David Benjamin

AES-GCM is indeed not a stream cipher. If you use it as that, you have a security bug because you're actually just getting a slightly different variant of CTR mode,...

OpenSSL has a lot of assembly code, which is written to the standard AAPCS64 calling convention. How does this new calling convention interact those? Do assembly function prototypes need to...

> Based on what I was looking at, it doesn't actually look like the windows targets in Openssl currently use any of the arm64 assembly code. I don't believe that's...

What happened here was: 1. OSS-Fuzz had an MSan regression. Annoying, but it happens. 2. We got a ton of false reports in OSS-Fuzz. 3. I spent 30 minutes of...

TBH I remain unconvinced we actually are going to need to turn the crank another time after HttpOnly but I agree with @sbingler that *if* we have to do it...

> it's redundancy to protect against attribute loss (from configuration errors and the like). I don't think that's *quite* right. Prefixes are redundancy to protect against cookie injection. The problem...

We're not using them. Some minor classification might be useful to us, since we *do* process flags programmatically, but nowhere near as fine-grained as the `bugType`s are now, and it...

Happy to put up a PR if it would help. Mostly I didn't want to collide with the work you're doing in #151.

Sure! Although in a similar unprecedented software engineer phenomenon, I realized that canonicalizing the indent with standard JSON implementations very quickly bumps into how to canonicalize the field order. On...

Hmm, okay, some other tricky things if I compare what Python did and look at the diff with all spaces removed: * AES-FF1 tests use lists of integers for their...