Look what I did during my Christmas-New Year vacation ➡️ https://github.com/TheBestTvarynka/grafily
(I am going to attach a screenshot of how it currently looks in the next post)

This is an Obsidian plugin for rendering family graph (family tree). The idea is to scan all pages in vault, extract metadata, build a relationships graph, and beautifully render it. Currently, it works well but has many limitations (see below).

It turned out to be a hard task. Rendering can be easily done using the reactflow library. But what about node positioning? There are plenty of libraries for positioning graph nodes, but all of them are bad for family trees. When we are talking about family trees, we have at least the following requirements:
🔴 All people from the same generation are in the same horizontal line.
🔴 Children are aligned below parents.
🔴 No nodes overlap and edges crossing.
🔴 Compact positioning.
🔴 ...and many other ones...

So, I decided to develop an algorithm for node positioning. I spent a few days on it without achieving any results. Then I googled for existing algorithms (yes, I should have done it before spending a few days on nothing).
I did not find the exact algorithm, but I found a similar (and much simpler) one: Reingold-Tilford Algorithm for rendering pretty trees. This algorithm is not suitable for me, because it is for tree node positioning. But all family relationships are a graph. This fact introduces a lot of complexity to the task, but I decided to implement the simpler version for a start and then improve it to meet my requirements.

So, I spent a few more days on its understanding, implementation, and debugging the Reingold-Tilford Algorithm. To make a tree, I cut out parallel family lines: siblings and their spouses for each person in each generation. I left only direct ancestors (parents, grandparents, great-grandparents, ...).

🔘 A Node-Positioning Algorithm for General Trees. September, 1989: https://www.cs.unc.edu/techreports/89-034.pdf. This paper is awful. It was useless for me. I understand a bit more than nothing from it. I mentioned it just for curiosity.
🔘 Reingold Tilford Algorithm Explained With Walkthrough: https://towardsdatascience.com/reingold-tilford-algorithm-explained-with-walkthrough-be5810e8ed93/. But this blog post is excellent 🤩. Good explanations, nice images, and illustrations. I recommend reading it.

During development, I noticed that my algorithmic skills level decreased a bit over the last years. I spent a lot of time thinking about how I should represent the tree in the code, how to implement a walk-through, how to calculate sub-trees shift, ... 😔

Btw, the current Reingold-Tilford positioning algorithm is implemented in one file: https://github.com/TheBestTvarynka/grafily/blob/b9d281d35fe5def9a3b0b260c82375d44f4755b1/src/layout.ts
Rate my TypeScript skills 😤😝 (haven't written so much TS in about 4 years)

When I was a child, I played many games on my Nokia C2-01. Last year, I tried to find the source code of those games, but unsuccessfully 😖.
It's sad. I really wanted (and still want) to read the source code of the wonderful moments of my childhood (my childhood wasn't sad, but it was still exciting to play on a 240x320 screen).

Totally agree that companies should open source their product if it has reached "end of support".

P.S. I know that they were written in Java, and I can easily decompile them, but it's not the same. I want the source code, not the decompiled code.

What have I done? 😭😭😭
Now I want my Nokia back. 🥺🥺🥺

Behold! The second version of the nodes positioning algorithm.

(actually, it is not a second version, but rather a complete rewrite with additional features)

Now it supports siblings and their spouses' positioning. Even more, it supports rendering sibling' spouse ancestors (parallel family tree).

Code ➡️ https://github.com/TheBestTvarynka/grafily/pull/1

Вчора я дочитав книгу КОЛОНІЯ Макса Кідрука. 9/10 👌💪
Рекомендую.

Це наукова фантастика (написана українцем) про майбутнє де люди уже колонізували Марс. В книзі описані екологічні, політичні, соціальні і ще куча різних проблем, які загрожують людстку вимиранню попри величезний технічний прогес.

Тут не має позитиву. Із кожним розділом усе стає тільки гірше. Дуже чекаю коли мені прийде друга частина (КОЛАПС) 😊.

Більше коротко сказати не можу. Я написав дуже великий лонгрід із моїми емоціями, враженнями, та думками: https://www.notion.so/thebesttvarynka/2e3c3e52142280439ca6f53cae424a0e (цей лінк дозволяє ставити коментарі, тому welcome 🙃).

What people won't do to avoid writing algorithms right away in Rust 🙄🙄🙄

Fast and Simple Horizontal Coordinate Assignment by Ulrik Brandes and Boris Kopf
https://scispace.com/pdf/fast-and-simple-horizontal-coordinate-assignment-2aawem94ts.pdf

A bit sad news from the Open-Source communities:
🔴 Tldraw pauses external contributions due to AI slop. Tldraw also has a blog post about it: Stay away from my trash!. (hn discussion)

For the good of the project, we're going to begin automatically closing pull requests from external contributors. We will of course continue to welcome issues, bug reports, and discussions.

🔴 cURL removes bug bounties. (hn discussions: [1], [2])

We will ban you and ridicule you in public if you waste our time on crap reports.

I am sure there are other.

I would say that this problem was very predictable. But closing PRs from non-maintainers IMO is a bit too radical.

I like how Ghostty does: https://github.com/ghostty-org/ghostty/issues/3558. They force you to create a discussion first:

Once a discussion reaches a point where a well-understood, actionable item is identified, it is moved to the issue tracker.

Of course, the solutions is not perfect. Maintainers still spend a lot of time into answering these discussions. But:
🟡 No AI-slop PRs. People, who submit AI-slop PRs, most likely will not want to discuss their contribution and prove that their "wOrK" is useful.
🟡 Discussing the idea/bug on early stages is easier than reviewing PR. Sometimes people tend to defend their code rather then prove that it is useful.
UPD: Hyprland adopted this approach too: https://github.com/hyprwm/Hyprland/issues/9854

Another idea that came to my mind is to charge people for PR 😁 Let's say, $5 per PR. Optionally, make it free if person has delivered more than 10 meaningful PRs.
🔘 Most of trash PRs will disappear. No one will want to pay their own money for something they do not care.
🔘 Even if someone will still submit a trash PR, maintainers would have 5 bucks 😎
Cons? 😌

Rust 1.93.0 ➡️ https://blog.rust-lang.org/2026/01/22/Rust-1.93.0/

rustup update stable

For me, there is nothing interesting except these stabilized APIs:
🔘 <[T]>::as_array / <[T]>::as_array_mut
🔘 <*const [T]>::as_array
🔘 <*mut [T]>::as_mut_array
🔘 char::MAX_LEN_UTF8
🔘 char::MAX_LEN_UTF16
🔘 std::fmt::from_fn

https://youtu.be/fCgNa-WwymY (8mins long)

tl;dw:
> fasterthanlime needed zed config file auto-completion
> fasterthanlime promised to send cookies to anyone who implements it (bsky post link and github issue comment)
> @Nereuxofficial has implemented it in https://github.com/zed-industries/zed/pull/43104
> @Nereuxofficial got his cookies
> Everyone is happy

Happy end ☺️