Show HN: Turtl – A secure, encrypted Evernote alternative

Post by orthecreedence » Wed, Feb 03, 2016, 5:53 PM UTC

ultramancool wrote: post #73

orthecreedence wrote: post #25

Earlier quoted context omitted.

Thanks for the honest feedback! The app IS a web frame, both on desktop an mobile. TBH it's pretty much the only way I can maintain it on 4 (and soon 5) different platforms efficiently. I will say this: Turtl will queue encryption/decryption in background threads so it does utilize cores while doing heavy processing. As far as how it utilizes cores in the main UI, that's really up to the underlying javascript engine,…

To your credit, the feature you're adding is far more important to me than any of the minor Evernote features the parent mentions. So don't sell yourself short. I currently find myself stuck syncing text files using SyncThing or ownCloud or making notes in my KeePass database just to get the level of privacy I want in this. I saw Turtl before and this makes Turtl highly appealing to me, I'll definitely be trying it o…

Post by sarciszewski » Wed, Feb 03, 2016, 6:13 PM UTC

logicallee wrote: post #82

sarciszewski wrote: post #69

Earlier quoted context omitted.

As attackers get better, we can just up the ante. More rounds, more memory usage, etc. If you're building software in 2016, you want to use one of the following for turning a password into a crypto key: - Argon2 - scrypt - bcrypt PBKDF2 should be your last resort. Don't fall back to a simple hash function.

that makes zero sense. something encrypted in 2001 isn't supposed to magically become plaintext in 2016 because "attackers get better." it's fundamentally not the promise of encryption. (I thought.)

Post by orthecreedence » Wed, Feb 03, 2016, 6:38 PM UTC

formichunter wrote: post #48

Is it possible, or maybe I should offer, to encapsulate this in docker using docker-compose? I find it easier and cleaner to try new things out running docker-compose up.

Post by orthecreedence » Wed, Feb 03, 2016, 6:54 PM UTC

sarciszewski wrote: post #49

https://github.com/turtl/js/blob/2ca59900d71284795e278e75585... Nothing says secure like PBKDF2-SHA1 with 50 rounds. https://github.com/turtl/js/blob/2ca59900d71284795e278e75585... ...or timing attacks on MAC validation. (Yeah, you switched to GCM, but a downgrade attack could potentially be used to find a valid MAC for a chosen ciphertext without the attacker knowing the key...)

Post by orthecreedence » Wed, Feb 03, 2016, 7:23 PM UTC

sarciszewski wrote: post #88

orthecreedence wrote: post #87

Earlier quoted context omitted.

> Nothing says secure like PBKDF2-SHA1 with 50 rounds. This code is deriving an encryption key and an HMAC key from a master key. The master key is already a random value, so even if you managed to "crack" either one of the derived keys, you wouldn't know it. > or timing attacks on MAC validation Good point, I'll fix that!

The correct tool for the key-splitting job is HKDF. If SJCL doesn't offer HKDF, I'll be surprised. It's not hard to implement, however.