<< 03-08-2014 >>
00:06:39*Matthias247 quit (Read error: Connection reset by peer)
00:07:46AraqVarriount|Mobile: "static is better than dynamic"
00:17:12Araq"it's the produced result that matters"
00:24:03Araq"statistics matter"
00:33:07Araq"programmers are smart"
00:58:12Araq"if meta programming does not solve your problem, you're not trying hard enough"
00:58:26Varriount:3
00:59:07VarriountI'll see if I can type up a "Zen of Nimrod" page based on those quotes.
00:59:28VarriountAraq: Any idea why this issue happens: https://github.com/Araq/Nimrod/issues/1442
01:05:27Araqhaven't looked at it yet
01:05:49Araqbut it surely looks strange
01:07:40VarriountAraq: Also, heads up - there's another issue concerning someone who wants uint32's to be implicitly convertible to ints
01:16:46flaviuI don't think thats a good idea
01:17:10flaviuCode will break if moved to a 32 bit machine
01:17:37Varriountflaviu: He says that it would be fine, because compile-time errors would be produced. :/
01:18:08flaviuSure, but one of Nimrod's things is cross platformness
01:18:22VarriountYeah. I'm editing my reply to point this out.
01:32:58AraqI said it'll convert to *int64*
01:33:01Araqnot to *int*
01:33:22VarriountDoes this reply look alright? -> https://github.com/Araq/Nimrod/issues/1441
01:34:01Araqsure
01:34:56VarriountIt's kinda hard to reply with anything more eloquent than "because it breaks cross-platform compatibility that much more"
01:44:42*brson joined #nimrod
01:58:51flaviuAraq: Where do symbol binding happen?
01:59:35Araqflaviu: lot's of places, but lookups.nim is a starting point
01:59:38*Fx00F quit (Quit: leaving)
01:59:50Araqthat's what everything uses that does a symbol binding
02:00:35flaviuthanks
02:15:32Araqgood night
02:29:45*xenagi joined #nimrod
02:32:14*Joe_knock quit (Quit: Leaving)
02:37:02*Varriount|Mobile quit (Ping timeout: 255 seconds)
02:37:55*Fr4n quit (Quit: Leaving)
02:53:04*q66 joined #nimrod
03:11:36*q66 quit (Quit: Leaving)
03:39:54*Demos quit (Read error: Connection reset by peer)
04:11:13*flaviu quit (Ping timeout: 255 seconds)
04:16:40*Fr4n joined #nimrod
04:30:14*BlameStross quit (Remote host closed the connection)
04:30:29*xenagi quit (Read error: Connection reset by peer)
05:25:11*BlameStross joined #nimrod
05:59:04*Trustable joined #nimrod
06:11:35*Trustable quit (Quit: Leaving)
06:28:42*io2 joined #nimrod
06:56:11*nande quit (Read error: Connection reset by peer)
07:15:47*zahary quit (Quit: Leaving.)
07:16:07*zahary joined #nimrod
08:10:47*brson quit (Ping timeout: 264 seconds)
08:12:21*brson joined #nimrod
08:21:13*brson quit (Quit: leaving)
08:29:37*Boscop joined #nimrod
08:29:37*Boscop quit (Changing host)
08:29:37*Boscop joined #nimrod
08:42:56*ARCADIVS joined #nimrod
08:59:55*ARCADIVS quit (Quit: WeeChat 0.4.3)
09:19:08*Matthias247 joined #nimrod
09:46:01*kunev joined #nimrod
10:47:57*ome joined #nimrod
10:54:35*Joe_knock joined #nimrod
10:54:58*Joe_knock quit (Max SendQ exceeded)
10:55:25*Joe_knock joined #nimrod
11:40:27def-Making a list comprehensions macro was easier than I expected
11:41:13def-https://github.com/def-/nimrod-unsorted/blob/master/listcomprehensions.nim
11:58:12dom96def-: awesome, put it in future.nim and make a PR
12:01:28def-dom96: I have the hope that it can still be improved. I would like to get rid of the "lc", and the type of the resulting seq should be inferable
12:03:47dom96you may need something to disambiguate it from `[]`(a, x)
12:04:48dom96not sure if you can do anything about that
12:15:13*Matthias247 quit (Read error: Connection reset by peer)
12:24:05dom96def-: actually, you may be able to make it an immediate macro. Detect if it's a list comprehension, output the original node if it's not
12:24:49def-dom96: nice idea, but i think [foo, bar] is invalid syntax anyway
12:25:28dom96I don't think so
12:25:48def-there always needs to be something before the [
12:34:11*tumak_ quit (Ping timeout: 255 seconds)
12:35:38*untitaker quit (Ping timeout: 250 seconds)
12:37:23dom96Hrm. indeed.
12:41:28*untitaker joined #nimrod
12:46:09*darkf quit (Quit: Leaving)
13:28:32*kunev quit (Ping timeout: 245 seconds)
13:45:20*Trustable joined #nimrod
13:47:22*q66 joined #nimrod
13:48:06*flaviu joined #nimrod
14:00:09*Matthias247 joined #nimrod
14:08:27*zahary quit (Quit: Leaving.)
14:31:23OrionPKwish we could just reimplement the entire compiler with macros :p
14:31:39OrionPKthen strip the core compile down to just if/else/while lol
14:55:16*Fx00F joined #nimrod
15:38:20flaviuOrionPK: Wouldn't work, there needs to be a way to introduce new top-level nodes with macros
15:44:22dom96http://www.codeproject.com/Articles/787668/Why-We-Should-Love-null
15:54:28Matthias247all the hipsters on reddit are already hatin it :)
15:56:55dom96yep lol
15:57:29Matthias247I actually like null ;)
16:15:26*nande joined #nimrod
16:18:50*Demos joined #nimrod
16:19:10flaviuThe problem with null is that its meaningless. Using Option or Maybe forces you to document *where* there can be a null, and *why* there is a null
16:21:04flaviuLike in his database example, I'd say that null is perfectly valid because its meaning is documented and storing 0 to represent unknown is ridiculous
16:23:39Demosthe thing is. Null literally /is/ zero
16:23:59Demosthe equivalent to null for something returning a number is something like returning -1
16:24:27flaviuDemos: -1 isn't a good idea either. Its too easy to ignore
16:24:46Demosyeah, I like null better since it crashes your program if you try and use it
16:25:10Demosbut for something like "search this array" returning -1 is better than using an optional type imo
16:25:54flaviuWhat about python-style array indexing?
16:26:03flaviuThat would mean the last element in the array
16:26:16Demoswell in python you would need soemthing different
16:26:25DemosI mean c++ returns an element off the end in that case
16:26:37Matthias247-1 is not good because it could be a valid number
16:27:22Matthias247but the Option stuff doesn't buy you much. Either you have lot's of overhead through pattern matching or you force-unwrap it which is the same as dereferencing null
16:27:32dom96Nimrod's 'not nil' will solve all our problems.
16:27:34Demosif I have something where it /could/ return a nullpointer I usually have one function that throws an exception instead and another called like wahteverOptional that can return nills
16:27:44*dom96 bows down to Araq
16:27:56flaviuMatthias247: Or you apply a function the the option and don't care whether it get executed
16:27:59Demosyeah not nil is a good idea
16:28:35flaviu-128 is a pretty good "not found" value
16:29:01flaviusince `-(-128)==0`
16:29:01Matthias247null pointer errors are also typically the most unproblematic errors that I encounter
16:31:08flaviuWell, the guy advocates nullable types
16:31:24flaviuWhich is essentially Option, but less powerful
16:32:59Matthias247I think the annotated nullable types from swift might be a nice middle way
16:35:00flaviuI don't see anything special with swift, except some syntax sugar that I'd say is of limited usefulness
16:38:39NimBotnimrod-code/packages master 0be1cb6 achesak [+0 ±1 -0]: Added imghdr and csv2json.
16:38:39NimBotnimrod-code/packages master f91e5ca Dominik Picheta [+0 ±1 -0]: Merge pull request #77 from achesak/master... 2 more lines
16:49:09Demosso len does not work on an openarray typeclass
16:54:01dom96Demos: What license is your vecmath module under?
16:55:44DemosI can release it under BSD, right now it is just sitting in another project.
16:57:08dom96Might be worth putting it in a separate repo. I might make some PRs then.
16:57:17Demoshm good idea
16:57:28Demosit would motivate me to split it up into multiple files as well
16:58:26Demosright now it has tons of stuff in that one file, vectors, matrices, AABB, ray intersection tests, Quaternions, and so on
16:58:43dom96do it
16:59:20dom96are you using any existing libraries as inpiration for the API of that module?
16:59:55Demossort of
17:00:12DemosI mean Eigen, but in general a lot of the stuff Eigen does can not work in nimrod
17:00:35DemosActually a few functions are straight up translated from Eigen or GLM
17:01:12dom96I'm going through a tutorial which uses GLM so I might translate some functions from there
17:01:26Demosmost should already be there
17:01:47Demoswe like the MIT license right?
17:02:11flaviuDemos: Sure, same as BSD but with less ambiguous naming
17:04:05dom96yeah
17:04:08dom96MIT might be better
17:05:19Demoscan you make a file in a babel package that is just a utility module, and is not avalible when the package is installed?
17:06:03flaviuinclude maybe?
17:06:32io2mpl2 is my favorite because it is the best mix of everything :)
17:07:31flaviuio2: Its copyleft though...
17:07:51io2flaviu: incorrect; you can use it in your commercial projects without an issue
17:08:03Demosdom96, you wrote babel right? so can you make a "private" module?
17:08:20io2"Eigen is Free Software. Starting from the 3.1.1 version, it is licensed under the MPL2, which is a simple weak copyleft license. Common questions about the MPL2 are answered in the official MPL2 FAQ. "
17:08:42dom96Demos: yes. And yes, but only by convention.
17:08:45dom96Put it in a private dir
17:08:54flaviuio2: Oh, I see. If you modify the software, you have to disclose the source, but you don't have to share anything else
17:09:00dom96as in a dir literally named 'private' :P
17:09:19flaviuie, submit your bugfixes upstream
17:09:29io2flaviu: pretty much, but way less ambiguous than lgpl stuff
17:09:32flaviuThat does seem pretty cool, I might move to that
17:09:48io2flaviu: i could not find another language that was compatible to metaprogramming
17:09:50io2license*
17:10:21flaviuI'm not sure I understand, now is metaprogramming special?
17:10:23io2since I do a lot of it, I prefer using mpl2 because it simply makes sense; the Eigen developers choose wisely.
17:10:45io2flaviu: imagine code generating code and mixing with your "proprietary" code
17:10:59io2in a non-linking fashion; it simply generates things that go in
17:11:08flaviuOh, I see
17:11:19io2that's one little detail that is very evil
17:12:05io2you currently don't have something better than mpl2 imvho for these cases, so I switched to that instead of making GPL exceptions etc.
17:12:40io2BSD/MIT are kind of confusing as to intent and derivative works as I see it. The tl;dr is: mpl2 is metaprogramming-compatible.
17:16:35DemosI had it all set up and then I git reset --hard :D
17:23:22dom96Isn't MIT basically "do whatever you want, and don't hold me responsible"?
17:23:29dom96"(if something breaks)"
17:23:29*flaviu quit (Read error: Connection reset by peer)
17:24:05DemosOK PR on packages is up
17:25:19Demosplease note that private/intrinsic contains a function that is needed for using the ray intersection code, you can provide a GCC version of the intrinsic or just comment that ray intersection code out on your computer (I am not even sure it works as is, I just wrote it recently)
17:25:34*flaviu joined #nimrod
17:31:25dom96Demos: you should update your git email
17:31:38dom96it doesn't match your github email
17:32:10dom96Thanks for separating it out!
17:36:42NimBotnimrod-code/packages master 50efb41 Charlie Barto [+0 ±1 -0]: added vecmath
17:36:42NimBotnimrod-code/packages master 0292ee1 Dominik Picheta [+0 ±1 -0]: Merge pull request #78 from barcharcraz/master... 2 more lines
17:38:17Demosno problem, I have been meaning to split out a lot of code from this graphics stuff
17:55:18Demosdom96, watch out with vecmath that you are calling the correct functions, there are seperate multiply functions for vec3s and vec4s and whatnot but the compiler WILL NOT emit an error if you use the wrong one
17:55:38Demosthe C compiler sometimes will though
17:59:54dom96hrm, ok. Why is that?
18:03:30Demoscompiler bugs
18:03:40Demosthey are reported
18:06:41*bogen joined #nimrod
18:07:32*ome quit (Quit: Connection closed for inactivity)
18:07:52bogendoes nimrod have user defined attributes for procedures? (accessible by ctfe)
18:08:36dom96hello bogen, welcome to #nimrod
18:08:50dom96Perhaps pragmas are what you have in mind?
18:09:00bogenhi dom96
18:09:22dom96dumpTree: proc x: int {.myPragma.} = 5
18:09:57dom96You'll see that the AST contains 'Pragma Ident !"myPragma"' or something like that.
18:10:07dom96(dumpTree is defined in the macros module)
18:11:21bogenpossibly. I'm exploring the possibily of using nimrod, there are some things I wan to know if it can do first. At compile time could I traverse a module and make a string map of all procedures that match my critera. so, a pragma might work for identifying which procedures I want to put into the map.
18:12:53bogenit does not have to be a compile time (building the map), it can be at run time if the type of reflection is possible. I'll look at dumpTree
18:13:32*flaviu quit (Read error: Connection reset by peer)
18:15:42*flaviu joined #nimrod
18:17:31dom96What you will need to do is wrap the whole module in a macro call.
18:18:19*nande quit (Remote host closed the connection)
18:19:22Demosdom96, he could use a macro pragma right?
18:19:29Demosand a compile time map of some kind
18:19:49dom96yeah, that too.
18:19:59dom96I wonder if you can 'push' a macro pragma.
18:20:33Demosthat would be ideal when I write my direct3D de-uglification macro
18:20:57Demosthing.method(parms) instead of thing.lpvtbl.method(thing, params)
18:21:08Demosbut I only want to to happen to my d3d procs
18:22:04bogenwrapping the whole module in a macro call... I don't know if like how that sounds, unless you mean referencing the source of the module or something in the macro call
18:22:49Demosjust putting macro: at the top, the pragma idea would work as well
18:23:13Demosbut instead of going through the module in the macro you just add the proc that has the pragma to your tracking structure
18:23:38Demosyou could probably also use the #! thing somehow
18:24:00dom96You might be able to just write {.push myMacroName.}
18:24:03dom96at the top of the module
18:24:10dom96bogen: what is your use case?
18:27:00bogenone use case, without going into all the details, is that I need to be able to accept proc names from outside my application (at runtime) and run procedure of the same name. I'd like an automatic mapping (via a hint, a pragma should work as you have mentioned), without having to have a seperate table in the code.
18:28:11bogencan macros retain state?
18:28:58bogencould I have a collecting macro in each function that records the function name, then later on dump the collection to a map via another macro?
18:29:08bogeneh, procedure...
18:29:28bogen(like I said, I'm exploring the possibilities of nimrod)
18:30:17*vendethiel joined #nimrod
18:30:18Demosyou should be able to generate some kind of mapping from string to procedural type
18:30:32Demosyou gotta be careful with the params though
18:30:58dom96You can get the macro to generate the table for you.
18:31:11Demosalternitively you can use your platform's shared library tools to do this
18:31:16bogenyeah, I know that, the call gate would take care of that (the params)
18:31:18Demosthat is pretty much what a shared lib is
18:31:50dom96I don't think macros can retain state at compile-time.
18:33:14*silven quit (Ping timeout: 250 seconds)
18:33:14bogenso compile time code execution is limited to "pure functions" (no side effects) I take it?
18:35:24dom96there are some exceptions, you can for example execute a process and get its output
18:35:29dom96at compile-time
18:35:35Demosbogen, not quite. What you can do is have a structure that is accessable to the code that calls the macro (so like a module global map) and then the macro can emit calls to add stuff to the map
18:35:37Varriountdom96: What about global variables?
18:36:19dom96Varriount: at compile-time?
18:36:29VarriountYeah
18:37:39dom96I don't think so.
18:38:56DemosI don't think macros have a very well defined sense of ordering, so allowing state changes from inside macros would be hard to control
18:39:19Demosor reduce the optimization possibilities in the compiler
18:43:12bogenLooking at the objdump -t output of a nimrod program, one "could" parse that output one oneself at runtime and generate the map. In that case my custom id would be in the function names. I was hoping to avoid that, as that does not make nimrod stand out over other languages that can do that. :)
18:43:19*io2 quit (Ping timeout: 255 seconds)
18:43:46bogenthat can resort to doing it that way...
18:44:13Varriountbogen: You could also write/read to an output file
18:44:27bogenat compile time? that would work. :)
18:46:03DemosI have a library that lets you use openGL by loading a program and then doing like program.uniform1 = obj1 program.uniform2 = obj2 program.vertices = somearray draw(program)
18:46:07Demoswhat should I call it?
18:46:48dom96Varriount: Are you sure there is a way to write files at compile-time?
18:47:12flaviuDemos: BWFOGL :P
18:47:28Araqhi bogen welcome
18:47:38bogenHello Araq
18:47:43Demosflaviu, erm where did that come from?
18:47:45Araqyou can do this with macros, but it's horrible ;-)
18:47:52flaviuDemos: Better Wrapper For OpenGL
18:48:08Demosit is not a wrapper though, it does not let you use a lot of features
18:48:21Demosfor example I require that ALL non-sampler uniforms be UBOs
18:48:30flaviuBetter Almost Wrapper For OpenGL
18:48:38Demosheck if my graphics card had bindless textures I would requite all uniforms to be UBOs
18:49:05Demosalso, I do not support getting vertex data from more than one array unless you want to take ownership of the VAO and VBO
18:49:11Araqbogen: IMO you're much better off with an .exposed pragma macro that you add to the procs that are to be exposed
18:49:17dom96Demos: EasyGL?
18:49:24Demosthat is good
18:49:37DemosFastGL
18:49:41Demosyeah I like FastGL
18:50:05Araqbogen: that said, you can do parseStmt(staticRead("module.nim")) to get the AST of a complete module
18:50:30Demosbecause they way I model the API can be substantially faster than regular GL once I implement the "bleeding edge extensions" version
18:50:38flaviuDemos: What about "Phosphor"?
18:50:58flaviusounds nice and sort of describes it.
18:51:02bogenArap: ok, and then from that AST filter out what I want and build the map
18:51:12Demosapperently Phosphor is the name of an FPS game
18:51:29flaviuYou'll never find a name if you don't reuse names
18:51:50VarriountAraq: Is there a current way for macros to store global state without resorting to output files and such?
18:52:12AraqVarriount: what do you think .compileTime global vars do?
18:52:19flaviuWhat about a dummy `discard "serialized data"`?
18:52:25VarriountOh, I forgot about those.
18:52:29DemosI like Phosphor
18:52:37Varriountbogen ^^^
18:53:20dom96flaviu: how does Phosphor describe it lol?
18:53:38bogenVarriount: yeah, so a .compileTime global var could be a map I'm building at compile time?
18:53:46VarriountYes.
18:54:00flaviudom96: It describes how computer monitors work, and GL is displayed on computer monitors
18:54:30Varriountbogen: Just be careful/avoid using the variable at runtime. I just did a runtime repr() on a compileTime seq[string] and got C errors.
18:54:44flaviuhttp://stuff.h-i-r.net/blogstuff/wx1.jpg
18:55:45*kunev joined #nimrod
18:56:05bogenVarriount: well, I assume I'd need followup code (after all the collections) to create the data structure that the runtime would use.
18:56:37Varriountbogen: Yeah, as long as you programmatically create that data structure as an AST somewhere.
18:58:09bogenVarriount and Araq: ok, so I have two options to explore 1) tag with a pragma then do a parseStmt on the module 2) collect with macro then generate the data structure
18:59:37VarriountI'd try doing #2 first.
18:59:49Varriountbogen: What exactly is your larger goal?
19:04:33bogenin simplest terms the app is a slave to another app. the procedures take no input arguments and don't return anything. state is maintained elsewhere. since there will be a lot of these procedures I don't want to have to maintain a seperate table. (Added it here, forgot to to add it here).
19:04:44*silven joined #nimrod
19:05:50*Trustable quit (Quit: Leaving)
19:10:21bogenis system.c generated uniquely for my app? or can I link against that dynamically?
19:12:03Araqyou can link against the runtime library dynamically, but system.c is unique
19:12:15Araqit's all documented
19:13:23Araqunfortunately you still have to read it and I'm serious here. I often wonder if excessive documentation puts people off moreso than no documentation.
19:17:05*io2 joined #nimrod
19:18:41VarriountAraq: It's better when one can search for something easily.
19:22:01*io2 quit (Ping timeout: 255 seconds)
19:23:53bogenAraq: Thanks. I understand (about needing to read it). I was just curious and as Varriount said, if would be easier for newcomers if one could search the main documentation site.
19:29:01Araqgoogle site search?
19:29:19AraqI guess we simply need a button for that
19:35:58Demoshm I have a generic proc that ends up calling contains via in on a TSet
19:36:07Demosthis fails to work unelss I import sets at the callsite of the proc
19:36:20Demoseven though I import sets in the module where this proc is defined
19:36:26VarriountKnown issue, I think
19:36:33bogenyeah. however, docs are under the top level of nimrod-lang.org, so searching again nimrod-lang.org gets a lot of noise from build.nimrod-lang.org (if one is only looking for something in the documentions). Just like the forum and build are at their own top level, a docs.nimrod-lang.org would help in searching the documentation.
19:36:44DemosVarriount, it is kinda a large issue
19:39:06Demoslooks liek there are some similar problems with the method call syntax
19:42:07Demosthis must be a new regression
19:43:56Araqhmm I fixed that recently, are you on devel?
19:44:19Demosyeah, but I have not updated in a few days
19:44:39Demoscome to think of it I should probably update /before/ bitching on IRC
19:46:28*io2 joined #nimrod
19:48:56DemosOK it still happens on the latest devel...
19:49:09Demosgit hash 915d329
19:53:10Demosdoing contains(set, item) works
19:53:15Demosbut set.contains(item) does not
19:53:30Demosand item in set also does not work
19:55:20Araqreport it asap
19:58:50dom96Demos: The indexing starts at 1?!
19:59:00Demosdom96, yes
19:59:03dom96why?
19:59:21Demosbecause that is common in math
19:59:34Araqdon't do that
19:59:50dom96ugh
20:00:28DemosAraq, why not?
20:01:33AraqDemos: programmers really do not like that
20:02:47DemosAraq, well too bad for them. The data array does start from zero I think
20:02:47dom96Yep. This just completely turned me off your module.
20:03:41*Varriount|Mobile joined #nimrod
20:05:49Demoshonestly I do not see how it is such a big deal, like for arrays there is no standard so you often start at 0 but matrices are almost always 1 based in maths
20:06:46AraqDemos: I can sympathize but this means people will roll their own module instead
20:06:50DemosOK that issue is up
20:07:40dom96One disadvantage is that copying code from GLM is more difficult.
20:09:04Demosdom96, well starting from zero makes copying code from maths texts, or from the notepad you just did some math on harder
20:09:57dom96You would be manually copying the code from maths texts and notepads anyway.
20:10:06DemosI guess I can consider changing it, but I did not expect anyone would care. In fact I figured people would like that it lines up with what you see in math
20:10:20Demoswell you are not going to just copy C++ code and expect it to work
20:10:56dom96Yes, but I can copy the boring stuff.
20:10:56Demosand if you are following a tutorial using GLM you should not be copying the code anyways
20:11:31dom96But now I must manually add 1 to each index.
20:13:00dom96and that's not even trivial: I have to start from the highest index and work my way down replacing the digits.
20:13:22Demos%s/\[(\d)\]/\=submatch(0)+1/g?
20:13:48Demosor something along those lines
20:14:44Demosor use <C-A>
20:15:01Demoswhich will increment the value under the cursor
20:15:29dom96I'm using Aporia.
20:16:20DemosI usually use Visual Studio, but when I need to translate code like this or write a wrapper I always use vim, for this reason
20:16:37dom96meh. I'll just write a quick vector module of my own I think.
20:18:17*milosn_ quit (Remote host closed the connection)
20:19:40Demosugh
20:20:06*milosn joined #nimrod
20:23:04*def- quit (Ping timeout: 240 seconds)
20:28:59Araqdom96: it really is trivial but the result is often too ugly and then some error prone rewriting happens
20:30:53*kunev quit (Ping timeout: 240 seconds)
20:36:08*jbe joined #nimrod
20:36:58jbehi nimrods :)
20:37:07Araqhi jbe
20:37:27jbethanks again for a sweet language
20:37:28dom96Araq: sure, replacing is trivial but cumbersome. The problem is indeed the possibility of introducing errors.
20:37:35dom96hey jbe
20:38:06jbeare there shared memory seqs yet?
20:38:07AraqDemos: it's math textbooks vs "code in other programming languages out there"
20:38:08*tinAndi joined #nimrod
20:39:11DemosAraq, you are right. I should probably switch to zero based indexing.
20:39:55Araqjbe: you can use allocShared and roll your own and I know several people do that, but nothing was added to the stdlib
20:41:06jbeah ok.. yeah that's what i do now. i saw the old post about the new (now old?) concurrency model that got bumped and became curious
20:42:02flaviuAraq: How about another feature request? :P
20:42:34Araqflaviu: how about you fix some bugs?
20:42:40flaviuarray[0..100, T] should be equivalent to array[1..101, T]
20:44:31Araqjbe: the new concurrency model makes shared memory easier in general
20:44:59Araqand its implemenation is ~70% finished
20:45:35*tinAndi quit (Quit: ChatZilla 0.9.90.1 [Firefox 31.0/20140716183446])
20:45:43Araqflaviu: which definition of "equialent" do you use?
20:46:15flaviuTheir types are equvilent
20:46:17jbevery nice..
20:46:41flaviueg, I can pass an array[0..100, T] to something that expects an array[1.101, T]
20:46:50flaviuLike tuple field names
20:49:07flaviuWait, that works
20:49:20flaviuWell, just ignore me then lol
20:50:28*def- joined #nimrod
21:12:12*jbe quit (Quit: Page closed)
21:23:32DemosI would just like to point out that fortran defaults to 1 based matrices
21:23:52AraqI know and so does Lua
21:30:16*Demos quit (Ping timeout: 255 seconds)
21:32:11*Demos joined #nimrod
21:38:25*q66_ joined #nimrod
21:40:54*q66 quit (Ping timeout: 260 seconds)
21:43:45Varriount|MobileWe should bypass the 0-based-indexing vs 1-based-indexing entirely, and use an alphabetical mapping >:) .
21:45:18Varriount|MobileIndexes will go foo["a"] to foo["z"] , to foo["aa"] etc.
21:50:08*q66_ is now known as q66
21:59:07*io2 quit (Quit: ...take irc away, what are you? genius, billionaire, playboy, philanthropist)
22:22:11*noam quit (Ping timeout: 264 seconds)
22:27:06*Matthias247 quit (Read error: Connection reset by peer)
22:27:07*dom96 quit (Excess Flood)
22:28:28*JStoker quit (Ping timeout: 244 seconds)
22:29:09*Roin quit (Ping timeout: 240 seconds)
22:29:51*Zuchto quit (Ping timeout: 250 seconds)
22:29:56*JStoker joined #nimrod
22:29:57*JStoker quit (Changing host)
22:29:57*JStoker joined #nimrod
22:31:09*dom96 joined #nimrod
22:31:28*oddmunds quit (Ping timeout: 250 seconds)
22:38:49*Zuchto joined #nimrod
22:40:07*oddmunds joined #nimrod
22:42:28*jez0990 quit (Quit: No Ping reply in 180 seconds.)
22:42:35*jez0990 joined #nimrod
22:56:35*Roin joined #nimrod
23:02:52*darkf joined #nimrod
23:24:42*Varriount-Mobile joined #nimrod
23:24:54*Varriount|Mobile quit (Read error: No route to host)
23:26:33*q66_ joined #nimrod
23:28:42*q66 quit (Ping timeout: 260 seconds)
23:28:54*Varriount-Mobile quit (Ping timeout: 244 seconds)
23:32:11*dom96 quit (Excess Flood)
23:33:21*vendethiel quit (Read error: Connection reset by peer)
23:33:39*dom96 joined #nimrod
23:34:35*vendethiel joined #nimrod
23:34:53*vendethiel quit (Read error: Connection reset by peer)
23:37:55*vendethiel joined #nimrod
23:45:51DemosVarriount, I like the alphabetical idea :D
23:46:11Demosalso people should realize that THE main linear algebra library is written in FORTRAN
23:58:36*Joe_knock quit (Quit: Leaving)