Show more

BTreeMap vs. BtreeSet with custom Ord vs. intrusive-collections's RBTree.

Analysis paralysis is kicking in again. :blobblush:

To my past self's defense, I was fully aware back then that the code was a horrible mess, I simply hadn't a clue how to do it well without using bloated frameworks that would have long been out of fashion by now. But nowadays we have much better options, such as Elm.

Just spent an hour fixing a tiny bug in some Javascript code I wrote 8 years ago. Back then I still thought it was a good idea to write complex UIs with plain DOM manipulation and a tangled mess of global variables.

Even my C code from that time is infinitely more maintainable.

Yet another very interesting discussion on how the stable Linux kernels are managed.
lwn.net/ml/linux-kernel/201812

I love those threads, they show how incredibly hard it is to provide reliable and stable software when everything keeps changing all the time.
(The fix, of course, is to stop changing things all the time, but that's unlikely to happen anytime soon 🙄)

Turns out I was (probably) wrong about the memory allocation part. Both the time and humantime crates have types that implement the Display trait, these can be used in a non-allocating manner and are (presumably) more efficient than allocating new strings.

(I haven't benchmarked a thing yet)

I want to format a timestamp obtained from a system call.

C: strftime() is available everywhere and no memory allocation needed.

Rust: Nothing in the standard lib, choose between various crates that can do this, figure out how to convert between SystemTime and whatever the crate has. Everything returns a newly allocated string.

All in all I wouldn't want to go back to C, but not everything is easier in in . :blobshrug:

Ayo boosted

*takes on 1000 projects, doesn't cull them enough*

*looks at TODO list for one single project in isolation* why am I not getting things done on this one single project

I made this awesome drawing to show how I envision that TorFS is going to work.

The system is specifically designed for people who like dogs but don't own any themselves.
code.blicky.net/yorhel/torfs

Is there a free noob and phone-friendly way to get a persistent IRC client running? How can I possibly get my friends off of WhatsApp?

I haven't been to in the past few years, always found a last-minute excuse not to go.

I just booked a hotel, I've no excuse this year.

Random observation: There is a big focus on (human-readable) naming systems in distributed networks - and this is indeed a hard problem - but I wonder if there's even much value in such systems. Looking at the web, most people I know go to websites through a Google search or link or bookmark, even when they already know the domain.

The downside of using cryptographic primitives in a way that apparently nobody else has done before: I have neither test vectors nor a reference implementation that I can compare with. :blobsweat:

Why programming makes me feel stupid, episode $n+1: I needed a base58 codec and figured I'd write a small/simple one myself. I've written codecs for base-10,16,32,64 before, so this ought to be easy!

Then I realized that a naive implementation needs integer division on the entire input, and for larger-than-64-bit values that's... not trivial at all.

Ended up just grabbing an existing library. :blobblush:

I'm (still) surprised that Merkle tree based hash digests are not more widely used as file checksums.

I've been looking around for a solution for my Torfs idea, but the only hash tree that is in "common" use for this purpose that I've found is still TTH.
code.blicky.net/yorhel/torfs/s

I was working out an idea of implementing something DC++-like over , and I think I actually came up with a pretty realistic project.

I wrote some details down at p.blicky.net/YcxaCt1Ben5j

Feedback appreciated!

I posted a comment on HN today.

I'm sorry.

I will not do it again.

Are there any cool privacy-preserving & distributed networking projects in need of additional client/server/tooling implementations?

I kinda feel like writing yet another DC++ client, but that network is pretty dead by now. Time to move on to something that still has potential.

The two sequels to Margaret Atwood's Orix & Crake (The Year of the Flood and MaddAddam) were pretty disappointing. The characters weren't as interesting, there weren't many new developments, it felt rather long-winded, the antagonists have only gotten shittier (for no reason whatsoever, we never get their part of the story), and there's a bunch of forced drama mixed in for good measure.

Such a pity, the first novel was brilliant.

How email reputation works in practice: Someone's sent me an email but I'm not allowed to reply because their server doesn't trust me. :blobthinking:

Show more
niu.moe

Welcome to your niu world ! We are a cute and loving international community O(≧▽≦)O !