Nickolai Zeldovich

I seem to recall that when Clang compiles a "delete" statement, it generates LLVM bytecode to check for whether the pointer being deleted is null (and if so, to not...

Oh, that's right, the default build of Clang doesn't put in those basic block names.. Forgot all about that.

I think hard links are a bad idea in the context of fscq. Hard links make the spec much more complicated (i.e., the file system state cannot be thought of...

The metadata that Vuvuzela tries to protect is "which users are communicating", rather than "which users are connected to Vuvuzela" or "what is this user's Vuvuzela username".

Indeed, I agree it would be useful in some cases to provide pseudonymity. It might be interesting for us to think about how to provide pseudonymity. One place where the...

@derbear and I were just talking about how to solve this. The original plan that @davidlazar and I were considering was to build some TCP-like acknowledgment and retransmit protocol on...

Indeed, was just poking at this, and it's easy enough to get Goose to emit the right code for this case (and even avoid changing any of our existing Goose'd...

@Lodifice I also ran into this problem with 7.1.1-42962, on a DS418. As far as I can tell, there's something buggy with the resulting `wg` kernel module. I'm not quite...

Seems fine, in the sense that it doesn't hurt, but also I don't think this really improves the state of security at all. The systemd service runs an algod binary...

@mxmauro might be sad about this, though this may be the right plan if Ledger thinks the U2F tunneling is a dead end.