05:55 <+bridge> [ddnet] interesting 06:03 <+bridge> [ddnet] 504, now 503, eh? … 06:07 <+bridge> [ddnet] now 502 06:07 <+bridge> [ddnet] at this rate, not long before 200, heh? 06:09 <+bridge> [ddnet] (ah, now it's done) 08:10 <+bridge> [ddnet] yo dont mind my times i always start 2 hours later to get some sleep 08:11 <+bridge> [ddnet] just admit it took you two hours to get the solution 08:12 <+bridge> [ddnet] lies! 08:22 <+bridge> [ddnet] No points for day 1 due too technical problems :> 08:42 <+bridge> [ddnet] > If you want to learn these techniques then start by carefully reading Jeff Preshing’s introduction to lock-free programming or This is Why They Call It a Weakly-Ordered CPU, and then consider joining a monastery or nunnery instead. 08:42 <+bridge> [ddnet] https://randomascii.wordpress.com/2020/11/29/arm-and-lock-free-programming/ 08:54 <+bridge> [ddnet] @deen is that an answer to something? 08:59 <+bridge> [ddnet] No 09:12 <+bridge> [ddnet] just woke up :monkalaugh: 09:12 <+bridge> [ddnet] time to code it 09:13 <+bridge> [ddnet] @heinrich5991 yesterday i did my first code golf answer, maybe u can improve it (in rust) https://codegolf.stackexchange.com/a/215799/58802 09:13 <+bridge> [ddnet] `(2..).for_each(|x|(1..x).for_each(|y|print!("{} ",y)))` 09:13 <+bridge> [ddnet] :monkalaugh: 09:14 <+bridge> [ddnet] i didnt know u could go to the infinite with (2..) 09:14 <+bridge> [ddnet] before that 09:41 <+bridge> [ddnet] how about 09:41 <+bridge> [ddnet] `rand="0"` 09:41 <+bridge> [ddnet] 9 bytes with newline 09:41 <+bridge> [ddnet] in the Cargo.toml and 09:41 <+bridge> [ddnet] `loop{print!("{} ", rand::random::())}` 09:41 <+bridge> [ddnet] + 42 bytes = 51 bytes @Ryozuki 09:42 <+bridge> [ddnet] idk if u can use packages 09:44 <+bridge> [ddnet] also it just spits out random numbers 09:44 <+bridge> [ddnet] huh i mean else you wouldnt be able to solve any problems which play with random numbers in rust 09:44 <+bridge> [ddnet] it doesnt solve the problem 09:44 <+bridge> [ddnet] why not? 09:44 <+bridge> [ddnet] i would say it solves it just as well as your solution 09:44 <+bridge> [ddnet] did u read the codegolf challenge? 09:45 <+bridge> [ddnet] yea 09:45 <+bridge> [ddnet] oh fuck yea replace `i32` with `u32` 09:45 <+bridge> [ddnet] it needs to write them ordered btw 09:45 <+bridge> [ddnet] like 09:45 <+bridge> [ddnet] 1 09:45 <+bridge> [ddnet] 1 2 09:45 <+bridge> [ddnet] 1 2 3 09:45 <+bridge> [ddnet] not really 09:45 <+bridge> [ddnet] > Any way of writing such a sequence is acceptable. 09:45 <+bridge> [ddnet] Notice that we write all naturals from 1 to N for all N ∈ ℕ. 09:46 <+bridge> [ddnet] @Patiga it means the format 09:46 <+bridge> [ddnet] seems fine for me 09:46 <+bridge> [ddnet] https://codegolf.meta.stackexchange.com/questions/1061/loopholes-that-are-forbidden-by-default 09:46 <+bridge> [ddnet] the example output is just an example 09:46 <+bridge> [ddnet] all other answers arent random like this 09:46 <+bridge> [ddnet] its not ok imho 09:46 <+bridge> [ddnet] why not? 09:47 <+bridge> [ddnet] thats why the loophole rule exists 09:47 <+bridge> [ddnet] which loophole rule do you mean? can you link to a specific one? 09:51 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/783253987997515816/unknown.png 09:51 <+bridge> [ddnet] ok its valid 09:51 <+bridge> [ddnet] but i dont like it 09:51 <+bridge> [ddnet] :c 09:55 <+bridge> [ddnet] > Because of an outage during the day 1 puzzle unlock, day 1 is worth no points. 09:55 <+bridge> [ddnet] oh rip 10:00 <+bridge> [ddnet] ```rust 10:00 <+bridge> [ddnet] use std::fs; 10:00 <+bridge> [ddnet] 10:00 <+bridge> [ddnet] fn main() { 10:00 <+bridge> [ddnet] let r = fs::read_to_string("input").unwrap(); 10:00 <+bridge> [ddnet] let x: Vec = r.split("\n") 10:00 <+bridge> [ddnet] .filter(|x| !x.is_empty()) 10:00 <+bridge> [ddnet] .map(|x| x.parse::().unwrap()).collect(); 10:00 <+bridge> [ddnet] 10:00 <+bridge> [ddnet] for a in x.iter() { 10:00 <+bridge> [ddnet] for b in x.iter() { 10:00 <+bridge> [ddnet] for c in x.iter() { 10:00 <+bridge> [ddnet] if a + b + c == 2020 { 10:01 <+bridge> [ddnet] println!("{}", a * b * c); 10:01 <+bridge> [ddnet] return; 10:01 <+bridge> [ddnet] } 10:01 <+bridge> [ddnet] } 10:01 <+bridge> [ddnet] } 10:01 <+bridge> [ddnet] } 10:01 <+bridge> [ddnet] } 10:01 <+bridge> [ddnet] ``` 10:01 <+bridge> [ddnet] my rust solution 10:01 <+bridge> [ddnet] `x.iter()` → `&x` 10:02 <+bridge> [ddnet] ? 10:02 <+bridge> [ddnet] `for a in x.iter()` → `for a in &x` 10:02 <+bridge> [ddnet] it's the same and shorter 😉 10:02 <+bridge> [ddnet] ah ye 10:02 <+bridge> [ddnet] i was not golfin here xd 10:03 <+bridge> [ddnet] for some reason split returned some empty strings 10:03 <+bridge> [ddnet] probably after the last line? 10:03 <+bridge> [ddnet] i guess 10:03 <+bridge> [ddnet] because the last line is still terminated with a \n 10:03 <+bridge> [ddnet] is there some linux tool to view newlines ez 10:03 <+bridge> [ddnet] https://doc.rust-lang.org/std/primitive.str.html#method.lines 10:03 <+bridge> [ddnet] is what you want, probably 10:03 <+bridge> [ddnet] `hexdump -C` 10:04 <+bridge> [ddnet] oh it has a lines method 10:04 <+bridge> [ddnet] (and yes, on non-windows, there's a newline at end of file) 10:04 <+bridge> [ddnet] you need to do `.trim()` probably though 10:04 <+bridge> [ddnet] because `.parse::()` probably chokes on whitespace 10:05 <+bridge> [ddnet] lines() work without the filter 10:05 <+bridge> [ddnet] no need to trim 10:05 <+bridge> [ddnet] k 10:10 <+bridge> [ddnet] https://github.com/not-a-seagull/breadx 10:10 <+bridge> [ddnet] :poggers: 10:12 <+bridge> [ddnet] https://codegolf.stackexchange.com/a/215831/99502 10:51 <+bridge> [ddnet] wow rust seems to be actually kinda cool 11:13 <+bridge> [ddnet] If I have a vector `v` in rust I can create a slice with `&v[x..y]` but `v[x..y]` also works. Does the second one copy the vector? 11:16 <+bridge> [ddnet] ah maybe when `&v` is the same as `v.iter()` then `&v[x..y]` is the same as `v[x..y].iter()`? 11:16 <+bridge> [ddnet] no. v[x..y] cannot be used in a non-reference context 11:16 <+bridge> [ddnet] it's always the reference 11:17 <+bridge> [ddnet] `v[x..y].iter()` autoreferences `v[x..y]` to `&v[x..y]` 11:17 <+bridge> [ddnet] `.` can autoreference one level 11:18 <+bridge> [ddnet] okay, I guess i need to read further in "The Book" if I really want to understand what I'm doing 11:30 <+bridge> [ddnet] @timakro did u know rust auto derefs 11:31 <+bridge> [ddnet] nope 11:31 <+bridge> [ddnet] https://stackoverflow.com/questions/28519997/what-are-rusts-exact-auto-dereferencing-rules 11:32 <+bridge> [ddnet] well its on chapter 15.2 11:32 <+bridge> [ddnet] 12:10 <+bridge> [ddnet] I love this guy’s blog, always a great read 12:45 <+bridge> [ddnet] @Ryozuki Prolog is easier :D 12:45 <+bridge> [ddnet] ``` 12:45 <+bridge> [ddnet] solve(L, Z) :- member(W, L), member(X, L), member(Y, L), 2020 is W + X + Y, write(X), Z is W * X * Y. 12:45 <+bridge> [ddnet] ``` 12:46 <+bridge> [ddnet] @Ryozuki Prolog is easier :D 12:46 <+bridge> [ddnet] ``` 12:46 <+bridge> [ddnet] solve(L, Z) :- member(W, L), member(X, L), member(Y, L), 2020 is W + X + Y, Z is W * X * Y. 12:46 <+bridge> [ddnet] ``` 12:50 <+bridge> [ddnet] :justatest: 13:02 <+bridge> [ddnet] autoref and autoderef makes rust even harder to follow imo 13:04 <+bridge> [ddnet] Though I don't like anything auto so my opinion is not really relevant. Compilers and interpreters making assumptions is just a no no for me, unless I explicitly tell it to think for me. 13:07 <+bridge> [ddnet] make the holy learath2 bible on how a language should be made 13:07 <+bridge> [ddnet] :monkalaugh: 13:08 <+bridge> [ddnet] One day when I have more time I will make my own language that compiles down to LLVM IR 13:09 <+bridge> [ddnet] It would probably end up very similar to D or zig though 13:09 <+bridge> [ddnet] Honestly, zig I think has everything I would want in a language 13:14 <+bridge> [ddnet] I've been wondering why every modern language moved the return type to the end. I mean it's easier to parse but it's just so silly to read and introduces an extra token that conveys no extra information 13:14 <+bridge> [ddnet] everyone did their homework today 13:14 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/783305033810837534/unknown.png 13:14 <+bridge> [ddnet] :monkalaugh: 13:14 <+bridge> [ddnet] (usually more than 1 extra token, some languages even have a ->) 13:15 <+bridge> [ddnet] why is ryo only 4th 13:15 <+bridge> [ddnet] ranking is by alphabet 13:15 <+bridge> [ddnet] since day 1 doesnt give points 13:15 <+bridge> [ddnet] due to a outage 13:15 <+bridge> [ddnet] Not like any of us will get any points ever 13:15 <+bridge> [ddnet] ryo will 13:15 <+bridge> [ddnet] @Learath2 we will get points in the private leaderboard 13:16 <+bridge> [ddnet] ah, is it separate in private? 13:16 <+bridge> [ddnet] thats why private leaderboards exist 13:16 <+bridge> [ddnet] since not all ppl can wake up at utc-5 13:16 <+bridge> [ddnet] u can make friendly rules with ur friends i guess 13:16 <+bridge> [ddnet] I hate this kind of ranking. So america centric 13:16 <+bridge> [ddnet] ye 13:16 <+bridge> [ddnet] utc-5 is new york i think 13:17 <+bridge> [ddnet] I guess I also hate these kinds of challanges 13:17 <+bridge> [ddnet] I never enjoyed challanges that only reward obsessive people. Look at the people doing the google challanges on youtube 13:19 <+bridge> [ddnet] It's like the same 5 obsessive people every time hogging the top5 spots, golfing every day to know all the latest theoretical questions that have 0 applications in reality 13:20 <+bridge> [ddnet] They spot a word in the question, instead of solving the question they refer to a database of question types in their heads, and just pattern match the answer 13:22 <+bridge> [ddnet] xd 13:23 <+bridge> [ddnet] well i enjoy the challenge of the later days, which require you to think about how you want to approach the problem 13:23 <+bridge> [ddnet] the first few days are practically only about time 13:25 <+bridge> [ddnet] @Learath2 i think you forget that you can also enjoy those challenges without looking at the top ranks. enjoy the difficulty (that is introduced later, you can skip the first days if you like) and participate in the friendly competition of private leaderboards if you like that 13:27 <+bridge> [ddnet] The later problems are about time too. The reason these kinds of people are unbeatable is it's O(1) for them to come up with a solution no matter how complex the problem 13:28 <+bridge> [ddnet] @Patiga oh yeah, I enjoy friendly competition. I just think the global leaderboards are just not indicative of anything 13:28 <+bridge> [ddnet] They are probably reading books full of problems/solutions 24/7 13:28 <+bridge> [ddnet] well let em be 13:28 <+bridge> [ddnet] ah thats nice :) 13:29 <+bridge> [ddnet] @Learath2 ur more the type to know the ins and outs of compilers and stuff i guess 13:29 <+bridge> [ddnet] other ppl like to do challenges fast 13:29 <+bridge> [ddnet] @Ryozuki it's a pet peeve of mine, never liked these people in high school either. It just stuck with me, I guess the part I despise the most is their perceived success 13:29 <+bridge> [ddnet] A childish jealousy wired deep inside my brain 😄 13:29 <+bridge> [ddnet] :yoo: 16:15 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/783350660062117888/unknown.png 16:32 <+bridge> [ddnet] Is this some window joke im too gnu+linux to get? 16:33 <+bridge> [ddnet] it's a joke you're too windows to get 16:35 <+bridge> [ddnet] :monkalaugh: 16:43 <+bridge> [ddnet] looks i see them the skin (Redboppenomred) but they still see me default 17:02 <+bridge> [ddnet] what did u expect? that u hacked their game xD 17:07 <+bridge> [ddnet] @hussainx3 if u want others to see ur skin, u gotta add it to the database 17:20 <+bridge> [ddnet] does someone here know how to use config_store.exe? I was able to extract a map's server settings .cfg and edit it but putting the new edited config inside the map doesn't seem to work 17:21 <+bridge> [ddnet] I think you drag-and-drop the map onto the `config_store.exe` to store the config with the same name into the map 17:22 <+bridge> [ddnet] @Ravie ^ 17:24 <+bridge> [ddnet] that's what I'm doing but it simply doesn't change the server settings 17:24 <+bridge> [ddnet] I even tried replacing everything with just `sv_deepfly 0` to be sure I'm not seeing things wrong 17:26 <+bridge> [ddnet] could you open the executable from the console (cmd.exe) to see what it outputs? 17:26 <+bridge> [ddnet] you can drag-and-drop the executable and the map filename into the console, I think 17:26 <+bridge> [ddnet] to get the complete paths in there 17:27 <+bridge> [ddnet] and then drop the map file into it as well? 17:28 <+bridge> [ddnet] yes 17:28 <+bridge> [ddnet] with a space between the executable name and the map name 17:33 <+bridge> [ddnet] @heinrich5991 17:33 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/783370284337463306/Untitled.png 17:34 <+bridge> [ddnet] couldn't open … for writing 17:34 <+bridge> [ddnet] sounds like you lack permissions to write to that file(?) 17:38 <+bridge> [ddnet] I can try running as administrator and see if that works 17:39 <+bridge> [ddnet] do you run your client as administrator? 17:39 <+bridge> [ddnet] you could try it on another file, or right-click → properties it and check if there are weird permissions 17:41 <+bridge> [ddnet] it looks like I have all permissions on map files 17:41 <+bridge> [ddnet] hmmm 17:42 <+bridge> [ddnet] do you have python installed? 17:42 <+bridge> [ddnet] (to debug this issue furtherly) 17:43 <+bridge> [ddnet] don't think so, I'm not that advanced but I can try if you give me instructions 17:43 <+bridge> [ddnet] https://www.python.org/ 17:44 <+bridge> [ddnet] download python 3.9, install it 17:44 <+bridge> [ddnet] What will you even do with python? 17:44 <+bridge> [ddnet] check if I can open the file for writing 17:44 <+bridge> [ddnet] python is my syscall tool on windows&linux, because I know that it behaves kinda sane 17:44 <+bridge> [ddnet] while we're on the topic of server settings, a simple way to rearrange the commands like this would be very welcome 17:44 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/783373054365401168/unknown.png 17:44 <+bridge> [ddnet] and I can trigger syscalls and get errors 17:45 <+bridge> [ddnet] @Ravie if importing/exporting to text file worked, would it still be an issue? 17:46 <+bridge> [ddnet] well it's still rather clunky to deal with files like this and I don't think this would be too hard to implement 17:46 <+bridge> [ddnet] but for now this is the only way 17:46 <+bridge> [ddnet] well, I think the endgame is implementing a text editor there 17:46 <+bridge> [ddnet] and that sounds rather hard\ 17:46 <+bridge> [ddnet] A small text editor there would be so nice 17:47 <+bridge> [ddnet] I think it's alright to use with the new Mod feature, I'm just missing the ability to rearragne everything neatly once all the commands are done 17:47 <+bridge> [ddnet] and you pretty much can't just write them in order from the start, never know what you'll end up needing 17:48 <+bridge> [ddnet] Well arrows should be trivial to add 17:48 <+bridge> [ddnet] Maybe I'll do it quickly tonight 17:49 <+bridge> [ddnet] 👍 17:49 <+bridge> [ddnet] well then I guess we don't need to debug config_store.exe, since I'm the only one with this problem 17:49 <+bridge> [ddnet] 🤷‍♀️ 17:49 <+bridge> [ddnet] > do/did you run your client as administrator? @Ravie 17:50 <+bridge> [ddnet] @Patiga this is an external .exe that does something to map files 17:50 <+bridge> [ddnet] yeah, question still stands tho 17:50 <+bridge> [ddnet] well no I don't 17:50 <+bridge> [ddnet] Well if you opened it with an administrator client. It would save it with wrong ownership 17:51 <+bridge> [ddnet] Anyway, odd issue :P 17:51 <+bridge> [ddnet] got a better idea on how to debug if the file can be opened for writing on windows btw? 17:52 <+bridge> [ddnet] Have you checked that we close the file after reading it? 17:52 <+bridge> [ddnet] Windows can be picky with it's exclusive opens 17:53 <+bridge> [ddnet] yes, we do 17:53 <+bridge> [ddnet] at least if CDatafileReader::Close closes the file 18:59 <+bridge> [ddnet] @Learath2 18:59 <+bridge> [ddnet] ```c++ 18:59 <+bridge> [ddnet] void SomeFunc() 18:59 <+bridge> [ddnet] { 18:59 <+bridge> [ddnet] // throws an exception with all the error info if something goes wrong 18:59 <+bridge> [ddnet] DoSomething(); 18:59 <+bridge> [ddnet] } 18:59 <+bridge> [ddnet] 18:59 <+bridge> [ddnet] void SomeFunc() 18:59 <+bridge> [ddnet] { 18:59 <+bridge> [ddnet] if (!DoSomething()) 18:59 <+bridge> [ddnet] { 18:59 <+bridge> [ddnet] // obtain the error info and report it to the caller somehow 18:59 <+bridge> [ddnet] } 18:59 <+bridge> [ddnet] } 19:00 <+bridge> [ddnet] ``` 19:00 <+bridge> [ddnet] 19:00 <+bridge> [ddnet] properly used exceptions = clean code 19:00 <+bridge> [ddnet] or no? 19:06 <+bridge> [ddnet] Gist of it is in limited contexts I am sort of okay with the idea of exceptions. In those limited contexts `try{ SomeFunc(); } catch(whatever) { handleerror; }` is equivalent to `if(!SomeFunc()){ handleerror; }`. I think the second is cleaner, I also think limiting them to such contexts (e.g. exceptions that can only go back 1 stack frame) makes exceptions not the correct solution 19:07 <+bridge> [ddnet] Exceptions also lead to "unexpected" code-paths which is why I much prefer something like rusts result or optional/variant types. 19:08 <+bridge> [ddnet] A compiler checked error type that can't be implicitly propagated upwards is what I would investigate for a language I would make 19:10 <+bridge> [ddnet] you mean something like checked exceptions in Java? 19:11 <+bridge> [ddnet] that's not the right way probably, but i know what you mean 19:11 <+bridge> [ddnet] it's good idea 19:11 <+bridge> [ddnet] Well the same idea but not as overengineered a solution, leading to both prettier assembly and allowing actual error recovery 19:12 <+bridge> [ddnet] instead of putting try catches everywhere throwing upwards and pretending you do "error checking" well 19:12 <+bridge> [ddnet] sometimes it's really hard to do error handling with exceptions properly 😄 19:12 <+bridge> [ddnet] that's true 19:13 <+bridge> [ddnet] And a single code path, leading to easier analysis and reasoning about flow. If you want to see the mess exceptions make look no further than disassemblers trying to visualize the code flow 19:13 <+bridge> [ddnet] If a program given minutes can't reason about the flow, how am I supposed to 19:13 <+bridge> [ddnet] yes, exceptions in assembly looks terrible 19:14 <+bridge> [ddnet] all those "cold clone" functions etc 19:14 <+bridge> [ddnet] *look 19:14 <+bridge> [ddnet] returns are so nice, idk why people would want anything else 19:15 <+bridge> [ddnet] because it makes the code harder to understand 19:15 <+bridge> [ddnet] They are efficient, they are well defined ABI-wise, they look clean in code and in assembly, they give such clean entry and exit points 19:15 <+bridge> [ddnet] do you have a sane ABI for multiple return values? 19:16 <+bridge> [ddnet] also interesting: RVO is inhibited if you have a Result-like type 19:16 <+bridge> [ddnet] native support for exceptions fixes that issue, at least 19:16 <+bridge> [ddnet] I don't get how `try { whatever(); } catch (Exception e){ handle(); }` is easier to understand than `if(!whatever()) { handle(); }`. That's just silly 19:17 <+bridge> [ddnet] because there's usually much more ifs than try-catch blocks 19:17 <+bridge> [ddnet] @heinrich5991 there is a sane ABI for returning pointers, which is what you would return a pointer to your variant type return 19:18 <+bridge> [ddnet] huh no, that would imply an allocation 19:18 <+bridge> [ddnet] that is inefficient 19:18 <+bridge> [ddnet] Not any more space inefficient than an exception 19:19 <+bridge> [ddnet] A try block with multiple things that can fail is exactly what's wrong with exceptions 19:19 <+bridge> [ddnet] i think error handling code after each function call (the ifs) distracts when you read the code and trying to understand what it does 19:19 <+bridge> [ddnet] How are you supposed to handle something failing if you can't know what exactly failed? 19:19 <+bridge> [ddnet] more inefficient if you assume that the error case happens more rarely 19:20 <+bridge> [ddnet] Hm, I guess it does require more ABI support 19:23 <+bridge> [ddnet] some errors even cannot be handled properly without exceptions - Let's say you call some function from some library. That function opens some file internally and it fails. How do you report such error back to the calling code outside the library without exception? Note that you need the filename to create meaningful error message for the user. 19:23 <+bridge> [ddnet] you have an error type that contains the necessary information for the user to handle the error 19:24 <+bridge> [ddnet] ^ 19:24 <+bridge> [ddnet] throwing a FileNotFoundError from deep inside the library isn't actually that helpful for the user of the library 19:24 <+bridge> [ddnet] @heinrich5991 I have a feeling RVO is only viable because of the current "crash and burn" approach to programming. If you actually figure out a way to recover from the exception it would cost way more than the optimization gained you 19:24 <+bridge> [ddnet] Result best ☕ 19:25 <+bridge> [ddnet] but name of the file is highly helpful for the end user 19:25 <+bridge> [ddnet] @Comrade You return an error type. The internal function puts it in another error type and returns it to the user 19:25 <+bridge> [ddnet] `FileNotFoundError(/var/cache/firefox/acbjksbefjksf)` what do you do now? 19:25 <+bridge> [ddnet] you can always do one thing - google that 😄 19:26 <+bridge> [ddnet] fair 19:26 <+bridge> [ddnet] Anyway, error types don't prevent you from signaling that 19:26 <+bridge> [ddnet] I'll do you one even better. I think that FileNotFoundError should never make it to the user 19:26 <+bridge> [ddnet] but might still do, due to bugs 19:26 <+bridge> [ddnet] I think it should be `LibraryError(FileNotFound: /path/to/file)` 19:27 <+bridge> [ddnet] it's so annoying when app shows something like "File not found" and nothing else 😄 19:27 <+bridge> [ddnet] with exceptions you can easily avoid that 19:27 <+bridge> [ddnet] @heinrich5991 the compiler can check that none of these errors make it out of your library 19:28 <+bridge> [ddnet] ah yes 19:28 <+bridge> [ddnet] but they might still make it out as "AssertionError: FileNotFound" 19:28 <+bridge> [ddnet] or similar 19:28 <+bridge> [ddnet] "InvalidStateError: FileNotFound" 19:28 <+bridge> [ddnet] I don't mind them wrapping it again and returning it. As long as it's an explicit decision, that's fine 19:29 <+bridge> [ddnet] maybe logging is the best error handling 😄 19:30 <+bridge> [ddnet] Something like `if-error(e : fopen()) { return wrap_unrecoverable(e); }` 19:30 <+bridge> [ddnet] but you always have to return that thing even if no error occurs 19:30 <+bridge> [ddnet] that's bad 19:30 <+bridge> [ddnet] or if you don't mind annoying the new people I wouldn't even introduce a new token for that 19:31 <+bridge> [ddnet] @Comrade again, this is a hypothetical language, you don't "have to" do anything 19:31 <+bridge> [ddnet] The return type could be `Type?` e.g. where either an object of type 19:31 <+bridge> [ddnet] is returned, or an error 19:33 <+bridge> [ddnet] interesting ideas 19:33 <+bridge> [ddnet] @Comrade hav u used rust? 19:33 <+bridge> [ddnet] not yet 19:33 <+bridge> [ddnet] ! 19:34 <+bridge> [ddnet] One more example: `try { foo(); goo(); boo(); doo(); } catch (Exception e) { ... }` Each of foo, goo, boo and doo have recovery routines on error. You caught an e, how do you recover? 19:34 <+bridge> [ddnet] RAII 19:35 <+bridge> [ddnet] The usual answer is that you a) Introduce exception hierarchies b) Introduce new exceptions so you can handle different types c) You put each call in it's own try block 19:36 <+bridge> [ddnet] a and b give you a lot more exceptions you just might forget to handle at each point and for b in the case of checked exceptions huge throws clauses. c is no more intuitive or clear than the if block you didn't want in the first place 19:38 <+bridge> [ddnet] by the way, there's one more problem - throwing exception in destructor 😄 19:38 <+bridge> [ddnet] that never worked in C++ anyway 😛 19:38 <+bridge> [ddnet] In java i use catch(Exception e) and call it a day :BASED: 19:38 <+bridge> [ddnet] Yeah, that's how 99% of the Java developer population does "error handling" 19:38 <+bridge> [ddnet] it does - you can mark destructor with `noexcept(false)` 19:39 <+bridge> [ddnet] Have u seen how absurdly long function signatures get when adding exceptions 19:39 <+bridge> [ddnet] Java is a circus 19:39 <+bridge> [ddnet] yes 😄 19:39 <+bridge> [ddnet] but Kotlin doesn't look that bad 19:39 <+bridge> [ddnet] but they're part of the signature 19:40 <+bridge> [ddnet] (except the underlying JVM crap) 19:40 <+bridge> [ddnet] @Comrade marking a destructor noexcept(false) is how you break a lot of things 😄 19:40 <+bridge> [ddnet] yes, that's true 😄 19:40 <+bridge> [ddnet] it's evil 19:40 <+bridge> [ddnet] but it works 😄 19:40 <+bridge> [ddnet] @Ryozuki the exceptions people will tell you that you should be using exception hierarchies 4head 19:41 <+bridge> [ddnet] btw, annoy linux about their errors in `close(2)` 19:41 <+bridge> [ddnet] That controls the exponential growth of exception numbers somewhat 19:41 <+bridge> [ddnet] doesn't fit RAII 19:41 <+bridge> [ddnet] ```java 19:41 <+bridge> [ddnet] public static int primeraPuntuacio(String ruta) throws ParserConfigurationException, IOException, 19:41 <+bridge> [ddnet] SAXException, XPathExpressionException 19:41 <+bridge> [ddnet] ``` 19:41 <+bridge> [ddnet] 😆 19:41 <+bridge> [ddnet] @heinrich5991 logging that is the only sane thing you can do i think 19:41 <+bridge> [ddnet] xml in java :justatest: 19:42 <+bridge> [ddnet] @Comrade but it's an error, you have to handle it 19:42 <+bridge> [ddnet] and it doesn't document anything like "fdatasync before close makes sure that close doesn't error" 19:42 <+bridge> [ddnet] yes, but you can't really handle it 19:42 <+bridge> [ddnet] only kernel can 19:42 <+bridge> [ddnet] huh? why not? 19:42 <+bridge> [ddnet] you can handle any file writing error, why not that one? 19:42 <+bridge> [ddnet] What is the problem with close()? 19:42 <+bridge> [ddnet] it fails in certain cases 19:43 <+bridge> [ddnet] Ah, it shouldn't be allowed to, true 19:43 <+bridge> [ddnet] They should provide a RAII compliant version of it too. 19:43 <+bridge> [ddnet] or, linux should provide a function that does the error bookkeeping so you can call that before close(2) 19:43 <+bridge> [ddnet] if you want your write to succeed 19:43 <+bridge> [ddnet] it's like `malloc` returning `NULL` - what can you do with that? 😄 19:43 <+bridge> [ddnet] @Comrade handle it like any file writing error? disk might be full e.g. 19:44 <+bridge> [ddnet] you also have to handle that 19:44 <+bridge> [ddnet] @Comrade depends, many of my programs handle it quite gracefully 19:44 <+bridge> [ddnet] (the error in close(2)) 19:44 <+bridge> [ddnet] > In practice, close should never be retried on error, and the fd you passed to close is always invalid (closed) after close returns, regardless of whether an error occurred. In some cases, an error may indicate that data was lost (certain NFS setups) or unusual hardware conditions for devices (e.g. tape could not be rewound), so you may want to be cautious to avoid data loss, but you should never attempt to close the fd again. 19:44 <+bridge> [ddnet] https://stackoverflow.com/questions/33114152/what-to-do-if-a-posix-close-call-fails 19:44 <+bridge> [ddnet] :justatest: 19:44 <+bridge> [ddnet] Though the OOM Killer will more than likely nuke your program along with many others, and malloc will probably never even return null on most modern OSs 19:44 <+bridge> [ddnet] only linux I think 19:45 <+bridge> [ddnet] they prefer to hand out a pointer anyway and either swap things around to let you use that virtual memory or just nuke you on page fault if swap is not available 19:45 <+bridge> [ddnet] is swap needed in these modern times 19:45 <+bridge> [ddnet] @Ryozuki depends on your workload 19:45 <+bridge> [ddnet] is that so for windows and macos, @Learath2 ? 19:46 <+bridge> [ddnet] let's say `malloc` returns `NULL`, so you log that first, but then you realize that your logging function calls malloc internally - how to fix that? 😄 19:46 <+bridge> [ddnet] @heinrich5991 I think it's similar on macOS, not sure about windows at all 19:46 <+bridge> [ddnet] Android studio 19:46 <+bridge> [ddnet] Simple, my logging functions don't allocate memory 19:46 <+bridge> [ddnet] that's not so easy 19:46 <+bridge> [ddnet] It's quite simple. Step 1 is to not allocate memory 19:46 <+bridge> [ddnet] There is no step 2 even 😛 19:47 <+bridge> [ddnet] as far as i know even `printf` may call `malloc` in glibc implementation 19:47 <+bridge> [ddnet] That's printfs business how it handles malloc 19:47 <+bridge> [ddnet] so, your allocation fails, logging fails, what's next? 19:48 <+bridge> [ddnet] do you achieve this by going like `char[256] buffer` to avoid malloc? or what 19:48 <+bridge> [ddnet] roll back the current transaction, I guess? 19:48 <+bridge> [ddnet] return a 500 error as a http server? 19:49 <+bridge> [ddnet] well, if all the libs you use can do that without malloc, then yes 19:49 <+bridge> [ddnet] Depends, if there is a way to recover, then you recover, if not then you just die 19:49 <+bridge> [ddnet] @Learath2 hav u seen the try catch finally mess u get in java when opening and closing a streamwriter, u hav to add another try catch inside the finally block, luckily in new java versions u have a nicer try 19:49 <+bridge> [ddnet] it's not entirely clear that just because malloc failed for one allocation, it'll fail for all others, too, btw 19:50 <+bridge> [ddnet] ```java 19:50 <+bridge> [ddnet]