:48:20	New-Guy	Ah, only 2 am over here.
05:57:29	fowl	New-Guy, we talk more off topic on #nimrod-offtopic
05:57:51	New-Guy	fowl: Ah, I see. That explains it.
06:06:33	Skrylar	right now its just me ranting about how google improperly implements imap over there <_<
06:07:13	NimBot	Araq/Nimrod devel ee0cbbe Charlie [+0 ±1 -0]: renamed UINT in windows.nim to WINUINT
06:07:13	NimBot	Araq/Nimrod devel e02abc6 Varriount [+0 ±1 -0]: Merge pull request #1233 from barcharcraz/fixedWindows... 2 more lines
06:09:59	*	Varriount|Mobile joined #nimrod
06:10:26	Varriount|Mobile	Hi New-Guy
06:11:48	Varriount|Mobile	New-Guy: It's true - most of the developers here live in Europe
06:12:34	New-Guy	Varriount: Ah, so it's dead in the middle of night for you guys.
06:15:43	*	boydgreenfield joined #nimrod
06:17:15	Varriount|Mobile	New-Guy: Well, I live off the East coast of the US, but yeah. Here it's 2:16 in the morning, and for the others it's 4-6
06:18:13	Varriount|Mobile	I'm just awake because of my weird sleeping habits
06:19:56	New-Guy	Varriount: Yeah, East Coast and bad sleeping habits here as well.
06:32:26	*	Skrylar headscratches
06:32:34	Skrylar	its always weird when its cheaper to buy a new license than buying an upgrade
06:34:38	*	bjz joined #nimrod
06:40:17	New-Guy	Skylar: What is it?
06:43:37	Skrylar	New-Guy: upgrading a mail client
06:44:38	Skrylar	i'm one of those weirdos who still uses desktop mail clients
06:50:56	Skrylar	sleeeeeeep time.
06:51:10	Skrylar	i'll go back to finishing up the nimrod texture loader tomorrow afternoon
07:01:48	*	New-Guy quit (Ping timeout: 240 seconds)
07:14:38	*	ehaliewicz quit (Remote host closed the connection)
07:20:35	*	boydgreenfield quit (Quit: boydgreenfield)
08:11:19	*	kunev joined #nimrod
08:26:08	*	Varriount-Mobile joined #nimrod
08:29:49	*	Varriount|Mobile quit (Ping timeout: 240 seconds)
08:30:18	*	Varriount-Mobile quit (Ping timeout: 240 seconds)
09:10:49	*	core-ix joined #nimrod
09:13:16	def-_	Hi
09:13:26	def-_	Is this supposed to compile?: var x: proc (x, y: bool): bool = `and`
09:42:49	*	nande_ quit (Read error: Connection reset by peer)
09:49:43	*	freezerburnv joined #nimrod
10:05:32	dom96	def-_: I think it should.
10:05:36	dom96	def-_: report it on github
10:22:40	*	dLog_ joined #nimrod
10:24:19	*	dLog quit (Ping timeout: 240 seconds)
10:53:49	Varriount	Good morning dom96
11:00:15	*	freezerburnv quit (Quit: freezerburnv)
11:00:29	dom96	hey
11:12:32	*	EXetoC joined #nimrod
11:31:31	*	dymk quit (Ping timeout: 240 seconds)
11:31:47	*	freezerburnv joined #nimrod
11:34:02	EXetoC	so did implicit symbol binding ever become the default? the manual claims that explicit binding is rarely needed now
11:37:05	EXetoC	only this is being considered "proc fail[T](x: varargs[T, `$`]) = discard" when unittest.test invokes 'fail', which is also defined as "template fail*" in unittest.
11:40:18	EXetoC	it's a little cryptic if you're trying to call the former, and you don't know that the latter exists: "Error: cannot instantiate: 'T.
11:41:11	EXetoC	explicit binding in the unittest module solves this
12:13:08	*	untitaker quit (Ping timeout: 258 seconds)
12:19:07	*	untitaker joined #nimrod
12:34:52	Araq	def-_: no, this shouldn't work, 'and' is a builtin
12:35:12	Araq	and in general a 'procvar' annotation is necessary for this to compile
12:50:49	BitPuffin	hey is this getting merged or what? https://github.com/Araq/Nimrod/pull/1228
12:51:32	BitPuffin	Araq: do we want the ability to use cast as a.cast[type] to be consitent with generic procedure invokations?
12:53:20	Araq	no we don't want that
12:53:39	Araq	BitPuffin: if it makes you happy I'll merge it. Reads like Javadoc though
12:54:15	Araq	public Foo(Bar bar) // creates Foo using Bar bar.
12:55:43	BitPuffin	Araq: well not all of it, but it's better than nothing :)
12:56:29	Araq	I dunno. If you don't want to read, how does MORE text help?
12:56:36	BitPuffin	I do want to read
12:56:57	BitPuffin	I'm all for being as clear as possible
12:57:30	BitPuffin	parMessage is a good example of something that needs clarification
12:57:43	Araq	so rename it instead
12:57:59	BitPuffin	yeah that'd probably be good
13:00:02	BitPuffin	What's a good name.. echoParMsg?
13:00:37	Araq	finding good names takes longer than writing it in the first place
13:00:49	BitPuffin	yes
13:00:51	BitPuffin	agreed
13:00:54	BitPuffin	but it's worthwhile
13:02:07	BitPuffin	I'd say something like echoParserMessage or printParserMessage but I have a fealing you would think that's too much to type :P
13:02:43	Araq	typing is not my major concern, I'm pretty good at typing
13:02:49	Araq	I don't want to *read* it
13:03:08	Araq	I want the compressed versions of things
13:03:28	Araq	so that I can overview much and have a better chance of spotting the bugs
13:04:30	Araq	anyway you convinced me
13:04:38	BitPuffin	well you have the advantage of already knowing what everything (but zahary's code) does. Us others who want to help spot bugs need to be able to guess what is happening without knowing the details
13:04:48	Araq	and we should have a renamefest
13:04:52	Araq	however
13:04:53	BitPuffin	I think we should
13:05:07	Araq	that means cherrypicking becomes harder
13:05:14	BitPuffin	elaborate
13:05:15	Araq	and branches will cause many conflicts
13:05:25	Araq	I have 2 branches I'm working on
13:05:41	Araq	plus we have the master vs devel distinction
13:05:42	BitPuffin	well, not if you merge it back into your wip branches
13:05:51	BitPuffin	that's usually what I do
13:06:07	BitPuffin	to avoid the conflicts
13:06:18	BitPuffin	and yes, it will be work, but it is worth it
13:06:39	BitPuffin	the result will hopefully be more PRs :)
13:06:56	Araq	I agree
13:07:00	BitPuffin	Araq: would you be opposed to moving for example the code generators in to its own subdirectory, and the vm, to get more organization?
13:07:18	Araq	BitPuffin: that's what I have in mind too :P
13:07:23	BitPuffin	okay nice!
13:07:31	BitPuffin	I thought you hated subdirectiories :)
13:07:48	Araq	I do, but I'm willing to attract fresh blood
13:07:53	BitPuffin	haha
13:08:11	BitPuffin	well again, they can be painful when you already know what every file is etc
13:08:20	BitPuffin	but not for someone trying to grok what is what
13:08:59	*	dymk joined #nimrod
13:09:07	Araq	well I think nobody uses 'spawn' yet
13:09:16	Araq	so I might as well merge my branch into devel
13:09:50	BitPuffin	new_spawn?
13:10:46	BitPuffin	what's new about it :P
13:11:43	BitPuffin	also when are we letting stuff like sempass2 become just sempass
13:13:17	Araq	hey, it's the 2nd semantic pass
13:13:27	Araq	why should it become sempass ?
13:13:30	BitPuffin	oh
13:13:39	BitPuffin	I thought it was a "new" semantic pass
13:13:43	BitPuffin	to replate the old one
13:13:44	BitPuffin	haha
13:13:46	BitPuffin	:D
13:14:07	BitPuffin	well okay docgen2 then
13:14:12	BitPuffin	that's what I think it is isn't it?
13:14:30	BitPuffin	yeah by the looks of it it is
13:17:32	BitPuffin	while we are at it renaming things we should try and go through the stdlib and rename things to be consistent everywhere, and deprecate the old ones, or something like that
13:18:05	BitPuffin	watch the world burn, and things
13:27:34	*	core-ix left #nimrod ("Leaving...")
13:29:19	*	Demos joined #nimrod
13:30:49	*	darkf quit (Quit: Leaving)
13:32:13	Araq	BitPuffin: new_spawn is a branch which has an improved thread pool and 'spawn' supports return types
13:32:32	Araq	the return value is transformed into a promise automatically
13:33:28	Araq	furthermore it adds a 'parallel' statement which makes Cilk look like a toy :P
13:55:07	*	Demos quit (Ping timeout: 240 seconds)
14:13:15	*	caioariede joined #nimrod
14:29:02	*	nande_ joined #nimrod
14:29:59	*	Jehan_ joined #nimrod
14:30:58	BitPuffin	Araq: awesome
14:31:14	BitPuffin	Araq: improved thread pool eh, goodbye go
14:31:23	*	bjz quit (Ping timeout: 252 seconds)
14:33:39	Jehan_	Araq: Why rename futures to promises? Unless they're actually promises now?
14:34:53	Araq	Jehan_: we already have Future for the async stuf
14:35:02	Araq	Promise is for inter thread communication
14:35:16	Araq	I'm not happy with the names but thought long about better ones
14:35:21	Araq	and didn't find any
14:35:22	Jehan_	Yeah, but it's confusing if it's not what promises are generally understood to be.
14:35:39	Araq	well it's not too late
14:35:46	Araq	so make a suggestion
14:35:50	Jehan_	Mind you, I'm also not happy with research slang becoming actual programming language vocabulary … :)
14:36:05	Jehan_	Depends on what a promise currently is?
14:36:22	Jehan_	I haven't looked at the code yet, I just scanned through the IRC logs.
14:36:39	Araq	it's a ptr with a condition variable and a pointer to the owning thread
14:36:46	Araq	you can read from it
14:37:03	Araq	this blocks and transfers the value into your heap
14:37:04	Jehan_	Umm. That's the implementation, it doesn't tell me what the semantics are?
14:37:19	Jehan_	So, basically, it is a future?
14:37:49	Araq	I don't enough official defintions to tell
14:37:55	Araq	but yes, I think so
14:38:18	Jehan_	Hmm. I'll have to have a look at the code, I think.
14:38:27	Araq	it's also very close to a dataflow variable
14:39:41	Jehan_	I've generally been in favor of using "task" instead of "future", like what C# does (assuming that it's actually appropriate, technically not all futures are tasks, though 99% of implementations are).
14:40:03	Jehan_	Task is a term that people can understand fairly easily, unlike future (which is just CS slang).
14:40:09	Araq	"task" doesn't convey the right idea
14:40:27	Jehan_	That may be, I'll have to look at the actual code to see what it does precisely.
14:41:20	Araq	a task would be some function pointer + data (closure) that you pass to the thread pool and tell it "work on that"
14:41:46	Araq	promise is simply a wrapper around the returned result
14:42:31	Araq	Delayed[T] would be an alternative name but that's not even a noun
14:43:13	Jehan_	Araq: Okay, any need to expose that part to the user?
14:43:31	Jehan_	I mean, the usual purpose of a future is to have it filled in by some computation.
14:45:37	Jehan_	Wait, is what you have actually a synchronization variable?
14:46:04	*	flaviu1 joined #nimrod
14:47:13	*	freezerburnv quit (Quit: freezerburnv)
14:47:31	Jehan_	Brb.
14:48:24	Araq	Jehan_: yeah but it can only be assigned once and read once
14:51:25	BitPuffin	Araq: is there a way that I can cheat with the generics and implement the matrix dimensionality using peano numbers (to work around generic limitations) but still have the same api, ie where you just say var m: TMatrix[int, 3, 3]
14:51:34	BitPuffin	TRM maybe?
14:53:08	BitPuffin	I hardly think anyone would be happy to type var m: TMatrix[int, TSucc[TSucc[TSucc[TZero]]], TSucc[TSucc[TSucc[TZero]]]]
14:53:26	Araq	you can hide it with a macro
14:53:31	Araq	but it remains a bad idea
14:53:52	BitPuffin	yeah I don't like it
14:53:59	BitPuffin	but I think it might not freak out
14:54:07	BitPuffin	like static[T]'s atm
14:55:21	Jehan_	Araq: Yeah, but then it's really not a future, either. More a low-level construct that can be used to emulate futures, synchronization variables, etc. (at least in a limited way).
14:56:34	Jehan_	I guess after all, promise may still be the closest approximation to what it actually is. :)
14:57:02	Jehan_	I still don't like the terms "future" and "promise", regardless of their accuracy.
14:57:12	Araq	I even thought about naming it Doom[T] ...
14:57:26	BitPuffin	lol doom
14:57:28	BitPuffin	:D
14:57:39	Araq	but programmers have no humor
14:58:17	BitPuffin	guess I am no programmer then
14:58:19	Araq	(we already have gorge and slurp and nobody likes these names)
14:58:19	Jehan_	Hmm, DeferredValue? Too long?
14:58:26	flaviu1	Araq: Have you seen caioariede's question? Even though he isn't here, I'm curious how you'd implement that
14:58:42	Jehan_	Araq: Well, my only problem with them is that they pollute the main namespace. :)
15:02:25	Araq	flaviu1: dunno, perhaps I would hack the TAny implementation to support copying. but I usually use strings for C#'s Object root
15:04:19	*	kunev quit (Quit: leaving)
15:06:17	Araq	Jehan_: I should name it a GoVariable :P
15:06:30	OrionPK	araq - how would i invoke a C++ static member function in nimrod
15:06:31	Jehan_	Araq: Heh. :)
15:06:56	Araq	OrionPK: simple. wrap it like this: importc: "Foo::bar"
15:07:00	Jehan_	Araq: Or SingleElementNonReusableChannel. :)
15:07:05	OrionPK	ahh, beautiful thanks
15:07:25	OrionPK	araq what do you think of a library that makes a kind of DSL for defining C++ interfaces
15:07:44	OrionPK	as an alternative to pragmas
15:08:16	Araq	lets see: C++? check. DSL? check. Interface? check.
15:08:43	Araq	--> awesome idea.
15:08:58	OrionPK	also for C
15:09:02	OrionPK	Im thinking 2 libs
15:09:16	Araq	you need more buzzwords
15:09:34	OrionPK	lol
15:10:34	Araq	Jehan_: using a *channel* for synchronization is already a misnomer
15:10:46	Araq	when you think about it
15:10:49	Jehan_	Why?
15:11:05	Jehan_	It has an infinite consensus number, similar to CAS.
15:11:06	Araq	well ok, not really a misnomer
15:11:49	Araq	but a channel is almost a queue and in my book message passing is inherently asynchronous
15:12:03	Araq	maybe I'm weird
15:12:30	OrionPK	araq e.g. http://pastebin.com/KvUhHSK8
15:12:44	OrionPK	generates the importcpp/importc's necessary
15:12:47	Jehan_	channels can block on read (and write, if they have finite capacity).
15:12:50	Araq	OrionPK: I like it. make it happen
15:13:10	OrionPK	araq alrighty, I'll work on it as I have time. I started on it a couple days ago
15:14:27	Jehan_	In fact, unbounded channel functionality is a superset of semamphore functionality.
15:14:40	Jehan_	A semaphore is equivalent to a channel of unit.
15:15:15	Araq	a channel with a capacity of 1 is weird, IMHO
15:15:17	flaviu1	Jehan_: Got it. SuperSemamphore? :P
15:15:47	Jehan_	A channel with a capacity of 1 is actually a pretty useful construct for a number of things. :)
15:16:23	Jehan_	flaviu1: sem_post() is equivalent to write(chan, ()), and sem_wait() is equivalent to read(chan).
15:16:25	Araq	yes, but it's a weird "channel"
15:18:29	Araq	or maybe not ... I conflate channel with queue
15:19:08	*	Roin_ quit (Ping timeout: 252 seconds)
15:22:17	*	freezerburnv joined #nimrod
15:26:11	Jehan_	A one-element channel can be used to emulate a mutex. :)
15:26:22	Jehan_	write() = lock(), read = unlock().
15:26:55	Jehan_	Channels are in principle an enormously powerful synchronization mechanism.
15:27:25	Jehan_	Also in general less efficient than more specialized constructs, of course. :)
15:28:00	Araq	yeah and that's even preferable because you shouldn't use shared memory to communicate but use communication to share memory ...
15:28:15	*	Araq is confused
15:28:40	Jehan_	It depends on what you're trying to accomplish.
15:28:47	*	boydgreenfield joined #nimrod
15:29:10	Araq	I'm trying to recall Go's claim
15:30:05	Jehan_	Araq: Go's claim is bogus.
15:30:16	Araq	oh really?
15:30:20	Araq	now that's news :P
15:30:38	Jehan_	They say that they use channels to ensure that only one thread (well, goroutine) has access to data at any time.
15:30:50	Jehan_	The problem is that their channel implementation doesn't ensure that at all.
15:31:00	Jehan_	Unless they fixed it since I last looked at it.
15:31:03	Araq	yeah I'm aware
15:31:41	Araq	also they make it very easy to pass closures to goroutines
15:32:16	Jehan_	Well, they assume that they have a shared memory GC, which DOES make such things easy.
15:32:25	Araq	heck they encourage it as a goroutine cannot return anything
15:32:34	Jehan_	Of course, it's not easy to write an efficient shared memory GC.
15:32:45	Araq	that is not the problem
15:33:06	Araq	closures enable innocent looking sharing like nothing else
15:33:39	Araq	it's an invitation to data races
15:34:46	Jehan_	You can make that safe.
15:35:27	Jehan_	Data races are in principle trivial to avoid.
15:35:54	Jehan_	Ensure that (1) each memory location is associated with exactly one lock and (2) ensure that the lock has been acquired before accessing the location.
15:36:10	Jehan_	The only tricky part is making this efficient.
15:36:19	Jehan_	Which can be accomplished in a number of ways.
15:36:37	Araq	(2) is a huge problem
15:36:45	flaviu1	Jehan_: It can never be more efficient than not having a data race
15:37:18	Araq	I get your point
15:37:26	Araq	you can easily check for (2) at runtime
15:37:32	Araq	but this is overly restrictive
15:37:33	Jehan_	Araq: Yup.
15:37:44	Jehan_	Not so much restrictive as lazy. :)
15:37:57	Jehan_	But even a runtime check beats having a data race.
15:38:04	Araq	often I can ensure it's not racy without acquiring that lock
15:38:16	Jehan_	It's not like we statically check array accesses all the time.
15:38:22	BitPuffin	OrionPK: make sure that you mention webscale and big data somewhere in the lib
15:38:33	Jehan_	Araq: Yup, a lot of the stuff can be optimized away.
15:38:45	Jehan_	In fact, that's something that I'm working on for my own project.
15:39:00	Araq	it's like making me insert assert i < a.len for every a[i]
15:39:17	Araq	otherwise a[i] fails at runtime
15:39:36	Araq	but hmm
15:39:53	Araq	oh I see
15:40:12	Araq	you don't do it this way
15:40:17	Jehan_	Incidentally, there are ways to alleviate the problem with closures.
15:40:29	Jehan_	Such as making a copy of the environment.
15:40:44	Araq	instead you insert the acquire operation, right?
15:40:50	OrionPK	bitpuffin what? I dont see how it's buzzywordy :P
15:41:02	Jehan_	Araq: You mean like Eiffel does?
15:41:06	OrionPK	you guys are all talking about promises and futures, thats way more buzzwordy :P
15:41:33	Jehan_	No, I mean literally guaranteeing that there is a check before every memory access where I can't statically guarantee that it's safe.
15:41:49	Jehan_	The trick is to statically guarantee it is often as possible.
15:42:02	Jehan_	Which also necessitates a language design to support that.
15:42:28	Araq	copying the closure changes the semantics
15:43:01	Jehan_	Yup.
15:43:08	BitPuffin	OrionPK: seriously, big data and webscale not buzzwords? :P
15:43:13	Araq	Jehan_: ah so that's what Eiffel does :P
15:43:21	BitPuffin	cloud is not a buzzword
15:43:25	BitPuffin	BY THAT LOGIC
15:43:41	Araq	I barely remember
15:43:44	Jehan_	But only for non-functional languages, and one can argue that this is actually truer to the functional way of doing things.
15:44:04	Jehan_	Didn't C# have to fix its closure semantics recently because of a related problem?
15:44:22	Araq	yeah for some meaning of "related"
15:44:46	Jehan_	Araq: Eiffel (SCOOP) acquires any "separate" object that's passed as an argument to a routine when the routine is called and the object hasn't been acquired yet.
15:45:23	Araq	Eiffel's Scoop is fundamentally broken
15:45:33	Jehan_	Somewhat simplified, the actual semantics are a bit more involved (especially once you start considering precondition).
15:45:46	Jehan_	Araq: In what way?
15:46:00	OrionPK	BitPuffin I just want a less painful way to wrap all those C++ libraries out there ;)
15:46:03	Jehan_	The original spec had holes, but you could mostly infer the intent.
15:46:13	OrionPK	that has nothing to do w/ big data / web scale
15:46:21	Araq	they map existing stuff to some bizzare concurrenc model and claim it's simple because they ended up requiring few new keywords
15:46:32	Jehan_	I also think it tries being too hard being minimalist.
15:46:37	BitPuffin	OrionPK: yeah, we need to make a variant of c2nim that generates output for your lib :)
15:46:44	OrionPK	agreed :)
15:47:02	OrionPK	once I have a prototype working I'll put it up on github
15:47:03	Jehan_	Ah, yeah. I agree with that. Minus the bizarre part, the underlying concurrency model is mostly straightforward.
15:47:08	OrionPK	hopefully we can get some contributions
15:47:38	Jehan_	At the same time, it was about the only language at the time that actually tried avoiding data races.
15:47:50	Jehan_	MInd you, it still is only one of very few that actually does.
15:48:05	Araq	Jehan_: yeah, ok, that's a result of their rather poor documentation
15:48:12	Jehan_	Java and C# in particular are little better than assembly language in that regard.
15:49:13	Jehan_	Neither is C++, but then C++ doesn't aspire to be high level.
15:49:19	flaviu1	Wow, UTF8 is pretty nice, you don't even have to care that its utf8, it keeps all the weird stuff above 0x7F
15:49:33	Araq	flaviu1: yes.
15:49:35	Jehan_	flaviu1: Yeah.
15:49:48	Jehan_	There were a few articles recently about that, even.
15:50:26	Jehan_	Speaking of which, I'm happy that Nimrod defines a string as basically a sequence of 8-bit characters.
15:50:52	Araq	you're quite alone with this opinion
15:51:05	flaviu1	Jehan_: Do you have a link to one of those articles?
15:51:59	Jehan_	Araq: Possibly. I still think that burdening one of the core data types in a language unnecessarily with all the unicode semantics is inherently broken.
15:52:09	Araq	but I'm happy too and I know more about character sets than the people who dislike nimrod's solution
15:52:23	Jehan_	A language design should be a pretty static document and the Unicode standard is too much of a moving target, just for starters.
15:52:56	Araq	the unicode standard will be dumbed down eventually
15:52:57	Jehan_	Even if it weren't, weighing down the core datatypes with all the unicode stuff does not convey any benefits.
15:54:07	Jehan_	I'm honestly not sure what the benefits of strictly enforcing a constraint that each string is valid utf8 would be.
15:54:21	Jehan_	Or whatever it is that people want.
15:54:34	flaviu1	Why should it be an issue that the unicode standard changes? I doubt they make major changes all the time, so it just requires rerunning the code generation scripts.
15:55:07	Araq	flaviu1: the tables are HUGE though and growing
15:55:47	Jehan_	flaviu1: Just implement isUpper(), for example. That's actually pretty simple.
15:55:47	Araq	at the same time many human languages are moribund
15:55:56	flaviu1	The computer doesn't care, the code generation can deal with it.
15:56:07	Jehan_	Or try to benchmark StartsWith() in C#.
15:56:25	flaviu1	Jehan_: Implementing toUpper() might be a good learning experience.
15:56:27	flaviu1	Not arguing against strings being byte sequences, I like that.
15:56:58	Jehan_	Don't get me wrong, comprehensive Unicode support is a good thing to have.
15:57:21	Araq	toUpper itself is a flawed concept...
15:57:34	Jehan_	Araq: Yup. See ß and friends. :)
15:57:58	Jehan_	Which is why I specifically said isUpper(). :)
15:58:31	Araq	ah
15:59:22	Jehan_	I actually had that discussion with a Google interviewer once.
16:02:47	flaviu1	I'm not allowed an \n in a character? That's atypical, usually \n is the newline character, not \n\r or \r\n
16:03:41	Araq	flaviu1: we don't really abstract away the different newline styles and so don't need to distinguish between "binary" and "text" files
16:04:06	*	hoverbear joined #nimrod
16:04:37	Jehan_	flaviu1: I think \l is what you need instead.
16:04:52	flaviu1	I know, still a bit odd
16:04:52	Araq	no you should really use "\n"
16:05:26	Jehan_	Araq: can't use \n in a character literal.
16:05:43	Araq	Jehan_: yes. that means he shouldn't use a character literal.
16:05:49	Araq	but a string literal
16:06:09	Jehan_	Depends on the application.
16:06:16	flaviu1	Araq: Well, I need to use a char literal, so I'm using a '\l'
16:06:35	Jehan_	For parsing strings character by character, it'll have to be '\l'.
16:07:15	flaviu1	^ Exactly my usecase
16:09:55	Araq	no
16:10:06	Araq	for parsing strings you need to do something like:
16:10:29	Araq	if s[i] == '\C': inc i; if s[i] == '\L': inc(lineNumber); inc i
16:12:05	Jehan_	What's the difference between \L and \l?
16:14:37	flaviu1	Case insensitive I think
16:26:05	Skrylar	Araq: i had to make a binding for setjmp/longjmp because jpeg and png use those for error handling; do you want the module for those functions in std?
16:40:24	Araq	Skrylar: hmmm nimrod uses that too for its exception handling
16:40:32	*	shodan45 joined #nimrod
16:40:42	Araq	maybe we can make it compatible
16:41:30	Araq	but no, I don't want it in the stdlib, I already have those :P
16:43:33	*	Matthias247 joined #nimrod
16:46:57	Skrylar	i didn't see public ones when i was grepping :\
16:53:17	*	q66 joined #nimrod
16:53:17	*	q66 quit (Changing host)
16:53:17	*	q66 joined #nimrod
17:17:11	*	Johz joined #nimrod
17:17:45	*	icebattle quit (Remote host closed the connection)
17:21:35	*	shodan45 quit (Quit: Konversation terminated!)
17:31:11	*	brson joined #nimrod
17:40:38	Araq	Skrylar: well they are not public
17:47:32	*	hoverbea_ joined #nimrod
17:50:51	*	hoverbear quit (Ping timeout: 252 seconds)
17:54:24	BitPuffin	Alright I will do what nobody else has done
17:56:14	BitPuffin	that we've needed for a while
17:58:55	*	Joe_knock joined #nimrod
17:58:58	Joe_knock	Hello
17:59:34	BitPuffin	hey Joe_knock
17:59:37	BitPuffin	welcome if you are new
18:00:05	Joe_knock	Hello BitPuffin, thank you. Yes I am. I've come to Nimrod via Python
18:01:13	*	Demos joined #nimrod
18:05:31	Araq	hi Joe_knock welcome
18:06:07	*	Demos quit (Ping timeout: 245 seconds)
18:06:31	BitPuffin	I shall update the htmlgen and dom modules
18:06:41	BitPuffin	and possibly write a css module
18:06:58	Joe_knock	Araq: Thank you
18:07:16	Joe_knock	What is/are the current primary use-cases of nimrod?
18:07:55	BitPuffin	Joe_knock: it's general purpose
18:08:58	Joe_knock	BitPuffin: Can I find some type of list that demonstrates where nimrod is being used in production?
18:08:59	BitPuffin	you could write an operating system, a web browser, a web application, a game, a photo editor, AI, etc
18:09:07	BitPuffin	anything :)
18:09:21	BitPuffin	I dunno, is there a list currently?
18:10:49	dom96	http://forum.nimrod-lang.org is a good example
18:10:55	Joe_knock	I'm keen on learning a "close-to-machine" language and I like the idea of porting over my understanding of the python syntax
18:11:04	BitPuffin	dom96: no it's not because it looks like a ballsack :P
18:11:15	BitPuffin	and is terribly insecure
18:11:18	BitPuffin	but either way
18:11:24	dom96	BitPuffin: Hey, I designed it :(
18:11:32	BitPuffin	dom96: hugs?
18:11:52	dom96	BitPuffin: Nope. Never gonna forgive you for that </3
18:11:58	BitPuffin	dom96: your blog looks good though, did you design that?
18:12:05	dom96	BitPuffin: Yes!
18:12:11	BitPuffin	dom96: so you got better :)
18:12:30	Araq	it's more secure than most commercial software that I've seen, but that means nothing :-)
18:12:31	dom96	But nah I kid. I agree the forum design looks meh.
18:12:46	dom96	But I wanted to get it up and running quickly
18:12:49	BitPuffin	Araq: yeah sure :P
18:12:53	BitPuffin	I can imagine
18:12:54	Joe_knock	Is Araq Andreas?
18:12:57	BitPuffin	yes he is
18:13:06	Araq	it used to look cool but it didn't age well ...
18:13:08	BitPuffin	dom96: yeah it's obviously rushed :)
18:13:29	Araq	there are designs which keep looking good and then there are other designs
18:13:41	dom96	It's a practical design :P
18:13:45	BitPuffin	I can't imagine the forum ever looking good lol
18:13:49	dom96	It serves its purpose
18:14:10	Joe_knock	To you guys that designed the forum, if it user friendly over design, screw the design, ill choose something i can use better over a fancy bootstrap thing
18:14:25	dom96	See ^ :D
18:14:25	BitPuffin	well alright it's not the most shitty looking thing in the world
18:14:31	BitPuffin	some things are ok-ish
18:14:33	BitPuffin	:P
18:14:39	dom96	Time to add more <blink>!!!
18:14:45	BitPuffin	dom96: he said IF it's user friendly :P
18:15:21	Araq	Joe_knock: I'm afraid there is no list of users who use it in production and this list is very small anyway
18:15:21	dom96	I'll make it so that it only uses blink tags for your IP
18:15:33	Araq	*this list would be
18:15:35	dom96	It'll still be user friendly, but it'll also be BitPuffin unfriendly :D
18:15:40	Joe_knock	what is the status of concurrency with regards to nimrod? Are you guys building that in like the Go and Rust folk?
18:15:50	BitPuffin	dom96: it doesn't even have search :P
18:16:19	dom96	BitPuffin: Google.
18:16:22	Jehan_	BitPuffin: That's what Google is for. :)
18:16:54	Araq	Joe_knock: yes. it pushes the state of art. (Jehan_: feel free to disagree vehemently)
18:16:55	Jehan_	Honestly, the only thing I'd like to see improved is account management. Actually using the forum to, like, write stuff is just fine.
18:17:00	BitPuffin	that is literally the most pathetic reply I've gotten twice in a long time
18:17:07	Jehan_	Araq: Heh. :)
18:17:14	BitPuffin	and why does it tell me how fast it generated the page
18:17:29	BitPuffin	Jehan_: kind of a difficult thing to get wrong though
18:17:32	Jehan_	Araq: I'm not disagreeing, but that's a low bar. The state of the art is a disaster area. Anything remotely sane is an improvement.
18:17:40	dom96	BitPuffin: why not? Reddit does too
18:17:43	BitPuffin	<textarea></textarea> done
18:18:07	BitPuffin	dom96: the forum isn't reddit, and reddit doing it is stupid as well
18:18:17	BitPuffin	YAGNI principle dom96, YAGNI
18:18:18	dom96	BitPuffin: Don't diss reddit bro
18:18:32	BitPuffin	reddit sucks dick
18:18:34	BitPuffin	:P
18:18:47	Joe_knock	Well if you guys don't mind, it seems like the community is relatively small, so perhaps I could contribute positively and not be a cog who gets ignored concerning the future of other languages.
18:18:48	BitPuffin	all software and websites suck today okay
18:19:05	BitPuffin	Joe_knock: yeah that's the advantage of our community indeed
18:19:28	BitPuffin	Although Araq can be a grumpy cat at times, however, with good reason
18:19:38	dom96	Jehan_: Do you think Araq's promises should be merged with my async futures?
18:20:03	Jehan_	I honestly haven't looked at the async stuff yet, so I can't tell.
18:20:22	dom96	IMO users will find the subtle differences between the two confusing.
18:20:33	BitPuffin	true
18:20:36	Araq	the difference is hardly subtle
18:20:43	Araq	they are completely different things
18:20:45	Jehan_	My actual usage of Nimrod for practical work mostly involves batch processing, which takes a bunch of data, feeds it through an algorithm, and spits something out.
18:20:53	BitPuffin	dom96: did you publish the article yet btw?
18:21:05	BitPuffin	Jehan_: nimrod should be more than fine for that
18:21:07	Jehan_	So, sockets etc. are not a priority for me at the moment.
18:21:13	Araq	the only confusing thing is that C++ uses the same terms and does it differently
18:21:14	BitPuffin	they exist though
18:21:19	Jehan_	BitPuffin: It is. That's why I'm using it.
18:21:19	BitPuffin	when you'll need those
18:21:23	BitPuffin	oh wait
18:21:27	BitPuffin	thought you were Joe
18:21:29	BitPuffin	haha
18:21:31	BitPuffin	:P
18:21:43	Joe_knock	aah crap, no tutorials for Jester either. Perhaps I should tinker and write a few.
18:21:48	BitPuffin	Araq: that is no good, we have to think of our own buzzwordy name
18:22:11	Araq	so make a suggestion. We all failed to come up with something better
18:22:16	Jehan_	BitPuffin: You don't have to sell me on it. I'm a person who uses Nimrod for actual paid work, after all.
18:22:17	dom96	BitPuffin: Araq has put an embargo on it :P
18:22:37	BitPuffin	Araq: hyperdimensional vow
18:23:09	dom96	Joe_knock: That would be awesome :)
18:23:12	BitPuffin	or maybe just vow
18:23:15	BitPuffin	lol
18:23:17	BitPuffin	:P
18:23:18	Joe_knock	Jehan_: You're obviously using Nimrod to process something equivalent to massive hordes of data?
18:23:36	Jehan_	Joe_knock: No, actually small amounts of data.
18:23:43	BitPuffin	Jehan_: well that's why I said I thought you were Joe
18:23:44	Joe_knock	oh shit, dom96 is the author of Jester haha
18:23:48	Jehan_	Mostly tooling for a project.
18:23:49	BitPuffin	athorwise I wouldn't be selling it to you
18:23:58	dom96	Joe_knock: indeed :D
18:24:10	BitPuffin	Araq: pledge
18:24:12	Joe_knock	What are the benefits that nimrod brings? Jehan_
18:24:36	Jehan_	Joe_knock: More productive than doing it in C++.
18:24:38	BitPuffin	anticipate
18:25:13	Jehan_	Joe_knock: I need a statically typed, imperative, garbage-collected language where I'm not drowning in boilerplate.
18:25:26	Jehan_	There aren't a whole lot of options in this area.
18:25:26	BitPuffin	Araq: prospect?
18:25:31	*	boydgreenfield quit (Quit: boydgreenfield)
18:25:51	Joe_knock	As long as it serves you well, Jehan_
18:25:56	BitPuffin	Araq: prognose?
18:26:04	BitPuffin	l :p
18:26:17	Araq	BitPuffin: 'plegde' is too positive; 'promise' has the benefit of implying "yeah ... right"
18:26:18	flaviu1	Does nimrod really have opportunities for boilerplate except for C bindings?
18:26:35	Joe_knock	Could an application that manages electronic flying devices be considered "soft real-time" ?
18:26:59	BitPuffin	Araq: what about vow then, that's short even
18:27:04	BitPuffin	so you should like it
18:27:18	Jehan_	flaviu1: I want to *avoid* boilerplate.
18:27:19	flaviu1	Joe_knock: If the electronic flying device doesn't get any commands for 200ms, will it crash?
18:27:44	flaviu1	Jehan_: I know, but I'm curious if there are any cases where you can't avoid it
18:28:08	BitPuffin	Joe_knock: what kind of flying device do you mean
18:28:09	Jehan_	flaviu1: It depends on your definition of boilerplate.
18:28:12	Joe_knock	flaviu1: Based on a flight height of at least 100 meters, no. Perhaps only take-off and landing will be the issue.
18:28:35	BitPuffin	Joe_knock: in the worst case you can also skip using a GC
18:28:36	Joe_knock	BitPuffin: Like an RC helicopter
18:28:56	flaviu1	soft real-time basically means that you configure an interval, and the GC limits its pauses to that interval. http://build.nimrod-lang.org/docs/gc.html has more detail
18:28:59	BitPuffin	that way you can even get the memory usage down if that's a scarce resource
18:28:59	Jehan_	flaviu1: I'm thinking of stuff like http://en.wikipedia.org/wiki/Rule_of_three_(C%2B%2B_programming)
18:29:36	Araq	Joe_knock: depends on your spec entirely. ask me in a few months again, then I can tell you more.
18:30:00	Joe_knock	flaviu1: So soft real-time is a provision, it isn't a constraint of nimrod?
18:30:27	Araq	would I personally use Nimrod for *hard* realtime systems? yes, but then I know what I do ;-)
18:30:27	BitPuffin	Araq: homework? XD
18:30:49	BitPuffin	"The Nimrod Homeworks"
18:31:08	BitPuffin	goal maybe
18:31:19	Joe_knock	I take it you guys are not yet experiencing oss politics.
18:31:31	BitPuffin	Joe_knock: hmm?
18:31:55	Joe_knock	BitPuffin: Eg. People disagreeing on every decision
18:32:17	BitPuffin	Joe_knock: we all disagree at times, but Araq is our almighty dictator so he always has the final say
18:32:22	flaviu1	Joe_knock: We have quite a bit of bikesheding, but we also have a dictator
18:32:49	Joe_knock	http://www.btbytes.com/notebooks/nimrod.html "For me, Nimrod is a vast upgrade over dynamic languages like Python and Ruby without sacrificing a whole lot of expressiveness." Fascinating
18:32:54	Jehan_	Until Guido runs him over with a bus or however that goes. :)
18:33:23	*	def- joined #nimrod
18:33:41	Joe_knock	Aah the great Guido.
18:33:41	flaviu1	Joe_knock: I'm not sure what you're asking, but you can get the GC to effectively take as long as it'd like by setting the max pause high.
18:34:08	flaviu1	Also, 200ms is quite a large overestimate, according to the docs it can do as low as 2ms pause times
18:34:37	Joe_knock	flaviu1: I assumed that when the site said soft real-time that it meant that as a constraint, but Araq has clarified that it can be used for "hard" real-time too.
18:34:38	Jehan_	flaviu1: There is a bit more explanation necessary.
18:35:17	*	nande_ quit (Read error: Connection reset by peer)
18:35:44	BitPuffin	Joe_knock: it can be used for anything, as you don't even *have* to use the GC
18:35:52	*	BitPuffin doesn't always use the GC
18:36:28	Araq	--gc:none gives warnings when you use a language construct that involves the GC
18:36:35	*	def-_ quit (Ping timeout: 252 seconds)
18:37:03	BitPuffin	plus for controlling a little helicopter or whatever wouldn't you know the environment you are working with with the sensors etc? So why would you need the heap
18:37:05	BitPuffin	:P
18:37:42	BitPuffin	I guess it depends on what sort of data you get back
18:37:53	BitPuffin	I'm not too familiar with coding for those
18:38:16	Jehan_	Araq: What would you say about Pending[T] instead of Promise[T]?
18:38:26	BitPuffin	that's a pretty good word
18:38:27	Joe_knock	Using the sensor data to re-calibrate the helicopter position would be important.
18:38:37	Joe_knock	But let me not get ahead of myself there :P
18:38:54	flaviu1	I didn't know that --gc:none warns, awesome
18:38:56	BitPuffin	Joe_knock: yeah but I mean when you poll from the sensor, do you always get the same amount of data?
18:39:07	Araq	flaviu1: that's new with 0.9.4
18:40:26	Joe_knock	Yes, that would be correct. The quantity of the data won't change, it's just getting the sensor data to return as many ping calls with feedback as possible
18:41:08	BitPuffin	Joe_knock: yeah, so you could just use the stack then, no GC needed
18:41:28	Araq	Joe_knock: when you have the memory just use the GC and be happy. I can't imagine it will cause problems for your use case.
18:42:15	Joe_knock	thanks for the feedback guys.
18:42:29	Araq	ping Varriount
18:42:31	flaviu1	Does RTTI still work with the GC off?
18:42:49	Araq	flaviu1: yes
18:44:09	*	boydgreenfield joined #nimrod
18:45:11	BitPuffin	Araq: should I just go crazy with the existing htmlgen module or should I make a html5gen module instead
18:45:40	Araq	the htmlgen module is widely used. don't touch it.
18:45:56	*	BitPuffin didn't think anyone uses it
18:46:04	BitPuffin	okay so html5gen it is
18:46:32	Araq	okay. I'll kill you if you use immediate templates/macros.
18:46:45	Araq	they caused lots of problems
18:46:48	BitPuffin	that's a bit harsch
18:47:02	Araq	fits your character
18:47:03	Jehan_	Heh. :)
18:47:20	BitPuffin	the only thing I'm doing is updating it to support all the html5 tags :P
18:47:22	flaviu1	Whats wrong with macros?
18:47:34	BitPuffin	Araq: I'll use generics of course
18:47:41	Jehan_	Immediate macros, he said, and I was wondering the same thing.
18:47:41	BitPuffin	and do fancy stuff
18:48:41	BitPuffin	fits my character to die o.O?
18:48:41	*	boydgreenfield quit (Ping timeout: 255 seconds)
18:49:09	BitPuffin	I actually barely use macros, dom96 is the guilty one there, and he's apparently not having problems, which I find hard to believe
18:50:06	Araq	nothing but *immediate* macros don't participate in overload resolution and exporting immediate macros like 'a', 'p' is asking for trouble
18:50:22	BitPuffin	ahahahaha
18:50:26	BitPuffin	I bet
18:50:58	BitPuffin	but what about using immediate macros to define the tags?
18:50:59	Jehan_	Yeah, I remember running into problems with trying to define an immediate template with varargs.
18:51:37	Araq	BitPuffin: never ever export an immediate symbol. what you do locally doesn't matter
18:51:46	flaviu1	BitPuffin: I've done stupider stuff with macros than dom96 has :P
18:52:16	BitPuffin	Araq: well why is everything in htmlgen module an immediate macro then XD
18:52:17	Joe_knock	"I'll kill you if you use immediate templates/macros." :O
18:52:29	Jehan_	Speaking of which, is it worth requesting an alternative syntax to "from foo import nil"?
18:52:32	BitPuffin	Joe_knock: see what I mean about grumpy cat? :)
18:52:37	Jehan_	Or is that set in stone?
18:52:40	Joe_knock	lol
18:52:51	BitPuffin	Joe_knock: I hope he doesn't talk like that to his kids
18:52:52	Araq	Jehan_: what's wrong with it?
18:53:28	Joe_knock	from ... import is pythonic
18:53:43	Joe_knock	"from ... import ..."
18:53:57	Jehan_	Araq: Umm, several things, starting with having to use a separate import statement for each module.
18:54:35	*	Jesin quit (Quit: Leaving)
18:54:43	Araq	so use your own privateImport macro :P
18:54:45	flaviu1	Jehan_: Some whitespace doesn't hurt anything
18:55:07	Jehan_	Araq: Yeah, that's where I ran into problems with variadic templates.
18:55:17	Jehan_	Which is why I remembered and mentioned it.
18:55:20	Araq	interesting
18:55:35	Jehan_	Basically, immediate templates can't have varargs parameters.
18:55:45	Araq	yep
18:56:00	Jehan_	And if I don't make them immediate, the compiler tries to resolve the module names.
18:56:07	Araq	you can use a macro though
18:56:31	Jehan_	Okay. I may, if I feel strongly enough about it again at one point.
18:56:52	Jehan_	I decided to resolve the situation back then with just copying and pasting the import statement.
18:57:12	BitPuffin	Araq: I think what I'll do is to create an immediate macro that creates exported templates or something
18:57:28	Araq	the compiler doesn't try to resolve module names anymore for 'expr' but the situation is ... hardly ideal
18:57:47	Jehan_	Heh.
18:57:50	flaviu1	Jehan_: The best solution may be to just put up with it. Macros for non-standard syntax is
18:57:54	flaviu1	't very good
18:57:59	*	boydgreenfield joined #nimrod
18:58:01	Joe_knock	Jehan_: I dont know if you're a fan of fancy text editors, but I'm sure by using Sublime Text, it can autofill those words for you
18:58:12	Jehan_	Well, it's hardly the most urgent problem in the world.
18:58:57	Joe_knock	Jehan_: when you say "import 1 module only" Do you mean that: "from foo import lib1, lib2, lib3, lib(x+1)" on the same line won't work?
18:59:02	Jehan_	Joe_knock: My concern was more with the readability of the code, less with having to type it.
18:59:17	Jehan_	Joe_knock: I was talking about "from foo import nil".
18:59:43	Jehan_	Which is sort of the same as "import qualified foo" in Haskell.
18:59:51	dom96	BitPuffin: write a new htmlgen module using the syntax Araq showed in his emerging langs talk
18:59:52	Joe_knock	What change would you propose? Jehan_
19:00:23	Jehan_	Note the "nil", which makes it different from: "from foo import x, y, z"
19:00:39	Jehan_	My complaint was about not being able to do "from foo, bar import nil".
19:01:13	Joe_knock	I see now. Okay that is a different situation
19:01:26	Araq	BitPuffin: I'm still surprised how good the old VM worked. And how much work the new VM was... but yeah, people run into macro bugs all the time.
19:02:22	Jehan_	Joe_knock: And it's … umm, not all that important. But I remembered that because of the immediate template stuff and since I had Araq around, I figured I might as well ask. :)
19:03:12	Araq	BitPuffin: I had to lookup "vow". can't be right for programming
19:04:03	Araq	and I'm afraid nobody except Jehan_ knows the definition of "promise" anyway :P
19:04:32	Jehan_	Araq: Yeah. As I said, I'm not a big fan of using CS slang.
19:04:47	Jehan_	Araq: The biggest offender in my mind is the type "unit".
19:05:01	Jehan_	Which makes a lot of sense if you're familiar with type theory, but totally confuses everybody else.
19:05:22	Araq	heh.
19:07:09	wan	How about "PinkieSwear"?
19:07:47	Araq	the biggest offender is C++'s "vector" for a flexible array
19:07:56	fowl	speaking of macro bugs
19:08:05	fowl	i made the vm crash yesterday
19:08:09	Araq	vectors in math are usually of a fixed size
19:08:23	Araq	fowl: so report it please
19:08:32	fowl	reporting issues is so 2013
19:08:44	fowl	i'll report it on twitter and reddit
19:08:46	Araq	yeah, report it on the forum instead
19:09:33	Araq	but start the post with "argh I forgot my password so I had to re-register"
19:10:32	fowl	lol
19:10:48	Jehan_	Araq: vector isn't CS slang, though, it's just a misnomer.
19:10:52	flaviu1	Araq: Shouldn't nimrod's arrays be called vectors then? :O
19:11:34	Jehan_	I'm talking about terms that were developed in some paper in a context where they make sense, but are difficult to understand outside that context.
19:12:12	Jehan_	As to fixed size, umm, take K[[x]] as a vector space. :)
19:12:50	Jehan_	Or, well, just K[x].
19:13:26	Joe_knock	I remember vectors, that would be linear algebra
19:14:22	*	flaviu1 quit (Remote host closed the connection)
19:14:44	fowl	sweet
19:14:47	fowl	i got issue #1234
19:16:07	*	flaviu1 joined #nimrod
19:17:56	Araq	fowl: you're the typical programmer :P using every single feature that nimrod offers. :-)
19:21:09	*	brson quit (Ping timeout: 252 seconds)
19:35:30	flaviu1	Any thoughts on syntax like `doSomething(_ + 1)` for lambdas?
19:35:39	*	onibaka joined #nimrod
19:35:56	Araq	hi onibaka welcome
19:36:13	Araq	flaviu1: we already have enough syntax for lambdas
19:36:33	Araq	more sugar is the point of the upcoming => macro
19:37:28	flaviu1	I know, but I don't see how `_ + 1` could be implemented without compiler support
19:39:17	Araq	you can always put it into a string literal, but yeah you're right
19:39:24	Araq	that's currently not possible
19:49:05	*	boydgreenfield quit (Quit: boydgreenfield)
19:56:16	Jehan_	flaviu1: If you want short inline expressions, the sequtils.nim approach may be an alternative.
19:56:26	flaviu1	hmm, 'got (proc (GenericParam): auto), expected proc (char): bool{.closure.}'
19:57:03	flaviu1	Jehan_: Thanks, that might work
19:57:36	BitPuffin	reactormonk: how do I install your nimrod-mode?
19:57:42	Jehan_	The only problem I see with the sequtils.nim approach is that the variable naming needs to be consistent.
19:59:25	flaviu1	Jehan_: If I do that, I'd probably do a macro such that `fn: _ + _` works
19:59:40	flaviu1	But I wonder how underscores will behave
19:59:56	Araq	the compiler rejects single underscores
19:59:58	Jehan_	Well, _ is not a valid identifier.
20:00:02	flaviu1	maybe using alphabet letters
20:00:36	Jehan_	That's why you may just as well use something like "it" for a single argument.
20:01:00	Jehan_	It's not really different from using "self" or "this" in single dispatch OO languages.
20:01:28	flaviu1	Jehan_: But in some cases I might have multiple variables, like a fold
20:01:31	Jehan_	A convention for multiple arguments would be nice.
20:01:38	Jehan_	Yeah, I was getting to that. :)
20:01:49	BitPuffin	Araq: it would be awesome if you could create your own customized effect tracikng
20:02:08	flaviu1	so I'd probably do a, b, c, ...
20:02:12	Jehan_	val1, …, valn or arg1, … argn are obvious candidates, but a bit verbose.
20:02:18	Araq	BitPuffin: you mean override the inference rules to something unsound? :P
20:03:16	Jehan_	Hmm, doesn't Groovy use "it" too to allow you to elide a single argument parameter list?
20:03:17	BitPuffin	Araq: so for example, in linagl, I would like to track if a matrix variable has been become unorthogonal, then I could make super cool optimizations, such as when calling the inverse proc on a matrix that is orthogonal, it simply transposes it instead
20:04:17	BitPuffin	Araq: basically I somehow create my own effect called orthogonal, and mark procs that ruin the orthogonality as making it fals, and orthogonalize as setting it to be orthogonal
20:05:08	Araq	there is a point in life where bug prevention ends up taking much more time than fixing bugs
20:05:20	BitPuffin	well this is for optimization
20:05:40	BitPuffin	without having to have more types
20:05:42	flaviu1	Araq: And I want to annotate data structures with performance information, so that if I say that something has N performance, it fails to compile if it turns out that it actually uses another method with N^2 performance
20:06:43	BitPuffin	Araq: I guess it doesn't have to be an effect, but a compiletime-only variable associated with a type
20:07:57	Araq	flaviu1: that's what the D guys try to do
20:08:26	Araq	"you must not overload 'in' when it takes O(N) time"
20:09:25	flaviu1	That's a bit insane, my version would be nothing more than statically checked verification.
20:09:39	Araq	that's not our way. Often I know N is small and O(N) is meaningless.
20:09:50	flaviu1	s/verification/documentation/
20:09:50	BitPuffin	I wonder if something like that can be created with macros
20:10:24	flaviu1	BitPuffin: Yes, but it'd be easier if I had a way to apply a pragma to all procs
20:10:57	BitPuffin	flaviu1: are you talking about your thing or my thing
20:10:57	flaviu1	If a hole appears, it becomes useless
20:11:00	Araq	flaviu1: you can do that with .push iirc
20:11:23	flaviu1	BitPuffin: My thing
20:11:24	flaviu1	Araq: Neat, I'll look into it later
20:11:52	Araq	flaviu1: there is a much easier way to do it though
20:12:17	flaviu1	hmm?
20:12:18	Araq	use the effect system to prevent recursions
20:12:26	Araq	it's a planned feature
20:12:45	Araq	gives you something different but workable
20:13:23	Araq	you only need to check nested for loops then to estimate the O and carefully inspect every 'while' loop
20:14:18	fowl	how can i importc a generic type in the cpp gen
20:15:25	Araq	fowl: you can't
20:15:27	flaviu1	fowl: http://en.wikipedia.org/wiki/Name_mangling#Standardised_name_mangling_in_C.2B.2B
20:15:39	BitPuffin	Araq: so do you think compiletime only properties to types is possible to implement with macros?
20:16:12	Araq	BitPuffin: we already have these, might be a bit buggy for now though. Check out the manual
20:16:36	BitPuffin	oh
20:17:04	BitPuffin	what are they called?
20:17:37	Araq	generics with default values
20:17:48	Araq	or something like that
20:18:28	fowl	BitPuffin, link me when u find it plz
20:19:08	Araq	type Foo[T; S = 3] = Bar[T]
20:19:09	BitPuffin	I'm lookin
20:19:20	BitPuffin	Araq: and those are mutable?
20:19:47	Araq	no, what does that mean?
20:20:51	BitPuffin	type TMatrix[T; R, C: static[int]; O = false] = ...
20:22:49	BitPuffin	proc orthogonalized[T, R, C, O](m: TMatrix[T, R, C, O]): TMatrix[T, R, C, true]
20:22:55	BitPuffin	well
20:22:58	BitPuffin	guess it doesn't really matter
20:23:13	BitPuffin	sinec I don't mutate
20:24:24	fowl	Araq, this module makes heavy use of macros, templates, tables and seqs at runtime and compiletime, and everything seems to work :D https://bitbucket.org/fowlmouth/entoody/src/502e02abd9932a933ad5c35da58a094c668616fb/src/entoody.nim?at=master
20:24:24	BitPuffin	Araq: but in a case where someone would have a matrix library that does mutate, ie the matrices on the heap, when they have proc orthogonalize(m: PMatrix)... = m.type.O = true ...
20:24:49	BitPuffin	fowl: did you go hg? :D
20:25:02	fowl	no
20:25:11	fowl	i dont know hg any more than i know git
20:25:30	flaviu1	BitPuffin: Main usecase for bitbucket is the free private repos
20:25:37	BitPuffin	flaviu1: I know
20:29:17	Araq	fowl: ok ... that doesn't mean it's a good idea though ;-)
20:29:35	BitPuffin	can't seem to find it in the manual
20:29:40	BitPuffin	are you sure it is documented?
20:30:47	*	brson joined #nimrod
20:32:09	*	boydgreenfield joined #nimrod
20:41:35	*	freezerburnv quit (Quit: freezerburnv)
20:41:36	*	ehaliewicz joined #nimrod
20:42:36	*	eigenlicht quit (Ping timeout: 240 seconds)
20:48:04	fowl	Araq, what do you mean? its nice to have for debugging info
20:51:36	*	filwit joined #nimrod
20:52:20	BitPuffin	o/ filwit
20:52:44	filwit	hey folks, is there a reason deriving a ref type from another causes an error? bug? or something else?
20:52:50	filwit	heya BitPuffin
20:53:12	filwit	type Foo {.inheritable.} = object ...
20:53:20	filwit	type Bar = ref object of Foo ...
20:54:04	filwit	# causes segfault or sometimes some odd behavior
20:55:34	filwit	woah, damn. 324 issues... i remember when there was just two pages..
20:56:44	BitPuffin	filwit: how's the website coming along?
20:57:23	filwit	good, i meant to have everything done last weekend, but got caught up doing other things
20:58:03	*	eigenlicht joined #nimrod
20:59:11	filwit	i'm shooting for this weekend. HTML/CSS is mostly done, just working on exporting some of the graphics still
21:00:43	filwit	i've just had a lot of other projects i'm working on, and Nimrod in general has taken a bit of a back-seat. Then when i want to work on it, I usually just start coding on my own Nimrod project instead of working on the web stuff
21:01:09	BitPuffin	cool!
21:02:25	*	onibaka quit (Quit: onibaka)
21:03:03	filwit	yeah, i'll try and stop by more often for encouragement on completing that. I usually try and avoid this place when I can't work on Nimrod stuff, cause I always end up spending an hour debating a minor syntax issue or something here
21:03:12	*	oddmunds quit (Ping timeout: 245 seconds)
21:03:37	*	seubert quit (Ping timeout: 245 seconds)
21:04:42	BitPuffin	hehe
21:04:46	*	seubert joined #nimrod
21:04:46	*	oddmunds joined #nimrod
21:06:32	filwit	My brother and I are working on a game for a MS competition (Unity3D) which is taking most of my free-time ATM. But I've also been working on my Nimrod engine a lot, with the goal of making a 3D iOS/Android/OUYA/SteamBox game (before i complete the editor and release the code).
21:07:37	filwit	right now my heads been wrapped up in engine design and animation code. I had to write my own Blender exporter cause Assimp doesn't work with Blender multi-textures at all (or rather, Blender doesn't export Collada multi-textures)
21:13:31	*	filwit quit (Ping timeout: 240 seconds)
21:18:06	*	Varriount|Mobile joined #nimrod
21:21:01	*	DAddYE joined #nimrod
21:26:31	*	Varriount|Mobile quit (Quit: AndroIRC - Android IRC Client ( http://www.androirc.com ))
21:26:57	*	filwit joined #nimrod
21:27:45	filwit	stupid internet...
21:28:08	filwit	was talking for a few minutes to no-one apparently :\
21:28:56	filwit	$90/month and the thing still drops at least 6 times a week on average
21:30:06	*	freezerburnv joined #nimrod
21:30:28	filwit	anyways, is Araq still around, or has he hit the hay already?
21:32:11	*	onibaka joined #nimrod
21:32:22	Araq	hi filwit
21:32:31	BitPuffin	filwit: ah cool, I wonder if there is a way to get the multi textures in there
21:33:01	filwit	oh, great, hi Araq
21:33:01	EXetoC	filwit: 1gb/1gb is it? :p
21:33:01	dom96	yo filwit
21:33:05	filwit	damn, everyone at once :D hi all
21:33:13	EXetoC	*gbit
21:33:45	filwit	Araq: is my earlier question a bug or expected behavior? (can't make 'ref object' derived of another object)
21:34:42	BitPuffin	afaik ref object should be no problem
21:34:49	BitPuffin	but what do I know really, I mean REALLY
21:35:52	filwit	BitPuffin: i'll be releasing a Blender exporter with my engine stuff eventually. You can find the current code on my brother's ReignSDK repo now though (https://github.com/reignstudios/ReignSDK/tree/master/Tools/BlenderPlugins) it supports multi-textures, animations (no curve-handles yet), and a bunch of other stuff.
21:36:21	Araq	filwit: my guess is that your problem is something else
21:36:21	filwit	Araq: no, i did a very simple test (Foo/Bar level stuff) and it caused a segfault
21:36:38	filwit	Araq: but it did also behave a bit differently when i made my OOP macros export the code
21:36:53	filwit	Araq: so i'm not entirely sure. Are you saying it should work?
21:37:08	filwit	Araq: if so, i'll do more tests, and report any bugs i find.
21:37:19	Araq	it's worth reporting either way
21:37:34	BitPuffin	filwit: well if it is just collada then shouldn't you try and get it upstream with blender=
21:37:36	BitPuffin	?
21:37:44	Araq	tbh I'm not sure what ref + .inheritable means :P
21:37:55	Araq	why not inherit from TObject?
21:38:19	filwit	because my OOP macros don't spit out a TObject.. they just spit out a `type T = ref object`
21:38:33	BitPuffin	like real men
21:38:48	filwit	so i'm extending to spit out `type T = ref object of U`, but it fails
21:39:37	filwit	yes BitPuffin, real men are direct, and don't need to pre-define the object when all we need is a ref variant :P
21:40:05	filwit	EXetoC: sorry, i didn't understand your question from earlier
21:41:21	filwit	EXetoC: oh, where you talking about internet speed?
21:42:30	filwit	EXetoC: idk, mines like 60Mbps (or however you say) but the router sux ass
21:43:11	BitPuffin	:)
21:43:12	EXetoC	really expensive
21:43:35	filwit	EXetoC: it's Brighthouse networks (some south east company) but it's my only option here (not even Verizon FiOS yet)
21:44:07	filwit	yeah, i hear internet in other places is cheaper, but gotta live with what's available i guess
21:44:49	filwit	it's not a huge deal really, just annoying when you're trying to upload something to a server and your connection drop the whole thing
21:46:16	filwit	BitPuffin: about collada, I kinda need my own format anyways, since I'm making an editor which will eventually be able to save files, and prefer to have control over things.
21:46:26	flaviu1	filwit: Can you make a git repo for your kate configurations?
21:46:56	flaviu1	I have a half-working indenter that I took from python and I'd like to send a PR
21:47:05	*	Jehan_ quit (Quit: Leaving)
21:47:08	BitPuffin	filwit: ideally you want a binary format
21:47:55	filwit	flaviu1: awesome, I've already uploaded to here: https://gist.github.com/PhilipWitte/11196561
21:48:26	filwit	flaviu1: but i need to update that, as some of the colors and stuff aren't there yet, and I've made two more (Visual Studios 2013 light & dark colors)
21:48:33	*	Roin joined #nimrod
21:48:57	flaviu1	That's a gist though, I can't send a PR. I can set up a repo for it if you'd like.
21:49:12	filwit	BitPuffin: yes, binary format will come eventually, but XML for now (since my brother and I are sharing the format and C#'s JSON isn't the best)
21:49:23	filwit	flaviu1: ah, i see
21:49:31	BitPuffin	filwit: yeah, much easier to start with
21:49:36	BitPuffin	filwit: and to write dummy data for
21:49:44	filwit	flaviu1: okay i will make a official git repo for it tonight
21:50:00	flaviu1	Thanks
21:50:34	flaviu1	I've been reading about sqlite, they claim it works as a file format. Have you considered that?
21:50:47	filwit	no
21:50:52	filwit	don't know anything about it
21:51:18	filwit	BitPuffin: i'm considering making an JSON version (already started) over XML anyways, just cause i prefer the format
21:51:31	BitPuffin	filwit: yeah
21:51:47	BitPuffin	filwit:I wonder if yaml is fitting for that type of data
21:52:01	filwit	plus, it would be easy to make a JSON -> BSON converter for binary formats
21:52:23	filwit	if Nimrod doesn't already have something like that, it would be helpful for more than just my project
21:52:36	Araq	I'd use sqlite
21:52:40	Skrylar	i talked about a 'pandoc for data' thing a while back
21:52:42	BitPuffin	I wouldn't
21:52:54	filwit	i guess i'll have to look into sqlite
21:52:57	Araq	BitPuffin: why not?
21:53:13	BitPuffin	Araq: well I guess it depends on the kind of game
21:53:31	BitPuffin	Araq: But I don't really think storing stuff like vertex data in a relational database is a good idea
21:53:32	flaviu1	I haven't done much with SQL, but it seems like fun to program in
21:53:56	Araq	BitPuffin: if the alternative is XML ...
21:54:09	BitPuffin	Araq: however perhaps things like level structure etc could be fitting
21:54:26	BitPuffin	you could do something where the metadata is in sqlite and the actual data is in binary files
21:54:35	BitPuffin	that would probably work quite well
21:54:35	flaviu1	BitPuffin: Why not store vertexes in sqlite?
21:54:39	Araq	a component model is already close to a relational model
21:55:00	Araq	and most games are heading for a component model
21:55:03	filwit	flaviu1, i've only done simple things with SQL.. but i wouldn't call it "fun" unless your idea of fun is playing MS Excel :P
21:55:16	BitPuffin	flaviu1: why not store music samples in sqlite
21:55:27	*	freezerburnv quit (Quit: freezerburnv)
21:55:32	BitPuffin	flaviu1: why not store pixel data in sqlite
21:55:33	Roin	did somebody just compare an SQL database with MS Excel?
21:55:36	flaviu1	BitPuffin: Ok, so store the vertex structure as a binary blob.
21:55:37	Roin	geez...
21:55:42	BitPuffin	flaviu1: yeah
21:55:54	Araq	hi Roin. glad you're still around
21:55:59	Roin	Heya Araq o/
21:56:03	Araq	what happened to underscore?
21:56:10	Roin	oh he sits right next to me atm
21:56:20	BitPuffin	Like I said, you could do something where you just store the metadata in sqlite (this object has this material, and it has a parent that is this object)
21:56:22	Roin	Still studying unlike me lol
21:56:24	filwit	Roin: yes i did.. probably shows my ignorance :P
21:57:16	Roin	filwit: hm well while you can use MS Excel as a rudimentary database it is... not a database, MS Access is a database but Excel.. nope :P
21:57:33	Roin	database management system actually duh hoh
21:57:42	filwit	the problem with storing Bitmaps, Sounds, etc in a SQL table (compared to the filesystem) is that it sux for content creators
21:57:44	Roin	Araq: I havent coded anything in ages, feel like I should do it again
21:57:51	Roin	so my brain doesnt dry out
21:58:12	Araq	filwit: yeah, don't do that
21:58:20	Roin	Araq: I'm working as an IT/Infrastructure Specialist for SAP HANA and BWA, so I do lots of Linux and infrastructure stuff and sometimes a bit of scripting but not much
21:58:33	Roin	filwit: databases are not meant to store such data
21:58:35	BitPuffin	I also don't know if I would use sqlite during the actual simulation etc
21:58:51	BitPuffin	I would probably store the base level metadata in sqlite, and then load it all into memory
21:58:59	Roin	You can use them as a reference table as in where to find such data on the file system or w/e but you really shouldnt store the data itself in a database methinks
21:59:13	BitPuffin	doing a bunch of mutations in sql probably wouldn't be very fast :)
21:59:41	flaviu1	It isn't very hard to use multiple database files with sqlite, but the best approach might be to load it into memory so you get predictable performance
21:59:47	*	Roin used SQLite to store lots of temperature data from his smart meter to write a web based application that plots the actual electricity usage of my systems
21:59:53	filwit	well it's "okay" for a release build where you want to stream data quickly (in which case, a database might be a performance benefit.. i think)
22:00:23	Roin	streaming data sounds more like you want a high performance file system
22:00:33	filwit	but for editing, it's silly. and when you consider that most games want the ability to mod nicely, it also doesn't make much sense
22:00:47	filwit	Roin, yeah i agree
22:01:03	BitPuffin	Roin: well we weren't talking about streaming data :) at least I wasn't :o
22:01:11	flaviu1	How would sql negatively impact modablity?
22:01:42	filwit	flaviu1: because it forces you to use an external tool to access/modify the data, rather than just the filesystem
22:01:59	BitPuffin	filwit: well you have to use a text editor to modify an xml file
22:02:04	flaviu1	You can't use notepad, sure. But sqlite has a nice repl
22:02:06	Roin	BitPuffin: oh ok I was just thinking someone was doing stuff like life editing of a stream of data (video data for example) with persistance in between like TV stations do for example
22:02:16	BitPuffin	and you are gonna have a more difficult time finding the relevant stuff
22:02:20	Araq	flaviu1: it requires modders to know about SQL. And they don't.
22:03:02	BitPuffin	Roin: ah. No I was just thinking more like serializing your metadata for levels and game objects to an sqlite file
22:03:11	BitPuffin	and then load it all to memory
22:03:11	flaviu1	Araq: The modders that can't learn the very basics of SQL are the ones that won't be releasing any worthwhile mods
22:03:36	Araq	flaviu1: I use this argument myself, but in this case it is really wrong :P
22:03:39	Roin	BitPuffin: my experience if this results into one big table that has to be read through even loading the whole sql file into memory is slow in the end
22:03:47	filwit	BitPuffin, flaviu1: say i want to modify a single texture, with everything saved in a familiar format on a filesystem, i just open with Gimp/Photoshop and edit. but with a SQL system, i need to lauch a tool, duplicate the asset to a format Gimp can read, edit/save, then reimport that asset
22:04:15	BitPuffin	Roin: yeah I'd use my own binary format but I was trying to explain how Araq's suggestion is somewhat valid
22:04:24	Roin	oh ok
22:04:25	flaviu1	filwit: Fair enough, non-blob data is much better for sqlite
22:04:48	BitPuffin	filwit: well that's why I say don't store vertex/image/sound data in sqlite
22:04:51	Roin	since I didnt see the initial statement I'm totally with you anyway, just trying to throw in my 50 cent ;P
22:04:52	BitPuffin	that's fucking dumb :D
22:05:07	BitPuffin	Roin: :)
22:05:16	filwit	BitPuffin: then what *would* i be storing in sqlite?
22:05:31	flaviu1	I would still store things like level data or gameplay state in sqlite, if only because I haven't really used it before
22:05:48	BitPuffin	filwit: for the fourth time, metadata, level structure (this object has this material, this parent, these components etc)
22:05:58	filwit	sure i guess
22:06:05	BitPuffin	:D
22:06:07	flaviu1	At least its better than creating a new blob format and dealing with that
22:06:16	filwit	though I'm happy with just JSON/BSON for that
22:06:25	BitPuffin	I think your own blob format is worth it in the end
22:06:27	filwit	cause one is human readable, and the other is optimized
22:06:32	BitPuffin	but sqlite is probably a good thing to start with
22:06:34	Roin	filwit: basically relational data... which results in text, numbers and so on that arent too big :P
22:06:36	BitPuffin	if you don't want to deal with that
22:07:13	flaviu1	filwit: You could implement json or bson exporting from sqlite
22:07:30	flaviu1	You could also use something like protobuf or capt'n crunch
22:07:30	Roin	hm in Python this was relatively easy
22:07:47	Roin	like 5 lines to get data from sqlite, turn it into a datastructure that you could dump to json
22:07:57	filwit	flaviu1: what's the point of sqlite at that point tho? if BSON is alread binary, what does putting it into an SQL table accomplish?
22:08:27	flaviu1	You get to use the SQL language to manipulate the data
22:08:32	BitPuffin	getting a repl
22:08:35	Roin	in sqlite tables you can haz structure and relations and stuff, json does not know relations
22:09:19	BitPuffin	yeah
22:09:24	Skrylar	json/bson have heirarchial relations and flexible schemas
22:09:27	BitPuffin	the way you'd do it is to put in an id or something
22:09:36	Skrylar	SQL has fixed schemas and relationships are a little more indirect
22:09:37	filwit	meh, not enough. Most of the actual data manipulation in that regards has to be done with the game editor anyways (to edit the internals) so adding a bit of extra file management isn't much
22:09:38	BitPuffin	which is more expensive to verify the integrity of
22:10:22	Roin	and you can haz indexes and other nice things
22:10:24	filwit	i'm just liking JSON/BSON approach, so i'll stick with if for now
22:10:37	filwit	but i'll look into sqlite and stuff later, just to be sure
22:12:05	*	onibaka quit (Quit: onibaka)
22:12:08	filwit	Araq: you mentioned that my Monokai colors where too colorful for your tastes last time I showed you. What colors *do* you like?
22:12:32	filwit	Araq: i've added VS 2013 light and dark colors, and the dark is now my favorite. You like those?
22:13:01	flaviu1	The VS colors are very nice
22:13:55	filwit	yeah, the light one is pretty good, but the dark one is just perfect IMO
22:14:42	filwit	i keep trying to switch just to try my other colors, but the VS dark is just my favorite. They did a good job picking those colors out
22:15:42	Araq	filwit: I need to see a screenshot to be able to tell
22:15:53	*	boydgreenfield quit (Ping timeout: 255 seconds)
22:16:27	filwit	one sec
22:16:27	flaviu1	Araq: http://blogs.msdn.com/cfs-filesystemfile.ashx/__key/communityserver-blogs-components-weblogfiles/00-00-01-29-92/3286.VS2013Preview_2D00_UX7abc_2D002D00_2.png
22:17:30	Roin	Hm I thought you were talking about a game lol
22:17:41	Roin	already wondered what VS stood for
22:17:49	Araq	what's the topic again? colors for the new documentation?
22:18:13	filwit	http://www.reign-studios.net/philipwitte/nimrod/kate/Syntax_VSDark.png
22:18:31	flaviu1	Araq: Kate color scheme
22:19:12	flaviu1	filwit: What KDE theme are you using?
22:19:20	filwit	Araq: :P i'm working on the new docs. shooting for having stuff to show this weekend (i know i said that about last weekend)
22:19:42	filwit	flaviu1: it's a modified Elementary Flat theme
22:19:57	Araq	filwit: it's much better yeah. but I still prefer my own scheme
22:20:44	flaviu1	Araq: I'm looking for color schemes, mind sharing?
22:22:16	*	boydgreenfield joined #nimrod
22:24:25	Araq	flaviu1: no time for screenshots sorry, it's the "darknim" scheme in aporia
22:25:08	flaviu1	Thanks
22:26:38	filwit	holy crap that's a horrible color scheme, lol
22:27:17	filwit	i now know that we will forever be at odds in terms of colors if this is your preference, haha
22:27:29	flaviu1	I like colorful color schemes
22:28:28	filwit	i like colors which indicate types and stuff, like distinguishing varname/typename/property/function-call/etc
22:29:03	flaviu1	I wonder if there's some way to get Kate to use the compiler for highlighting information
22:29:16	filwit	but beyond just being completely uniform, that color scheme is way to "stark white on pitch black"
22:30:04	filwit	flaviu1: there might be. though a better challenge would be to get Kate to suggest names based on semantics (like the D plugin someone wrote)
22:30:42	flaviu1	I know, I've been considering that. I'll probably do it in python though, I don't really want to figure out C++
22:31:02	*	noam_ is now known as noam
22:31:28	filwit	why not do it in Nimrod, then simply make a C++ wrapper where needed
22:31:46	filwit	well.. i supposed that might not be possible if you need to start up the GC
22:32:21	filwit	then again, they're doing it for Python, so it should be a similar challenge
22:32:53	flaviu1	Because nimrod and C++ don't have several modules I'd need, and I don't want to end up shaving yaks.
22:33:24	filwit	shaving yaks? term i'm unfamiliar with (though i can guess it's meaning)
22:34:02	*	filwit read that as "shaving _yanks_" at first
22:34:16	flaviu1	I've just reversed "yak shaving". The second result on google is pretty good, I had no idea what it was a month ago
22:41:18	*	Johz quit (Quit: Leaving)
22:41:31	*	def-_ joined #nimrod
22:44:38	*	def- quit (Ping timeout: 252 seconds)
22:47:21	*	perturbation joined #nimrod
22:48:03	perturbation	hi all... i'm working my way through the tutorials
22:48:07	perturbation	and I kind of have a dumb question
22:48:23	filwit	hello. shoot
22:48:31	perturbation	how do I know when to use methods (dynamic binding) vs proc's (static binding)
22:48:50	dom96	That is not a dumb question at all
22:48:53	perturbation	it seems like I could make overloaded procs and it should be able to look up which to use by object type
22:49:16	EXetoC	not at runtime
22:49:18	filwit	when you want to invoke a overload from a "base object" reference
22:49:29	EXetoC	(polymorphism)
22:49:38	Skrylar	filwit: yak shaving is when you have to do work so that you can do work so you can do the work you originally wanted to do
22:49:56	filwit	Skrylar: yeah, i looked it up and was pretty much what i was expecting
22:50:23	filwit	perturbation: let me make a simple gist for you real quick, to explain things. one sec
22:50:25	flaviu1	Its origin story is still interesting
22:50:46	perturbation	thanks filwit, EXetoC
22:51:02	flaviu1	I saw a guy on the reddit post say that we should replicate the python tutorial in nimrod, I think that's a good idea
22:51:31	dom96	we should write a "by example" tutorial
22:54:13	flaviu1	Yeah, by example is much faster and easier
22:55:07	flaviu1	GitBook is really nice
22:56:47	filwit	perturbation: https://gist.github.com/PhilipWitte/81f92d2da57d42109b43
22:57:10	filwit	perturbation: let me test that real quick, just to make sure i'm not lying to you :P
22:57:46	perturbation	heh, no worries... really conceptually I just need to wrap my head around it conceptually
22:58:48	filwit	hmmm... well it doesn't work, one sec let me fix it before you study it..
22:59:44	Araq	good night
23:00:00	filwit	later
23:00:20	filwit	perturbation: okay, fixed it. now it compiles and works right
23:00:33	perturbation	thanks filwit
23:00:46	filwit	np.
23:01:21	Araq	perturbation: simple. never use inheritance and then you never need 'method'
23:01:59	perturbation	well I confess I am coming at this from a stone-age C perspective
23:02:11	perturbation	so structs are all I need really :D
23:02:41	filwit	perturbation: the key thing to note is that the list `things` is of the base type `Foo`, but since `Bar` is derived of Foo, you can add Bars to the list. Then when you call `baz` on them, it correctly select the appropriate overload based on their type (even though the references are all of type Foo, not Bar)
23:02:52	EXetoC	makes sense. you just need the features that C has :p
23:03:02	flaviu1	perturbation: If you only use them as structs, you'll never need to use the method keyword
23:03:41	perturbation	filwit: ohhhhh. Because I can make it either, *that*'s why it can't know which is which at compile time
23:04:38	filwit	perturbation: yes, exactly. When you have a list of base types (but potentially have derived types in that list), you need methods to correctly overload (at runtime) procedures
23:05:27	Araq	well there are alternatives filwit
23:05:46	filwit	perturbation: though it's not *just* lists. You can also have just a `var f: ref Foo` and make it point to a `Bar`, then call `baz` on it and it will correctly call the right overload
23:06:01	filwit	Araq: yes, i know about procvars
23:06:11	filwit	Araq: i'm just explaining methods
23:06:19	Araq	yeah sure
23:06:50	flaviu1	Araq: The only way to process the `adfs"rawstring"` sugar is with a indefinite lookahead, right?
23:07:45	Araq	not really
23:07:56	Araq	and nimrod's lexer doesn't do it this way
23:07:56	*	xenagi joined #nimrod
23:09:12	filwit	hmmm... wtf, now `type T = ref object of U` appears to be working in my basic tests...
23:09:19	perturbation	for the examples on http://nimrod-lang.org/tut2.html#properties, you can define setters/getters as procs b/c they don't have any objects that inherit from the examples?
23:10:43	EXetoC	inherit
23:11:18	EXetoC	?
23:11:31	*	ehaliewicz quit (Remote host closed the connection)
23:11:59	*	ehaliewicz joined #nimrod
23:12:24	EXetoC	"proc p(x: TSomeObj): int = 42" -> echo someObj.p
23:12:35	EXetoC	or p(someObj). it's your choice
23:12:39	perturbation	any other types defined as 'object of' the 'TSocket' so no ambiguity
23:12:57	perturbation	?
23:13:03	*	nande_ joined #nimrod
23:14:01	flaviu1	I assume that `object of TSocket` would be downconverted to a TSocket for that method
23:14:02	filwit	perturbation: "properties" in nimrod are just procs/methods/templates.. it's just referring to languages like C#/D which have special syntax for properties (which allow you do use `T.p` & `T.p = 0` instead of `T.p()` & `T.p(0)`).. in Nimrod, all methods can be called without he parentasees and you have the special syntax "proc `p=`(v:T)" to define setters
23:14:48	filwit	the parentheses**
23:14:55	EXetoC	no conversion needed
23:19:15	filwit	so, there's a bug with {.inheritable.} it seems.. or at least different behavior
23:20:12	filwit	when i do `type Foo {.inheritable.} = ref object ...` I get the error: "Only final objects can be inheritable"
23:20:45	EXetoC	only final wut
23:20:48	filwit	however, if I do `type Foo = ref object of TObject` (which i was underneath impression was exactly the same as the former) it works.
23:21:32	Araq	the error message is clearly wrong
23:21:35	filwit	there error is: "inheritance only works with non-final objects"
23:21:37	perturbation	one sec... let me put a paste bin to try and narrow down where I'm getting confused... you guys are awesome though
23:21:56	Araq	but the point is "only *objects* can be inheritable"
23:22:05	Araq	you're using it on a 'ref'
23:22:19	Araq	but yeah, we can make it work, it's pedantic
23:22:27	flaviu1	Araq: Since you're still awake, is '\"' just a carryover from pascal?
23:22:29	*	nande_ quit (Remote host closed the connection)
23:22:37	filwit	Araq: okay, but it actually works (functions correctly) if i do `type Foo = ref object of TObject` then make a Bar derived of it
23:22:43	Araq	flaviu1: yes
23:23:09	Araq	filwit: .inheritable is really for low level hacks
23:23:42	filwit	Araq: okay, good to know. So always use 'of TObject' for inheritance then. That's probably my problem with the OOP macros then.
23:23:53	Araq	quite possible
23:24:02	perturbation	http://pastebin.com/kPCRMzPV
23:25:20	*	hoverbea_ quit ()
23:25:29	Araq	gah it's late
23:25:35	Araq	flaviu1: x <= y implies a <= b if x <= a and y <= b
23:25:42	Araq	right or wrong?
23:26:42	Araq	x <= y implies a <= b if a <= x and y <= b
23:27:00	Araq	that's the correct rule ... I think
23:27:05	flaviu1	I know, I have to write it out
23:28:25	filwit	perturbation: in your example (since you're using procs, not methods) then the procedure which is deduced to be associated with the type (in this case, TSocket since that's what `s` is) is called.
23:29:15	*	Matthias247 quit (Read error: Connection reset by peer)
23:29:24	flaviu1	Araq: It looks true, I'll see if I can find a theorem prover to plug it into
23:29:39	*	darkf joined #nimrod
23:30:21	filwit	perturbation: if you where to use methods, it would call the procedure appropriate for what the *actual* type is. In this case, it would still be the one for TSocket, because that's what `s` is. But if you had made TSocket a references type (aka, `ref object`) and then made it point to a TSockObj instead of a TSocket, then it would call the method bound to that type instead.
23:31:46	filwit	perturbation: to be clear, calling a 'method' looks at the reference types type-information and dynamically selects appropriate method to invoke.
23:33:59	perturbation	okay, I think that makes sense :)
23:34:10	perturbation	I'll play around with it some more just to get a better feel for it
23:37:12	flaviu1	Araq: Correct. http://rise4fun.com/Z3/SjGL
23:37:23	filwit	perturbation: your example might be confusing you since you're not working with reference types (just objects on the stack), which means methods don't come into play (since they only apply to a reference which points to a "higher level" type, aka, a type which is a derived which 'of' the actual type of the reference).
23:37:53	flaviu1	It seems that asserting false will return unsat, and that returns sat so I think its correct
23:38:08	filwit	perturbation: sorry, my typos might have made that a bit unclear
23:38:13	flaviu1	No, nm
23:39:42	filwit	perturbation: the point is that I could make a `var s: ref TSocket = newTSockObj()`, meaning (to the compiler) the type is `ref TSocket`, however the actual memory it points to is of type `TSockObj`
23:39:45	perturbation	filwit: I think I get what you're saying: it's a reference (pointer), so the actual type has to be looked up at run time, just like when you have a sequence which can have either type
23:39:50	Araq	flaviu1: it's more complex than 4 asserts
23:40:33	perturbation	(to misquote Arthur Weasley) I just don't trust a programming language unless I can see where it keeps its brain ;)
23:40:47	flaviu1	Araq: How so? In the first two I'm checking the precondition, then the condition, and then that the implication is true.
23:41:21	filwit	perturbation: yes exactly. Whenever you have a pointer which points to a type that is actually different from what the compiler sees (which *should* only happen with inherited objects) then you need methods to tell the code to dynamically look up the appropriate method (based on the references type-info)
23:41:58	flaviu1	For example, if I replace the last assert with `(assert (> a b))`, it fails to prove
23:42:09	Araq	(assert (<= x a)) # note: swapped x and a here
23:42:11	Araq	(assert (<= y b))
23:42:12	Araq	(assert (<= x y))
23:42:14	Araq	(assert (<= a b))
23:42:22	Araq	is 'sat' but it's wrong anyway
23:44:02	Araq	maybe you need forAll here somewhere
23:44:21	Araq	but I'm sure you need "implies" or at least "or"
23:44:32	perturbation	filwit: thanks! this makes sense... I'll do some more Project Euler stuff and see if I get tripped up anywhere else working through the tutorials
23:44:56	Araq	you have 4 asserts combined with 'and' here, that can never express the original statement
23:45:59	filwit	perturbation: np, glad i could help. Keep in mind there are also other ways (function pointers) to do similar things without inheritance, and often that makes more sense.
23:46:15	filwit	perturbation: though I'm sure you already knew that, if you're a C programmer :)
23:53:10	Araq	good night