| 00:02:39 | * | thomasross_ joined #nim |
| 00:02:39 | * | thomasross quit (Killed (molybdenum.libera.chat (Nickname regained by services))) |
| 00:02:39 | * | thomasross_ is now known as thomasross |
| 00:24:24 | FromDiscord | <spaceshaman> Wait... all procedures are pointers? |
| 00:24:43 | FromDiscord | <spaceshaman> why though? |
| 00:25:29 | FromDiscord | <Elegantbeef> They're not |
| 00:25:36 | FromDiscord | <Elegantbeef> They're first class |
| 00:25:51 | FromDiscord | <Elegantbeef> > A procedural type |
| 00:25:56 | FromDiscord | <Elegantbeef> Is the important part |
| 00:32:45 | FromDiscord | <spaceshaman> What I mean is, when you use the identifier of a proc, you are getting a value that is internally an address. You dont do `my_func.addr()` |
| 00:33:09 | FromDiscord | <Elegantbeef> Right cause Nim has first class procedures |
| 00:33:27 | FromDiscord | <Elegantbeef> Calling named procedures does not create a procedure type |
| 00:33:39 | FromDiscord | <Elegantbeef> This works just like C |
| 00:33:42 | FromDiscord | <spaceshaman> you can have first class functions and procedures without that |
| 00:34:02 | FromDiscord | <Elegantbeef> They're not first class if you have to call `.addr` on them to get the pointer proc |
| 00:34:26 | FromDiscord | <spaceshaman> thats like saying that nim doenst have first class integers |
| 00:34:44 | FromDiscord | <spaceshaman> which is obviously wrong |
| 00:34:48 | FromDiscord | <Elegantbeef> Well no cause then `ptr proc()` is the type when you pass them around |
| 00:35:02 | FromDiscord | <spaceshaman> of course |
| 00:35:06 | FromDiscord | <Elegantbeef> Which means you can only have a proc inside of a pointer type |
| 00:35:14 | FromDiscord | <Elegantbeef> So they're clearly not first class |
| 00:35:32 | FromDiscord | <Elegantbeef> If you cannot just pass procedures as values I find it hard to believe they're first class |
| 00:37:28 | FromDiscord | <Elegantbeef> sent a code paste, see https://play.nim-lang.org/#pasty=YNPxhaFH |
| 00:37:38 | FromDiscord | <Elegantbeef> Just for clarity |
| 00:38:20 | FromDiscord | <Elegantbeef> It serves little benefit to have a non usable `proc()` type that you have to prefix with `ptr` imo |
| 00:38:56 | FromDiscord | <spaceshaman> if you are passing procedures as values, that means you are passing around the actual function body. |
| 00:39:07 | FromDiscord | <Elegantbeef> No |
| 00:39:11 | FromDiscord | <Elegantbeef> That's not what I meant |
| 00:39:24 | FromDiscord | <Elegantbeef> sent a code paste, see https://play.nim-lang.org/#pasty=MWENyFdr |
| 00:39:35 | FromDiscord | <Elegantbeef> If you need to do `i: ptr proc()` and `myFunc.addr` that is certainly not first class |
| 00:40:09 | FromDiscord | <Elegantbeef> There is `static proc()` which allows you to pass procedures into a procedure then expand them statically instead of with a pointer proc |
| 00:41:00 | FromDiscord | <Elegantbeef> Then there's also the issue that `ptr T` is generally to be avoided, but in the case of procedures it'd be enabling them. Then we also have closures which add even more headaches |
| 00:41:21 | FromDiscord | <Elegantbeef> This is at least if I understand you right |
| 00:41:42 | FromDiscord | <spaceshaman> is there no way to get a `ref T` from an lvalue? |
| 00:41:53 | FromDiscord | <Elegantbeef> A ref is ref counted |
| 00:41:58 | FromDiscord | <Elegantbeef> So of course not |
| 00:42:05 | FromDiscord | <Elegantbeef> You have to heap allocate then move the data |
| 00:42:22 | FromDiscord | <spaceshaman> so thats what it meant by that |
| 00:42:34 | FromDiscord | <Elegantbeef> `ref proc()` certainly is not the semantics you want for a pointer proc obviously |
| 00:42:39 | FromDiscord | <spaceshaman> ptr doent work with the GC? |
| 00:42:49 | FromDiscord | <Elegantbeef> ptr is an unsafe pointer |
| 00:42:53 | FromDiscord | <Elegantbeef> No different to a C pointer |
| 00:43:01 | FromDiscord | <spaceshaman> manual got it |
| 00:43:13 | FromDiscord | <spaceshaman> is there a gc pointer type? |
| 00:43:33 | FromDiscord | <Elegantbeef> `ref` |
| 00:43:38 | FromDiscord | <Elegantbeef> That's all there is |
| 00:43:57 | FromDiscord | <Elegantbeef> It's a ref counted heap allocated data block |
| 00:44:13 | FromDiscord | <Elegantbeef> There is no other method in Nim presently to have a memory safe pointer |
| 00:44:16 | FromDiscord | <spaceshaman> are the docs referring to reference counting when they mention a garbage collector? |
| 00:44:22 | FromDiscord | <Elegantbeef> well aside from `var` and `lent` |
| 00:44:37 | FromDiscord | <Elegantbeef> With Nim 2.0 and Arc/Orc yes |
| 00:45:10 | FromDiscord | <spaceshaman> ah, Im used to GC referring to a scanning GC |
| 00:45:11 | FromDiscord | <Elegantbeef> refc is a proper garbage collector but the default is now orc which is scope ref counting + cycle collector |
| 00:45:52 | FromDiscord | <Elegantbeef> The cycle collector is a mark and sweep collector that is only supposed to be invoked on cyclical types |
| 00:46:57 | FromDiscord | <spaceshaman> I have a pretty elementary understanding of GC types and terms |
| 00:47:40 | FromDiscord | <Elegantbeef> Good thing too my dear watson |
| 00:48:05 | FromDiscord | <spaceshaman> I definitely disagree on when functions are first class |
| 00:48:35 | FromDiscord | <Elegantbeef> > In computer science, a programming language is said to have first-class functions if it treats functions as first-class citizens. This means the language supports passing functions as arguments to other functions, returning them as the values from other functions, and assigning them to variables or storing them in data structures |
| 00:49:12 | FromDiscord | <spaceshaman> I would say the reason C doesnt have first class functions is that it lacks the ability to create them as values. And when I say first class functions, I mean first class function pointers |
| 00:49:16 | FromDiscord | <Elegantbeef> > In languages with first-class functions, the names of functions do not have any special status; they are treated like ordinary variables with a function type. |
| 00:49:33 | FromDiscord | <Elegantbeef> Well that's not what anyone agrees with |
| 00:50:05 | FromDiscord | <Elegantbeef> Passing a function by value means passing a pointer |
| 00:50:20 | FromDiscord | <Elegantbeef> Since a function is a labelled unit of code |
| 00:50:32 | FromDiscord | <spaceshaman> thats not by value |
| 00:50:42 | FromDiscord | <spaceshaman> a variable is a labeled unit of memory |
| 00:50:43 | FromDiscord | <Elegantbeef> That is by value |
| 00:50:47 | FromDiscord | <Elegantbeef> Cause a function's value is it's label |
| 00:51:03 | FromDiscord | <spaceshaman> a variable's value is not its label |
| 00:51:12 | FromDiscord | <spaceshaman> thats its address |
| 00:51:26 | FromDiscord | <Elegantbeef> Functions are pointer types though |
| 00:51:56 | FromDiscord | <spaceshaman> you can have function types that are not pointer types |
| 00:52:40 | FromDiscord | <Elegantbeef> In what way? |
| 00:56:13 | FromDiscord | <majortrips1763> You know .. realistically ... functions are just an agreed upon calling symantic on how we setup the registers and the stack before moving the instruction pointer around... |
| 00:57:31 | FromDiscord | <spaceshaman> sent a long message, see https://pasty.ee/ndhDuvdG |
| 00:57:50 | FromDiscord | <majortrips1763> like .. are we going to say that there is an idea of a load&store function vs an immediate in-so-far as the calling symantics? |
| 00:58:39 | FromDiscord | <majortrips1763> In reply to @spaceshaman "they are composed of": This isn't strictly true, at the ASM level you can jump to an absolute address .. the label is just a matter of convencience.. you can even setup a whole stack of offsets for different jmp's. |
| 00:59:00 | FromDiscord | <spaceshaman> its not strictly true, but its a good generalization |
| 00:59:12 | FromDiscord | <Elegantbeef> A function is the label though |
| 00:59:13 | FromDiscord | <majortrips1763> Any sort of function table, interrupt table, jump table, etc .. there doesn't need to be a label at all. |
| 00:59:30 | FromDiscord | <Elegantbeef> What the function does requires invoking it, so it's tangential to what a function is |
| 00:59:55 | FromDiscord | <majortrips1763> @ElegantBeef no, you don't need the label .. it can be fully optimized out. At the end of the day it is just an address somewhwere that you can point the IP at. |
| 01:00:21 | FromDiscord | <spaceshaman> In reply to @majortrips1763 "Any sort of function": but thats true of variables too, so... I think you havent read enough of the converation |
| 01:00:26 | FromDiscord | <majortrips1763> but, you can pass intructions as arguments and execute them |
| 01:00:27 | FromDiscord | <spaceshaman> (edit) "converation" => "conversation" |
| 01:00:43 | FromDiscord | <majortrips1763> see: executable stack |
| 01:00:56 | FromDiscord | <Elegantbeef> The address is the label |
| 01:00:57 | FromDiscord | <spaceshaman> major you're missing the point of the convo |
| 01:01:01 | FromDiscord | <Elegantbeef> I'm not using the ASM term label |
| 01:01:31 | FromDiscord | <majortrips1763> I think when we get down into ASM we are off in the woods .. the labels are sort of arbitrary as they are aliases for an address somewhere (be it an offset or not) |
| 01:01:43 | FromDiscord | <Elegantbeef> I'm saying a function is some location in the binary, and calling that a label, and passing "by value" makes no sense given a function is nothing more than a place in memory with no useful data |
| 01:01:47 | FromDiscord | <Elegantbeef> Other than tobe invoked it's pointless |
| 01:02:55 | FromDiscord | <spaceshaman> In reply to @majortrips1763 "I think when we": right, but there are conventions that I am referring to. |
| 01:03:26 | FromDiscord | <majortrips1763> In reply to @Elegantbeef "I'm saying a function": Yah, I generaly agree on the whole of it is generally impractical as hell to imagine passing function bodies as a param to a function .. BUT .. it can be done .. but at the same time .. pretty much nothing larger than a register can "truly" be passed to a function either... |
| 01:04:20 | FromDiscord | <majortrips1763> Like .. you really can't pass a string as an argument to a function .. when we get down to immidiates or indirects .. almost nothing is going to be trully passed "as a param" .. it will almost always be an address to some piece of memory somewhere. |
| 01:04:38 | FromDiscord | <spaceshaman> In reply to @Elegantbeef "Other than tobe invoked": and thats an ok thing to have. It means that you cant mistake the type of `foo` and `foo` where atm it could be a direct call or fetching the pointer |
| 01:06:37 | FromDiscord | <Elegantbeef> There is no mistaking though |
| 01:06:42 | FromDiscord | <Elegantbeef> `myFunc is proc()` |
| 01:07:14 | FromDiscord | <majortrips1763> In reply to @spaceshaman "and thats an ok": Again, we are back to the calling ABI where we agree that we have data types of limited sizes that can be passed directly as arguments (immediate), and everything else is either copied onto the stack, or is an address pointing off into the heap.↵↵Again . I think once we are down in ASM land the distinction become super weird because a function is going to suddenly change its behavio |
| 01:12:47 | FromDiscord | <majortrips1763> But .. it is also worth pointing out that it is totally possible to pass a "copy" of a function to another function at the C/ASM level. I dunno how practical it would be... maybe as part of some tooling for writing code that self-mutates binaries in memory? |
| 01:13:04 | FromDiscord | <Elegantbeef> We have macros for that 😄 |
| 01:13:11 | FromDiscord | <majortrips1763> Like, you really could pass copies of functions just like copies of strings.. |
| 01:13:17 | FromDiscord | <Elegantbeef> Or `static proc()` is the closest to passing by value, where it can inline calls |
| 01:13:23 | FromDiscord | <majortrips1763> I thought macros only worked on the AST.. |
| 01:13:41 | FromDiscord | <Elegantbeef> Right but you can pass a symbol to a proc then do things to it |
| 01:13:54 | FromDiscord | <Elegantbeef> Of course all methods are compile time limited |
| 01:14:34 | FromDiscord | <majortrips1763> I was more off in hypothetical land, trying to imagine doing things like having a common routine in which I could pass some code to "inject" into memory and have the routine sort of PIE load the code into a random address and pass me back a pointer to it. Stuff like that. |
| 01:15:50 | FromDiscord | <Elegantbeef> Yea practically runtime macros, which sounds horrendous |
| 01:15:51 | FromDiscord | <majortrips1763> Like .. I really can't imagine a practical use for it .. but it could certainly be done. Realistically .. stack injections are kinda like passing a function and moving the instruction pointer to a bad address .. though .. most stuff should have a non-executable stack anymore to avoid that mess. |
| 01:16:16 | FromDiscord | <Elegantbeef> Excuse me whilst my program requires me to catch sigill |
| 01:18:13 | FromDiscord | <majortrips1763> I think that sort of stuff would likely be only interesting to people trying to do dynamically obfuscated binaries.. which anymore is likely only people trying to write malware... I feel like the stranglehold on "licensed binaries" and obfuscation compilers is part of the days of old.. |
| 01:19:03 | FromDiscord | <Elegantbeef> I think the worst part on the userfront is it makes a type that is pointless 😄 |
| 01:19:52 | FromDiscord | <majortrips1763> Hmm .. I wonder if it would be possible to get Nim to transpile to popcorn/tal instead of C/ASM |
| 01:20:30 | FromDiscord | <Elegantbeef> Araq does like pointing out that the VM is only a few thousand lines of code |
| 01:20:42 | FromDiscord | <majortrips1763> https://www.cs.cornell.edu/talc/ |
| 01:21:09 | FromDiscord | <majortrips1763> Was an old project .. but it implemented a full Typed Assembly language and a C-like dialect (popcorn) that targetted it. |
| 01:23:12 | FromDiscord | <majortrips1763> Stuff like this was always interesting. |
| 01:32:17 | FromDiscord | <litlighilit> What about Hot Code Reloading(https://nim-lang.org/docs/hcr.html)? |
| 01:32:23 | FromDiscord | <Elegantbeef> It don't work |
| 01:32:53 | FromDiscord | <litlighilit> all right. I used to think it helps😅 |
| 01:33:45 | FromDiscord | <Elegantbeef> https://github.com/beef331/potato/ is my attempt at some HCR goodness, not that it's overly stable(and is linux only) 😄 |
| 01:34:24 | FromDiscord | <Elegantbeef> https://streamable.com/2mxktc It does partially work! |
| 02:17:09 | * | thomasross is now known as Guest5758 |
| 02:17:09 | * | thomasross_ joined #nim |
| 02:17:09 | * | Guest5758 quit (Killed (tantalum.libera.chat (Nickname regained by services))) |
| 02:17:09 | * | thomasross_ is now known as thomasross |
| 02:17:12 | * | thomasross quit (Remote host closed the connection) |
| 04:04:48 | * | SchweinDeBurg quit (Ping timeout: 246 seconds) |
| 07:05:50 | Amun-Ra | nice |
| 08:40:05 | FromDiscord | <whisperecean> I am trying to use disruptek's atoz package but I am getting a weird error: Error: invalid number: '25ae1f9e9aa35755ec5de0b1926a9031842726200eaf5db68815970cd60e7324'. Looks like something to do with the lfs (the generated APIs are big files). |
| 10:10:08 | * | strogon14_ quit (Quit: Auf Wiese geh'n!) |
| 10:11:09 | * | strogon14 joined #nim |
| 10:43:48 | FromDiscord | <firasuke> In reply to @firasuke "Is this still true?": ? 🙂 |
| 10:52:21 | FromDiscord | <nnsee> In reply to @firasuke "? 🙂": seems fairly easy to test |
| 11:15:53 | FromDiscord | <jns> sent a code paste, see https://play.nim-lang.org/#pasty=iPlvslQJ |
| 11:16:03 | FromDiscord | <jns> (edit) "https://play.nim-lang.org/#pasty=epOUdlde" => "https://play.nim-lang.org/#pasty=MvyEGBCI" |
| 11:16:18 | FromDiscord | <jns> (edit) "https://play.nim-lang.org/#pasty=kmYfssiK" => "https://play.nim-lang.org/#pasty=rvTQrUuC" |
| 11:25:08 | FromDiscord | <odexine> only certain keywords can be used as operators |
| 11:29:27 | FromDiscord | <jns> sent a code paste, see https://play.nim-lang.org/#pasty=NCufMpfb |
| 11:37:47 | * | SchweinDeBurg joined #nim |
| 11:42:14 | * | SchweinDeBurg quit (Ping timeout: 260 seconds) |
| 11:47:50 | * | SchweinDeBurg joined #nim |
| 12:03:18 | FromDiscord | <odexine> templates require return types as well |
| 12:06:33 | * | ryuukk quit (Remote host closed the connection) |
| 12:38:39 | * | ryuukk joined #nim |
| 12:50:05 | FromDiscord | <firasuke> In reply to @nnsee "seems fairly easy to": ok gotcha xD |
| 13:10:17 | * | strogon14 quit (Quit: Auf Wiese geh'n!) |
| 13:11:04 | * | strogon14 joined #nim |
| 13:42:46 | * | beholders_eye joined #nim |
| 13:44:02 | * | ntat joined #nim |
| 13:44:56 | FromDiscord | <double_spiral> How can i change which c compiler nim uses? |
| 13:45:34 | FromDiscord | <double_spiral> Like as a path to the binary since im using a custom one |
| 13:46:58 | * | ryuukk quit (Remote host closed the connection) |
| 13:51:23 | FromDiscord | <nnsee> In reply to @double_spiral "Like as a path": see https://nim-lang.org/docs/nimc.html#compiler-selection and the section below that (the one about cross compilation) |
| 13:51:29 | FromDiscord | <nnsee> shows how you can set custom paths |
| 13:58:29 | FromDiscord | <double_spiral> thanks |
| 14:07:03 | FromDiscord | <double_spiral> Oh theres already a thing for it↵https://nim-lang.org/docs/nimc.html#crossminuscompilation-for-nintendo-switch |
| 14:07:08 | FromDiscord | <double_spiral> Surprised im not the first oen |
| 14:07:11 | FromDiscord | <double_spiral> (edit) "oen" => "one" |
| 14:36:48 | FromDiscord | <double_spiral> yessssss https://media.discordapp.net/attachments/371759389889003532/1297206823652425848/image.png?ex=6715157f&is=6713c3ff&hm=d2ca8aee2c16c926b3880a871628209dd30cd9bf1c587cf2d21ad6f38d6463fd& |
| 14:41:52 | FromDiscord | <narimiran> For those who don't check the #nimconf channel:↵NimConf will be held next Saturday!↵↵See the schedule here: https://conf.nim-lang.org |
| 14:47:50 | FromDiscord | <whisperecean> Can I use Nim js backend to write a github action? |
| 14:49:04 | * | ryuukk joined #nim |
| 15:02:56 | * | ryuukk quit (Remote host closed the connection) |
| 15:04:44 | * | ryuukk joined #nim |
| 15:05:02 | * | ryuukk_ joined #nim |
| 15:06:44 | * | ryuukk quit (Remote host closed the connection) |
| 15:23:18 | FromDiscord | <knakas> sent a code paste, see https://play.nim-lang.org/#pasty=hXlhtInl |
| 15:31:36 | FromDiscord | <knakas> Nevermind figured it out. (It's because it's wrapping a C function, I need to update the init.Chr4c to have an extra layer.) |
| 15:37:06 | * | SchweinDeBurg quit (Ping timeout: 246 seconds) |
| 15:39:05 | * | SchweinDeBurg joined #nim |
| 15:49:49 | FromDiscord | <knakas> sent a code paste, see https://play.nim-lang.org/#pasty=xRGtIiHW |
| 16:10:54 | FromDiscord | <rakgew> thank you for the notify - looking forward to nimconf!↵is that `#nimconf` room bridged to matrix by? I cannot seem to find the handle in the matrix room search.↵(@narimiran) |
| 16:44:09 | * | coldfeet joined #nim |
| 17:31:33 | * | beholders_eye quit (Ping timeout: 248 seconds) |
| 18:17:31 | FromDiscord | <double_spiral> How can i echo a number's binary value? |
| 18:18:12 | Amun-Ra | std/strutils |
| 18:19:08 | Amun-Ra | to_bin |
| 18:26:26 | * | xet7 joined #nim |
| 18:45:15 | * | xet7 quit (Remote host closed the connection) |
| 18:46:20 | * | xet7 joined #nim |
| 18:51:12 | * | xet7 quit (Remote host closed the connection) |
| 18:54:03 | * | xet7 joined #nim |
| 19:17:16 | * | xet7 quit (Quit: Leaving) |
| 19:21:35 | * | xet7 joined #nim |
| 19:46:19 | * | coldfeet quit (Remote host closed the connection) |
| 19:54:10 | * | bgupta joined #nim |
| 20:04:04 | bgupta | When defining an importc proc with optional params, and call it while leaving the optional params unset but they are all at the end it works fine and sends along the defaults for any that are missing. However, if I set one of the later optional params, and leave some unset/default, I get missing param errors. I can add an extra wrapper function to address, but this seems a little wrong. Am I missing something? (Docs didn't help.) |
| 20:14:11 | FromDiscord | <spaceshaman> are types values? |
| 20:14:33 | FromDiscord | <spaceshaman> basically, is there a `Type` type? |
| 20:14:52 | FromDiscord | <Elegantbeef> there is `typedesc` but it cannot be instantiated |
| 20:17:59 | FromDiscord | <spaceshaman> sent a code paste, see https://play.nim-lang.org/#pasty=bzvZlcEo |
| 20:19:38 | FromDiscord | <Elegantbeef> bgupta the importc should match the ABI of the importing procedure |
| 20:19:38 | FromDiscord | <Elegantbeef> As such optional parameters make little sense |
| 20:19:54 | FromDiscord | <Elegantbeef> Alternatively make an issue and hope it's fixed 😄 |
| 20:23:42 | FromDiscord | <spaceshaman> In reply to @spaceshaman "I was wondering if": wait does this matter if I use template or proc here? |
| 20:25:46 | FromDiscord | <spaceshaman> like does by using proc here does this:↵`do_stuff(int)(55)`↵ basically become this?:↵`(static do_stuff(int))(55)` |
| 20:27:20 | FromDiscord | <Elegantbeef> Oh bridge dropped a message, but yes you can do that |
| 20:28:38 | FromDiscord | <Elegantbeef> That is afterall what generics are fore |
| 20:28:38 | FromDiscord | <Elegantbeef> sent a code paste, see https://play.nim-lang.org/#pasty=WxWYlvkl |
| 20:28:39 | FromDiscord | <Elegantbeef> fore\ |
| 20:28:39 | FromDiscord | <Elegantbeef> for\\\\\\\ |
| 20:31:27 | FromDiscord | <SmaamX> Any lib like fuzzywuzzy(python) for nim? |
| 20:56:46 | FromDiscord | <Elegantbeef> https://nim-lang.org/docs/editdistance.html#editDistance%2Cstring%2Cstring |
| 21:14:23 | bgupta | To be clear I mean by optional, we would define the default to pass into the C function, so as far as C goes it's getting all it's params, but nim which does understand optional/default values, can just send the default for any params that aren't in the call to the nim function. |
| 21:15:13 | bgupta | sorry that last should say "that aren't in the call to the C function." |
| 21:15:19 | FromDiscord | <Elegantbeef> Yea it's likely a bug that it doesn't work |
| 21:16:07 | bgupta | Ah ok. |
| 21:16:31 | * | ntat quit (Quit: Leaving) |
| 22:12:14 | * | GnuYawk joined #nim |
| 23:23:25 | FromDiscord | <monkey085395> Hi, can any forum mod help me sign up? Im getting `Couldn't send email` |
