<< 07-04-2014 >>

---

00:02:00	dom96	Unfortunately, I must sleep now. Good night.
00:02:26	NimBot	Araq/Nimrod devel e66d5c1 Araq [+0 ±1 -0]: updated news.txt
00:02:26	NimBot	Araq/Nimrod devel b41e3b0 Araq [+0 ±5 -0]: Merge branch 'devel' of https://github.com/Araq/Nimrod into devel
00:02:54	Araq	dom96, Varriount ok so lets release tomorrow
00:03:04	Araq	but really! then
00:03:23	dom96	lol
00:03:25	Varriount	Ok. If that's the case, then I need to reboot my computer now.
00:03:26	dom96	ok
00:03:38	*	Varriount quit (Read error: Connection reset by peer)
00:03:45	dom96	bye
00:03:48	renesac	Araq, my pull request with the compressed images for the website was merged on master
00:03:53	renesac	don't forget to pull it to devel
00:04:14	dom96	he'll be merging devel to master
00:04:20	dom96	so it'll be fine
00:04:22	dom96	bye
00:04:23	renesac	right
00:04:32	renesac	good night
00:04:39	Araq	good night
00:05:07	Demos	I am getting strange lookup behavior in a template
00:05:46	Demos	the template calls a generic proc and the generic proc is not able to find definitions that are in the definition scope but not in the scope of the template instantiation
00:13:02	*	ics quit (Ping timeout: 246 seconds)
00:21:47	*	rta quit (Ping timeout: 246 seconds)
00:22:37	*	BitPuffin joined #nimrod
00:29:57	*	Varriount joined #nimrod
00:34:36	*	brson joined #nimrod
00:35:49	*	rta joined #nimrod
00:36:56	*	Varriount quit (Quit: Leaving)
00:39:03	*	brson quit (Ping timeout: 255 seconds)
00:40:56	*	brson joined #nimrod
01:01:45	*	seagoj joined #nimrod
01:05:42	*	q66 quit (Quit: Leaving)
01:06:02	EXetoC	so does loadExtensions only load referenced extensions?
01:06:56	EXetoC	it tries to load some unneeded symbol but fails. I generated a module called gltest.nim and now I'm trying to use it
01:12:02	*	seagoj quit (Quit: Textual IRC Client: www.textualapp.com)
01:17:09	*	rta quit (Ping timeout: 245 seconds)
01:22:38	Demos	so command syntax/properties + getStackTrace() is pretty amazing
01:29:56	*	rta joined #nimrod
01:34:08	NimBot	Varriount/NimLime master fb6da3a Erik O'Leary [+0 ±2 -0]: Added match support for _ in tmpl's
01:42:30	*	Guest78262 quit (Ping timeout: 255 seconds)
01:43:07	*	Guest78262 joined #nimrod
02:08:07	*	Demos quit (Quit: Lost terminal)
02:18:09	*	BitPuffin quit (Ping timeout: 255 seconds)
02:24:13	*	Varriount joined #nimrod
02:28:38	Skrylar	lol the reference implementation of porter's stemmer is silly
02:28:42	Skrylar	global statics everywhere
02:29:17	Varriount	I finally got raid 1 working on my computer! :D
02:39:08	*	brson quit (Quit: leaving)
02:43:08	*	Demos joined #nimrod
02:50:10	*	zezba9000 joined #nimrod
02:51:32	zezba9000	Hey was just wondering if there was any idea of when multiple interfaces might be added into nimrod?
02:52:50	Varriount	zezba9000: What do you mean?
02:53:33	Varriount	A type can already fulfill multiple user defined type classes.
02:53:49	zezba9000	Like right now you can derive from a base class in nimrod, but not interfaces after that like so: "A : B, IC, ID, ect.."
02:54:03	Varriount	You mean, multiple inheritance?
02:54:19	zezba9000	yes, but only with interfaces support like C#, not C++
02:54:55	Varriount	I'm not quite sure I understand - I don't have much experience with C#
02:55:28	Demos	zezba9000: apperently it is possible (as in humanity knows how to do it in the context of multiple dispatch) but it is low priority
02:55:46	Demos	you can cook up some macros to make and fill vtables and whatnot if you like
02:55:55	zezba9000	Ok, well in C# I can do "class A : B, IC, ID {...}" C++ can do the same and more, but its not needed
02:56:09	Demos	but often typeclasses are a better solution
02:56:11	Varriount	I think filwit has some things for OO programming.
02:56:45	Demos	since with typeclasses the module with types "implementing" the typeclass need not depend on the typeclass
02:56:49	zezba9000	Demos: Well I may make a C# to Nimrod converter... and it would be so much better with multiple interfaces for translation
02:56:58	zezba9000	Varriount: Ya he is my brother
02:57:12	Demos	zezba9000: that would be neat. Perobably very hard
02:57:31	zezba9000	But it might make more sense to use standard nimrod, not his off-shoot
02:57:34	Demos	one idea is to use the new `.` overloading to access function in the vtable
02:58:09	*	darkf_ joined #nimrod
02:58:20	zezba9000	Demos: Iv'e seen how it can be done, but its so dirty
02:58:41	Demos	zezba9000: how do you mean? v-tables are how c#/java/c++ style OOP works
02:58:45	zezba9000	I would like it to be human readable. So it maps over well from C# to Nimrod visually
02:59:04	zezba9000	Demos: But not visually
02:59:35	zezba9000	It would make the translator cleaner if it didn't need to convert in that way is all.
02:59:37	Demos	well sure, you can write macros to make objects with vtables, you could even have the macro go through the type decl and change foo: proc(...) to a vtable entry
02:59:38	Demos	maybe
03:00:53	Demos	zezba9000: well nimrod is not going to copy c#'s object model, and if you are reimplementing all of c#'s syntax and semantics than I am not sure what you are getting
03:01:25	zezba9000	well, not sure. I don't want to make my life hard. I know C# very well, but changing how Nimrod works is just going to complicate my life lol
03:01:48	Skrylar	interfaces... hrmm. i don't think nimrod favors them to be honest
03:01:57	*	darkf quit (Ping timeout: 265 seconds)
03:02:04	Skrylar	you have generics code, and composition of smaller objects
03:02:52	*	Varriount doesn't get what the difference is between interfaces and typeclasses
03:03:34	Skrylar	Varriount: interfaces are dynamic
03:03:52	zezba9000	I code in a OO fashian and am not going to write large code bases outside that model. I don't care if Nimrod does or not, I care about its speed, but it would be nice to have some standard features in a way other langs have then like C#, C++ or D
03:03:55	Skrylar	IIRC the way go does it, when you say "x as IFoo" it actually makes a struct full of function pointers
03:05:05	Demos	well nimrod is not really all about OOP, I really like that nimrod does not bundle functions and data
03:06:04	zezba9000	Varriount: Example: http://pastebin.com/WBdcNgZh
03:06:31	zezba9000	Demos: Why not both...
03:06:34	Skrylar	i haven't actually come to a point where i needed interfaces oddly. subclassing yes, but i handled that by just shoving in a "super: Foo" field like go does it
03:06:48	Skrylar	i don't see why it couldn't be done with macros
03:06:51	zezba9000	C++ has both?
03:07:43	zezba9000	Varriount: Fixed typo: http://pastebin.com/tzQxVBqu
03:08:10	Demos	c++ lacks typeclasses. and having method calls be sugar for normal functions makes some kinds of code eaiser to write. Also if you really need dynamic binding you can use nimrod's first class functions
03:08:43	Demos	another problem with bundleing data and behavior together is that it means you almost always are thinking of your data one element at a time, which may not be ideal
03:09:21	zezba9000	But humans thing in terms of objects and data groups?
03:10:04	zezba9000	Even if the code doesn't have the concept, you apply it yourself. Makes it easier to visually see how you put it together in your head
03:10:31	Varriount	Yes, but not to the extreme. Unless you recieve prolonged exposure to Java.
03:10:35	zezba9000	Anyway how would I go about duplicating the example in C# to Nimrod?
03:11:08	zezba9000	Varriount: Both worlds have very good uses. To say ones doesn't is silly
03:11:35	Varriount	(I maintain the assertion that uncautious use of Java and it's restrictive OO paradigm damages the brain)
03:11:36	xenagi	is that even advisable? I may be wrong, but, being that C#/Java and Nimrod, are two different languages, they probably have different programming styles
03:12:12	xenagi	Why would you want to "duplicate" C# to Nimrod when you should probably write the Nimrod-idiomatic version.
03:12:26	zezba9000	Varriount: I disagree. I think your just not seeing how it makes life easier in large apps.
03:12:59	*	rta quit (Ping timeout: 245 seconds)
03:13:03	zezba9000	xenagi: Because I like C# over any lang in many areas. But I do game dev. And Nimrod is super F'in fast.
03:13:25	zezba9000	They both have a GC and thus its logical to want C# with the power of nimrod for me
03:13:43	zezba9000	I target .NET only things like Xbox 360 ect.
03:14:00	xenagi	ah I see
03:14:10	zezba9000	Indie SDKs in short. Like Vita PSM as well wich is .NET only
03:14:43	xenagi	Nevertheless, I'm not sure I'm qualified to offer Nimrod advice
03:14:47	zezba9000	But I would love to have that portibility with power on other platforms like PCs ect
03:15:03	Demos	zezba9000: you realize that idomatic c# is going to be pretty much just as fast in .net as it is in nimrod
03:15:26	Demos	also microsoft just announced a compiler that compiles c# to native code
03:15:41	Varriount	However it's not as fast as you might think Demos.
03:15:43	zezba9000	did tests with that
03:15:45	zezba9000	Its slow
03:15:51	Varriount	I've heard that it's slo-
03:15:56	Demos	Varriount: is it slower than .net?
03:16:04	zezba9000	Demos: Yes is slower which is funny
03:16:10	zezba9000	At least right now on x64
03:16:16	Skrylar	microsoft doing a bad job of something?
03:16:21	Skrylar	hold on, i have to write the media
03:16:28	Skrylar	:P
03:16:33	Demos	I mean nimrod is not fast just because it generates native code. Nimrod is fast because it has value types and a thread local gc
03:16:35	zezba9000	Hell if I know. I think something is just stupid with .NETs math libs
03:16:46	zezba9000	And they are casting to double when they should not be
03:17:09	Demos	on that note you will be unable to translate a lot of c# code since c#'s GC is not thread local, nimrod's is
03:17:10	Skrylar	well property getters are wildly prevalent and involve a function call
03:17:16	Skrylar	probably v-calls too
03:17:20	zezba9000	Demos: This is the test I use: https://github.com/zezba9000/RayTraceBenchmark
03:17:28	Skrylar	there's a whitepaper that talks about how 20% of a C++ program's time is spent in the vtable
03:17:54	zezba9000	Maybe C# suffers from the same?
03:18:05	zezba9000	On iOS C++ is 8 times as fast as Mono with LLVM
03:18:07	Demos	it really depends on what you are doing
03:18:25	zezba9000	Something about .NET is slow and I wish I knew what the hell it was.
03:19:04	zezba9000	In that benchmark Nimrod is almost as fast as C++
03:19:14	zezba9000	Thats what I would like to get out of C#.
03:19:17	Skrylar	Demos: they did the test by writing code that should be inlinable, then used a GCC flag to force all class calls to be virtual
03:19:18	Demos	one problem is that classes tend to grow without bound, and if your vtable has like 35 pointers in it that over 4 cache lines
03:19:25	Skrylar	thats how they determined the overhead of vtabling
03:19:54	Demos	Skrylar: well if you are using -fgenerate-shitty-code....
03:20:15	Demos	oh wait they just turned off devirtualization
03:20:32	Skrylar	yeah, they wanted to see what the overhead was
03:20:47	Skrylar	it was pretty significant
03:20:50	Demos	if you actually need dynamic dispatch vtables are pretty fast, sometimes eaiser to optimize if you can make modifying them UB
03:20:57	Skrylar	couple that with bad code gen -- no surprise that C# code compiles to slow
03:21:22	Demos	like GHC generates native code as well, but they need a slow GC because they are lazy
03:21:45	zezba9000	Skrylar: I don't know if its code-gen, or something to do with secerity in .NET method calls ect
03:22:00	zezba9000	There may be more overhead.
03:22:15	Demos	in cpp people think that "program to an interface" means use pure virtual functions, in reality it means to use a spec, dynamic dispatch is orthogonal
03:22:21	Skrylar	why would a native compiler care about managed security stacks
03:22:43	Demos	Skrylar: they do checking of various things
03:23:03	Demos	zezba9000: you could just use c#'s structs and unsafe blocks to write quick code in c#
03:23:04	zezba9000	Skrylar: You can set the securtiy of the method in C#, so the runtime will not let you dynamicly invoke it ect
03:23:26	Skrylar	zezba9000: my point is that if you're compiling C# to native code, that doesn't matter
03:23:30	Demos	zezba9000: also reflection carries a huge hit with it wrt code size
03:23:35	zezba9000	Demos: LOL I thought of that onces. I could litterally right C code in C# yes
03:23:40	Skrylar	whether or not something is dynamically invokable only means it goes in a table or not, thats not a runtime thing :\
03:23:52	zezba9000	But there is still the Math libs converting everything to doulbes when it should not
03:24:21	Demos	zezba9000: I would think with math libs the big speed difference would come from heap allocation
03:24:29	Demos	doubles are just fine
03:24:40	Demos	in fact nimrod's float type is in fact a double percision float
03:24:52	zezba9000	Either way its stupid to have to worrry about those things.
03:25:05	*	rta joined #nimrod
03:25:08	zezba9000	The point it to have something easy to use and is fast at the same time.
03:25:26	zezba9000	Demos: Does not the hardware use 80bit?
03:25:26	Demos	well that is why people use dynamic languages. Having everything be the same size is nice. although c# is mostally like that as well
03:25:47	Varriount	C# is a dynamic language.
03:26:16	zezba9000	When you chain sin(cos(...)) ect it should do everything in 80bit then it converts to 64bi at the end
03:26:45	zezba9000	Varriount: No its a strong typed lang
03:27:04	zezba9000	it does have the dynamic keyword an abylity to be though
03:27:08	Varriount	However it's dynamic in the sense that it is run in an interpreter.
03:27:18	zezba9000	No its not
03:27:26	zezba9000	only .NET Micro uses an interpreter
03:27:38	zezba9000	normal .NET JIT compiles to native instructions
03:27:39	Skrylar	i think everyone is talking in circles now
03:28:11	xenagi	Is anyone blogging about Nimrod? I'd love a tutorial blog on Nimrod :D
03:28:13	Varriount	zezba9000: Jit compilation means that parts of an interpreted language are compiled to native code at runtime.
03:29:02	Varriount	zezba9000: http://en.wikipedia.org/wiki/Common_Intermediate_Language
03:29:10	Skrylar	Varriount: c# isn't inherently dynamic
03:29:24	Skrylar	its built on a jvm-clone, but most of the language is pretty tightly bound and typed
03:29:29	zezba9000	Idk if thats the right way to say it? Its not staticlly compiled like Nimrod, but that doesn't mean its counted as an interpreted lang
03:30:34	Varriount	zezba9000: C# is interpreted by the CLR (common language runtime), and part's of it are JIT'd, just like Java and the JVM, and Pypy and it's JIT
03:30:35	zezba9000	The CLR determins how the IL code gets ran.
03:31:19	zezba9000	No all of it gets JIT compiled. Only .NET micro uses an interpreter
03:31:46	zezba9000	what part would not be JIT to native instructions?
03:31:46	Varriount	*shrug* if you say so. Anyway, back to finding out why starbound is reporting error.
03:32:20	Varriount	zezba9000: Then why isn't a Java program entirely compiled to native code?
03:33:36	zezba9000	You are not being clear. C# gets compiled to IL. Then the JIT compiles it to native instructions that in turn the work with the CLR.
03:33:56	zezba9000	C# is not interpeted by the CLR
03:34:18	zezba9000	Nor is Java
03:34:52	zezba9000	The reason for IL is hardware agnastic code as you and I both know.
03:37:42	*	ruzu left #nimrod ("http://quassel-irc.org - Chat comfortably. Anywhere.")
03:37:48	zezba9000	anyway my whole point here was just wondering about interface support in Nimrod...
03:38:02	zezba9000	...an ETA I mean.
03:38:46	zezba9000	but guess there is none as no one here uses OO design practices to that level.
03:40:38	Demos	damn fire alarm
03:41:01	*	xenagi quit (Quit: Page closed)
03:41:08	zezba9000	lol
03:42:14	Varriount	"Intel RapidStorage: Data 52% Complete"
03:42:31	Demos	zezba9000: if you just want compile time checking that you conform to an interface a proc like proc implements(typ: typedesc) = when not(typ is Interface): {.fatal: "does not implement".} could work
03:42:46	Demos	where Interface is a typeclass
03:44:10	zezba9000	But it would only support one interface no?
03:44:31	Varriount	use an "and"
03:45:10	Varriount	when not((typ is interfaceOne) and (typ is interface2) and...):
03:45:20	Varriount	Or a template
03:47:03	Demos	zezba9000: nope as many as you like
03:47:12	zezba9000	k well I need to learn more about Nimrod. First i'm using/learing Roslyn to convert C# to HLSL, GLSL ect.
03:47:14	Demos	the difference is that things are bound at compile time
03:47:26	Demos	so for example a seq[interface] would not work
03:47:42	Demos	although I bet there is a way to get around that
03:47:45	zezba9000	Then after I know more about Roslyn i'll do some other experements
03:48:50	zezba9000	But dudes, Nimrod should have interface support built in like C#, C++ and D yo!
03:49:01	zezba9000	Would attract more ppls to the lang
03:49:25	zezba9000	Make it easier to port exister code they may have ect
03:49:34	Skrylar	seems like something a macro could do
03:49:48	zezba9000	But not someone is going to take the time to write
03:49:55	Demos	I disagree, in fact I think zach and araq would like to remove inheratance totally and implement it as a library feature
03:50:10	zezba9000	90% of people are not going to do that who would like the feature to be there I think, just a guess
03:50:22	Skrylar	Demos: he wants to remove the dispatch trees, not inheritance IIRC
03:50:32	zezba9000	Demos: Thats build in to me :)
03:50:35	Demos	yeah true
03:50:36	zezba9000	**built
03:50:47	zezba9000	Even in .NET things are done that way
03:50:58	zezba9000	But people consider it part of the lang
03:51:11	Demos	zezba9000: my first language was java. But I have come to realize that OOP offers almost /nothing/ in terms of code re-use or modularity
03:51:25	zezba9000	I'm telling your it would be good thing to advertise on...
03:52:02	zezba9000	Then I would disagree.
03:52:42	zezba9000	Would you right a billing app in a non OO lang?
03:52:53	zezba9000	I don't think anyone would today.
03:53:27	zezba9000	**write
03:54:23	Demos	well you probably do not need fast native code for a billing app
03:54:31	zezba9000	The worst code I have ever seen was 'C' code ported to C# that kept everything non OO in it about C
03:54:41	zezba9000	But you do for games
03:54:46	zezba9000	and games should use OO
03:55:08	zezba9000	the only thing I can think of that does not need it (maybe) is writing service code
03:55:13	Demos	not really. OOP for games causes some problems, esp if you model each "thing" in the gameworld as one object
03:55:27	Skrylar	now we're arguing about buzzwords...
03:56:01	Demos	things tend to get lifted up through your inheratance tree and you get a really super big "player" or "ai" class (just some examples)
03:56:21	zezba9000	Demos: Thats not true
03:56:25	Varriount	Demos: Not to mention that it's bad for cpu caches
03:56:56	Demos	Varriount: that it is, but if you are in C# you are already screwed in that area
03:57:50	zezba9000	You also know C# has partial classes right?
03:58:00	Varriount	Demos: Why?
03:58:12	Demos	zezba9000: partial classes are not the same thing
03:58:21	Demos	Varriount: because people tend to heap allocate everything
03:58:43	zezba9000	You are going to have big player classes in Nimrod as well as C#
03:58:47	Demos	so I guess you still get pwned by all the extra data in each class, and you can fix that. But you can not really exploat prefetch
03:58:50	zezba9000	How does that make a diff?
03:58:59	Demos	zezba9000: you dont have a player class
03:59:17	Demos	you have a position class, and a health class (or just a distinct int) and so on
03:59:26	zezba9000	Then your code is not organized...
03:59:31	zezba9000	lol you have the same in C#
03:59:46	zezba9000	but you also have a player class that just contains them
03:59:49	Demos	no it is, the API is very similar to something you might do in ruby (with mixins and so on)
03:59:52	zezba9000	Or derives from them
04:00:19	zezba9000	You should have a player class...
04:00:44	*	Guest78262 quit (Ping timeout: 255 seconds)
04:00:55	zezba9000	You don't free float objects around. You end up duplicating arrays that don't need to exist and cause bugs
04:01:31	Demos	yeah, I organise my objects in arrays, one for each type of object
04:01:33	zezba9000	That reminds me of when I was in Middle school and used C
04:01:39	zezba9000	horible code
04:02:33	zezba9000	Demos: So for a player you have a class health array and class position array ect?
04:02:41	*	Guest78262 joined #nimrod
04:02:55	zezba9000	That are not contained/grouped in a player class?
04:03:09	zezba9000	You have now fragmented all your logic
04:03:24	Demos	well I have arrays of (EntId, type)
04:03:32	zezba9000	When you could have a single array or players that you can quickly add and remove logic from
04:03:33	Demos	so they are linked in "entities" by a simple ID
04:03:49	Demos	zezba9000: actually logic is really easy to add how I have it set up
04:03:51	zezba9000	ok dude, you are doing this all wrong
04:03:56	Demos	my data has no logic
04:04:00	Demos	these are just arrays of data
04:04:09	zezba9000	You in your mind have the concept of a player no?
04:04:33	zezba9000	So why not visualise that day in the code how you project the abstract concept
04:04:43	Demos	I mean sure, but when I am writing code to render the player I do not give a crap about her health
04:04:56	zezba9000	So...
04:05:07	zezba9000	You have added on a huge layer of complexity
04:05:18	zezba9000	As a game dev I used to try thing like you suggest
04:05:33	zezba9000	bad idea as it makes far more bugs and you end up with more code
04:05:52	zezba9000	Instead of one Count variable, you now have 5 or more
04:06:09	Demos	why would I have a count variable, what am I trying to count?
04:06:26	Demos	I don't store any more data than I would if I were doing OOP
04:06:28	zezba9000	When you add a feature to the player you now have to updated every peace of logic that handles array counts instead of just changing one
04:06:39	Skrylar	i'm bored now... i'll be back later.
04:06:46	*	Skrylar quit ()
04:06:50	zezba9000	count vars are built in the array types
04:07:02	zezba9000	You should use them
04:07:06	Demos	zezba9000: I do not follow... there are lengths, I do use them
04:07:14	Demos	actually I use iterators, which is nice
04:07:29	zezba9000	Let me guess you have one global Length var?
04:07:34	Demos	what?
04:07:36	Demos	no
04:07:48	Demos	length of what?
04:08:13	zezba9000	For everything that represents the aspstract obj array of player?
04:08:55	zezba9000	Look i'm telling you, don't do what you are doing. Make a player class. When code starts to get complex it will save your life
04:10:01	zezba9000	No game dev job would ever let you do what you are doing
04:11:02	Demos	OK lets get concrete. Lets say my player has 5 "fields". Health, Texture, Mesh, and Position. I then have 5 arrays and when I make a new player I add an item to each array. I have 5 lengths, but tons of other stuff has meshes and positions and whatnot
04:11:31	Demos	besides of all the things in your game storeing array length is never going to cause memory problems
04:11:43	zezba9000	They would never allow that in game dev
04:11:52	zezba9000	very bad thing to do
04:12:19	Demos	http://macton.smugmug.com/gallery/8936708_T6zQX#!i=593426709&k=BrHWXdJ ?
04:12:29	Demos	mike acton is a game developer, he ships lots of games
04:12:33	zezba9000	I would would have my boss looking for someone else if I started doing that
04:12:43	Demos	warning: some of those slides are pretty gnarly PS3 perf stuff
04:13:46	zezba9000	I see some messed up code, but not what we are talking about
04:14:58	zezba9000	What you are doing should not even be done in C code.
04:15:42	Demos	what specific reason is "what I am doing" wrong. Why will it be harder to maintain/slower/worse/sad etc
04:16:56	zezba9000	http://pastebin.com/v9Hqeaij
04:18:12	Demos	nononononononononono why would i have a playerSize array, that is just bad
04:18:19	Demos	I have a size array and a health array
04:18:43	zezba9000	Show me how you would make the player in pastebin?
04:18:51	Demos	and storeing like your mesh and material with health is just bad
04:19:00	Demos	https://gist.github.com/barcharcraz/f7ed665fa352fa7917cf
04:19:18	Demos	that is a static mesh, not a plater
04:19:21	Demos	but you get the point
04:19:38	zezba9000	can you give an example with player?
04:19:46	zezba9000	I don't know how those objects are used in the system
04:21:30	renesac	zezba9000, I'm writing this whitespace FAQ clearing up misconceptions about it: https://github.com/Araq/Nimrod/wiki/Whitespace-FAQ
04:21:38	Demos	https://gist.github.com/barcharcraz/89949faacc8d88ba884d
04:21:59	Demos	and so on
04:22:37	renesac	Demos, the 'EntityID' would be the player class/struct?
04:22:53	Demos	EntityId is literally type EntityId = distinct int
04:22:58	Demos	but yeah
04:23:11	renesac	how you call 'add' on that?
04:23:19	Demos	I have some stuff that can pack up an entity into one block of data if you need to like transmit it over the wire
04:23:42	Demos	add is just proc add*[T](ent: EntityId, item: T)
04:24:58	renesac	hum, that proc adds the item on it's own array, but refering to the EntityId?
04:25:22	*	darkf_ is now known as darkf
04:25:24	Demos	I map the name of each componet type (the T type param) to an identifier
04:26:03	Demos	the arrays are private arrays in the module with add
04:26:16	Demos	so their storage is global, but their scope is not
04:26:19	*	OrionPK joined #nimrod
04:27:02	zezba9000	This is just silly. Why? Just contain it in a Player object and call genEntity(Types.PlayerType);
04:27:41	zezba9000	No you have to search an array object just to get a reference to a obj property?
04:27:46	zezba9000	**now you...
04:29:35	zezba9000	genEntity is your obj pool correct?
04:29:51	zezba9000	...is pulling from a pool?
04:30:19	Demos	yeah you do. But usually you care about all the components of a given type, and I sort by entity id so you can go though all the entities that share a common set of components in linear time. Note that this is /not/ true of the OOP approach, If I want everything that has a position and a mesh I can do that eaisly
04:30:31	Demos	well it is pulling from the pool of 64-bit integers
04:30:44	Demos	remember EntityId is literally an integer
04:31:48	zezba9000	This seems slow...
04:32:00	zezba9000	You have to search the array to find properties...
04:32:16	Demos	right, but you access them all in order
04:32:19	zezba9000	When you could directly access them
04:32:20	Demos	that is the entire point
04:33:05	zezba9000	what you still need a player class, even with this
04:33:26	zezba9000	So when you access transform its no entity[0].positon
04:33:35	zezba9000	its player.transform.position
04:33:47	Demos	no
04:34:09	Demos	it is transforms[some index].position
04:34:24	Demos	you can have the entity id be the index into that array, but I do not
04:34:58	renesac	Demos, things can get deleted from that array, right?/
04:35:39	renesac	then a index you had would not be valid anymore, opposed to a pointer to a player struct that would still be valid
04:35:40	zezba9000	renesac: yes what happens when you delete things?
04:35:40	Demos	for me yes, but often it is better to just "disable" it
04:36:00	zezba9000	But then the index EntityID is wrong
04:36:11	Demos	yeah, I said you can do that, I do not
04:36:22	Demos	I store stuff densely
04:36:30	renesac	so you binary search in this case?
04:37:02	Demos	yeah, although if you have an index you can search for like 4-5 items around that index to see if you find your item
04:38:00	Demos	to make this faster you can store another array of EntitiyIds, each index is the entity id of the component in that index of the component array
04:38:08	OrionPK	lib/core/macros.nim(208, 21) Error: for a 'var' type a variable needs to be passed
04:38:15	zezba9000	To make this faster you could just have a Player obj...
04:38:36	zezba9000	and it would be more human readable
04:38:45	renesac	Demos, but then on each deletion you would have an O(n) operation updating the array of EntityIds
04:38:51	Demos	another nice thing there is that the stuff in your component array is just plain data, you can actually copy that whole array right into a memory on your graphics card
04:39:20	Demos	renesac: right, but deletions are rarely fast, even with OOP, you need to rebuild all kinds of structures
04:39:20	zezba9000	Demos: But you can't just do that
04:39:24	zezba9000	You must batch it
04:39:42	zezba9000	So only the obj transforms in order get passed and drawn
04:40:25	Demos	that is not batching zezba9000, that is ordering by depth, which is not a great thing to order on unless you need transparancy
04:40:58	zezba9000	No, I mean occlusion culling
04:41:07	renesac	sure is nice to have homogeneous arrays, if you usually access one element type at time, it is more cache efficient
04:41:14	zezba9000	You must order an array of ONLY what needs to be drawn
04:41:24	zezba9000	and pass it to the GPU
04:41:47	Demos	and you do that with no perf loss. The code to figure out what needs to be culled needs to go though ALL the meshes/renderables anyways, and so you get a cache win from not pulling in stupid data. Then the renderer can look at that data and fetch the right thigns to draw
04:42:16	renesac	but the code does seems more complex than if you did group things in a player object/struct
04:42:41	Demos	it is simpler code! it does not depend on anything aside from just what it needs
04:43:02	zezba9000	yes, you could do the same thing with a player obj. But when the time comes to draw, you run through the objects and batch them into array pools then..
04:43:24	Demos	right, but what if players are not the only thing in your scene
04:43:29	zezba9000	It is more complex code to read when you start doing a lot all over the place with 100s of files
04:43:58	zezba9000	Demos: The player is derived from the Entity obj
04:44:11	Demos	and the entitiy object grows without bound
04:44:17	zezba9000	sure why not
04:44:37	zezba9000	But the player obj just references the entity array with real names
04:44:46	zezba9000	So I can do Player.Transform
04:45:05	zezba9000	but I can only create a Enemy derived from Entity
04:45:13	zezba9000	and both have the dynamic array you want
04:45:15	Demos	zezba9000: why not is because if you have a huge entity object you will DESTROY your cache
04:45:51	Demos	and having an hierarchy does not work great, actually the thing I do is kinda an implicit hierarchy that is better than one I could design myself
04:46:35	zezba9000	Ok then have a single cache/pool, but you still should have a player obj
04:47:00	zezba9000	that derives from an entity that handles where it is in your global cache/pools
04:47:17	Demos	a cache/pool is not going to help you
04:47:39	Demos	because you still need to actually load all that data, which bings it into your CPU cache
04:48:45	Demos	http://www.gdcvault.com/play/1911/Theory-and-Practice-of-the is a persentation on how this kind of thing was actually used
04:48:53	zezba9000	anyway you look at it, you can have a player obj reference the data rather then index it
04:49:33	zezba9000	Yes you are making pools.
04:49:43	zezba9000	But you should still reference those pools by objs
04:50:35	renesac	zezba9000, but then the data will not be contiguous in memory, and it would have one or two extra indirections to iterate over all instances of the same type of data
04:50:40	zezba9000	That optamisation layer should exist. But should expose itself via objs
04:50:50	Demos	I have an exam tomorrow and need to sleep. I dont think OOP is a good way to design software. In particular I do not think modeling objects after the real world is a good idea in almost any situation
04:50:57	*	Demos quit (Remote host closed the connection)
04:52:12	renesac	if you do the same optimization than Demos, then your Player object would be the array index, and you would need to update all them in case of a deletion
04:52:38	renesac	well, deletions seems costly on demos model, as you have a O(n) array reordering anyway
04:53:32	renesac	unless you use some rope/btree based array, but he did mentioned other costs
04:54:12	zezba9000	Yes you would have to have an entity pool I guess
04:54:40	zezba9000	But when you get back your player from the pool it link up to the transform array ect
04:55:19	zezba9000	There is also going to need to be a key.
04:55:44	zezba9000	Its going to need to be a dictionary, but one thats made of two arrays for each entity type
04:56:27	zezba9000	anyway i'm getting tierd too
04:59:30	renesac	did you get a look at my whitespace faq?
05:00:11	zezba9000	what about it
05:00:50	renesac	you were talking about the lack of braces in nimrod another day
05:01:17	renesac	is there some worry that you have that isn't covered in the faq?
05:02:36	zezba9000	On the line "const DeBruijnNumbersTable: array[32, int8]" is it required to tab/space out the lines below?
05:03:01	renesac	"After a continuation, the code can be positioned quite freely. The only rule is that the continuing line must be indented at least one level above the first line."
05:03:22	renesac	so you can't have them be on the same level as the first line
05:03:25	renesac	you need some spaces
05:03:28	zezba9000	Can I do this: http://pastebin.com/DgeMAxp5
05:03:55	renesac	no, you need one level of indentation
05:03:58	zezba9000	So the answer is no then
05:04:07	zezba9000	Well thats why I don't like whitespace then
05:04:19	zezba9000	Its why I gave the example I did the other day too
05:04:37	renesac	is two spaces that troublesome?
05:04:55	zezba9000	If the IDE could handle white space then I would not care about brackets
05:04:57	renesac	select a block and key 'tab'
05:05:12	zezba9000	yes it makes the code no longer a block
05:05:23	renesac	?
05:06:00	zezba9000	https://github.com/reignstudios/ReignSDK/blob/master/Platforms/Common/Reign.Core/Math/Matrix4.cs#L57
05:06:05	zezba9000	Thats the example I gave before
05:06:23	zezba9000	I like it when I can make a block with the brackets
05:07:27	zezba9000	I just find it easier to read when its on the same line
05:07:35	renesac	I don't understand, if you take the brackets out (and of course changed function declaration, etc), this would be valid nimrod
05:07:59	renesac	ah
05:08:05	renesac	the ( );?
05:08:27	zezba9000	so the 'return' can be on the same line as '(' ?
05:08:47	zezba9000	you said it must has at least one tab
05:08:52	renesac	yeah
05:09:00	zezba9000	well why though
05:09:21	zezba9000	the parser should be able to tell that '(' means what comes next
05:09:24	zezba9000	not a white space
05:09:48	zezba9000	Brackets define what white space fails to do
05:10:14	zezba9000	Or you could have an IDE define it, but then it would relay on the IDE
05:10:46	zezba9000	I simply have more control over how I layout any given block
05:10:51	renesac	and actually, you would need a parenthesis on the first line
05:11:23	renesac	as the function name alone is already a complete statement (for first class functions)
05:11:33	zezba9000	I just don't see anything wrong with brackets. I do see something wrong with ';' though
05:13:06	zezba9000	Brackets define lines. Thus you have the power to layout something is it bests sutes you. Ending marks serve no purpose for the human involved as a human understands the end bracket is the end of the line
05:13:39	zezba9000	A human can't understand if there is a tab vs 4 white spaces
05:13:44	Varriount	Forgive me if I'm wrong, but in your example, it's the semicolon that allows you to do that.
05:13:45	renesac	brackets are redundant, add more lines of code and, as I said on the faq, can get out of sync with the indentation
05:14:00	Varriount	Not the brackets.
05:14:09	renesac	zezba9000, you don't have to worry about tab vs space on nimrod
05:14:20	renesac	it will only accept spaces as indentation
05:14:23	zezba9000	Varriount: Yes in C#, but I don't like the ';' as I stated its not needed
05:14:36	renesac	and Varriount has a point
05:14:58	renesac	it is the ; not the brackets that let you put the parenthesis there
05:15:09	zezba9000	renesac: But it should not be is the point.
05:15:18	zezba9000	Don't take C# as my perfect example
05:15:24	Varriount	zezba9000: There are plans for a nimrod varient with brackets (look in the compiler directory at pbraces.nim)
05:15:27	zezba9000	Its just the C style brackets I find usful
05:17:04	zezba9000	Varriount: https://github.com/zah/grip-lang/blob/master/compiler/pbraces.nim
05:17:05	zezba9000	That?
05:17:32	Varriount	Yes. It will be where the braces parser code is located.
05:17:49	Varriount	Operational word being the future tense "will"
05:18:04	renesac	Varriount, why again nimrod needs an extra level of indentation for a line continuation?
05:18:08	zezba9000	I don't see any '{..}' brackets?
05:18:14	renesac	given an unfinished statement in the previous line?
05:18:18	Varriount	renesac: Where?
05:18:26	renesac	as redundancy to be able to give better error messages?
05:18:52	zezba9000	Redundent means it no longer has a use
05:18:55	Varriount	renesac: It helps reduce ambiguity when parsing.
05:19:11	zezba9000	Brackets allow me to line up constructor blocks how I like them
05:19:15	zezba9000	thus its not redundend
05:19:42	renesac	zezba9000, it is not the brackets, it is the ;
05:19:49	renesac	unless you are adding brackets to every line
05:19:52	Varriount	zezba9000: Yet, as I pointed out in your example, it is not the braces that allow you to put those in a block, it's the semicolon
05:19:55	zezba9000	yes ';' is rudundent
05:20:16	renesac	no, they are not redundant in your example
05:20:31	renesac	they are what makes possible to put the open parenthesis in it's own line
05:20:36	Varriount	The C# parser knows that the 'return' statement continues until the semicolon.
05:20:38	*	nequitans quit (Ping timeout: 240 seconds)
05:21:17	zezba9000	But the C# should just look for '{' or '}' instead
05:22:17	Varriount	That's not how it works though.
05:22:22	zezba9000	I know
05:22:22	renesac	not for deciding the end of a line
05:22:27	renesac	{} delimit blocks, not lines
05:22:42	zezba9000	yes but block are what matters, not lines
05:22:58	renesac	but your return is a line
05:23:30	zezba9000	what do you mean?
05:24:01	renesac	the return, open parenthesis, elements, and close parentesis are a single line
05:24:06	renesac	for the compiler
05:24:10	Varriount	^
05:24:18	zezba9000	or ic what you mean
05:24:33	zezba9000	ok well yes
05:24:35	renesac	and coincidently the block is a single line too
05:24:47	zezba9000	not a method block
05:24:53	zezba9000	its has more then one line
05:25:13	zezba9000	it states here are lines for a given context
05:25:38	renesac	the 'FromMatrix3' has a single line for the compiler
05:25:46	zezba9000	A would say a line should break "IF" there is no '(...)' brackets
05:25:50	renesac	unless you are counting the } after the ; as another line
05:27:36	renesac	zezba9000, a better way to say it would be: you should have a line break when you reach the end of a line with a complete statement.
05:27:45	zezba9000	http://pastebin.com/J75QsrME
05:28:06	renesac	if it is not complete (an '(' w/o the corresponding ')' for example) you look at the next line
05:28:08	zezba9000	oops http://pastebin.com/c31wZ3q2
05:28:15	zezba9000	has ; marks in there
05:28:25	renesac	nimrod does essentially this, but adds the additional constraint of one extra indentation level
05:28:28	zezba9000	renesac: yes correct
05:28:49	zezba9000	What do you think of my example?
05:29:05	renesac	the second one would also be valid in nimrod
05:29:13	zezba9000	Line break with a RETURN ONLY happens if there was no '(..)' brackets
05:29:16	renesac	as you can't end a statement with a =
05:29:36	zezba9000	I think it should not as thats what the ';' mark was for
05:29:38	renesac	yes, you rule is more restrictive in that sense
05:30:20	renesac	why it should not be allowed?
05:30:47	zezba9000	Because you normally don't think that way. Keep in mind I have the use of {..}
05:30:57	renesac	the fact is that nimrod would treat those 3 as valid
05:30:58	zezba9000	Its not white space driven in my example
05:31:17	zezba9000	That example does not depend on white space
05:31:32	zezba9000	it depends on how the human thinks about a line
05:31:40	zezba9000	At least how I think people do
05:31:55	renesac	it depends on line breaks (whitespace)
05:31:57	renesac	right?
05:32:12	zezba9000	No it does no depend on whitespace in: http://pastebin.com/c31wZ3q2
05:32:26	zezba9000	It depends on brackets
05:32:53	renesac	how the compiler knows that the statement 'int i = 0' has ended?
05:32:56	zezba9000	a new line is value 'IF' there was no '(' bracket before it
05:33:06	zezba9000	it the '(' was not closed I mean
05:33:19	zezba9000	Because there was a return
05:33:20	renesac	because it has a *new line( after it, and no open parenthesis
05:33:25	zezba9000	yes
05:33:37	zezba9000	so if it has (int i = 0
05:33:42	renesac	so it depends on line breaks (whitespace)
05:33:43	zezba9000	the line would still be open
05:33:48	renesac	right
05:34:15	zezba9000	yes and normally when you hit return on the line "int i = 0" it would make sense to a human
05:34:30	renesac	you can allow for the dangling '=' in the same way
05:34:37	zezba9000	and so would "(int i = 0... not being the end of the line
05:35:01	renesac	if a line has a operator lacking it's right operator, then the line is still open
05:35:02	zezba9000	ya
05:35:09	renesac	*operand
05:35:24	renesac	then example two would be valid
05:35:25	zezba9000	ya operators don't effect lines
05:35:30	zezba9000	only brackets
05:35:47	zezba9000	no it would not
05:36:06	renesac	http://pastebin.com/DgeMAxp5 <-- so this would be invalid under your rule?
05:36:10	renesac	no brackets on the first line
05:36:15	renesac	so it has ended on the line break
05:36:18	zezba9000	because there was a return without there being an open bracket of {...
05:36:44	zezba9000	or (...
05:36:57	zezba9000	yes it ends on the line break without ( or {
05:37:26	renesac	your FromMatrix3 would also be invalid
05:37:48	zezba9000	no it would be valid
05:38:05	renesac	'return new Matrix4' <-- no ( { in this line
05:38:11	renesac	so it has ended
05:38:24	zezba9000	http://pastebin.com/yLULzSUq
05:38:27	zezba9000	thats valid
05:38:47	zezba9000	ignore the ';' I left in there
05:38:51	zezba9000	forgot to take that out
05:38:59	renesac	'return new Matrix4' <-- no ( { in this line
05:39:07	renesac	so it has ended
05:39:14	zezba9000	its an OR and AND
05:39:17	zezba9000	( OR {
05:39:19	renesac	the ( would be another statement
05:39:22	renesac	yes
05:39:30	renesac	no ( OR { in that line
05:39:34	zezba9000	but ( is required for method and if starts
05:39:45	renesac	?
05:39:56	renesac	so another rule?
05:40:04	zezba9000	there is a ( or { on every line.
05:40:11	zezba9000	No there is no other rule
05:40:15	renesac	I don't see any on line 3
05:40:33	*	isenmann joined #nimrod
05:40:39	Varriount	Hi isenmann
05:40:39	zezba9000	ya why would that need one?
05:41:00	zezba9000	renesac: { came before line 3
05:41:02	renesac	a line break w/o an opening ( or { means that the line has ended, righ?
05:41:07	zezba9000	so its still a line
05:41:26	renesac	the { is a block delimiter
05:41:38	zezba9000	no ) OR } OR 'Return' means the line ends
05:42:14	zezba9000	but 'Return' ONLY ends a line if ( OR { was not closed with ) OR }
05:42:32	renesac	hum, so you could not do 'var res = new Matrix 4 \n ( ... ) \n return res'
05:42:52	renesac	it would be special case for blocks consisting of a single return line
05:43:19	zezba9000	yes I guess that would have to be a special case
05:43:48	zezba9000	So if you are declairing something it expects ( OR { before the line ends
05:44:12	renesac	ok
05:44:24	renesac	I don't see that much gained from this special case
05:44:35	renesac	that justify {} around every block
05:44:50	zezba9000	you can the aility to make blocks as the Matrix example
05:44:57	zezba9000	**you gain the...
05:45:22	renesac	I don't remember needing to do something like that
05:45:53	zezba9000	Well I use it in my Matrix example
05:45:55	renesac	and I woudn't mind having to indent a bit more the parenthesis (and put the first one in the first line, actually)
05:46:12	zezba9000	most people don't
05:46:19	renesac	nimrod could implement this special case, but special cases are ugly
05:46:45	Araq	er I think we indeed special cased the closing ]
05:46:56	Araq	did you try?
05:47:11	Araq	but I only skimmed this conversation...
05:47:36	zezba9000	Araq: http://pastebin.com/c31wZ3q2
05:48:04	renesac	Araq, indeed the closing ] can be at the same indentation level as the firs line
05:48:07	zezba9000	I was saying how that example is valid and no white spaces needed
05:48:11	renesac	but the opening one cannot
05:48:32	renesac	zezba9000, you need new lines, that are whitespace
05:48:46	zezba9000	Better: http://pastebin.com/FZwH78LY
05:48:47	renesac	you only don't enforce indentation
05:49:13	zezba9000	yes no indentation is needed, as in it doesn't matter in that example
05:50:11	renesac	all those are valid in nimrod, as you give the needed two spaces from the first line
05:50:35	renesac	oh, in the last you need a ':' after the if
05:50:37	zezba9000	Araq: Most clear example: https://gist.github.com/zezba9000/10015396
05:51:11	zezba9000	renesac: The last example in the gist I posted is not valid in nimrod
05:51:27	Araq	zezba9000: nor should it be valid
05:51:34	Araq	as it makes no sense
05:51:36	Araq	so yes
05:51:39	zezba9000	Yes it does
05:51:42	zezba9000	You missed how so
05:51:52	Araq	we restrict your freedom to write crap
05:52:07	Araq	big deal...
05:52:35	zezba9000	Araq: No as you can do: https://gist.github.com/zezba9000/10015411
05:52:38	zezba9000	...with it
05:52:41	Varriount	Ohi Araq
05:52:47	zezba9000	Its crap
05:52:53	zezba9000	I just gave an extreme example
05:53:01	zezba9000	** Its NOT crap lol
05:53:12	Varriount	Note to self: Do not employ Araq as a diplomat
05:54:17	renesac	that last example is only possible after the last return keyword in a funcion in your hypothetical language, or in a language that needs ; to delimit line ends
05:54:53	zezba9000	No or a closeing bracket }
05:55:05	zezba9000	Either closes the line
05:55:06	renesac	ok, after any return keyword
05:55:49	renesac	but you theoretically don't need the }
05:55:54	zezba9000	yes you do
05:55:55	renesac	the indentation of the block could suffice
05:56:18	renesac	unless you want to go lower than the indentation of the first line
05:56:18	zezba9000	as it closes { that is required by the method.
05:56:29	zezba9000	Or yes you could make it so {..} is not required as well
05:56:33	zezba9000	if you wanted both I guess
05:56:38	renesac	yes
05:57:06	renesac	that being possible is not a virtue of {}, but of the special case around the return
05:57:16	zezba9000	so you can still do you white space stuff, but its just for feel and nothing more
05:57:59	zezba9000	ya guess so
05:59:21	zezba9000	So in that theoretical model, you can use white space like Nimrod is now, but you don't have to. You can only use the C way
06:00:15	zezba9000	So it a block formats better with things lined up on the X axis you can do so with the use of brackets (..)
06:00:21	renesac	the requeriment for an extra indentation makes it easier to spot continuation lines, and makes possible for the compiler to give better error messages
06:00:37	zezba9000	renesac: Same with ';'
06:00:45	renesac	it is not a bad trade of to get rid of ;
06:00:47	zezba9000	But humans don't need it
06:00:47	renesac	yeah
06:00:55	renesac	I agree
06:01:01	renesac	and I argue the same about {}
06:01:12	renesac	humans need the indentation, not the {}
06:01:29	zezba9000	I think both can be useful
06:01:48	zezba9000	I find it easier to find a block or give scope with {}
06:01:49	renesac	{} lets you go whack on indentation
06:02:17	zezba9000	Yes but no one does that, or should not.. you can write shit code in anything.
06:02:26	renesac	zezba9000, editor highlighting of indentation levels makes easy to spot a block
06:02:53	renesac	I have it enabled even when coding in C
06:03:17	zezba9000	but in this model you would be able to give a internal method scope too "scope {}
06:03:34	zezba9000	I mean with {} if you wanted them
06:03:36	renesac	?
06:03:44	renesac	nimrod have 'block:'
06:03:49	renesac	I'm not sure if that is what you mean
06:03:49	zezba9000	yes I know
06:04:03	zezba9000	scope block whatever :)
06:05:07	zezba9000	Problem is {} is required for a sub block in a method
06:05:25	renesac	in C#, not in nimrod
06:05:27	zezba9000	if white space is not to be considered {} is required
06:05:44	zezba9000	I mean in the hypothetical lang
06:05:47	renesac	indentation you mean
06:05:52	zezba9000	yes
06:05:54	renesac	whitespace is too general
06:06:25	zezba9000	k so yes with no indentation taken into account {} is needed for sub method blocks
06:06:35	renesac	yes
06:06:44	renesac	for any type of block actually
06:07:05	renesac	well, nimrod has (;) , but that is equivalent to {;} in C#
06:07:10	zezba9000	Well it would make it consistent to keep them reqardless
06:07:46	zezba9000	So you can use ; in Nimrod, but you don't need to?
06:07:53	renesac	basically, you can chose if you want your language to delimit blocks with indentation or {}
06:07:57	renesac	zezba9000, yes
06:08:14	renesac	you can end any line with it
06:08:17	zezba9000	What I thought Nimrod could not use {}
06:08:19	Varriount	Araq: I'm about to go to bed - Is there anything you want me to do?
06:08:29	renesac	yes, {} are for sets in nimrod
06:08:50	zezba9000	but if Nimrod does use {} it still required the line indent for the first {
06:09:01	renesac	but you have: http://build.nimrod-lang.org/docs/manual.html#statement-list-expression
06:09:10	renesac	but not everything can be written inside it
06:09:26	renesac	not sure whether it is a bug or a planned limitation
06:09:45	zezba9000	So you can't do: https://gist.github.com/zezba9000/10015517
06:09:55	renesac	yeah, I think indentation is still required
06:10:09	renesac	they are used most for stuffing things in a single line
06:10:21	zezba9000	Hmm well if Nimrod supports {} that should negate indentation
06:10:28	zezba9000	then you could do both
06:10:38	renesac	I don't think code should behave diferently from what it looks
06:10:57	zezba9000	Not sure what that means
06:11:07	renesac	humans look first at the indentation to see block boundaries
06:11:08	Varriount	Anyway, I'm off. Good luck renesac
06:11:10	zezba9000	How code looks mades it means differnet things
06:11:17	zezba9000	**makes it mean
06:12:04	zezba9000	renesac: Yes I agree, but they also look at brackets to give context
06:12:21	zezba9000	Like on a single line for ((1+2) * 3)
06:12:33	renesac	I've already got C code to read with mixed spaces and tabs..
06:12:49	renesac	banged my head for a while trying to understand it till I noticed the mixing
06:13:14	zezba9000	Visual Studios will fix this for you
06:13:20	renesac	then I needed to make a search and replace to get rid of tabs
06:13:24	renesac	so I could read the code
06:13:28	zezba9000	You should set your IDE to use TABS
06:13:30	renesac	{} alone don't cut it
06:13:35	zezba9000	then you don't have the issue
06:13:50	renesac	it was code I downloaded
06:13:56	renesac	with mixed tabs and spaces
06:13:57	zezba9000	In VS you can just have it auto do it for you
06:14:15	zezba9000	Well you just use auto formatters
06:14:21	renesac	VS discovers what tab stop the original coder used?
06:14:29	renesac	ah, right
06:14:38	zezba9000	no it formats to your style settings
06:14:46	renesac	yes, I use it too
06:14:53	renesac	on code blocks though
06:15:02	zezba9000	Its Edit->Advanced->Format Document
06:15:27	renesac	useful especially when getting a code improperly pasted in a forum or blog comment
06:15:37	renesac	that is also unreadable w/o indentation
06:16:18	zezba9000	lol what the code you downloaded?
06:16:49	renesac	an disk bwt implementation
06:17:49	renesac	anyway, in python3 and nimrod you don't have those problems
06:18:38	renesac	no mixing of tabs and spaces, and the code you see on internet will aways be readable (otherwise it would be useless, and the person posting would have fixed it)
06:18:40	zezba9000	I hate Phyton lol
06:19:00	zezba9000	You don't have that problem in C++ either if you set your IDE to use Tabs
06:19:13	zezba9000	I don't know why the hell its not set that way by default
06:19:30	renesac	I aways set my IDEs to convert tabs to spaces
06:19:39	zezba9000	Well thats your problem
06:19:47	renesac	why?
06:19:58	zezba9000	Its way easier to backspace C#/C++ code with tabs
06:20:01	renesac	when set to use tabs, it is more likely a space get added by accident
06:20:03	zezba9000	in VS anyway
06:20:24	zezba9000	Set it to block mode
06:20:29	renesac	well, most editors have the option to go one indentation level down when backspacing on the start of the line
06:20:40	renesac	even when using spaces
06:21:05	renesac	and when working with blocks of code, you can aways 'shift' + 'tab' or equivalent
06:21:19	zezba9000	Still its bad, because it confuses the IDE is its mixed
06:21:24	zezba9000	THus keeping tabs is best
06:21:30	renesac	I don't mix
06:21:32	renesac	I use only spaces
06:21:38	renesac	in python that is PEP8
06:21:54	renesac	in nimrod is enforced by the compiler
06:22:15	renesac	and in C is the coding guideline in many places, the kernel for example (8 spaces)
06:22:21	zezba9000	Python pisses me off. How you access base classes ect
06:22:26	zezba9000	so ugly
06:22:35	renesac	super?
06:22:49	zezba9000	no its like __class__ or something right?
06:23:37	zezba9000	renesac: Well I disagree with the C standard for C and the kernel then :)
06:23:43	renesac	maybe, I've never used it though
06:24:13	zezba9000	O well I used it for making a Blender plugin... god I knew I didn't like it before, but after that I was like gag
06:24:29	renesac	zezba9000, the problem of using tabs is that you then want to align something
06:24:47	renesac	like my first example:
06:24:47	renesac	someReallyLongProc(withMany, commands,
06:24:48	renesac	that, may, be, broken,
06:24:48	renesac	in many, lines)
06:24:48	zezba9000	It makes it easier
06:24:58	renesac	it will look aligned to you
06:25:01	zezba9000	I don't align things like that
06:25:15	zezba9000	I disagree with how a lot of C++/C# devs algin things
06:25:27	renesac	but anyone with a different tab stop than you will see this wrong
06:26:11	renesac	well, I often do align things like that, when it don't mean many more lines of code
06:26:26	renesac	so I use space instead of tabs
06:26:34	renesac	so I have the tab key as an quick way to add spaces
06:26:37	zezba9000	sure but unless i'm working on someone elses code base with there standard, they can suck it. I would do that like: https://gist.github.com/zezba9000/10015607
06:27:07	zezba9000	I like making it fit into a block. Easy to see
06:27:11	renesac	you forgot the ';'
06:27:13	renesac	:P
06:27:21	zezba9000	Well I don't like that stupid thing lol
06:27:30	renesac	it isn't return for that special case to work
06:27:34	zezba9000	Do you know how many times in C++ that stupid thing has fucked me up
06:27:45	zezba9000	Fuck the ';' mark lol
06:27:58	renesac	I also hate ';'
06:28:02	zezba9000	When you do intaliser lists, the ';' mark makes no sense
06:28:13	renesac	but w/o it you can't write function calls this way
06:28:20	renesac	that you like
06:29:06	zezba9000	well yes but {..} is more important to me then having to do ';'
06:29:26	zezba9000	In C# the ';' doesn mess you up
06:29:39	zezba9000	Because the real time parser tells you its an error
06:29:53	zezba9000	but in C++ holly god dam it will go off in no-mans land
06:30:37	renesac	well, if you had to work with a language w/o ';', how you would do that function call?
06:31:07	zezba9000	I would do it like I posted: https://gist.github.com/zezba9000/10015607
06:31:20	renesac	it would not work
06:31:23	zezba9000	I would group each line in context of what those vars being passed in mean
06:31:34	renesac	how it will know to continue the first line/
06:31:34	renesac	?
06:31:39	zezba9000	In the hypithetic lang it would I gave you
06:31:48	zezba9000	**hypothetical
06:32:07	renesac	I don't follow
06:32:16	renesac	I would group each line in context of what those vars being passed in mean <-- ?
06:32:28	zezba9000	Because the syntax tree contains "someReallyLongProc"
06:32:34	renesac	you have a function name in the first line
06:32:34	zezba9000	So it knows its a method
06:32:42	zezba9000	C# gets around confix using this all the time
06:32:54	renesac	your language don't have first class functions/methods?
06:33:06	renesac	C# gets around because it has ';'
06:33:56	zezba9000	But mine gets around it with {} OR ()
06:34:46	renesac	in nimrod 'var x = someReallyLongProc', x() would now be an alias for someReallyLongProc()
06:34:52	zezba9000	"someReallyLongProc" is a method defines in the class your typing in. So when "someReallyLongProc" gets hit it requires '(' before the line fully starts
06:35:00	renesac	you would need 'var x = someReallyLongProc('
06:35:12	zezba9000	No
06:35:13	renesac	for nimrod know that it needs to look at the parameters in the next line
06:35:31	zezba9000	Becuase "someReallyLongProc" is a method the compile now requires '(' to be next
06:36:08	renesac	how would you pass methods as callbacks in your language?
06:36:09	zezba9000	any whitespace before '(' mark doesn't matter if a method came before it
06:36:49	renesac	or alias an method name
06:36:50	renesac	?:
06:37:09	zezba9000	If you are already in a method block (...) then its uses differently
06:37:25	renesac	like 'var p = some.long.namespace.printingfunction'
06:37:32	zezba9000	You mean like "Foo (someReallyLongProc) {...}" ??
06:37:50	zezba9000	someReallyLongProc is passed to a function pointion
06:37:53	zezba9000	**pointer
06:38:04	renesac	yes
06:38:09	renesac	p would be a function pointer
06:38:19	renesac	but nimrod has type inference
06:38:27	zezba9000	o ic your point
06:38:29	renesac	so you can type just what I did
06:39:11	zezba9000	Well I would require a sybole for that then maybe "var p = &some.long.namespace.printingfunction"
06:39:13	zezba9000	Or something
06:39:30	zezba9000	**symbol
06:40:21	zezba9000	Or you could do "var p = delegate(some.long.namespace.printingfunction)"
06:40:35	zezba9000	or "var p = delegate some.long.namespace.printingfunction"
06:41:06	zezba9000	so "delegate" would define that a method name without (..) is required
06:41:33	renesac	right
06:41:51	renesac	yeah, that would be a design choice
06:42:12	zezba9000	I think it would give context to what is going on
06:43:10	Araq	so ... your problem is {}is already taken in the default syntax, zezba9000 ?
06:43:18	renesac	but I guess it is too late to change it in nimrod, even if it indeed does not have any other downside
06:43:32	renesac	Araq, he wants to call a function like this:
06:43:32	renesac	https://gist.github.com/zezba9000/10015607
06:43:56	renesac	with the ( ) in the same column as the first line, in separated lines
06:44:42	zezba9000	Araq: not sure I follow what you mean by {} being already taken? {} would be used in my idea
06:45:21	Araq	never mind. I keep thinking you're trying to be helpful.
06:45:45	zezba9000	renesac: Or: https://gist.github.com/zezba9000/10015517
06:46:23	Araq	we could allow that but I don't see the point
06:46:24	zezba9000	lol I'm getting hammered with difficult question from renesac
06:46:38	zezba9000	To see if there are any flaws in my idea
06:47:25	renesac	Araq, the point is that it is his style preference, and in some cases it indeed does look nice
06:47:44	renesac	I personally don't really care though
06:47:45	zezba9000	Araq: The point being, you have the option to layout a block of code how you feel it bests fits and is most human readable. No indentation required
06:48:06	renesac	oh, and I'm agains the 'no indentation required' part
06:48:30	zezba9000	Well Don't get me wrong, you should indent a method and many thing
06:48:59	zezba9000	I think there is something to be said for this layout in some cases though: https://gist.github.com/zezba9000/10015517
06:49:04	Araq	zezba9000: and yet the cost is a redundancy which leads to questions like "what if the compiler and the layout disagree"
06:49:27	Araq	and also it leads to cascading }}}} which take up valuable screen space
06:49:32	zezba9000	Its not redundent if I have use for it. It helps me parse whats going on faster
06:49:39	zezba9000	To me its not redundent
06:49:40	Araq	welcome to world of trade-offs
06:50:08	Araq	it's obvjectively redundant, there is no "to me" here
06:50:09	zezba9000	No cascading happens with or without }}}
06:50:25	zezba9000	You simple don't use the } symbol and have a return whitespace instead
06:50:46	zezba9000	To you there is a "to me"...
06:51:04	Araq	"return whitespace" doesn't cascade
06:51:14	zezba9000	The same amount of space is the same in both
06:51:36	zezba9000	Do you litterally mean people put }}}} in a row?
06:51:39	zezba9000	No one does that
06:51:47	Araq	no that would be great
06:51:54	zezba9000	aww sry but I see your point
06:52:02	Araq	but indeed nobody does that so it takes up pages of space
06:52:03	renesac	https://gist.github.com/ReneSac/f3afa585841d72df2d02
06:52:08	zezba9000	no that would be so hard to parse
06:52:19	renesac	<-- you can't have this density with } in their own lines
06:52:58	Araq	zezba9000: the Lisp people do use ))))) without the newlines and it's easy to read, for me
06:53:00	zezba9000	Having a little extra space is blown way out of perpotion here. Its not a bad thing when it makes it easier to sperate whats going on
06:53:40	zezba9000	I'm going to convert renesac code rly quick to compare
06:54:00	Araq	it's not "little" extra space, it makes a factor of 1,5 to 2 for many things
06:54:38	zezba9000	I don't know what code your looking at. When you have huge blocks like that its best to give them space requardless of {}
06:55:34	zezba9000	renesac: is [A] a template method?
06:55:57	renesac	A is a generic type
06:56:06	zezba9000	k cool thought so
06:56:09	renesac	yes, the proc is generic
06:56:26	renesac	this is from tables.nim in the std library
06:56:44	zezba9000	well let me see if I understnad enough to convert it
06:57:14	renesac	it was the first relatively deep nested thing I found
06:58:16	Araq	zezba9000: this number comes from converting thousands of lines of Pascal to Nimrod where Pascal takes up the same amount of screen space as a C-like language
06:59:04	zezba9000	I'm sure its small, I'm just saying that super compressed code doesn't mean its a good thing in some cases.
06:59:18	zezba9000	It can be so compressed its hard to parse whats going on
07:00:57	zezba9000	let me finish the conversion so we have something better to compare
07:01:50	zezba9000	Is the "div" = "/" ??
07:02:06	renesac	integer division
07:02:06	renesac	yes
07:02:10	zezba9000	k
07:02:15	renesac	'/' is aways float division
07:02:24	zezba9000	what?!
07:02:43	zezba9000	Why is it not the one that doesn't loose resolution?
07:03:04	zezba9000	**lose
07:03:39	zezba9000	"var 1.0f / 1.0d" should return a double in my mind
07:03:42	renesac	you do have to convert explicitly the operands to float (by float I mean 32 or 64bit float)
07:03:47	zezba9000	That sounds like what VB .NET does
07:04:02	zezba9000	You end up getting bad performance
07:04:07	renesac	'float' in nimrod is usually a 64bit float
07:04:19	renesac	or a double as called in many languages
07:04:31	zezba9000	ok ic
07:04:34	renesac	you have float32 and float64 to specify
07:04:38	renesac	the exact size
07:04:54	renesac	float is the 'fastest float', and is float64 in x86
07:04:59	zezba9000	but if I do "var i = (float32)1 / (float64)1" is it 64 or 32 bit
07:05:10	zezba9000	what bit is 'i'
07:05:20	renesac	I think 64bit
07:05:24	Araq	float64 but it helps to learn the syntax first
07:05:41	zezba9000	ok cool
07:06:02	Araq	but I know, there is no need to learn when you can argue about minor asthetic details instead
07:06:30	Araq	renesac: actually the spec changed and 'float' is always 'float64' now
07:06:33	zezba9000	I understnad what is going on. This is all about the details
07:06:43	renesac	oh, that change got throught?
07:06:51	zezba9000	Araq: How come
07:06:52	Araq	yeah
07:06:57	renesac	I remember seeing the issue on github
07:07:04	zezba9000	I like the sound of float being the bit of the build?
07:08:00	renesac	zezba9000, you can aways alias your own float type, I guess...
07:08:01	zezba9000	if float was changed to 64bit all the time, it might be useful to have a 'Num' keyword or something idk
07:08:10	Araq	there were issues with operator overloading and the fact that math uses 'float' everywhere but wraps C's double functions
07:08:24	zezba9000	float being CPU agnositc makes so much sense though :)
07:08:26	Araq	*with overloading
07:08:41	Araq	yeah we need floatx for performance
07:09:00	zezba9000	aaa yes floatx is the best name for it
07:09:08	zezba9000	but then why not keep float at 32bit?
07:09:16	zezba9000	would make porting so much easier
07:09:35	renesac	by 'floatx' he meant 'float32' or 'float64'
07:09:48	Araq	renesac: no I did not
07:09:53	renesac	oh
07:09:56	zezba9000	what rly?
07:10:00	zezba9000	what did you mean
07:10:03	renesac	ok, I missunderstood
07:10:27	Araq	type floatx = when defined(x86): float64 else: float32
07:10:56	Araq	zezba9000: I wanted to improve the naming of things and "double" is a horrible name
07:11:00	zezba9000	then how is that not what renesac said?
07:11:14	Araq	argh
07:11:18	Araq	communication is hard
07:11:20	zezba9000	32bit for x86 and 64bit for x64
07:11:23	renesac	zezba9000, I though 'x' was placeholder for a number
07:11:24	zezba9000	yes I know lol
07:11:45	renesac	I'm going sleep
07:11:51	renesac	too late already
07:11:54	Araq	nowadays I would have picked double and float and half
07:12:03	Araq	since that's what IEEE uses
07:12:14	*	nande quit (Read error: Connection reset by peer)
07:12:14	zezba9000	o well whether or not you ment it. floatx is a great name for a compile determined type like IntPtr in C# or soemthing
07:12:31	Araq	I meant it. and thanks. lol
07:12:44	zezba9000	ya thats whats shader langs use to.
07:13:10	renesac	Araq, and 'byte, short, int, long'?
07:13:12	renesac	:P
07:13:14	zezba9000	or err HLSL uses float16 instead of half
07:13:30	renesac	(that is D's naming convention, which I like too)
07:13:33	zezba9000	renesac: I'll post the converted file so you can see it later
07:13:34	Araq	renesac: perhaps
07:14:07	renesac	I would be pro it, if it wasn't already that late
07:14:14	zezba9000	I like float = dynamic at compile-time. float32, float64, float16, float10 ect
07:14:17	Araq	what we have now is nice, but confusing for beginners
07:14:22	renesac	well, now really going
07:14:37	Araq	bye
07:24:00	zezba9000	renesac: This is how I would right it if I got everything right: https://gist.github.com/zezba9000/10016041
07:26:33	zezba9000	anyway i'm getting off here too
07:31:33	*	BitPuffin joined #nimrod
07:32:58	*	zezba9000 left #nimrod (#nimrod)
08:08:06	*	faassen joined #nimrod
08:29:07	*	io2 joined #nimrod
08:39:37	*	silven_ is now known as silven
09:30:20	*	shodan45 quit (Quit: Konversation terminated!)
10:03:53	*	supporters quit (Remote host closed the connection)
11:02:35	BitPuffin	ping fowl
11:03:56	Araq	pong BitPuffin
11:13:00	BitPuffin	Araq: hey fowl
11:19:10	*	io2 quit (Quit: ...take irc away, what are you? genius, billionaire, playboy, philanthropist)
11:29:02	BitPuffin	was fowl KIA?
11:40:17	*	delian66 quit (Read error: Operation timed out)
12:11:58	*	io2 joined #nimrod
12:22:14	NimBot	Araq/Nimrod devel be48988 Araq [+1 ±2 -0]: fixes #1063
12:25:44	*	ics joined #nimrod
12:34:38	EXetoC	ok so I forgot to enable dead code elimination and that's why it tried to load every OpenGL proc
12:45:59	BitPuffin	EXetoC: stOPID
13:05:15	*	BitPuffin quit (Ping timeout: 255 seconds)
13:14:58	*	BitPuffin joined #nimrod
13:18:31	*	darkf quit (Quit: Leaving)
13:46:03	EXetoC	BitPuffin: yeah maybe, but you are fat and stuff
13:46:12	NimBot	Araq/Nimrod devel 45817bc Araq [+0 ±1 -0]: make tests green again
13:46:29	*	faassen quit (Remote host closed the connection)
13:53:45	*	vendethiel quit (Ping timeout: 255 seconds)
13:54:07	*	Mat3 joined #nimrod
13:54:21	Mat3	hi all
13:54:30	*	vendethiel joined #nimrod
13:57:00	*	vendethiel quit (Read error: Connection reset by peer)
13:57:13	*	Guest78262 quit (Ping timeout: 240 seconds)
13:59:01	*	gabber808 joined #nimrod
13:59:10	*	Guest78262 joined #nimrod
14:00:35	gabber808	Hi everyone. I am finding the indentation rule a little stifling. Lets suppose I have a with events loop (sdl), and within that a while loop. Therefore I have two indentations. How should I be managing the indentation? Four single spaces? I have set the tab on my editor to represent two spaces, yet the compiler doesn't seem to like this.
14:01:28	Araq	yeah use four single spaces
14:01:39	Araq	make your editor emit spaces when you press tab
14:01:56	Araq	if your editor can't do this, get a real one
14:02:55	gabber808	OK, thanks Araq. BTW, with help from Dom (& your earlier help) I eventually got the SDL working. DOM managed to find all of the 1.2 sdl dll files
14:03:11	Araq	yeah I know, I read the logs
14:03:42	gabber808	error: tabulators not allowed.
14:04:01	*	vendethiel joined #nimrod
14:04:03	gabber808	Does this mean there is something wrong with my use of tabs
14:04:08	Araq	yes
14:04:18	Araq	make your editor emit spaces when you press tab
14:04:38	gabber808	do you mean omit?
14:04:59	Araq	emit as in "emit"
14:05:17	Araq	in fact, juse use Aporia
14:05:19	Araq	*just
14:05:43	gabber808	I am using Jedit. I couldn't seem to compile Aporia
14:06:41	Araq	http://nimrod-code.org/download/gtk.zip
14:06:49	Araq	http://nimrod-code.org/download/aporia.zip
14:06:59	Araq	old but working build of Aporia
14:07:33	gabber808	Thanks
14:07:41	*	vendethiel quit (Remote host closed the connection)
14:08:58	gabber808	Where should I place the gtk files?
14:09:21	Araq	unzip both into a common directory
14:10:00	Araq	bbl
14:10:24	gabber808	There are 165 files with the same name (when I unzip both into a common directory). Is it ok to overwrite?
14:12:40	*	vendethiel joined #nimrod
14:15:00	renesac	gabber808, I think you need to enable 'Soft (emulated with spaces) tabs' in jEdit editing options
14:16:08	renesac	and go in 'Edit -> Indent -> Tabs to Spaces' to fix what you did so far
14:18:24	gabber808	Hi renesac. I tried that. Let me try again
14:22:04	gabber808	I am still getting invalid indendation error. I have "automatic indentation: full" , "tab width: 2" and "indent width: 2" + soft
14:22:08	renesac	don't forget to convert the tabs you already typed
14:23:19	gabber808	yes, I deleted them and retabbed
14:23:33	renesac	paste your code somewhere for we to see
14:23:38	renesac	like a gist
14:24:24	EXetoC	are you able to select individual spaces?
14:24:44	gabber808	Thanks. It is complaining about line 10:
14:24:46	gabber808	http://pastebin.com/4c1EhqjQ
14:25:16	EXetoC	complaining about what?
14:25:34	gabber808	invalid indentation
14:26:29	EXetoC	hm, x++?
14:27:28	renesac	yeah, that is the problem
14:27:38	renesac	nirmod don't have postfix operators
14:27:44	renesac	(except [])
14:28:03	gabber808	I see, thanks both. So it wasn't actually an indentation error, despite it reporting that !
14:28:03	renesac	you want x.inc
14:28:08	EXetoC	yeah, deal with the errors in series
14:28:27	renesac	or inc(x)
14:28:37	renesac	or x += 1
14:29:01	renesac	I made a `++` prefix operator in my bithacks module though
14:29:04	EXetoC	inc is the most common for ints. "inc x" works too. I prefer that one
14:29:33	gabber808	or x=x+1
14:29:51	renesac	yes
14:30:30	gabber808	well, my animation is nearly working. Just need to find how to clear the sdl surface
14:31:49	EXetoC	'graphics' can't be used alone?
14:32:08	EXetoC	I haven't used it so I wouldn't know
14:33:14	BitPuffin	EXetoC: :'(
14:33:28	BitPuffin	EXetoC: now that's not very nice ;_;
14:33:39	*	Demos joined #nimrod
14:34:00	gabber808	What is graphics exactly? Doesn't it use SDL anyway? Is it a different set of commands?
14:34:23	BitPuffin	gabber808: believe it's a higher level api that uses sdl in the background
14:35:05	gabber808	We could really do with a bit more documentation on how to use these things
14:36:14	BitPuffin	it's there somewhere
14:36:26	*	severholding joined #nimrod
14:40:35	gabber808	Found it under "other". I'd have thought it would be under graphics !
14:43:44	Demos	yeah the graphics package could use more docs. There are some tests and examples at the end of graphics.nim though
14:44:03	Demos	I know directing you to look at the source of the module is not ideal, but that is how it is
14:45:10	EXetoC	BitPuffin: :p
14:45:44	gabber808	And here is my first animation if anybody is interested ! ,..
14:45:45	gabber808	http://pastebin.com/4ny1aLFM
14:46:40	Demos	gabber808: protip: I bet you can render a cairo memory buffer with the graphics module, that could be good if graphics lacks something you need
14:48:16	*	severholding quit (K-Lined)
14:48:17	gabber808	I think graphics/sdl will be fine. The only thing that is holding me back a little is having to compile/debug through the console everytime to find errors.
14:50:53	EXetoC	that's a given ;)
14:51:12	Demos	gabber808: I think nim-lime and zach's vim plugin let you compile
14:51:30	Demos	I do know that syntastic works if I have nim-lime (not that well, but it catches some errors)
14:51:37	Demos	nimrod's compile times are wonderful as always
14:51:52	EXetoC	nim-line?
14:51:55	EXetoC	it works pretty well imo
14:51:58	Demos	sorry, nim-lime
14:52:18	Demos	oh no, syntastic
14:52:18	EXetoC	you sure? :p
14:52:45	Demos	and I have not noticed nim-lime's idetools driven completion, but my project has subfolders and that could confuse it
14:53:36	EXetoC	where do I find this ungoogleable thing that is nim-line?
14:53:40	EXetoC	lime
14:56:18	gabber808	i couldn't get nimlime to work
14:56:32	EXetoC	nevermind
14:57:48	EXetoC	I don't know why I quoted "nim-lime", but apparently it matters in this case
15:00:28	Demos	I usually just seach on github
15:03:33	*	Guest78262 quit (Ping timeout: 240 seconds)
15:04:14	*	Guest78262 joined #nimrod
15:12:53	*	rta quit (Ping timeout: 246 seconds)
15:13:45	*	Demos quit (Quit: http://quassel-irc.org - Chat comfortably. Anywhere.)
15:13:47	*	ics quit (Quit: My MacBook Pro has gone to sleep. ZZZzzz…)
15:13:55	*	Demos joined #nimrod
15:17:01	Mat3	is the new version out now ?
15:17:48	*	Guest78262 is now known as psquid
15:19:18	EXetoC	Mat3: no. it might get released today
15:20:32	OrionPK	https://github.com/Varriount/NimLime
15:25:10	dom96	hello
15:25:40	Demos	so last night I was debugging and I un-exported an object field and made two "property" functions that used the getStackTrace() proc. It was epic
15:26:00	EXetoC	dom96: morning
15:26:21	*	rta joined #nimrod
15:26:40	EXetoC	Demos: indeed. I forgot all about it
15:26:44	dom96	EXetoC: afternoon
15:26:48	dom96	hello rta
15:26:57	Demos	EXetoC: about getStackTrace()?
15:27:06	EXetoC	yeah
15:27:55	dom96	I really would love to see a version of getStackTrace which doesn't return a string.
15:29:17	Mat3	EXetoC: Sublime is one of these newer online editors ?
15:30:11	EXetoC	nope
15:30:26	*	ics joined #nimrod
15:32:05	*	Demos quit (Quit: http://quassel-irc.org - Chat comfortably. Anywhere.)
15:32:09	EXetoC	it's a native app, first released about 6 years ago apparently
15:32:15	*	Demos joined #nimrod
15:32:47	Mat3	hmm, looks like a nice editor, however not open-sourced
15:33:06	Demos	I use it on windows, although I may switch to vim
15:33:55	dom96	Mat3: Have you tried Aporia?
15:35:18	Mat3	no, I can not compile it (and if so, I would miss a language translation)
15:35:49	dom96	what's the error?
15:36:26	dom96	oh, you would like the menus etc to be in your native language?
15:37:17	Demos	wow all the khronos group websites are down
15:39:31	*	io2 quit (Quit: ...take irc away, what are you? genius, billionaire, playboy, philanthropist)
15:39:56	EXetoC	it was not down yesterday but no requests could be completed
15:40:28	Mat3	dom96: Yes. If I remember correctly, the error was related to the GTK development packages in Ubuntu
15:41:16	Mat3	have you ported Aporia to GTK 3 ?
15:43:02	dom96	no, it's still gtk2.
15:43:21	*	Mat3 is using gEdit for Nimrod related development
15:44:19	dom96	You should really try to get Aporia working then.
15:44:37	dom96	Aporia is similar to gedit, but has nicer Nimrod support IMO
15:46:03	Mat3	hmm, ok - I try to compile it next week (hope to find some free time)
15:46:37	Mat3	probably I can add some translations
15:48:16	Demos	hm reading a bit about scala, it seems insane. I hope nimrod never grows that complex
15:48:37	dom96	what in particular seems insane about it?
15:49:55	Demos	there are like 15 different kinds of types (like T, T+, T- T#i (I think that is like nimrod's T.i or type(thing).i not sure), and so on) what looks like two different ways to define functions. The whole path dependent type thing seems un-nessassary
15:58:06	*	gabber808 quit (Quit: Page closed)
16:00:16	Mat3	Demos: Reads like a Hindley–Milner like type system
16:01:08	*	Mat3 founds the implicit logic of type systems general questionable
16:02:14	*	Demos quit (Ping timeout: 246 seconds)
16:03:50	EXetoC	too many hidden aspects?
16:07:49	Mat3	who knows
16:14:31	*	Demos joined #nimrod
16:24:34	Demos	Mat3: but I was under the impression that mixing subtypes into a HindleyMiller causes problems
16:34:07	*	psquid quit (Ping timeout: 252 seconds)
16:34:14	*	Mordecai joined #nimrod
16:34:33	*	Mordecai is now known as Guest27830
16:38:13	renesac	http://build.nimrod-lang.org/docs/tut1.html#result-variable <-- this is a bad example
16:38:31	renesac	nobody writes result = 42
16:38:31	renesac	return
16:38:54	renesac	that return is redundant in this case
16:39:26	renesac	in old versions of nimrod it was needed?
16:41:08	renesac	that bad tutorial was the reason I didn't understood 'result' at first, and even though it was useless.
16:41:09	EXetoC	either that or someone just forgot to include that variation
16:41:34	renesac	it also isn't clear that you can't do 'var result = x'
16:42:17	renesac	I think it is better to include a full function as example
16:43:38	*	shodan45 joined #nimrod
16:43:45	EXetoC	we simply need more volunteers
16:44:37	renesac	I think the factorial function is a good example
16:45:38	renesac	I'm editing it
16:46:00	renesac	I will also add the 'var parameter = parameter' idiom in the next topic
16:46:13	dom96	I think a string builder is a good example for result.
16:46:34	Demos	I was all confused because I thought the primary motivation of result was to ensure return values need-not be copied
16:46:47	renesac	that I didn't know it was possible until I heard in irc
16:46:53	renesac	dom96, do you have it written?
16:47:02	renesac	that string builder function?
16:47:25	dom96	No.
16:47:31	dom96	Should be pretty trivial though.
16:47:45	renesac	so write it, otherwise I will put the factorial function
16:48:08	dom96	result = ""; result.add("Foobar"); result.add(myVariable)
16:48:23	dom96	vs. return "Foobar" & myVariable
16:48:29	dom96	meh. that's a crappy example
16:48:33	dom96	nvm
16:48:33	renesac	yeah..
16:48:46	renesac	the cool thing about factorial is that it ends in a for loop
16:48:54	*	Varriount is awake
16:49:03	renesac	so you save a line
16:49:39	dom96	where is Araq? It's release day!
16:49:52	Varriount	Have the branches been merged yet?
16:50:49	dom96	No.
16:51:04	dom96	Varriount: Can you try to figure out why the tester reports 12 more tests in win-x86?
16:53:36	Mat3	Demos: indeed
16:59:11	Varriount	dom96: Babel packages
16:59:25	dom96	ahh
17:00:12	*	DAddYE joined #nimrod
17:00:14	Varriount	I don't know why they aren't showing up in the 64 bit version though.
17:02:43	*	Demos quit (Ping timeout: 252 seconds)
17:07:38	Araq	dom96: I'm here now
17:10:22	*	q66 joined #nimrod
17:15:31	Araq	so Varriount are you ready?
17:15:44	Varriount	Yes.
17:15:47	Araq	:O
17:15:53	Araq	what's your local time?
17:15:59	Varriount	1:15 pm
17:16:29	Araq	ok good, well I'm trying to make async IO stable but even if I fail, we'll release tonight
17:20:17	*	BitPuffin quit (Ping timeout: 246 seconds)
17:27:31	*	nequitans joined #nimrod
17:27:52	*	askatasuna joined #nimrod
17:35:10	*	zezba9000 joined #nimrod
17:37:37	*	nande joined #nimrod
17:39:24	reactormonk	Araq, decided on git purge?
17:40:02	Araq	did you make an announcement on the forum? sent emails warning the people who have forks?
17:40:06	Varriount	^ I can send bug reports to all fork owners
17:40:40	Varriount	But first, a shower.
17:41:37	reactormonk	good, I'll write the announcement.
17:41:46	reactormonk	but first, I have to figure out how to deal with patches.
17:42:00	Araq	alright thanks
17:42:13	*	bitcrusher joined #nimrod
17:42:26	EXetoC	how significant is the hasKey+mget overhead on large tables compared to just querying the table once and getting a pointer?
17:42:38	EXetoC	not that it'll matter often
17:42:44	Araq	hi bitcrusher
17:42:56	EXetoC	I guess it's obvious, assuming that there's no caching going on
17:43:07	bitcrusher	hello Araq, just checking out the language
17:43:09	Araq	;-)
17:43:43	Araq	I still wonder if we should do 0.9.4.0 ...
17:44:02	Araq	4th version number would simply be honest, I think
17:44:33	dom96	What software do you know of that uses 4 version numbers?
17:45:00	Araq	microsoft often uses 4 numbers
17:45:34	Araq	if 0.9.6 breaks some code in comparison to 0.9.4, and it WILL do that
17:45:43	Araq	then we have no patch number left
17:46:04	EXetoC	major.minor.revision.cosmetics
17:48:12	Araq	we could also drop the leading 0 ...
17:50:17	Araq	oh well patches go to master, new development to devel
17:50:21	reactormonk	Araq, pushed a branch purged_master
17:50:34	Araq	once 0.9.4 is out, that is
17:50:38	EXetoC	should "" != nil work?
17:51:53	Araq	EXetoC: I think so
17:52:23	reactormonk	Araq, I suppose devel and master should have the same ancestry?
17:52:44	reactormonk	or will you discard master and use devel from now on?
17:53:02	Araq	we can do that
17:53:16	Araq	discard master, introduce 'stable'
17:53:28	Araq	er ugh, no
17:53:35	EXetoC	Araq: ok, will report it. I get "gen.nim(11, 14) Error: internal error: getStr"
17:53:42	reactormonk	the question is - will you discard the current data of master?
17:53:52	reactormonk	or do I have to make sure that devel and master share a common ancestry
17:53:52	Araq	"git origin master" is an idiom... I think
17:54:05	reactormonk	it's not about the naming, but the data here
17:54:24	Araq	the data of master is irrelevant by now
17:54:36	Araq	unless I missed some commit that's not in devel
17:54:53	reactormonk	ok
17:55:19	reactormonk	there's your purged_devel
17:55:36	reactormonk	now to find out how to rebase
17:55:40	NimBot	Araq/Nimrod purged_devel 5a448cd Araq [+0 ±5 -0]: 'bind' default for clean templates
17:55:40	NimBot	Araq/Nimrod purged_devel c773d60 Araq [+0 ±3 -0]: made some tests green
17:55:40	NimBot	Araq/Nimrod purged_devel cdcbfa0 Araq [+0 ±1 -0]: attempt to improve line information for '!=' etc. templates
17:55:40	NimBot	Araq/Nimrod purged_devel cc4b951 Araq [+0 ±11 -0]: compiler bootstraps with new VM
17:55:40	NimBot	662 more commits.
17:56:00	reactormonk	.... go figure.
17:56:37	dom96	could we maybe worry about purging after release?
17:56:59	reactormonk	dom96, sure, but it makes it easier doing it at the split - at least to my understanding
17:58:51	*	brson joined #nimrod
18:07:50	reactormonk	a bit more messy than I expected :-/
18:15:04	fowl	662 commits :D
18:15:16	Varriount	Araq: Do you want me to start sending out the notifications?
18:15:23	fowl	hello whoever pinged me
18:15:45	Araq	Varriount: wait for reactormonk please. he said it's "messy"
18:35:58	*	BitPuffin joined #nimrod
18:42:26	*	Varriount_ joined #nimrod
18:42:55	Varriount_	Ok, I'm currently writing a generic issue message for the rewrite - besides deleting and re-forking the main nimrod repository, what other methods are available to recover from the re-written git history?
18:43:23	*	Varriount quit (Disconnected by services)
18:43:30	*	Varriount_ is now known as Varriount
18:43:44	reactormonk	Varriount, git rebase
18:43:55	Varriount	Any specific commands?
18:46:22	*	io2 joined #nimrod
18:51:33	Varriount	Wait, was the purge done while I was disconnected?
18:51:53	Varriount	Also, I never noticed that forks lacked issue areas before. :/
18:55:13	*	BitPuffin quit (Ping timeout: 240 seconds)
18:57:32	*	q66 quit (Ping timeout: 252 seconds)
19:00:56	Varriount	Araq: With the absense of an issues page on forks, and emails on only about a quarter of the users who have forked nimrod, I don't think I'll be able to notify them all.
19:06:37	EXetoC	hopefully you can at least enable it explicitly..
19:07:21	*	Mat3 is now known as Mat3-work
19:07:27	EXetoC	yeah, you can
19:09:13	Varriount	Does nimrod have an official email address or something?
19:09:35	dom96	Just announce it on the forum.
19:09:50	*	Matthias247 joined #nimrod
19:09:50	*	q66 joined #nimrod
19:09:54	Varriount	-_-'
19:09:59	*	ics quit (Quit: My MacBook Pro has gone to sleep. ZZZzzz…)
19:13:38	Varriount	Araq: Is this message ok? -> https://gist.github.com/Varriount/51165a76caa69415cb22
19:24:33	renesac	what about asking github help for this?
19:24:46	Varriount	Huh?
19:24:58	renesac	they have a contact form
19:25:17	renesac	maybe they can help rebasing nimrod
19:25:19	Varriount	This isn't really help.. it's more like.. communication.
19:25:28	renesac	like, how to send a notification to all forks
19:25:47	renesac	what is the perfect git command line for forkers to use after this
19:25:48	renesac	etc
19:26:14	renesac	it is their interest also new forks not having to download 300mb of data
19:26:23	Varriount	Well, at the moment, I'm gathering email addresses and contacts. Feel free to modify that message.
19:26:50	Matthias247	if you do a git fetch you would get informed anyway that it was a forced update
19:27:04	Araq	Varriount: I think the text is just fine
19:28:57	Matthias247	my normal update process is git fetch and then git merge origin/master --ff-only. This would at least give me an error that something happened
19:31:32	*	Demos joined #nimrod
19:31:40	renesac	Araq, 0.10 ?
19:32:12	Varriount	I'm almost done with contact gathering.
19:32:28	Araq	renesac: hmm good idea but also ... meh
19:32:39	Araq	this way people never know when it ends
19:32:53	Varriount	0.0000001
19:34:17	EXetoC	yeah it's inconsistent
19:34:56	*	q66 quit (Ping timeout: 252 seconds)
19:35:20	renesac	2.6.32-431.3.1.el6 ?
19:35:28	renesac	(a linux version)
19:35:57	renesac	a - instead of a third dot
19:36:37	dom96	0.9.5-2014.04.07
19:36:54	*	q66 joined #nimrod
19:36:54	dom96	I still think the idea of a date is nice.
19:37:07	renesac	or just release the 1.0 for no reason other than getting out of this . hell, like linux did
19:37:08	renesac	XD
19:37:17	dom96	Won't be compatible with babel versions though.
19:37:36	renesac	dom96, it isn't clear if the date is something additional, or if it is also specifying a more specific version
19:37:55	Araq	we'll have master aggregate the bugfixes and devel will be cutting edge
19:38:01	renesac	*linus
19:38:38	renesac	Araq, nice
19:38:47	dom96	nimbuild already provides latest builds
19:39:06	Araq	and the versioning can stay as it is
19:39:08	renesac	should the download page redirect to nimbuild?
19:40:16	Demos	wow that debate last night went on for another /two hours/
19:40:19	dom96	The download page will contain a table with a button for each cpu/os combo and it will point to nimbuild. I have some things in NimBuild that I need to modify before that is possible though.
19:40:31	Demos	wadler's law in action
19:40:51	EXetoC	Demos: about the syntax?
19:41:07	Demos	http://www.haskell.org/haskellwiki/Wadler%27s_Law
19:41:29	dom96	We'll probably spend another couple of hours arguing with redditors about syntax once we release.
19:42:14	EXetoC	/ignore :>
19:42:57	Demos	we will also get c++ "programmers" (I suspect these people do little actual programming) bitching about how we rely on a GC and you should NEVER EVER EVER use manual management
19:43:10	EXetoC	so do you know when the optimal time to announce is?
19:44:51	NimBot	Araq/Nimrod devel 7780d81 Dominik Picheta [+0 ±1 -0]: Change the expected path in tfailedassert.... 4 more lines
19:44:51	NimBot	Araq/Nimrod devel f0ea793 Dominik Picheta [+1 ±3 -0]: Merge branch 'devel' of github.com:Araq/Nimrod into devel
19:44:51	NimBot	Araq/Nimrod devel 0ccea55 Dominik Picheta [+0 ±2 -0]: Disable async macro code output. Added some WSA error codes to winlean.
19:44:59	EXetoC	I just took that at face value before. It is of course a nonsensical argument
19:45:36	Demos	EXetoC: took what at face value, the GC hate or the wierd "pointers are bad but wrapping them in a class makes them OK" thing
19:46:22	*	BitPuffin joined #nimrod
19:46:31	EXetoC	the former
19:47:16	EXetoC	I guess many people use java as a reference
19:47:34	Demos	yeah same. Apperently I had a prolonged argument with my lisper friend over this... although neither of us remember that argument
19:47:38	fowl	i am shooting myself in the foot: https://gist.github.com/fowlmouth/10002354
19:47:43	Demos	but it is a right tool for the right job thing
19:49:03	Demos	can you actually draw those JSON trees now?
19:49:12	reactormonk	Varriount, figuring them out
19:49:31	fowl	Demos, they get parsed into widgets, but yes
19:49:42	Demos	gtk widgets?
19:49:52	fowl	no
19:50:42	fowl	https://gist.github.com/fowlmouth/ff6b7698b76a8cc408ae
19:51:17	Varriount	reactormonk: Figuring what out?
19:51:57	*	BitPuffin quit (Ping timeout: 255 seconds)
19:52:04	reactormonk	Varriount, the correct commands for everyting
19:52:10	reactormonk	git clone [email protected]:Araq/Nimrod.git --single-branch -b purged_devel # <- this gives you the new repo
19:52:19	Demos	I do something similar, except I only have buttons and my storage format is binary (which I think it "A Very Bad Idea")
19:52:35	Varriount	Araq: As a last resort, do you want me to contact users through the issues pages of any of their own repositories which use nimrod as a langauge?
19:52:47	Demos	esp since any pointers in your data will cause lots and lots of pain
19:53:35	Demos	fowl: I think you could set all the procs in your vtbl to {.nimcall.} to save 8bytes per proc (big whoop I know)
19:54:08	Varriount	dom96: You sure that you don't have any email software setup on on the server? It might look a bit odd for a student to be emailing all these people.
19:54:18	Demos	and yes I am that nitpicking arsehole
19:55:20	dom96	Varriount: No, sorry I don't.
19:55:29	Varriount	Then again, I could always forgo using my personal email, and use the email my college gave me...
19:55:32	fowl	Demos, thanks
19:55:41	Varriount	That might look even more odd though.
19:56:13	fowl	Demos, im going to open a new repos for this, i write a middle-man connection manager for enet that i like a lot
19:56:42	Demos	yeah, another idea is to have a `.` that can call methods and do all the relevant casting
19:59:27	Araq	Varriount: the "official" email address is my personal email address lol
19:59:59	Araq	but just use your own, it's fine, you are part of the core team
19:59:59	Varriount	:/
21:08:30	*	NimBot joined #nimrod
21:09:20	*	Varriount|Mobile joined #nimrod
21:09:20	*	nequitans joined #nimrod
21:09:20	*	Skrylar joined #nimrod
21:09:20	*	ics joined #nimrod
21:09:20	*	bstrie_ joined #nimrod
21:09:20	*	q66 joined #nimrod
21:09:20	*	Demos joined #nimrod
21:09:20	*	Matthias247 joined #nimrod
21:09:20	*	io2 joined #nimrod
21:09:20	*	Varriount joined #nimrod
21:09:20	*	brson joined #nimrod
21:09:20	*	bitcrusher joined #nimrod
21:09:20	*	askatasuna joined #nimrod
21:09:20	*	DAddYE joined #nimrod
21:09:20	*	shodan45 joined #nimrod
21:09:20	*	rta joined #nimrod
21:09:20	*	vendethiel joined #nimrod
21:09:20	*	isenmann joined #nimrod
21:09:20	*	OrionPK joined #nimrod
21:09:20	*	fowl joined #nimrod
21:09:20	*	comex joined #nimrod
21:09:20	*	Raynes joined #nimrod
21:09:20	*	silven joined #nimrod
21:09:20	*	noam joined #nimrod
21:09:20	*	zahary1 joined #nimrod
21:09:20	*	eigenlicht joined #nimrod
21:09:20	*	runvnc joined #nimrod
21:09:20	*	renesac joined #nimrod
21:09:20	*	Amrykid joined #nimrod
21:09:20	*	Araq joined #nimrod
21:09:20	*	tumak joined #nimrod
21:09:20	*	cark joined #nimrod
21:09:20	*	Zuchto_ joined #nimrod
21:09:20	*	flyx joined #nimrod
21:09:20	*	EXetoC joined #nimrod
21:09:20	*	reloc0 joined #nimrod
21:09:20	*	rixx joined #nimrod
21:09:20	*	eximiuswastaken joined #nimrod
21:09:20	*	oxful joined #nimrod
21:09:20	*	Trixar_za joined #nimrod
21:09:20	*	krusipo_ joined #nimrod
21:09:20	*	Ycros joined #nimrod
21:09:20	*	musicalchair joined #nimrod
21:09:20	*	zielmicha_beta joined #nimrod
21:09:20	*	phI||Ip joined #nimrod
21:09:20	*	seubert joined #nimrod
21:09:20	*	mal`` joined #nimrod
21:09:20	*	dom96 joined #nimrod
21:09:20	*	Boscop joined #nimrod
21:09:20	*	reactormonk joined #nimrod
21:09:20	*	betawaffle joined #nimrod
21:09:20	*	iNode001 joined #nimrod
21:09:20	*	JStoker joined #nimrod
21:09:20	*	Roin_ joined #nimrod
21:09:20	*	epsylon joined #nimrod
21:23:38	Araq	so ... what's left to do?
21:25:08	*	Guest27830 joined #nimrod
21:25:50	*	BitPuffin joined #nimrod
21:31:40	dom96	let's release already
21:38:57	OrionPK	fix osx?
21:39:15	Araq	OrionPK: what's wrong with osx?
21:39:33	OrionPK	as of last night my proj wasnt building
21:39:50	OrionPK	i pasted the error up in here
21:39:57	OrionPK	log doesnt go that far back though
21:40:18	Araq	well repeat the error message please
21:40:21	OrionPK	saw someone fixed selectors, but then another compiletime error happened
21:40:31	OrionPK	k, when I get home. Im at work and dont have access to my mac atm
21:41:25	Demos	I would test but I have a chemestry exam...
21:41:29	Demos	speaking of which, I gotta go
21:41:31	*	Demos quit (Remote host closed the connection)
21:44:24	OrionPK	if someone scrolls up, it's probably one of the last things i said
21:44:55	Araq	well async IO is not ready at all in this release
21:45:08	Araq	we have lovely random crashes when stress testing things
21:45:55	OrionPK	ah
21:46:28	OrionPK	idk if im even using asyncio, just using another module that uses asyncio, and not using its asyncio functionality
21:46:41	OrionPK	a stdlib module
21:46:44	Araq	oh
21:51:50	*	shodan45 quit (Quit: Konversation terminated!)
21:52:33	*	Guest27830 quit (Ping timeout: 255 seconds)
21:53:08	*	nequitans quit (Ping timeout: 245 seconds)
21:53:19	*	Guest27830 joined #nimrod
21:57:36	Araq	ping Varriount, Varriount|Mobile
22:00:31	Varriount|Mobile	Hm?
22:01:12	Varriount|Mobile	Araq: Yes?
22:02:10	Araq	I think we should purge the repo today, now but do the release somewhat later
22:02:28	Araq	it annoys me too much asyncio is *almost* ready
22:02:45	Varriount|Mobile	Ok then, purge away
22:03:04	Araq	also we didn't test on macosx enough, I think
22:03:08	dom96	asyncdispatch not asyncio
22:03:33	Araq	though zahary1 likely tested everything on mac
22:03:57	Araq	well reactormonk didn't write the forum message
22:04:13	Araq	did you send the messages around, Varriount|Mobile ?
22:04:52	Varriount|Mobile	I wasn't going to send them till the purge actually occurred
22:05:42	Varriount|Mobile	The composed email is open on my desktop, unsent.
22:08:54	dom96	I don't think we can do the purge as long as there are PRs open
22:09:38	NimBot	Araq/Nimrod devel fcb472d flaviut [+0 ±1 -0]: Add renderRstToJson in docutils
22:09:38	NimBot	Araq/Nimrod devel 2dd98ee flaviut [+0 ±1 -0]: Modify renderRstToJson to use the json library
22:09:38	NimBot	Araq/Nimrod devel c473ede Andreas Rumpf [+0 ±1 -0]: Merge pull request #1069 from flaviut/devel... 2 more lines
22:13:18	fowl	hrm
22:14:19	fowl	is this normal: https://gist.github.com/fowlmouth/d54a36d05fb1878083fe
22:14:31	fowl	/tmp/aporia/a26.nim(2, 5) Error: undeclared identifier: 'a28'
22:16:26	dom96	nope, that's a bug.
22:17:12	Araq	no it's a feature but the compiler should tell you
22:17:23	Araq	you can't forward module names
22:18:06	dom96	oh, I was thinking of this bug: https://github.com/Araq/Nimrod/issues/1029
22:19:18	fowl	thats whack >_>
22:27:32	Skrylar	Araq: if you think my coding style is erratic, look at the ansi c version of the porter stemmer o_o
22:27:38	Skrylar	crazy brace soup
22:27:53	*	io2 quit (Quit: ...take irc away, what are you? genius, billionaire, playboy, philanthropist)
22:31:07	Araq	Varriount|Mobile: save your email, dom96 is right I need to merge PRs ...
22:31:13	OrionPK	araq here's the error: lib/core/macros.nim(208, 21) Error: for a 'var' type a variable needs to be passed
22:31:31	dom96	OrionPK: What's your code?
22:31:53	dom96	or is that just compiling macros.nim?
22:31:56	OrionPK	sec, im determining the piece thats causing it
22:34:13	BitPuffin	Araq: tell me something nimrod can do that C++ can't :D
22:34:25	BitPuffin	I'm trying to win people over to nimrod \o/
22:34:49	NimBot	Araq/Nimrod devel 7fc5ae9 Araq [+0 ±5 -0]: preparations for easier debugging
22:34:49	NimBot	Araq/Nimrod devel 231a096 Araq [+0 ±4 -0]: Merge branch 'devel' of https://github.com/Araq/Nimrod into devel
22:35:17	BitPuffin	so far I've said term rewriting macros
22:35:26	Araq	BitPuffin: my usual answer: Everything that I do in my emerging langs talk is either very hard or downright impossible in C++
22:36:00	dom96	BitPuffin: Who are you trying to convince?
22:36:05	OrionPK	hmm
22:36:08	OrionPK	I guess it's the template stuff
22:36:29	OrionPK	i.e. https://github.com/onionhammer/onion-nimrod/tree/master/templates
22:36:47	BitPuffin	Araq: you mean strangeloop?
22:36:54	BitPuffin	dom96: an old irc buddy
22:36:59	fowl	src/gamestates.nim(136, 29) Warning: Cannot prove that 'evt' is initialized. This will become a compile time error in the future. [ProveInit]
22:37:05	fowl	how do you get past this
22:38:06	Araq	fowl: relax. the future is still away. report a bug if you think it is one.
22:38:26	*	bitcrusher quit (Write error: Connection reset by peer)
22:38:28	Araq	BitPuffin: yes
22:38:44	*	bitcrusher joined #nimrod
22:42:02	*	mal`` quit (Quit: ERC Version 5.3 (IRC client for Emacs))
22:43:33	BitPuffin	Araq: hehe, when I explained term rewriting macros with the simple example of rewriting x * 2 to x + x, he said that compilers usually catch that. But then I was like sin(x) + cos(x) = 1, let's see your compiler catch that :D
22:45:51	Araq	BitPuffin: no no no you're doing it wrong
22:46:16	BitPuffin	Araq: waii kawai
22:46:19	Araq	x * 2 -> x+x # does C++ for built-in types but not for your bignum library. That is the difference.
22:46:34	BitPuffin	ah true
22:46:38	BitPuffin	I should mention that
22:47:42	EXetoC	optimizers have a hard time with sin(x) + cos(x)?
22:48:08	BitPuffin	EXetoC: they probably do
22:48:14	EXetoC	sin(x) + cos(x) -> fsincos?
22:48:24	BitPuffin	EXetoC: sin(x) + cos(x) = 1
22:48:43	*	mal`` joined #nimrod
22:49:34	BitPuffin	Araq: btw we should add the term rewriting macro for sin(x) + cos(x) in math.nim I think
22:49:39	BitPuffin	it would gief penis
22:49:47	EXetoC	no that's not what fsincos does
22:50:36	Araq	BitPuffin: optimizations are useful when they transform real world code
22:50:52	BitPuffin	Araq: well it can happen in real code
22:51:05	Araq	sin(x) + cos(x) is unlikely to occur in real code
22:51:13	BitPuffin	Araq: btw his answer to the 2*x thing was that you write code so that it is inlined, and therefor it does
22:51:19	EXetoC	{.error: "silly operation".}
22:51:26	BitPuffin	Araq: in scientific computing it might
22:52:48	BitPuffin	and probably more efficiently, hmm probably not
22:55:15	Araq	inlining doesn't perform x*2 -> x+x, that's just bullshit
22:56:45	BitPuffin	Araq: well I think he meant that inlining would perform x*2
22:57:11	BitPuffin	at compile time or something
22:57:12	fowl	Araq, proveinit always comes up when declaring a TEvent
22:57:23	BitPuffin	fowl: is the allegro wrapper done? I wanna code some nimrod
22:57:54	BitPuffin	I have a game that is perfectly sized project that would be awesome in nimrod :)
22:57:58	fowl	BitPuffin, aside from some audio functions, yes
22:58:11	BitPuffin	fowl: ah, well then I should at least eb able to start using it
22:59:31	fowl	BitPuffin, please fix it for windows
23:00:06	*	Matthias247 quit (Quit: Matthias247)
23:00:46	OrionPK	ok
23:00:50	OrionPK	node.add newCall("add", ident("result"), newCall("$", ident(identifier)))
23:01:23	OrionPK	lib/core/macros.nim(335, 21) Error: type mismatch: got (int literal(5), string)
23:01:55	OrionPK	ident in this case is const x = 5
23:02:27	OrionPK	it expected: system.add(x: var string, y: string)
23:02:39	fowl	did you import macros
23:02:57	OrionPK	yes, this code works on main but not devel
23:03:17	OrionPK	newCall("$", ident(identifier)) should be translated to $x where x = 5
23:03:43	OrionPK	code is up on this repo: https://github.com/onionhammer/onion-nimrod/tree/master/templates
23:04:05	Araq	OrionPK: please make a proper bug report
23:04:31	BitPuffin	fowl: but I barely use winblows
23:05:29	OrionPK	araq I want to narrow it down further than "this huge codefile doesnt work anymore"
23:05:35	OrionPK	just seeing if anyone else had any insight
23:05:46	*	CARAM joined #nimrod
23:06:06	BitPuffin	fowl: other than for playing games for let's plays
23:10:51	*	Guest27830 quit (Ping timeout: 255 seconds)
23:12:39	fowl	OrionPK, last time i wrote a macro the error line was off by one
23:13:57	fowl	OrionPK, where is the macro
23:14:05	OrionPK	in templates.nim
23:14:17	*	Varriount|Mobile quit (Remote host closed the connection)
23:14:22	OrionPK	line 251 is pretty much the heart
23:14:25	OrionPK	compile time proc
23:16:34	*	darkf joined #nimrod
23:17:26	OrionPK	I think it has to do w/ the insertionPoint
23:19:26	dom96	good night
23:21:09	BitPuffin	dom96: or is it!?!?!?!
23:21:10	*	Demos joined #nimrod
23:21:11	BitPuffin	NUNUNUNUUU
23:22:27	*	Guest27830 joined #nimrod
23:22:31	*	Guest27830 is now known as psquid
23:22:38	*	CARAM quit (Remote host closed the connection)
23:22:44	Demos	we released yet?
23:23:04	*	psquid quit (Client Quit)
23:23:25	*	psquid joined #nimrod
23:24:22	EXetoC	in a couple of days I think
23:24:25	Demos	Araq: is x+x even faster than 2*x I can recall benchmarks where they are within the margin of error of each other
23:24:29	*	nequitans joined #nimrod
23:24:37	NimBot	Araq/Nimrod devel 7dc8ab8 Araq [+1 ±12 -0]: merged better html links #850
23:24:50	Araq	Demos: it's only an example
23:25:08	BitPuffin	Demos: they are pretty much the same on modern CPUs
23:25:17	Araq	yeah
23:25:33	*	CARAM joined #nimrod
23:27:47	Araq	good night
23:28:15	BitPuffin	OR IS IT!!==!=!!?!!??!"?!#?"!=)#
23:28:18	BitPuffin	DUN DUN DUUUUUUUUUN!
23:49:52	*	CARAM quit (Remote host closed the connection)
23:54:26	Skrylar	BitPuffin: i thought the trick with twos was to bit shift
23:54:52	Skrylar	i remember at least on the JVM they said that using << instead of multiplying by a power of two was faster, and some of the peephole optimizers do this to jar files
23:55:53	*	BitPuffin quit (Ping timeout: 240 seconds)
23:57:25	Varriount	fowl: I'm available to try out the allegro wrapper on windows.
23:58:14	*	CARAM joined #nimrod