I Coded in a Random Programming Language Everyday, For Science
Вставка
- Опубліковано 30 лис 2023
- Advent of Code 2023 is UPON US! What better way to spend the holiday season, learn to program, and test your skills against your friends. This year, I've decided to try something kind of insane. I'll be choosing a language off the wheel every day.... let's see what happens.
🏫 COURSES 🏫 Learn to code in C at lowlevel.academy
📰 NEWSLETTER 📰 Sign up for our newsletter at mailchi.mp/lowlevel/the-low-down
🙌 SUPPORT THE CHANNEL 🙌 Become a Low Level Associate and support the channel at / lowlevellearning
🔥🔥🔥 SOCIALS 🔥🔥🔥
Low Level Merch!: lowlevel.store/
Follow me on Twitter: / lowleveltweets
Follow me on Twitch: / lowlevellearning
Join me on Discord!: / discord - Наука та технологія
A normal youtuber uses an online wheel of fortune, an embedded programmer builds on themself
No array languages on the wheel. Hoping for some APL or BQN
“Formally verified”, they said
Statement 1: "A normal youtuber uses an online wheel of fortune"
Statement 2: "An embedded programmer builds on themself"
These statements don't directly connect and lack context. Here's a breakdown of possibilities for each statement:
Statement 1:
Online wheel of fortune: Many UA-camrs use online tools or websites that offer virtual "wheel of fortune" games. These tools allow them to select random options, challenges, or topics to incorporate into their videos, adding an element of chance and audience engagement.
Normal UA-camr: This is a subjective term. Defining what constitutes a "normal" UA-camr is difficult as the platform hosts a vast variety of content creators with diverse styles and approaches.
Statement 2:
Embedded programmer: This could refer to a programmer who focuses on developing software components designed to be integrated into other applications or systems.
Builds on themself: This phrase is unclear and grammatically incorrect. It's likely meant to convey an idea of personal growth, self-improvement, or continuously building upon one's skills and knowledge.
"C++ is trying to be a Rust clone" is probably one of the most scathing remarks on the language.
i meeeeeeeeeeeean
Not to mention wrong.
@@LowLevelLearning It's a circular dependancy.
Fitting. C++ is development tetanus.
Idk to me it seems like rust is trying to be the c++ clone with where it's going at this rate. They are both taking inspiration from each other though.
I think the "two-pass" solution is actually more efficient, since you can break as soon as you hit the first and last numbers. Whereas the other solution has to go through the entire string every time.
I thought the same thing. And considering that there is at least one number on each string, the two pass solution would in the worst case scenario read 1 more character then the one-pass solution, so it would be O(n+1) in the worst case scenario, not O(2n). And as you said, it would most times be more efficient than the single pass solution if there are two digits or more
It would be more efficient if you knew the size of each string beforehand. If you were to implement it in C, you would have to first calculate the size of the string (that's one pass) so you don't access invalid memory, and then do the "two-pass" solution. Doing the "one-pass" solution, you wouldn't need to know the size of the string.
Definitely. I originally solved it with practically the exact same 1 pass algorithm. Then I made a 2 pass and got about 3x performance boost.
It's actually not ( i think ), and the reason is that the 2-pass solution actually has 3 passes, of which one is done by std::getline().
If you use getchar instead you can do (the first part) in that same pass you find the newline in, and without allocating any memory too.
The second part you can also do in that same pass, but i'm not sure of the effect that has on performance:
( this solution is untested. It merges strcmp's internal loop into the getchar loop, eliminating a lot of duplicate comparisons, but as a result it will fail if a number string overlaps with itself. fortunately that can't happen for 0-9 in English )
uint indices[sizeof numbers] = {0}
for c in input:
(EOF/null/newline check)
(check against literal number)
for i in 0 to sizeof(numbers)-1
if !numbers[i][indices[i]]
handle match
if numbers[i][indices[i]] == c
indices[i]++
continue
indices[i] = 0
The one pass is always O(n) with n being the length of the string. While the 2 sided is at MOST O(n), but likely less as you can early stop once you encountered a digit.
My dude says that c++ was intended to be c with classes, but casually proceeds to use typedef struct😂😂
he doesn't know C++. Let him stay ignorant
@@1337Kupo Doesn't know it for a reason lol. Not worth knowing (unless absolutely required)
@@VivekYadav-ds8oz It is worth knowing if you wanna make statements about it
Braindead take@@VivekYadav-ds8oz
You don't need to close the file because the destructor closes it.
Some programming languages that I would love to see on the wheel of (mis)fortune:
- Java: because it's very popular, i was surprised it wasn't on the wheel at the start
- Lua: because it's used for multiple video games
- Smalltalk: this was the language they used at my university when we started learnong OOP, sounds fun for a challenge
- Pascal: first language i saw at university, probably torture at this point
- Scratch: it has colours.
I thing some lisp would be torture part 2
How about Brainfuck?
T-SQL / PL/SQL / PL/pgSQL
C#, specifically using COSMOSOS to make an OS, or maybe some esolangs.
As a java developer I can say that I was quite disappointed to not see it on the wheel...
using C++ but writing it like its C 😂
This is such a cool series idea! Keep up the great content!
Thanks so much!
Let's get some Fortran in here!
Seeing you, a C dev, throw around -1 in uint32s caused irreparable emotional damage😢
Since halloween is already over probably better to compile with -Wall
"-1 in uint32s" Yeah that hurt me.
@@bb-sky yes it does exactly that, the compiler allows it because you need to explicitly enable warnings during compilation ( -Wall, -Wextra)
Nitty picky detail: Should O(n) and O(2n) not be the same? Afaik there is a scalar in the definition of Landau big-O that captures any positive scalar, as the only requirement is that it is real and positive. I think the correct terminology would be scales like n, scales like 2n.
"good programmers close your files". Good programming languages, like C++, close your files for you so you don't have to remember to do it manually. Also good programming languages have standard libraries that find the first number and last number in a string like C++'s std::find_if. std::atoi does exist in C++ in the header, so you could have used it since you were basically writing C with fstreams anyway. This was fun to watch, I'm looking forward to Zig.
Fantastic series idea! I had never heard of AOC before. We are doing it at work as a contest and I'm hooked!
I did earlier aoc today (day 1) in c and I came up with a solution that was just a translation of what you did (at least I think so). It's always fascinating how on these easier problems a lot of solution end up looking and working similarly !
I’m quite curious about your thoughts on Swift, that’s why I was a little bit saddened by not seeing Swift on the wheel. 😅
People tend to overlook Swift because of misconceptions like it’s only Apple platforms etc etc. Basically if you like the philosophy of Rust, Swift maybe for you. Swift was created by Chris Lattner and he along with few other brilliant minds laid pretty impressive foundations for the language, compiler and philosophy
🙏🏻 CHRISTMAS IS ABOUT GOD 🙏🏻
🙏🏻 VOTE FOR HOLY C 🙏🏻
literally never gonna happen 🤣
Let’s do COBOL :)
I tried cobol once, I think I'm still recovering from the experience lol
some fortran would be fun too
@@woofcaptain8212 try handling abends everyone will become sober
More like COOLbol, right guys? 😂
"C++ is trying to be a rust clone" lmao
This is like saying a modern young celebrity popularized a fashion style that's been around, and in and out of style, since the 80's haha
Please finish this series. Don’t stop half way through
Bro literally said C++ is trying to be a rust clone, As the proposal of type traits in C++ was made in 2004 and the concept of owner ship(smart pointers) was made in 1990s, when rust is not even a thing💀
Chat is definitely gonna force you to use eso langs like IuseArchbtw or Brainfuck
this is what im afraid of
@@LowLevelLearning someone is going to make you learn APL syntax
Loving this! Can't wait to see more.
It is funny, that you use an uint32_t to store -1 aka 0xFFFFFFFF since it unsigned, than you compare it to an int literal of -1, you are really lucky, that the underlying bytes are exactly the same, since no real conversion happens, but the result just happens to be correct xD
Agree about cout. Its a torture to use it if you want pretty formatted output. In my personal programs switched to libfmt, or just using C prints.
FYI, in C++20 introduced std::format which is inspired by libfmt. It formats the string the same way as libfmt does and then you use cout/printf to output it.
EDIT: And in C++23 we get std::print which equivalent to libfmt's print!!! Hooray.
@@sledgex9 Ik, However, if u need to compile it for old platform (debian buster as example), you better go with additional 3rd party dependency, that trying to use c++20.
Btw, it is not inspired, but it is actually proposed by the author of libfmt to add it to the standard library.
Love this! Great Series
Love your part 2 solution!
Entered to see what language you got, left to do those exercises myself, and came back to finish the video.
This mans out here setting a uint32_t = -1 smh
😏
C# is actually a really nice language
Not nearly as nice as F# though.
@@Akronymus_ Both are good
I watched a video yesterday that had me deciding to do AoC this year, and then as I'm catching up on my subs feed I see you doing it too. You make me feel so much better about the off by one error I had that stumped me for half an hour as I did part two. I ended up going with C because it's my favorite language, but I used a trie for the second part with the reversed strings of the numbers to do the last numbers backwards. Your attempts at doing C++ are funny, but I hope that wasn't a genuine struggle. Also, K&R bracing is disgusting, consider Allman next time.
Where's Haskell on that wheel? Erase PHP and write in Haskell!
Great video and a great idea!
In case you feel a bit adventurous, maybe Forth is something you want to try if you haven't done it already ?
There are dozens of us!
i was doing this with the exact same approach as you but in c, but was getting an answer too big and I could not figure it out. I finally caved and watch this video and then at 10:23 when you mentioned atoi i realised i had been adding the ascii values and not the integer values.
oh hes definitely writing in dreambird this month
I did this last year, it was a lot of fun!
Might as well do a pass from either end. Still n and not 2n since in the worst case the number is in the middle and has to approach it from both ends (assuming you break once you find it). It will actually be faster since if they are different numbers you didn't need to iterate over the whole line.
This was fun to watch!
Some folks have mentioned grabbing the last digit by iterating backwards, but I don't think that would work if you had a line ending in, say, "oneight." Single pass is just easier and more reliable.
I can tell you're not the most comfortable writing C++, but I appreciate that you stuck it out and made it work. Hopefully you enjoy the next language you land on! 😂
As someone who implemented a backwards lookup, it does work. My testing function for both forward & backward took a char*. If it was a digit, it returned the int value of that digit and I stopped iterating. If it wasn't a digit, I compared the substring starting at that char*, returning an integer based on which string it matched if any. I stopped iterating as soon as I found an answer, so as soon as the char* walked backwards to the `e` in eight, I stopped looking backwards, and the fact that the `one` was also part of it didn't matter. Same with forward iterator: as soon as it saw `one`, it stopped looking
Hah, saying c++ gone a "little overboard" is probably the understatement of the year :p
I was thinking that going just one "pass" from left to right would _always_ be n operations, while going from left to right to find first number, then right to left to find last number would be n _or less,_ since if the numbers are at the first and last index you avoid dealing with the rest of the input. So the average case for "two passes" would be n/2 instead of the average of one pass that is n.
6:37 Yes, iostreams aren't nice to use (or rather the opposite).
But at least use _std::format_ for formatting (C++20 needed).
If compilers would have implemented it already, you could do _std::print("x = {}
", x);_ instead of _std::cout
Why not std::print on 23?
@@heavymetalmixer91 I wrote that, but some compilers didn't implement it yet.
no way they got the rust println, that's awesome
@@VivekYadav-ds8oz technically it's from Python
You could have had gTable be a string array, and returned i + 1 in the match function. The index of the string in the array corresponds to the contents of the string! :)
Very cool! I hope you'll be able to do all days!
Wow. I remember doing the 2022 puzzles, haven't completed them yet, but now the 2023 ones are here...
I always love dreaming of the python solution that takes 3 lines and then coming back to reality where im writing a 100 line program in c++ to solve the same problem
I love this idea
Wow, you went so much complicated than I using the same language. I just abused the C++23 ranges, most of my code is oneliner predicates and pipes
I suggest you add TI-basic to your list, could be fun
And intercal
And Unlambda
Nicely done. I liked your solution for the two star. I don’t dislike C++ but I’m with you. C is the GOAT. I hope you do a challenge in ZIG. I’m curious about it but have been too lazy to try it yet.
Also if you fail on a challenge, I’ll vote for you to do your solution in Java 😈😈😈
Brainfuck, white space or chicken lang would be fun as well
Zig is cool, you can embed the input file at compile time so a lot of the input processing is done before you even run. Not sure how flexible the standard library is though.
this is the most C looking C++ I've seen
he actually meowed for us
LOLCODE sounds like a good one to add to the wheel. 😜
assembly might be fun
what desktop environment is it? I love it ❤
I hope one of these days he lands on Odin. I love it. It would be cool to see what he'd think of it
Wow a different language every day, was just the idea this game me lol.
I teach c++ and completely agree. cout/cin drives me nuts. There are just lots of weird and unintuitive things that obfuscate what the code is really doing, and the lack of safety always makes the pointer section fun to teach. I hope we move to rust or golang soon.
Anyway, good work! Excited for this series. I'm learning game development so I'm doing this year's AOC in Godot.
I'm already having regrets 😂
If you have access to C++20/C++23 take a look at std::format and std::print. If not, then take a look at "fmtlib".
C++23 is bringing std::print and std::println which I'm going to use in a heartbeat over std::cout
The lack of safety? Isn't it safer than printf, though?!
@@assyyn4139 Yup cout is safer than printf. I have no idea what is obfuscated by cout that isn't obfuscated by printf, other than the fact that you don't dangerously use the wrong type specifier anymore. However I think the C++ teacher above is refering to pointer safety. Perhaps they are not teaching their students to use smart pointers over raw owning pointers.
iostream is a pain to use (cin/cout), but it does one C++ thing - it is generic. You can change the type and it will work, there is no need to change the format string.
Ooh boy, finally! The above zero chance of seeing Nim code on this channel.
should include prolog on that list.
I can't wait for the fortran one
can't wait for you to solve a puzzle in malbolge
I like to think of this as "25 days of random programming language crash course"
Please consider explaining the logic behind your code. I saw your video after finishing the puzzle myself and couldnt for the life of me figure out what you did
How would you handle a scenario like "oneight"? Are all valid subsets supposed to be numbers, or are you supposed to interpret that as one number and some nonsense letters?
This is supposed to be interpreted as one and eight. I solved the problem using a Trie and the edge case was “seight”, “ninine” And things like that. I solved it by adding all those corner cases to the trie.
I think you will be of good if chat decides just sometime like VB and such. It’s the internet, you will end up with more and more esoteric programming language.
They'd straightup force him to do it in Malbolge
my first thought was Brainf*ck :D
"I got the first part done on C++ guys, I don't feel bad doing it this way" Last famous words before doing it in C-style way
For part 1, wouldn't the 2 pass be slightly more efficient in some cases..?
if you consider the line 1aaaaa:
you'd get to 1 in 1 unit of time
when you work backwards you'd get to 1 in 5 units of time.
the total time in O(n) = 6 (worst case)
best case is e.g. 1aaaaa2
you'd get to 1 from the front in 1 unit and 2 from the back in 1 unit so it's much quicker.
with 1 pass you'd always have to check all n elements, so wouldn't it be slower..?
thats actually fivehead.
I thought the same thing. My idea was to do it in one pass at first, but then when you mentioned walking forward and back I realized it would be faster.
In 2 passes, the worst case would be only one digit in the string -> walking from the front takes you d characters, walking from the right takes you N-d + 1 characters -> worst case, you walk the whole line only once in total. I don't think there were lines without digits (those would be 2N, but who cares anyways). In the end, who cares, because it's still linear, as is the reading of the input.
personally what i did (in Go) is just made a slice of all the digits in the line and then simply `sum := digits[0]*10 + digits[len(digits)-1]`
I definitely want to see you write a solution in Brainfuck.
10:33 the redundant else killed me
lol this is awesome!
I really hope someone puts ATS and / or lean on the wheel :D
I have a doubt, isn't the complexity O(n) even if you do the iterate from the left and then iterate from the right method, because, think about it, the longer it takes for the left iteration the lesser it will take for right iteration.
The one pass approach smells like premature optimization and putting too much weight on O-notation. Searching from the left AND right will both abort when they find a digit, so they are sure to never look at more than N elements (in case there is only one digit). Otherwise depending on how close to the edge the digits are, they might both abort much sooner.
Of course we could then start worrying about how bad a reverse search is for the cache and other low level considerations. But purely on an "algorithmic" level, the one pass approach is always worse and equal at best.
we want more the dog
C++ finally got its own `std::print()` and `std::println()` functions in C++23 and are based on {fmt}. Also, you not using the standard algorithms hurts me as a C++ dev. In part 1 you could've use `std::accumulate()` and put your (inner for) loop logic in a lambda (probably a named one to make it easier to read) saving values in a `std::pair`. This would return the pair at the end which can be destructured to more easily do the sum calculation.
1:49 Nahhhhhh no way he did statans favorite language just like that 💯💯🗣🗣
Imagine the chat selecting malbolge or brainf*ck as programming language.
Regex replace to replace the set of strings with digits, then run your other algorithm as-is?
The problem with that approach was the string 'eightwo' in the test data. Replacing the 'eight' with '8' will remove the 'two' and yield an incorrect solution. It's tricky to use a find/replace algorithm because it has to be done separately over two copies of the string.
@@tyoda A good workaround! You can then use the code from part one to find the solution. I found the problem to be much trickier then past first nights and I was glad to finish as well, and I went to bed a little late that night.
amazing
The second solution truly was a C++ is C with classes solution.
P2 was brutal for this one
One really weird language I would like to see, just because: Prolog
By defenition O(2n) = O(n)
We need Assembler on the wheel
Hope chat votes for Haskell
You know all the languajes of the roullete? That's amazing :0
lol i dont thats the problem
@@LowLevelLearningOhhhh, that makes sense LMAO
woah you really don't know much of C++! (I am writing this in a friendly manner). Typing the struct as a typedef made me chuckle.
why did you not do index + 1 as the word value instead of using a struct and specifying it manually?
im triying to make the second challnege in c# and im getting so confused with how to choose which one is the rightmost and leftmost XDDD i think im gonna do it tomorrow cause its one am
btw you can also write value = c-'0'. No need to know ASCII.
The "sixteen" is so troll 😂
Love the idea - put Perl on there. You won’t. 👀💯💯
9:00 O(n) and O(2n) are the same complexity
Good Luck
What do you think about using regex in such a scenario?
I had a different answer but still got the correct answer. Is that typical?
shouldn't that be
Did I hear functional programming on day 24? Jelly, bqn, apl, ... have fun :)
Are the inputs different for everyone? I got a different answer and passed the test.
yup!
Xcode X-STEP MACHINA LANG U A G E
Do Smalltalk next (the mother of all that is object-oriented).
Simula?
@@mihaelasimerea9700 That as well.