Nikolai Konovalov

P.S. I'm tired of transferring ASN.1 schema to TS like 10 times and catching errors because of fucking CHOICE

ISO/IEC 39794 is good replacement for ISO/IEC 19794, but i have never seen so many "pyramid" ASN.1 structures. tsemrtd will implement this standard, but only for mandatory fields and only for DG2 now

Ask me about my dream and i'll answer "stały pobyt"...

"Many thanks" to Roskomnadzor, cuz now I can't open sites that using @cloudflare.social without using VPN.

btw GPL (and AGPL, LGPL) sucks, use MIT

Especially funny situation with Ukrainian algorithms (DSTU)

Kupyna (DSTU 7564:2014) - modified Grøstl;
Kalyna (DSTU 7624:2014) - modified AES;
Strumok (DSTU 8845:2019) - modified SNOW 2.0;
Skelya (DSTU 8961:2019) - modified NTRU;
Vershyna (DSTU 9212:2023) - modified CRYSTALS-Dilithium.

Screenshot from my Github profile with a list of implementations of cryptographic algorithms from post-USSR countries

Welcome to the "fascinating" world of cryptographic algorithms in post-USSR countries...

Screenshot from my GitHub

It was very painful...

Im fucking hate @jsr.io #2

Why do I have to specify an explicit type for constants?
I understand that this should work with variables, but constants just got me...

В продолжение поста

Коммиты на русском языке be like

Никаких вам "Initial commit", только "Начальное закрепление"

I fucking hate JSR...

Context: Project has no dependencies at all, but JSR changed format of Bun lock file (.lockb -> .lock), and also throwed error 🫠

How do know which Swarm is serving specific Session ID?

- Split Session ID into 4 parts of 8 bytes and XOR each other.
- Next, you need to find closest Swarm ID value in blockchain

Example: d31609a18228b69e -> d300000000000000 (closest value in blockchain)

Specifications >> The XEdDSA and VXEdDSA Signature Schemes This document describes how to create and verify EdDSA-compatible signatures using public key and private key formats initially defined for the X25519 and X448 elliptic curve Diffie-Hellman functions.

Remarkably, Signal had same problem when developing the XEdDSA and VXEdDSA algorithms, but they solved it by forcibly zeroing the sign bit. (check 2.3 section of specification)

P.S I especially like that "libsession-util" repository intentionally disables issues

Messenger Session in new Blinded ID algorithm makes same mistake as in old algorithm.

Again failed if-condition is used to check the sign bit (pk[31 & 0x80) to generate hash of the server's public key, as result we again have 2 variants of possible public key when converting key pair

Also keep in mind that the current Blinded ID algorithm is reversible, i.e. it is possible to get the user ID knowing only the public key of SOGS.

A practical implementation of this algorithm is - github.com/theinfinityw...

It's pretty funny that Session messenger uses 16 bytes, obtained from the seed phrase and not modified in any way, as the Ed25519 private key for your profile 🫠