tpankaj

That's not what that column is supposed to mean afaict. The way I read it is it's showing that if the website requires hundreds of different parallel backend service calls to serve the page load, what's the probability a page load hits the p99 instrumentation latency?

We have a similar chart at my job to illustrate the point that high p99 latency on a backend service doesn't mean only 1% of end-user page loads are affected.

There's a difference between legal reasons and decency

I don't want my own office. If I wanted to work in a room by myself, I would work from home. The whole reason I go to the office at a job that lets me choose is because I find it easier to be productive when sitting with my teammates.

You do not have the right to force your personal religious views onto other people though. Tolerance does not mean everyone else has to allow you to dictate how they live their lives.

I don't think it's fair to complain that you're not interested in reading about the state of venture capitalism on an article titled "The State of Private Markets". The article delivered exactly what it promised in the title.

Well-equipped soldiers may be fine but low-income civilians in Khartoum would not have the means to obtain P100 respirators or gas masks.

By law, if the FDIC takes losses to the insurance fund to pay out on uninsured deposits, they have to charge an extra premium to the member banks. So it's still the banks paying for it.

Why would banks take more risk with their deposits? The bank executives, shareholders, and bondholders wouldn't be personally affected by the loss of deposits anyway so I'm not sure they would care either way. If deposits aren't guaranteed, they lose their investment, and if they are, they still lose their investment.

> Unless the insurance is paid for by the banks themselves

It is though. The FDIC insurance fund is funded by premiums paid by banks.

This is also assuming that the sale of SVB's assets doesn't end up covering the full cost, and that the remaining cost is large enough to result in a meaningfully sized special assessment on FDIC member banks. My understanding is that neither of these outcomes are a given or even likelier than not.

Taxpayers did not pay for the FDIC protections. And shareholders are wiped out, so they didn't gain from this collapse.

While the goal sounds admirable, it seems like it would inevitably mean discriminating against young, child-free employees.

Maps has become buggier and buggier over time. These days I can't figure out how to use the UI and my navigation to work has become completely wrong (it used to be correct for years) out of nowhere.

I think the fear is a bit overblown. California already doesn't allow non-competes as far as I know.

My understanding is they would have to prove criminal intent.

The process is quite complex so it's usually too heavy handed for a project of this size. That's why people look for an existing nonprofit to take it on. But if kloch mainly wants a couple other people who care about the project to be around to make sure it continues, I'm onboard.

I'm 25 and I'd totally take it on, but you're probably looking for an actual nonprofit if you want true survivability. As someone who has used this tool for at least 10+ years (as long as I've known what an IP address is), I'd love to help make sure it stays around.

My team is 3 right now and 33% time oncall is still pretty bad.

There are a couple of blog posts and associated papers on TAO:

https://engineering.fb.com/2013/06/25/core-data/tao-the-powe... https://research.facebook.com/publications/flighttracker-con... https://engineering.fb.com/2021/08/18/core-data/ramp-tao/ https://engineering.fb.com/2022/09/07/open-source/taobench/

I work on TAO consistency at FB and we've put out a bunch more articles/papers in the past couple years. Until 2 years ago, there was just the 2013 paper on TAO itself.

Where did you move to?

That's a pretty narrow viewpoint. I can tell you from personal experience that it is very difficult to graduate college almost straight into a (pandemic-induced) remote job in a new city while trying to make local connections and ramp up in my job during a global pandemic.

How does DynamoDB guarantee p99 latency? At first glance it sounds impossible to guarantee something like that, especially accounting for random network congestion or queuing on the client. (Also hi Lu!)

This is why production services should have health checks that actually verify whether the service is doing its job in the system, not whether it's merely running the process. A process can be completely broken while still technically running, and vice versa (at least with janky ps-parsing scripts).

I noticed some other weirdness like functions taking a double pointer argument and then freeing the memory and replacing it with a newly allocated piece of memory. Seems rather dangerous to do that.

And zeroing out data by freeing the memory and callocing new memory. Why not just write zeros into the memory instead of reallocating it from scratch only to overwrite it with zeros anyway?

I believe another contributor wrote a Go user-space implementation and Jason decided to build off that for the PC user-space implementations and Android and iOS. So basically, because it was there.