| 00:02:49 | skrylar_ | why do math heads write less readable things like "for x in 1 <= x <= n" instead of just "for each x" |
| 00:03:14 | skrylar_ | in this case it's not possible for a range to be outside of x, so the extra runes just lower the legibility of the paper |
| 00:03:20 | krux02 | that is not math heads |
| 00:03:31 | krux02 | and by the way that would be invalid code |
| 00:03:50 | skrylar_ | "For each row r where 1 ≤ r ≤ n and r ≠ j ," |
| 00:04:13 | krux02 | I have no idea who would write that |
| 00:04:18 | skrylar_ | it's in a math book. |
| 00:04:50 | krux02 | I would write (for r in 1 .. n: if r != j: [...] |
| 00:05:01 | skrylar_ | well they're not writing computer code, so |
| 00:05:10 | krux02 | pseudocode |
| 00:05:15 | krux02 | very cool stuff |
| 00:05:17 | skrylar_ | but its less readable than just saying "for x in [1, n]" or just "for every x" |
| 00:05:37 | krux02 | I realized that when you have a good programming language it is often much more readable that pseudocode |
| 00:05:49 | krux02 | (but nim is not there yet :() |
| 00:05:59 | skrylar_ | dunno, nim has a lot of hacky things i use |
| 00:06:09 | skrylar_ | you can make templates for just one function to use :> |
| 00:06:41 | skrylar_ | i think i do that in some neural net code because it allows you to read "current_layer" where that actually is somevar[someindex] |
| 00:08:16 | krux02 | honestly Nim does not encourage you to write clean code |
| 00:08:23 | krux02 | it lets you do a lot of horrible things |
| 00:08:32 | krux02 | that is also how the compiler code reads |
| 00:08:35 | krux02 | very badly |
| 00:08:35 | skrylar_ | dunno. its clean enough for me. but then i think using #defines is acceptable ;x |
| 00:08:52 | skrylar_ | have sometimes used defines in c/c++ that live for one function, just because less boilerplate |
| 00:09:03 | krux02 | a lot of injected code via dirty templates and no idea where actually variables come from |
| 00:09:24 | krux02 | well I did that, too, and I don't see it as dirty |
| 00:09:25 | skrylar_ | then use better templates :> |
| 00:09:28 | krux02 | I see it as very clean |
| 00:09:46 | krux02 | because it is very local and the definition of the define/template is also close to where it is used |
| 00:09:49 | skrylar_ | then again i don't use a lot of code i haven't written or written the bindings for, so i know where all this stuff came from |
| 00:09:55 | krux02 | it is a good usage of templates/defines |
| 00:09:57 | skrylar_ | but yea. macros/templates require discipline or your code and compile time goes to hell |
| 00:10:18 | krux02 | yes |
| 00:10:28 | krux02 | bit in nim it's not as bad as it is in c++ |
| 00:10:28 | skrylar_ | i mostly use templates for rewrites/D aliases |
| 00:10:42 | skrylar_ | D let you alias stuff like "colour" to "color" |
| 00:10:47 | skrylar_ | or "left" to "x" |
| 00:11:03 | krux02 | in c++ templates are instanciated for each compilation unit just so that the linekd removes all but one instance |
| 00:11:12 | krux02 | at lot of optimizations go directly to the bin |
| 00:11:15 | skrylar_ | NOW yea. it used to be worse |
| 00:11:48 | skrylar_ | i think my pipe dream would just be a lisp/smalltalk esque compiler tho |
| 00:12:06 | skrylar_ | do small jit compiles on the fly because you don't need to run these heavyweight optimizers every debug cycle |
| 00:12:08 | krux02 | well I would be happy if nim would remove a few features |
| 00:12:14 | krux02 | then nim would be my dream language |
| 00:12:23 | skrylar_ | what, like the gc? lol |
| 00:12:28 | krux02 | yes |
| 00:12:39 | skrylar_ | everybody always wishes for that |
| 00:12:40 | krux02 | I don't use it |
| 00:12:44 | skrylar_ | it's the least-bad GC i've seen yet |
| 00:13:05 | krux02 | well it is thread local, and therad local allocation is a smart default |
| 00:13:21 | krux02 | that is what it used to be when there was only one thread |
| 00:13:38 | krux02 | then with multitherading the wrong design chose has been made |
| 00:13:39 | skrylar_ | well. biggest bitch about GCs is that you don't control the stoppage. but you can build the incremental GC and use it as a sleeper |
| 00:13:48 | skrylar_ | which for games and some servers is just fine |
| 00:14:57 | skrylar_ | my complaint is that nobody's GCs talk to one another |
| 00:15:12 | skrylar_ | so you get bad situations like if Lua and Nim want to co-exist they end up oblivious to each other's memory usage |
| 00:15:34 | skrylar_ | in the case of Go i think they rewrote a lua interpreter just to have shared memory o_o |
| 00:17:55 | skrylar_ | meh. someone requested a toml4 reader |
| 00:19:09 | skrylar_ | krux02, i still use inheritance but its sorta eeh. careful use of it |
| 00:19:17 | skrylar_ | Go's inheritance mostly |
| 00:19:27 | krux02 | you mean embedding |
| 00:19:48 | skrylar_ | last time i used go was trying to do a whole gui engine as a component model |
| 00:19:55 | skrylar_ | so it was shallow inheritance and generic behavior objects |
| 00:20:00 | skrylar_ | Unusual but neat. |
| 00:21:42 | krux02 | yes that is also what c++ does |
| 00:21:51 | krux02 | but c++ also has virtual inheritance |
| 00:22:42 | krux02 | There is very little chance that a c++ developer actually fully understands virtual inheritance, because the syntax is cryptic and it's usage is so rare that is really is very hard to understand |
| 00:23:16 | skrylar_ | isn't that what everyone else just calls interfaces |
| 00:23:27 | skrylar_ | or what we'd use a method for. lol |
| 00:23:32 | krux02 | but since I developed scala, where that behaviour (or at least a very similar one) is the default and the syntax is compact and therefore very easy to play around with, I do fully understand it |
| 00:23:53 | skrylar_ | although i did get bitten by methods lack of a super; proccall was weird to find |
| 00:23:57 | krux02 | no virtual inheritance is not virtual methods |
| 00:24:20 | krux02 | it is for multiple inheritance when you inherit from two classes that have the same base class |
| 00:24:29 | skrylar_ | ah diamond inheritance |
| 00:24:33 | skrylar_ | i never did that |
| 00:24:34 | krux02 | yes |
| 00:24:44 | krux02 | well the default inheritance is like in go |
| 00:24:50 | krux02 | the super type gets embeddet |
| 00:25:10 | krux02 | and c++ can do multiple inheritance, so it really is very go like |
| 00:25:11 | skrylar_ | i prefer composition over inheritance usually |
| 00:25:29 | krux02 | but with it comes the problem that in that diamond case you have the base clas twice |
| 00:25:44 | krux02 | those fields are embedded twice |
| 00:25:54 | krux02 | and virtual inheritance "solves" it. |
| 00:26:03 | krux02 | by confusing everybody |
| 00:27:39 | skrylar_ | yea i just don't use that model x.x |
| 00:30:22 | skrylar_ | its so weird. when i started learning to code, dirty rects were a common thing in gui. now people are just disposing of geometry every frame |
| 00:32:41 | krux02 | well, most people have no idea about programming |
| 00:34:18 | skrylar_ | you don't save a whole lot with dirty rects on modern gpus |
| 00:34:37 | skrylar_ | i played around with immguis though. i understand the appeal but i'll just stick to scene graphs |
| 00:35:37 | skrylar_ | one of the hidden costs of immguis is that you're paying api call costs for each one so stuff like unity (or worse, lua without luajit etc) you get to pay a lot of round-trips to make disposable geometry. but retained apis are just like |
| 00:35:40 | skrylar_ | "this warning label = shown" |
| 00:40:25 | krux02 | well I like the idea of immguis for development a lot |
| 00:40:51 | krux02 | but I don't like the idea to rerender text each frame |
| 00:40:59 | krux02 | text rendering is still one of the slower things to do |
| 00:41:35 | krux02 | and immguis are one of the areas where I think that modern hardware power is spent well for development efficiency |
| 00:42:04 | skrylar_ | IMHO its bankrupcy caused by using crap languages that shunned macros |
| 00:42:33 | skrylar_ | i got to looking at the pros/cons, and they're almost entirely cons. artists don't like immgui, tooling doesn't like immgui, hardware doesn't optimize for immgui, but it does let solo devs play around with gui |
| 00:42:47 | krux02 | well there is still this other language made by Jonathan Blow that might be a serious competition to Nim |
| 00:42:54 | skrylar_ | that being said with a couple macros you can have the immgui experience with a retained api so |
| 00:42:57 | krux02 | unlike Nim it seems more focussed on what is important |
| 00:42:58 | skrylar_ | that's probably a worthwhile goal |
| 00:43:14 | krux02 | generally simpler in the implementation and that might pay off eventually |
| 00:43:23 | krux02 | but other than that it shares almost the same benefits |
| 00:43:28 | krux02 | but I haven't used it yet |
| 00:43:38 | skrylar_ | he's making a games language, nim is a systems language |
| 00:43:57 | skrylar_ | it might be neat some time after never when it comes out, but we have nim right now :> |
| 00:44:02 | krux02 | well, not sure where you see the difference |
| 00:44:49 | krux02 | well there is a lot of progress going on with it, and he already realead great software, so he knows how to get things finished |
| 00:45:25 | krux02 | I just made a funny PR: https://github.com/nim-lang/Nim/pull/6253 |
| 00:45:35 | krux02 | it is basically doing what rust does |
| 00:47:00 | skrylar_ | i dunno. it just looks more like nim doesn't have active developers and blow has enough money to coast on his own thing |
| 00:47:37 | skrylar_ | when i look around its like well we have a package manager now but half the codes not in it |
| 00:47:54 | subsetpark | can I fake choosenim -y with some piping and yes? |
| 00:48:34 | krux02 | not sure how important a package manager is though |
| 00:48:44 | krux02 | we have github and can search on our own |
| 00:49:12 | krux02 | Go can import from github, package manager? Who cares |
| 00:49:27 | krux02 | well they are installey eventually |
| 00:50:53 | skrylar_ | lot of people care considering "go get someproject" is becoming a standard install manta |
| 00:51:15 | skrylar_ | it's also standard for python (pip), ruby (gems), node (npm) |
| 00:52:29 | FromGitter | <krux02> I personally think go is a very good language to learn programming |
| 00:52:46 | skrylar_ | its a nice C |
| 00:52:56 | krux02 | well it has a lot of nice C things |
| 00:53:00 | krux02 | but no it's not a nice C |
| 00:53:10 | krux02 | it can't replace C |
| 00:53:13 | krux02 | so it's not a nice C |
| 00:53:25 | skrylar_ | nothing will ever replace c/c++ |
| 00:53:33 | skrylar_ | it's a cancer(++) that has metasticized too big to stop |
| 00:53:45 | krux02 | well not really |
| 00:54:05 | krux02 | I think C can't die |
| 00:54:36 | krux02 | just beacause it is the main compilation target for a lot of languages not just nim |
| 00:54:57 | krux02 | and it is used for drivers and stuff like that where low level control is important |
| 00:55:01 | krux02 | but c++? |
| 00:55:11 | skrylar_ | i ended up coming back to nim because Go lacked acceptable metaprogramming, and Red's priorities are bonkers |
| 00:55:18 | skrylar_ | well, go does "have" it |
| 00:55:26 | krux02 | For application development ther just needs to be a language that truly understands what is important in c++ and make a better languaeg out of it |
| 00:55:38 | krux02 | most of the time the new languages utterly fail at that task |
| 00:55:40 | skrylar_ | D1 did that, didn't take |
| 00:55:57 | krux02 | Java? just the parts of c++ that should be avoided, great language |
| 00:56:15 | krux02 | D was gc, too |
| 00:56:25 | skrylar_ | it's hard to impossible to compete against those langs because people expect you to ship gui builders and visual studio clones and its like even great langs can't make those from no budget |
| 00:56:30 | skrylar_ | D1 was gc optional |
| 00:56:35 | skrylar_ | you could shut it off and never touch it |
| 00:56:57 | skrylar_ | they botched that rule in D2 |
| 00:57:07 | krux02 | how that? |
| 00:57:20 | skrylar_ | wrote a lot of bad code and then canonized the bad code instead of fixing it |
| 00:57:30 | skrylar_ | "hey half our stdlib is bad so lets just rewrite history guys" |
| 00:58:13 | krux02 | well nim has also a lot of bad code in the stdlib |
| 00:58:43 | skrylar_ | GCs aren't evil. Some early GCs were bad. Some modern GCs lack tuning knobs. But there was a weird whitepaper that talked about one of the ML compiler's collectors outperformed boost smart pointers |
| 00:58:43 | krux02 | that is why I barely use the stdlib |
| 00:58:58 | Xe | krux02: what do you mean by "bad code"? |
| 00:59:08 | krux02 | I don't say GC is evil, but enforcing it, is |
| 00:59:16 | krux02 | GC should at least be optional |
| 00:59:27 | skrylar_ | it sort of is in nim, although i think silicing requires it |
| 00:59:36 | skrylar_ | isn't there an ARC branch now |
| 01:00:43 | skrylar_ | My main frustration with GC is 1) most aren't incremental, 2) they don't tend to respond well to manual collection windows and 3) they tend to think ignoring finalizers is ok |
| 01:01:08 | krux02 | Xe: https://github.com/nim-lang/Nim/pull/6253 |
| 01:01:11 | krux02 | sorry |
| 01:01:21 | krux02 | I mean: https://github.com/nim-lang/sdl2/blob/master/examples/sdl_opengl_example.nim#L125 |
| 01:01:23 | skrylar_ | if they are obedient i can at least set health flags so servers don't drop heavy requests on a node thats mid-collection, which is something .NET/Java fail at IIRC |
| 01:02:21 | skrylar_ | krux02, is that immediate mode GL i see |
| 01:02:24 | skrylar_ | disgraceful :e |
| 01:03:05 | skrylar_ | ah well it was a lot more newbie friendly at least |
| 01:03:19 | krux02 | the problem is gc is that the gc is a global thing, ant it can only be configured globally, but when the system grows and becomes more complex, i single point to config the GC is just not enough |
| 01:03:43 | krux02 | for one subsystem you need config A, and for the other one you need config B |
| 01:03:47 | krux02 | that can't work |
| 01:04:05 | skrylar_ | you shouldn't generate lots of garbage in the first place and its not a problem though |
| 01:04:13 | krux02 | that immediate mode GL is not the problem, the casting of the event types is |
| 01:04:28 | skrylar_ | well the immediate GL is deprecated for years tho |
| 01:04:48 | skrylar_ | as for casting events.. eeh |
| 01:05:17 | skrylar_ | It can't be helped without shuffling data out of the C class in to nim, or ultimately hiding it behind templates that typechecked and then cast it anyway |
| 01:05:43 | skrylar_ | the second would probably be my preference, since the typecheck can be made in to a debug assertion |
| 01:06:32 | krux02 | or here: https://github.com/nim-lang/Nim/blob/47a7f2fff795be3c4c28059cfa573f5d6069af44/lib/pure/htmlgen.nim#L196 |
| 01:06:40 | krux02 | use a keyword as a template name |
| 01:08:24 | krux02 | skrylar_: well normally I don't need to generate garbage at all if the seq type and the string type would not be implemented so badly |
| 01:10:30 | skrylar_ | send in a patch to use realloc i guess |
| 01:10:44 | skrylar_ | it does tweak me when i see C++ type code where they do malloc and free instead of reallocs |
| 01:11:07 | skrylar_ | sometimes the memory allocator had more space to give but you went ahead and trashed it with a new alloc because we're C++ so why do we care about memory fragmentation \o/ |
| 01:12:07 | skrylar_ | although the +1 with nim is you can make your own seqs if you just wanted some strange bucketing logic |
| 01:12:44 | * | chemist69 quit (Ping timeout: 246 seconds) |
| 01:22:50 | * | yglukhov joined #nim |
| 01:23:35 | * | mal`` quit (Quit: Leaving) |
| 01:23:57 | * | heinrich5991 quit (Ping timeout: 240 seconds) |
| 01:24:38 | * | mal`` joined #nim |
| 01:26:45 | * | chemist69 joined #nim |
| 01:26:57 | * | yglukhov quit (Ping timeout: 240 seconds) |
| 01:31:55 | * | heinrich5991 joined #nim |
| 02:08:15 | * | skrylar_ quit (Remote host closed the connection) |
| 02:17:44 | * | pilne joined #nim |
| 02:39:41 | * | vivus quit (Quit: Leaving) |
| 02:39:48 | * | v17d joined #nim |
| 02:44:30 | * | krux02 quit (Remote host closed the connection) |
| 02:59:55 | * | def-pri-pub joined #nim |
| 03:07:36 | * | dddddd quit (Remote host closed the connection) |
| 03:13:35 | * | rauss joined #nim |
| 03:15:26 | * | v17d quit (Ping timeout: 240 seconds) |
| 03:16:17 | * | endragor joined #nim |
| 03:28:05 | * | v17d joined #nim |
| 03:35:55 | * | pilne quit (Quit: Quitting!) |
| 04:32:04 | * | def-pri-pub quit (Quit: leaving) |
| 04:41:20 | * | rrgg joined #nim |
| 04:42:53 | * | rrgg left #nim ("WeeChat 1.4") |
| 05:51:29 | * | v17d quit (Ping timeout: 248 seconds) |
| 05:56:40 | * | rauss quit (Quit: WeeChat 1.9) |
| 06:14:00 | * | mahsav2 joined #nim |
| 06:15:24 | * | mahsav quit (Read error: Connection reset by peer) |
| 06:18:01 | * | mahsav2 quit (Ping timeout: 240 seconds) |
| 06:32:49 | * | ShalokShalom_ joined #nim |
| 06:35:35 | * | ShalokShalom quit (Ping timeout: 240 seconds) |
| 06:42:52 | * | nsf joined #nim |
| 06:43:06 | * | mahsav joined #nim |
| 06:58:47 | * | Vladar joined #nim |
| 07:01:00 | * | nattefrost joined #nim |
| 07:01:25 | * | v17d joined #nim |
| 07:25:23 | * | gangstacat joined #nim |
| 07:50:08 | * | gokr joined #nim |
| 07:54:49 | * | Arrrr joined #nim |
| 08:01:41 | * | yglukhov joined #nim |
| 08:03:41 | * | yglukhov quit (Read error: Connection reset by peer) |
| 08:04:04 | * | yglukhov joined #nim |
| 08:08:26 | * | yglukhov quit (Ping timeout: 240 seconds) |
| 08:08:27 | * | endragor quit (Read error: Connection reset by peer) |
| 08:10:59 | * | ch4tbridge joined #nim |
| 08:11:19 | * | endragor joined #nim |
| 08:11:22 | * | xet7 joined #nim |
| 08:24:26 | * | couven92 joined #nim |
| 08:25:09 | * | yglukhov joined #nim |
| 08:51:58 | * | Ven joined #nim |
| 08:52:21 | * | Ven is now known as Guest37379 |
| 08:55:04 | * | ShalokShalom_ is now known as ShalokShalom |
| 08:57:20 | * | Yardanico joined #nim |
| 08:58:55 | * | adeohluwa joined #nim |
| 09:00:33 | * | jinshil quit (Quit: Good-bye!) |
| 09:16:32 | * | dom96|w joined #nim |
| 09:22:23 | * | gokr quit (Ping timeout: 246 seconds) |
| 09:38:06 | * | gokr joined #nim |
| 09:45:30 | * | dom96|w quit (Quit: My Mac has gone to sleep. ZZZzzz…) |
| 09:47:23 | * | dom96|w joined #nim |
| 09:50:11 | * | BitPuffin|osx joined #nim |
| 09:53:41 | * | Guest37379 quit (Quit: My MacBook has gone to sleep. ZZZzzz…) |
| 10:17:55 | Yardanico | how can I create a unique procedure name inside of dirty template? |
| 10:19:00 | euantor | is the procedure always going to have a similar name format? You could always maintain a global counter and increment it so you have `someProc_1`, `somreProc_2`, etc |
| 10:19:31 | Yardanico | well I've tried that |
| 10:19:45 | Yardanico | the problem is - I'm trying to use a template inside of a macro, and after that getAst() on it |
| 10:19:53 | Yardanico | Araq said in IRC that templates are better than quote do? |
| 10:19:57 | * | v17d quit (Ping timeout: 240 seconds) |
| 10:20:40 | euantor | I'm not 100% sure, I haven't tried to do anything like that before |
| 10:21:02 | Araq | well if you use a .dirty template then by definition it doesn't support "gensym" |
| 10:21:10 | Yardanico | ah, ok then |
| 10:21:13 | Araq | you can pass the proc name as a template paramter though |
| 10:21:22 | Araq | and in the macro use genSym(...) |
| 10:21:37 | Yardanico | ah, ok |
| 10:22:28 | Yardanico | Araq, I pneed to pass proc name as "untyped" or "typed"? |
| 10:24:16 | Araq | it doesn't matter, getAst ignores the template's types |
| 10:27:41 | * | arnetheduck joined #nim |
| 10:30:18 | * | Matthias247 joined #nim |
| 10:33:41 | * | v17d joined #nim |
| 10:49:52 | * | ch4tbridge quit (Remote host closed the connection) |
| 10:55:50 | Yardanico | Can I add current filename to logger format? |
| 10:56:21 | Yardanico | e.g. if I call logging procedure from file foo.nim, it would print "[13:55:39][N][foo.nim] Something" |
| 10:58:16 | * | haha_ joined #nim |
| 11:01:48 | Araq | dunno, can you? |
| 11:03:06 | * | ch4tbridge joined #nim |
| 11:08:21 | * | adeohluwa quit (Quit: Connection closed for inactivity) |
| 11:09:32 | * | mahsav2 joined #nim |
| 11:10:03 | * | mahmudov joined #nim |
| 11:12:27 | * | mahsav quit (Ping timeout: 240 seconds) |
| 11:37:16 | * | mahmudov quit (Ping timeout: 255 seconds) |
| 11:41:02 | FromGitter | <faustinoaq> > nothing will ever replace c/c++ ⏎ ⏎ Perhaps not "ever" but "in a long time", *skrylar_:* Maybe Nim can compile to Rust in the future |
| 11:42:51 | FromGitter | <faustinoaq> Also, LLVM is a good backend https://github.com/arnetheduck/nlvm |
| 11:46:17 | * | enthus1ast- quit (Quit: Lost terminal) |
| 11:49:15 | FromGitter | <Smaehtin> https://gist.github.com/zacharycarter/844b9898e06def6d18e00e3d209aaf9d - hmm, flow-sensitive typing in Nim, or just an ugly hack? :D |
| 12:04:30 | * | dom96|w quit (Quit: My Mac has gone to sleep. ZZZzzz…) |
| 12:09:36 | * | mahmudov joined #nim |
| 12:13:27 | * | dave24 quit (Ping timeout: 246 seconds) |
| 12:14:33 | * | dave24 joined #nim |
| 12:20:53 | * | endragor quit (Quit: Leaving...) |
| 12:21:27 | * | dave24 quit (Ping timeout: 240 seconds) |
| 12:23:18 | * | dave24 joined #nim |
| 12:34:42 | * | couven92 quit (Quit: Client disconnecting) |
| 12:36:17 | Araq | Yardanico: I dunno if the logging module comes with this feature but it's pretty easy to do either way |
| 12:36:33 | Yardanico | macros.lineinfo? |
| 12:36:47 | Yardanico | (I mean get filename) |
| 12:36:54 | Araq | system.instantionInfo |
| 12:37:20 | Araq | you need to wrap your log call in a template for this but this is exactly what you need to do anyway |
| 12:37:28 | Yardanico | yeah |
| 12:37:35 | Yardanico | thanks |
| 12:37:46 | * | dom96|w joined #nim |
| 12:41:17 | * | mahsav joined #nim |
| 12:43:13 | * | mahsav2 quit (Ping timeout: 248 seconds) |
| 13:07:58 | * | couven92 joined #nim |
| 13:22:30 | * | endragor joined #nim |
| 13:24:11 | dom96|w | Maybe we could get some ideas from this: https://gist.github.com/lattner/31ed37682ef1576b16bca1432ea9f782 |
| 13:40:24 | * | rauss joined #nim |
| 13:44:17 | * | dddddd joined #nim |
| 13:47:12 | * | vivus joined #nim |
| 13:56:43 | * | haha_ quit (Quit: haha_) |
| 13:57:05 | FromGitter | <andreaferretti> What's up with the flow sensitive typing example? |
| 13:57:16 | FromGitter | <andreaferretti> Does the compiler know about options or what? |
| 13:57:18 | FromGitter | <andreaferretti> :-o |
| 13:57:28 | euantor | Posted that in #nim-offtopic earlier dom96|w, looks interesting |
| 13:58:17 | euantor | async/await makes sense, but it looks like they're going the C# way with it and generating a state machine at compile time which is a little bit complex |
| 13:59:15 | FromGitter | <andreaferretti> oh I see, the `x` was injected by the template |
| 13:59:27 | FromGitter | <andreaferretti> really an ugly hack! |
| 13:59:34 | FromGitter | <andreaferretti> :-D |
| 13:59:54 | FromGitter | <superfunc_twitter> yeah I was gonna go with hack haha |
| 14:11:58 | * | dddddd_ joined #nim |
| 14:12:05 | * | dddddd quit (Ping timeout: 240 seconds) |
| 14:23:05 | * | couven92 quit (Ping timeout: 240 seconds) |
| 14:24:10 | FromGitter | <Smaehtin> |
| 14:28:05 | * | PMunch joined #nim |
| 14:29:25 | * | PMunch quit (Client Quit) |
| 14:29:47 | * | PMunch joined #nim |
| 14:33:27 | * | dom96|w quit (Quit: My Mac has gone to sleep. ZZZzzz…) |
| 14:35:45 | * | dom96|w joined #nim |
| 14:48:26 | FromGitter | <andreaferretti> It's actually pretty cool! :-D |
| 14:48:28 | * | ofelas joined #nim |
| 14:53:02 | FromGitter | <Smaehtin> Yeah, I think so too! You can sort-of mimick flow-sensitive typing in that way |
| 14:56:12 | FromGitter | <Smaehtin> The same principle can be used to make something like: ⏎ ⏎ ```if widget.ofType(ListView): ⏎ widget.addItem(...) # widget is ListView here``` [https://gitter.im/nim-lang/Nim?at=5997000c2723db8d5eb9cd51] |
| 14:57:23 | Yardanico | there are variant types in nim also |
| 14:59:11 | * | mahmudov quit (Ping timeout: 248 seconds) |
| 15:03:15 | FromGitter | <krux02> TODO Nim In Action arrived, yaya |
| 15:03:17 | * | dom96|w quit (Quit: My Mac has gone to sleep. ZZZzzz…) |
| 15:04:43 | FromGitter | <krux02> not todo, I mean today and instead of yaya i mean yay |
| 15:04:45 | FromGitter | <krux02> I can't type |
| 15:08:18 | * | krux02 joined #nim |
| 15:13:54 | FromGitter | <Smaehtin> Of course ... Nim has everything! |
| 15:14:52 | FromGitter | <Smaehtin> I've had the e-book version of Nim in Action for a while, it's really nice! |
| 15:23:01 | * | nsf quit (Quit: WeeChat 1.9) |
| 15:37:03 | * | BitPuffin|osx quit (Ping timeout: 248 seconds) |
| 15:40:29 | * | Sembei joined #nim |
| 15:40:44 | * | Matthias247 quit (Read error: Connection reset by peer) |
| 15:42:32 | * | Pisuke quit (Ping timeout: 260 seconds) |
| 15:46:09 | FromGitter | <krux02> I am a bit happy now because of the book |
| 15:46:19 | FromGitter | <krux02> I already had my doubts that it will ever arrive |
| 15:46:22 | FromGitter | <krux02> but it did |
| 15:46:27 | FromGitter | <krux02> and I am happy |
| 15:59:18 | Yardanico | how to make my application use less ram? Well I don't have any issues with performance, just curious that I would need to do |
| 16:02:35 | * | Sentreen quit (Ping timeout: 240 seconds) |
| 16:03:47 | * | mahsav2 joined #nim |
| 16:05:35 | * | mahsav quit (Ping timeout: 246 seconds) |
| 16:05:57 | * | endragor quit (Remote host closed the connection) |
| 16:06:29 | * | endragor joined #nim |
| 16:07:30 | FromGitter | <krux02> well I once did set a limit for all applications that I start from the shell |
| 16:07:55 | FromGitter | <krux02> very useful, because istead of swapping and making my system horribly slow the program just crashes because of not enough memory |
| 16:08:18 | Yardanico | well my program takes ~10-30mb ram, but it was lower some time ago :) |
| 16:08:19 | FromGitter | <krux02> but generally, use efficient memory structures |
| 16:08:34 | FromGitter | <krux02> well my recommendation, don't use ref types |
| 16:08:37 | Yardanico | ah |
| 16:08:43 | Yardanico | maybe that's the problem? |
| 16:09:04 | Yardanico | I use them almost everythere. So instead of "let something = SomeRefType" I would do "var something = SomeObjType" ? |
| 16:09:14 | Yardanico | : instead of = |
| 16:09:20 | Yardanico | I think you understood me |
| 16:09:26 | FromGitter | <krux02> ref types have the overhead of the pointer that references them, and the header of the object itself |
| 16:09:43 | FromGitter | <krux02> well you can stull use let |
| 16:09:47 | Yardanico | well I know |
| 16:09:51 | Yardanico | but I often need to change objects |
| 16:10:10 | Yardanico | and what's the best GC if I don't care about amazing GC speed but I care about RAM? |
| 16:10:11 | FromGitter | <krux02> it's better coder, when you don't reed to |
| 16:10:21 | * | endragor_ joined #nim |
| 16:10:26 | FromGitter | <krux02> only change the ``result`` variable |
| 16:10:30 | * | MyMind joined #nim |
| 16:11:11 | FromGitter | <krux02> make a factory function and that one sets all members of result but when it returns, don't touch it, unless you really need to |
| 16:11:11 | * | endragor quit (Ping timeout: 248 seconds) |
| 16:11:34 | FromGitter | <krux02> well the fastest GC is when you don't need GC |
| 16:11:57 | FromGitter | <krux02> but GC itself does not affect memory usage |
| 16:12:12 | FromGitter | <krux02> just when all your types are reference types because you program in something like java |
| 16:12:25 | FromGitter | <krux02> then the fact that all types are reference types consts you a lot of memory |
| 16:12:26 | * | Sembei quit (Ping timeout: 240 seconds) |
| 16:13:09 | Yardanico | ah, ok then |
| 16:14:35 | * | endragor_ quit (Ping timeout: 240 seconds) |
| 16:15:38 | * | Sentreen joined #nim |
| 16:16:03 | * | Pisuke joined #nim |
| 16:16:37 | FromGitter | <andreaferretti> in general, I suggest to use `let` for things that you don't want to mutate and `var` for things that you need to mutate |
| 16:17:00 | FromGitter | <andreaferretti> then use `ref` if you want to allocate the object on the heap instaed of the stack |
| 16:17:19 | Yardanico | well I know about let vs var difference |
| 16:17:37 | * | MyMind quit (Ping timeout: 248 seconds) |
| 16:17:59 | FromGitter | <andreaferretti> this is usually because: ⏎ ⏎ 1) either the object outlives the current function, or ⏎ 2) the object is big and you fear that passing it by value will incur the cost of copying [https://gitter.im/nim-lang/Nim?at=59971337ee5c9a4c5fd4f4ac] |
| 16:18:21 | FromGitter | <andreaferretti> (but the compiler can actually optimize this by passing a pointer on the stack, at least in some cases) |
| 16:19:00 | FromGitter | <andreaferretti> another reason to use `ref` types is when the type refers to itself |
| 16:19:09 | FromGitter | <andreaferretti> as would happen in a tree or a linked list |
| 16:19:20 | FromGitter | <andreaferretti> otherwise, stick to allocating on the stack |
| 16:19:39 | FromGitter | <andreaferretti> it will be faster, more predictable and will save you memory |
| 16:20:01 | krux02 | http://ix.io/zgy |
| 16:21:08 | krux02 | don't use linked lists, and trees don't necessarily need pointers |
| 16:22:12 | krux02 | instead of pointers, one can also use indices to an array |
| 16:22:47 | krux02 | indices can be, depending on the expected array length much smaller than a pointer |
| 16:24:52 | krux02 | how do I deveop on the compiler, when the change in system.nim would break the current implementation of the compiler? |
| 16:26:25 | FromGitter | <andreaferretti> well, I was not suggesting to use linked lists, just making a point :-) |
| 16:26:48 | FromGitter | <andreaferretti> `ref` types allow recursion in type definitions |
| 16:29:46 | * | arnetheduck quit (Ping timeout: 255 seconds) |
| 16:30:15 | krux02 | yea I know, but linked lists they are the default beginners example, but in practice they should just be avoided |
| 16:30:28 | krux02 | I would prefer if it would be a less popular datastructure |
| 16:30:49 | krux02 | I think the c++ dequeue is the better linked list |
| 16:34:44 | krux02 | its implementation is unspecified, but it is probably several chunks of continuous memory |
| 16:36:46 | * | ch4tbridge quit (Remote host closed the connection) |
| 16:46:37 | Yardanico | Why I'm getting "Error: illegal capture 'msg'" after I've changed attachesproc attaches*(msg: Message, vk: VkApi): Future[seq[Attachment]] {.async.} = to proc attaches*(msg: var Message, vk: VkApi): Future[seq[Attachment]] {.async.} = (previously Message was a ref object, now it's an object) |
| 16:47:00 | Yardanico | If I try to use "msg" inside of this proc, I'm getting this error |
| 16:51:31 | FromGitter | <Smaehtin> @Yardanico you cannot use var parameters in async procs |
| 16:51:45 | Yardanico | lol |
| 16:52:25 | Yardanico | well so I need to stick to ref types then :( |
| 16:53:37 | * | nattefrost quit (Ping timeout: 276 seconds) |
| 16:53:40 | FromGitter | <Smaehtin> Or `ptr` :) |
| 16:53:41 | krux02 | Yardanico: maybe you can make a local copy and capture that |
| 16:53:48 | krux02 | yes or ptr |
| 16:54:39 | krux02 | how do I compile nim with a compiler from somewhere else? |
| 16:54:50 | krux02 | when I run ./koch it tries to bootstrap itself |
| 16:54:58 | krux02 | and that is something I break :D |
| 16:56:27 | * | vivus quit (Quit: Leaving) |
| 16:56:48 | Yardanico | krux02, wouldn't it be worse than ref types? If I would do local copies |
| 17:06:07 | krux02 | well I don't know you code, but local copies are normally not a bad thing |
| 17:06:20 | krux02 | a local copy only consumes temorary memory on the stack |
| 17:06:32 | krux02 | so it does never eat up all your ram |
| 17:06:52 | krux02 | in extreme cases it can cause a stack overflow, but tha is generally not the case unleass you allocate big arrays on the stack |
| 17:07:01 | krux02 | (or deep recursion) |
| 17:10:11 | * | Trustable joined #nim |
| 17:25:01 | krux02 | Araq: are you there? If yes, can you tell me the tool again that helps developing the compiler |
| 17:25:38 | krux02 | I want to change something in system.nim but then I can't run nim anymore, because it relies on system.nim |
| 17:26:05 | * | dddddd_ quit (Ping timeout: 240 seconds) |
| 17:37:29 | * | yglukhov quit (Remote host closed the connection) |
| 17:39:22 | * | dddddd_ joined #nim |
| 17:40:05 | * | nhywyll joined #nim |
| 17:43:23 | * | yglukhov joined #nim |
| 17:44:53 | * | yglukhov quit (Remote host closed the connection) |
| 17:51:37 | * | yglukhov joined #nim |
| 17:53:11 | * | acidx joined #nim |
| 17:56:05 | * | yglukhov quit (Ping timeout: 240 seconds) |
| 17:56:57 | * | haha_ joined #nim |
| 17:59:44 | * | haha_ quit (Client Quit) |
| 18:08:38 | * | nsf joined #nim |
| 18:09:16 | * | dddddd_ quit (Quit: Hasta otra..) |
| 18:09:37 | * | dddddd_ joined #nim |
| 18:09:42 | * | dddddd_ is now known as dddddd |
| 18:11:47 | Araq | krux02: that's not a question of tooling |
| 18:11:54 | Araq | here is what you can do: |
| 18:12:04 | Araq | # system.nim |
| 18:12:10 | Araq | when defined(newStuff): |
| 18:12:11 | Araq | A |
| 18:12:13 | Araq | else: |
| 18:12:16 | Araq | oldThing |
| 18:12:32 | Araq | koch temp c -d:newStuff program.nim |
| 18:12:38 | FromGitter | <krux02> ah, ok |
| 18:13:02 | FromGitter | <krux02> does make sense |
| 18:14:19 | * | sz0 quit (Quit: Connection closed for inactivity) |
| 18:18:52 | * | andrzejku joined #nim |
| 18:18:53 | andrzejku | hey |
| 18:18:55 | andrzejku | :) |
| 18:19:22 | Araq | hmm my disk is full with nothing obvious I could delete |
| 18:21:27 | Araq | I am gonna remove every directory that starts with a dot |
| 18:21:47 | FromGitter | <krux02> how about you use a disk usage analyzer |
| 18:21:59 | FromGitter | <krux02> then you will find out that you have a gigantic download folder |
| 18:23:01 | * | haha_ joined #nim |
| 18:23:10 | FromGitter | <krux02> whenever I do that, I find out that what my usual memory eaters are |
| 18:23:24 | FromGitter | <krux02> I did not have memory space problems anymore in years |
| 18:23:40 | FromGitter | <krux02> but when I did it normally was movies and games |
| 18:24:33 | FromGitter | <krux02> and other people collect hard drive dumps, that eats memory quickly |
| 18:24:56 | * | v17d quit (Ping timeout: 240 seconds) |
| 18:26:11 | * | haha_ quit (Client Quit) |
| 18:27:17 | krux02 | Araq: is the define(nimnomagic64) still relevant? |
| 18:27:46 | Araq | read compiler/condsyms.nim |
| 18:27:57 | * | haha_ joined #nim |
| 18:27:59 | Araq | iirc that's always defined now |
| 18:29:37 | Araq | disk usage analyser? what's that, I'm on OSX here. there will be 20 different programs that do this, every single one will be crap and in the end I'll copy some cryptic shell script and hope for the best |
| 18:30:32 | krux02 | well the first one I used was integrated in KDE |
| 18:30:44 | krux02 | it shows graphically what eats most bytes |
| 18:31:02 | Araq | and why is it that every website can now detect my adblocker? |
| 18:31:29 | Araq | aren't adblockers supposed to be undetectable? |
| 18:32:18 | Araq | no, I will not disable my adblocker for your crappy website. I'll use a different website then, thank you very much. |
| 18:33:08 | krux02 | I use baobab |
| 18:33:28 | krux02 | for disk usage |
| 18:33:37 | Araq | wow that's almost a word |
| 18:33:56 | krux02 | and for add blocking, I heared μBlock Origin is the best (not μBlock) |
| 18:34:03 | * | relaxbox joined #nim |
| 18:34:12 | krux02 | yea I had to look it up how it was called |
| 18:35:04 | FromGitter | <Smaehtin> μBlock Origin + anti-adblock filters + anti-anti-adblock filters + anti-anti-anti-adblock filters ... and you'll still get some sites that block you because you're using an adblocker |
| 18:35:15 | * | yglukhov joined #nim |
| 18:36:45 | krux02 | oh I just realized I have 52 GB in trash |
| 18:38:15 | * | haha_ quit (Quit: haha_) |
| 18:38:35 | Araq | and the tool only lets me see dirs and files, not delete anything, yay |
| 18:39:26 | * | yglukhov quit (Ping timeout: 240 seconds) |
| 18:39:47 | * | v17d joined #nim |
| 18:39:59 | * | haha_ joined #nim |
| 18:40:24 | krux02 | first world problems |
| 18:40:30 | krux02 | you get the information you need |
| 18:40:50 | krux02 | the kde tool was a bit better, because it was integratetd in the file browser |
| 18:42:42 | Araq | no, not "first world problems", given all the hype OSX got, I actually expect its braindead "Finder" program to do that for me |
| 18:42:51 | Araq | out of the box. |
| 18:42:54 | krux02 | Araq: is condsyms for new features, too? |
| 18:43:12 | Araq | it's the general mechanism we use for patching system.nim, yes |
| 18:43:28 | krux02 | well I don't really use file browsers anymore |
| 18:43:33 | krux02 | I have my launchers |
| 18:43:42 | Araq | I mean so that both old and new compilers can use the same system.nim |
| 18:44:16 | Araq | well ok but "it's so broken I stopped using it" is basically my point as well |
| 18:44:17 | krux02 | so I should add a new define in there |
| 18:44:24 | Araq | yes |
| 18:44:32 | krux02 | and old compilers will just not know it |
| 18:45:15 | krux02 | I think I am still most familiar with the windows XP file browser |
| 18:45:24 | krux02 | because I used that for a very long time |
| 18:45:50 | krux02 | on linux that thing switched so often and I never really liked one (except the one from KDE3) |
| 18:46:04 | krux02 | and eventually I just did not use them anymore |
| 18:46:14 | krux02 | drop down terminal |
| 18:46:16 | krux02 | yay |
| 18:46:48 | krux02 | well ok I still use the one for mounting usb drives, but that is rara |
| 18:46:50 | krux02 | rare |
| 19:05:22 | * | Jesin joined #nim |
| 19:07:08 | * | xet7 quit (Quit: Leaving) |
| 19:19:38 | * | gangstacat quit (Quit: Leaving) |
| 19:25:57 | * | haha_ quit (Quit: haha_) |
| 19:32:08 | * | endragor joined #nim |
| 19:36:26 | * | endragor quit (Ping timeout: 240 seconds) |
| 19:46:20 | krux02 | Araq: I currently try to get rid of int as a distinct type, so that functions don't need to be overloaded for int anymore |
| 19:46:55 | krux02 | but I broke something I did not expect to break |
| 19:47:58 | krux02 | Error: internal error: wanted: tyInt got: tyInt64 |
| 19:53:58 | * | rauss quit (Quit: WeeChat 1.9) |
| 19:58:15 | * | Vladar quit (Remote host closed the connection) |
| 20:01:18 | * | haha_ joined #nim |
| 20:06:55 | * | relaxbox quit () |
| 20:08:57 | * | RelaxBox joined #nim |
| 20:09:10 | FromGitter | <zacharycarter> I need a web app idea |
| 20:09:42 | krux02 | zacherycartey: your 3d engine in webassembly |
| 20:10:24 | FromGitter | <zacharycarter> krux02: not a bad idea |
| 20:11:15 | FromGitter | <zacharycarter> I'm trying to figure out an idea for a web app though as well |
| 20:11:16 | krux02 | it's is also an advantage to me, because it paves the way for me to publish to webassembly |
| 20:11:42 | krux02 | I think there will be some friction initially |
| 20:14:35 | FromGitter | <zacharycarter> I imagine so |
| 20:23:26 | * | ShalokShalom quit (Remote host closed the connection) |
| 20:24:03 | * | Yardanico quit (Remote host closed the connection) |
| 20:27:01 | * | Arrrr quit (Read error: Connection reset by peer) |
| 20:32:50 | krux02 | zacharycarter: well I try to focus on the things that are important |
| 20:33:54 | FromGitter | <zacharycarter> krux02: debugging web assembly compilation errors isn't the most joyous thing in the world |
| 20:34:51 | krux02 | well when you think about it, programming isn't the most joyous thing, because it contains a lot of reading compilation errors |
| 20:35:10 | krux02 | it just becomes fun as soon as you are successful in what you wat to do |
| 20:35:15 | FromGitter | <zacharycarter> true |
| 20:35:23 | FromGitter | <zacharycarter> it's just webassembly errors look like this |
| 20:35:32 | krux02 | so get good at webassembly and you will see joy in it, and it is an important technology |
| 20:35:45 | krux02 | it helps outside of the Nim community as well |
| 20:35:48 | FromGitter | <zacharycarter> ```code paste, see link``` [https://gitter.im/nim-lang/Nim?at=59974fa480d90ca0241326f8] |
| 20:35:53 | krux02 | you might be able to make money from it |
| 20:36:31 | FromGitter | <krux02> well I had problems getting to know how to debug with gdb |
| 20:36:39 | FromGitter | <krux02> it is horrible to use from the command line |
| 20:37:29 | FromGitter | <krux02> but then at some point it is just nice to use when you actually know what you are doing |
| 20:37:32 | FromGitter | <krux02> same with emacs |
| 20:42:29 | * | haha_ quit (Quit: haha_) |
| 20:49:10 | * | gokr quit (Ping timeout: 240 seconds) |
| 20:58:50 | * | ShalokShalom joined #nim |
| 20:59:23 | RelaxBox | what's the deal with slices and arrays / openarrays? Can they be sliced? |
| 21:01:24 | RelaxBox | looks like not, if I've learned how to read the type mismatch errors |
| 21:04:42 | * | gokr joined #nim |
| 21:13:35 | * | gokr quit (Ping timeout: 248 seconds) |
| 21:16:37 | FromGitter | <stisa> RelaxBox : looks like there's no overloaded `[]` with slices for openarray, don't know why really. You can do something like this I guess https://play.nim-lang.org/?gist=f2a7111712aadec58675af462197181b |
| 21:19:09 | FromGitter | <TiberiumN> @stisa or you can just convert openarray to seq :) |
| 21:19:20 | FromGitter | <TiberiumN> But yeah, this has a lot of overhead |
| 21:37:33 | * | xmonader joined #nim |
| 21:38:31 | xmonader | for parseopt2 library, ./echo hello "-n" -> why -n here is parsed as a switch even if it's enclosed in double quotes? |
| 21:40:43 | * | def-pri-pub joined #nim |
| 21:42:29 | * | andrzejku quit (Remote host closed the connection) |
| 21:42:34 | * | BitPuffin|osx joined #nim |
| 21:48:28 | * | RelaxBox quit (Ping timeout: 240 seconds) |
| 21:52:30 | FromGitter | <krux02> well that is the shell |
| 21:53:17 | FromGitter | <krux02> xmonader: the double quotes are something that are processed by the shell the program you launch doesn't know about the double quotes |
| 21:53:25 | FromGitter | <krux02> it just gets the arguments |
| 21:53:25 | * | BitPuffin|osx quit (Ping timeout: 240 seconds) |
| 21:54:14 | FromGitter | <krux02> eg: ./echo 'hello world' "foo bar" xyz\ abc |
| 21:54:40 | FromGitter | <krux02> will have the arguments list ["hello world", "foo bar", "xyz abc"] |
| 21:55:40 | xmonader | @krux02, You're correct, my mistake thank you :) |
| 21:57:39 | FromGitter | <krux02> I am listening to the Unreal soundtrack |
| 21:57:56 | FromGitter | <krux02> I kind of miss the style of such games. |
| 21:59:00 | FromGitter | <krux02> but on the other hand I recently got myself Dark Souls, that is what I like about old games |
| 22:00:10 | * | mahmudov joined #nim |
| 22:12:35 | * | endragor joined #nim |
| 22:16:55 | * | endragor quit (Ping timeout: 240 seconds) |
| 22:21:44 | * | couven92 joined #nim |
| 22:33:40 | krux02 | Araq: well I think I need help for this |
| 22:34:43 | Araq | I'm here, kinda |
| 22:34:53 | krux02 | cool |
| 22:35:25 | krux02 | I am in the nim compiler, and I try to get rid of the distinct int |
| 22:35:58 | krux02 | so that int is defined to be one of int8 int16 int32 or int64 |
| 22:36:27 | krux02 | but it breaks the int literal |
| 22:36:56 | krux02 | in the expression ``let x : int32 = 123`` I get a type mismatch |
| 22:37:09 | krux02 | and I don't understand the type matching process |
| 22:38:23 | Araq | well if you get rid of int, rename tyInt to tyOldInt and let the type checker make you fix it? |
| 22:39:20 | krux02 | how should I proceed then? |
| 22:39:33 | krux02 | integer literals are a bit complicated afaik |
| 22:39:54 | krux02 | meaning let x = 123 will be of type int == int64 |
| 22:40:08 | krux02 | bit let x : int8 = 123 should work, too |
| 22:41:05 | krux02 | so the literal can't be fixed to be type int64 |
| 22:48:21 | * | Jesin quit (Ping timeout: 240 seconds) |
| 22:49:20 | Araq | doesn't seem hard, the compiler uses tyInt and t.n.intVal for an integer literal |
| 22:49:31 | Araq | you could use tyInt32 and t.n.intVal instead |
| 22:51:02 | * | Jesin joined #nim |
| 22:51:29 | krux02 | well it might not be hard for you, but I don't exactly know how the literal encode the state that it does not yet have a fixed type or not. |
| 22:51:59 | Araq | I just told you how it works inernally |
| 22:52:17 | krux02 | you think, but I still feel puzzled |
| 22:53:55 | krux02 | I don't feel comfortable to use tyInt32 for "the node has an unspecified integer type" |
| 22:55:29 | Araq | well I dunno why tyInt is so annoying for you that you're gonna change ~80_000 lines of code |
| 22:57:09 | krux02 | well it is |
| 22:57:34 | Araq | but you should read the manual first |
| 22:57:41 | Araq | how it currently works |
| 22:58:02 | krux02 | well good, nim in action arrived today :P |
| 23:00:27 | * | couven92 quit (Quit: Client Disconnecting) |
| 23:01:05 | * | nhywyll quit (Ping timeout: 240 seconds) |
| 23:01:44 | krux02 | no I change this, for the reason that I think it is an improvement to that language by taking away something |
| 23:02:05 | krux02 | and it get's me to understand the internals of the compiler better |
| 23:02:15 | krux02 | so I do it |
| 23:02:43 | krux02 | I really want to make nim as my default language to use. |
| 23:02:52 | krux02 | And I use a little subset of Nim that I like |
| 23:03:05 | krux02 | and from that subset I would like to make sure that I really like it |
| 23:03:18 | krux02 | and I would like to make this subset a really good language |
| 23:03:39 | krux02 | and this includes taking away the distictiveness of the integer type |
| 23:11:13 | dom96 | you want to relax the integer type rules? |
| 23:13:59 | Araq | krux02: fine with me but it's lots of tedious work |
| 23:14:14 | krux02 | I know |
| 23:14:19 | Araq | I dunno what else I could explain to you |
| 23:14:29 | Araq | Nim has the notion of an "int literal" type |
| 23:14:38 | krux02 | yes |
| 23:14:40 | Araq | that's currently not a TypeKind |
| 23:14:51 | krux02 | how is it encoded then? |
| 23:15:05 | Araq | but just tyInt with t.n (the node part of the type) set to its value |
| 23:15:33 | Araq | you could use tyInt32 with t.n for this |
| 23:15:53 | Araq | or keep tyInt with t.n and all the other usages of tyInt become tyInt32 |
| 23:16:37 | krux02 | then how is this different to an int32 literal? |
| 23:17:09 | Araq | an int32 literal is a node with type tyInt32 and t.n == nil |
| 23:17:41 | Araq | an int literal would be a node with type tyInt32 and t.n != nil |
| 23:18:18 | krux02 | and where is the value stored in an int32 literal then? |
| 23:18:22 | Araq | but you could also rename tyInt to tyIntLit and make the code more obvious |
| 23:18:34 | Araq | an int32 literal is not a type! |
| 23:18:36 | krux02 | yes I will |
| 23:18:41 | Araq | it's a node! |
| 23:18:56 | Araq | an int literal is a node with a special "int literal" type |
| 23:19:00 | krux02 | ah the type has the value |
| 23:19:04 | krux02 | I get it |
| 23:19:05 | Araq | big difference, yes |
| 23:21:25 | krux02 | still very confusing to me |
| 23:22:09 | krux02 | when I have a int32 literal, I have a node with the value and the type with kind == tyInt32 |
| 23:22:44 | krux02 | when I have a literal thas is yet unspecified, the value is stored in the type, but then what value has the node then? |
| 23:25:02 | Araq | the value that you think it has |
| 23:25:08 | Araq | it exists twice |
| 23:26:02 | krux02 | ok |
| 23:26:06 | krux02 | so confusing |
| 23:26:43 | Araq | not really |
| 23:27:26 | * | Jesin quit (Read error: Connection reset by peer) |
| 23:28:28 | krux02 | well for me it is |
| 23:28:33 | krux02 | that is a bit abusive |
| 23:28:47 | krux02 | you spent a lot of time there and probably are totally used to it |
| 23:28:53 | krux02 | for me this is complicated |
| 23:29:09 | krux02 | and hard to grasp |
| 23:30:22 | krux02 | I really like Nim like it is, but I really want to smoth a few edges before I can truly say it is the language to go. |
| 23:30:54 | krux02 | e.g. I want to safely be able to disable the GC when I don't use ref types at all |
| 23:31:01 | krux02 | yea already said that |
| 23:32:06 | krux02 | btw for the PR with the compile time offsetof and sizeof, it works everything that it needs is that the builtin types have the values set correctly internally for alignment depending on the platform. |
| 23:32:31 | krux02 | you said you want that PR, as soon as that is done, tha PR will not fail anymore. |
| 23:40:03 | Araq | well there was a change in the plans, but I can merge it tomorrow. good night |
| 23:42:03 | * | xmonader quit (Quit: Leaving) |
| 23:55:25 | * | nattefrost joined #nim |
| 23:56:25 | * | Trustable quit (Remote host closed the connection) |