14:47:04 in case someone wants to address this issue: 14:47:04 accented passphrases from Cake restore to a different wallet in Monero. 14:47:04 would need a PR on Cake 14:49:12 > crypt() skips NFKD so Cake-encrypted seeds with accented passphrases restore to another wallet in Monero 15:23:33 Cake seems to take forever to respond to PRs 15:23:52 At this point, it might be worth contacting them privately to ask them to take a look at it 15:29:18 yeah they ignored my previous PR totally, didn't even run ci for it so I could potentially improve it 15:30:43 it wasn't a bug fix and I disclosed full AI usage because I never used dart back then, but still running ci for it was the minimum 15:37:27 There's also this known bug in their Dart Polyseed library which isn't getting merged 15:37:27 https://github.com/cake-tech/polyseed_dart/pulls 16:52:24 Well, I would be very surprised if they changed behavior here in any way. They probably would not risk to cause problems for their users. 16:54:15 I guess if it has to do with Unicode normal forms some simple accented characters like é and ö are still ok, it needs something more "exotic" before there is trouble? 16:59:49 The one I linked to is a legit bug where seeds with restore dates after ~2040 will be rejected 17:00:06 Obviously not an issue right now, but why wouldn't you fix it? It's also not in accordance to the canonical impl 17:00:14 @mrcyjanek0:matrix.org 17:00:27 Maybe talk to the dev 17:00:38 Forgot he was here :P 17:01:14 Its a small repo, posdible he just didnt see (or forgot about) the prs 17:01:40 Thomas has had a PR for the main Cake repo sitting for a few months 17:01:43 No response yet 17:17:59 probably busy with zano, zcash, and mweb exploits :P 19:03:14 Speaking of polyseed, does anyone else have an opinion on this? https://github.com/tevador/polyseed/issues/25#issuecomment-6170942720 19:59:36 @rbrunner7: é and ö are exactly the problem case lol 20:02:42 Oh, then it's bad. 20:04:50 And we don't have that problem already with the ages-old wallet passwords themselves? Only with Polyseed passphrases, because somebody had the genius idea to process that somehow first? 20:05:57 I mean wallet password compatibility between CLI wallet and Cake Wallet if the password has an accented or umlauted character in it 20:06:15 > * polyseed_dart crypt() derives the mask from utf8.encode(password) - polyseed C (Monero) NFKD-normalizes the password first. 20:06:15 > * With a passphrase, Cake's backup phrase is the encrypted seed, so Monero decrypts it with the NFKD bytes; if they differ → valid but different seed (no MAC) → empty wallet, no error. 20:06:15 > * Affected: anything NFKD changes: precomposed accents (é → e + U+0301, ö, ñ, й), Hangul, kana with dakuten, fullwidth. Unaffected: ASCII (up to 575 bytes), CJK unified ideographs. 20:06:15 > * A Monero-side fallback needs a non-normalizing crypt, which polyseed's API doesn't have. 20:08:57 Well, as there is no functional spec, I guess it's easy to overlook that normalization in the original Polyseed implementation in C, and then not doing that in a Dart port -> problems 20:12:55 What is this Dart function doing? utf8.encode(password) Doesn't Dart already use UTF-8 throughout for strings, why would it be necessary to "encode" the passphrase? 20:14:09 I guess if nothing else helps, restoring using the private key directly would be a last fallback 20:14:15 Dart doesn't necessarily use UTF-8 for strings 20:14:28 IIRC, they're UTF-16 / WTF-16 20:15:01 That surprises me now, given that Dart is a pretty young language ... 20:15:26 Well, since 2011 20:15:38 Older than Rust 1.0 20:15:39 Hmmm 20:17:09 Yeah, the internal character representation of C# is also not UTF-8, if I remember correctly - you don't often go down to that level. 20:18:16 Ok, enough Monero for today ... 20:18:37 theres no such thing as enough monero for today 20:18:49 havent seen the sun in days