3 ms·
I know you wrote the document; that's why I was surprised to see you agree with a claim that using WebCrypto is "more questionable" than serving and using SJCL
by giaour 10y ago
I know you wrote the document; that's why I was surprised to see you agree with a claim that using WebCrypto is "more questionable" than serving and using SJCL based on the differences in the libraries interfaces. I personally find SJCL's interface difficult to work with, since all bytes must be converted to an idiosyncratic format (sjcl.bitArray), whereas WebCrypto uses the unsigned int array primitives introduced to the language after SJCL was released. I'm pretty sure that's not what you mean when you say the WebCrypto interface is inferior, and I'll probably need to do some basic reading to figure out what exactly that means in the context of cryptography.
FWIW, WebCrypto should only be available in a secure context (https://lists.w3.org/Archives/Public/public-webcrypto/2016Sep/0007.html https://lists.w3.org/Archives/Public/public-webcrypto/2016Se...), though I believe only Chrome enforces that requirement at the moment.
- CiPHPerCoder 10y agoWebCrypto is also a potluck of primitives, designed my committee, with "secure end-to-end encryption" as a non-goal. I'd personally rather see browsers ship libsodium.js (or equivalent) than WC or SJCL.