This raises the question of if the writing can be too good.

That `const` performance cliff is brutal!

Tested various version in my shader, interesting how it becomes fully pathological when simply adding a `const`: gist.github.com/vassvik/621c...

It often feels like the hole software is digging for itself is inescapable as security concerns keeping offering shovels with which to dig deeper. It's surprising (to me at least) that sandboxing is so hard!

Yeah, I agree. I do sympathize with the desire to have visibility into what’s going on to help with estimates and availability, but I also think ever increasing visibility goes from diminishing returns to bureaucratic quicksand while tricking managers into thinking things are steadily improving.

Related, Boston Dynamics released a nice video on their current gripper design. Great to see folks challenging the reign of vacuum grippers! youtu.be/gS4rOqNDTBk?...

A corollary that plays a large role in my life philosophy is that a lot of successful papers are successful because of good engineering. People looking for silver bullets will be disappointed.

I’m referring to the initial rollout when my impression was that export of passkeys from one manager to another was not complete.

Congratulations, what an amazing fit!