This Data Structure could be used for Autocomplete

[@lame_lexem]

mom can we have emacs no we have emacs at home emacs at home: 42:05

[@kernelk1931]

looks like you've uploaded a brand new quality content, tsoding!!

[@larryd9577]

Instead of maintaining a bool in every node: A memory-efficient solution (for reasonably large trees) for marking the end would be to use a guard-noded (one should be enough) which is inserted as a leaf of every end of a word.

[@dingalong14]

Haven't watched in a couple of weeks, glad to see you've switched to the *objectively superior* text editor. Jokes aside, nice stream. I always pick up at least one or two little things I didn't know about previously.

[@Nick-lx4fo]

Gedit, you mean, is the objectively superior text editor

[@experiment0003]

So, this guy could work for any company he wants (Google, Tesla, Amazon, etc) and be a senior engineer at the least! instead, he chooses to make his living via streaming.

[@Chaosman88]

If I could just code 4 times slower than him, I would be a respected member of my team

[@venkateshgunda25]

It could be because at a large company, coding only comprises about 40 to 50 per cent of the time. Rest is projects, meetings, presentations and other corporate BS that eats up valuable time.

[@richienguyen2703]

not every software engineer wants to work for big tech companies. They're all big so their development process is more bureaucratic, and some people just purely love coding.

[@AndrewTSq]

agile ruins the fun of coding.

[@BboyKeny]

@@dill__pickles Well you are a God on earth. So all other people, aside from you, fit your description.

[@bart2019]

2k per node is not a lot for modern computers, but at 1024 nodes per tree that is 2MB, and that is a lot for modern CPU caches. On its own you won't notice it, but on a heavily loaded computer you might notice it because 2MB is easily a quarter of the entire CPU cache. In this case, a theoretically much slower approach but with a much much compact storage, for example with a simple comparison loop in a linked list, would put much less strain on the cache.

[@arroba3218]

THIS is what programming looks like. I'm into web development, I don't feel a programmer at all

[@JustSomeAussie1]

Because you're not :)

[@lonelyfloat2582]

Just do what you enjoy doing, doing anything else (unless it's for a job or something similar obviously) is stupid IMO

[@Leonhart_93]

Not sure what people work on in web dev to say that they don't do stuff like this. Writing elegant programs that do a specific task efficiently should be the target in anything you use. If I write a 1000 line JS program for a specific page to have some arcane functionality, then it's clear that I need to apply a lot of programming paradigms to do it well.

[@Rohan-bg8ci]

"That is much productive than Java enterprise edition developer" -chad Tsoding

[@N00byEdge]

Another way to indicate the end of a word is to set index 0 of the children array to something, just like you would have a null terminator in a c-style string

[@geovanniportilla7159]

Pure data structure, algorithm, and design. This is pure gold. I follow your content and here is an idea: maybe, would be nice to have a project that created a programming language for database management. I don't know something basic: relational algebra, join, or some k-tree search algorithms. May be is something you are not working in the past and it's challenging for you. Peace!!

[@lieryan9137]

For the use case of autocomplete, especially one that needs to work on Unicode, I think a regular binary search tree would work better than a trie. For finding candidate completions, you'll want a binary tree search implementation that returns all nodes in a range. For example, if you want to find all words that has a prefix of "sal", you'll want a search function that returns all nodes whose values matches `"sal"

[@edusandalo]

How would this approach work with Unicode? Would the search be `"...A"

[@keldwikchaldain9545]

To compress memory the primary way would be to use an array mapped trie, which trades memory footprint for increased memory allocations, but with pooling that can be reduced as well.

[@Teacher_Ali]

One of the coolest channels on internet world.

[@geovanniportilla7159]

This dude is just insane!!! Pure data estructure, algorithm and design. This is pure gold.

[@DJohn001]

Hello, what's the reason that you're not binding your local branch with your remote one? for example: `git push -u origin ` then you did have to type your branch each time you commit. Instead you can just use `git push`

[@froggoboom]

the man types fast

[@bigbluebiosphere]

commit-ee commit-ee pushy-ee pushy-ee made me LMAO. But nowhere near as much as cucumbersome did.

[@bonchonmaru]

1:47:12 is it possible to prevent stackoverflow in `print_autocomplete` for deep trees?

[@TsodingDaily]

You probably can by using a bigger stack. Which might mean increasing the size of the callstack itself or implementing a non-recursive solution that maintains its own stack. In the later case you can have whatever size of the stack you want.

[@lishugupta]

U earned a sub just by the title of video🔥🔥

[@petsesto]

34:17 children[69] ... Tsodin nooooo🤣

[@miniflint2423]

How could I achieve something like this guy do. He’s so good

[@archavez100]

How did you zoom and add that spotlight to your cursor? At 11:08

[@iuc7254]

He wrote an application in nim to help him do that.... you can find it on one of his github repos under the name of Boomer

[@MaxHeroGamer]

The go thing made me thing why Erlang is so brilliant.

[@arkamel5987]

you can mark the end of the word not with bool flag, but instead put empty node at \0 in children

[@derekburlingame213]

That doesn't work for cases where words like "dog" and "dogged" are both in the structure, since the first 'g' would have children, but also be the end of a word.

[@DylanNelsonSA]

@@derekburlingame213 They meant adding a node at index 0, not having 0 children. So it would be like treating the word as a c-style string with a null terminator.

[@coffee-is-power]

you dont need to create the project, when you push to an repository that doesnt exist and you have permission to create repositories, it creates automatically

[@baka_geddy]

How so? Can you share it?

[@lolcat69]

It would be interesting to use this data structure for AI, imagine like passing binary files to this and passing ut to an AI, it could figure out a file is X type based on the signature or smth and ut would learn to clasify files or smth like that

[@ItzAnameOk]

Amazing

[@tobeypeters]

Did you say, There’s no Switch...case in c?, around 1:18:00 ? Idk ... Anyhow would love to see how to do this in python.

[@TsodingDaily]

No, I didn't say that there is no switch-case in c. Please listen to what I've said again. You missed two important words.

[@ricoh51]

No switch for string 😉

[@anon_y_mousse]

@@TsodingDaily Fun fact, AMD's optimization manual had an example of how to simulate string switching by hashing constants ahead of time. With C++'s constexpr you could have base that allowed adding words with minimal effort. (Hopefully you'll be back and posting in the near future.)

[@ratherbyexploring4898]

Missed opportunity to call this repo "Tsrie"

[@tentative_flora2690]

I hadn't thought before about someone using recursion in an intuitive solution before. This was really cool.

[@yramagicman675]

Once you start working with recursion it becomes more and more intuitive. There are a number of problems that are actually easier to solve with recursion than they are with iteration. Zipping 2 arrays is one example of this that I used recently. I implemented the recursive algorithm for it in less time than it would have taken to figure out how to do it iteratively.

[@tentative_flora2690]

@@yramagicman675 I guess I am so used to trying to stuff things into while loops I miss out on a lot of good use cases.

[@yramagicman675]

@@tentative_flora2690 get out of your comfort zone and try a language that forces you to use recursion. Lisps are an easy way to do that, and the easiest one of those to get started with is probably Racket. I really enjoy Racket. It has taken Pythons place as my go to language for basic scripting.

[@yukwa8672]

🔥🔥

[@hrisheekeshdandekar5126]

What is the theme this guy is using???

[@L1Q]

29:03 cope

[@hindenburg3038]

Where I can get the link to his discord server?

[@csbnikhil]

Twitch About Page

[@trungthanhbp]

this guy is funny, haha

[@s000x]

Ur emacs config got me to patch my neovim so current line number is aligned with relnum lol

[@alexzander__6334]

you just moved to vim from emacs?

[@zahir.parker]

Which font is this ?

[@trejohnson7677]

We have the same brain but not the same muscle memory or gut.

[@rogo7330]

Not watched the whole video yet, but what i think is storing pointers to every character is kinda pointless. What we are really care about is branches. That saying we really wanna store pointers to each individual string (null-terminated, btw, that way we not gonna waste 2kibi on each character lol) and then just cat them together. Btw, we can do the same trick with null termination with our pointers! LMAO А потом всё превратится в ад из незнания сколько на самом деле памяти занимает наша программа, кек

[@anon_y_mousse]

Exactly, and then the index into the string and a length. That said, I really wish people would stop using those ldlotic prefixes. Other than storage device manufacturers who rip people off, no one ever meant kilobyte as 1000 bytes. I will never use those prefixes and just keep saying kilobyte and meaning 1024.

[@miezekatze3536]

Where is Emacs?

[@tokiomutex4148]

error: Where: command not found

[@sahilbaori9052]

@@tokiomutex4148 Lame.

[@tokiomutex4148]

@@sahilbaori9052 error: Lame.: command not found.

[@هلالمجيد-ظ3ك]

What kind of linux does he use?

[@narigoncs]

Debian

[@johndowson1852]

1:08 CAN YOUR GITHUB DO THIS?

[@kuwapt951]

Hello hello

[@dream11tatyabichoo92]

how to donate from India

[@MoeQ_]

discord link please ?

[@Q_20]

proud to be a reckless java code duplicator

[@tokiomutex4148]

So you're using Vim now

[@dibyojyotibhattacherjee4279]

Why no cpp?

[@latingm]

why cpp

[@dibyojyotibhattacherjee4279]

@@latingm cuz, it's an awesome language.

[@avananana]

@@dibyojyotibhattacherjee4279 brainfuck is an awesome language as well

[@dibyojyotibhattacherjee4279]

@@avananana Haven't used it before, gonna check it today, thanks for the suggestion 🙂

[@dibyojyotibhattacherjee4279]

@@SimGunther ohh, yea, Tsoding's subscribers are really nice people.

[@qerdcv]

Второй