392

Buy a whole load of the same model voice recorder, label them with tape for location, spread them out and record till batteries die, pull data to PC, repeat until get a sample? Someone above mentioned microsecond accurate clocks on raspberry pi being used for TDOA.

All of these are much more complicated and expensive than just sending someone off in a stolen car to fire some shots wherever you'd like the cops to be in 20 minutes or so. Convenient leash.

True in general but some (especially libs aimed at larger datasets processed in batch) are taking advantage of benchmarks like this to do things like: Try the fast way, if it works great Try the slow way, if above fails This makes the slow path 2x slower at worst (and you can advise to always use the slow way with optional params) but the fast path can be 10x faster

Hard to set up in Kubernetes. Seems to want you to know too many specifics for how to shard your db ahead of time. Don't remember how it handles schema evolution. I settled for Databend because it just worked easy and fast with nothing more than table definitions.

I use Firefox because they don't threaten to shut down power user extensions every few years. I find it faster and more reliable because it doesn't chew all of my RAM and lead to swap / early-oom-kill.

I'm quite confused by folks' opinion of Firefox in this thread because my only problem with it is that maybe once per year (or less) I find an online checkout that is subtly broken in it (sometimes it's just my adblocker). But I have occasionally found sites that are inverse, only working in Firefox.

I have tried unsuccessfully to get those setup on my k8s twice. The default configs don't seem sufficient for even a small shop's scale, I think I must be holding them wrong. But I'm ready to try this because it's all I need if it works.

I'll give installing it a shot after that is fixed!

Yes, and it's double the price of a module quoted in the article.

Two more clutches to wear and be replaced adds weight, space, maintenance, and cost. Motorcycles are an art in throwing away assemblies that aren't necessary.

No database is fully SQL compliant. Though clickhouse can at times be painfully so.

https://stackoverflow.com/questions/784900/why-does-no-datab...

>The sleep period consistently occurred during the nighttime period...was not interrupted by extended periods of waking...

Or, you know, take five minutes to read the article.

The map is not the terrain.

Chiming in here as someone who used to play in competitive GTA leagues. I quit playing because of too much time spent in load screens. It was worse than just this slow loading bug: random disconnects that triggered loading again got more and more common over time, too.

I actually just read a paper where some new file format achieved far faster and smaller writes than the Parquet used in all of these modern big datalakehouses, and it was done by guessing like 20 different simple compressions on a sample to figure out what will work best for each chunk of data. Sounds like manual feature detection to me!

It doesn't, maybe you mean something like Dagster or Airflow managing insertion time dependcies? Although iceberg does provide transactions across tables I believe.

Same. I wasted a month or so of off time trying to get that old stack to work well enough to let me just insert data, left unhappy. Had Databend up and running in an hour, figured it will get easier to do it right in the future once there's a Rust impl (for portability vs Java/Hive)

I don't see it on the Play Store, sadly :-)

I'd start by reading the papers mentioned below and then implementing a real distributed kv store without any fluff, just hard code your test cases into a simple client. Then try load testing by running a ton of your client. The problems you run into will be practical and real and you will have to debug them yourself with real tools. Then after that basic exercise I think Maelstrom will be more useful to you. TLA+ is an awesome design tool but imo implementation practice is more helpful for a beginner, you'll be more familiar with the problems TLA+ can help you solve.

https://omscs.gatech.edu/cs-6210-advanced-operating-systems

They're not really, but I can see why Beltone stores would give you that impression. Go see an audiologist.

I found changing my phone display from Vivid colors to Flat helped.

For a long time `npm` would do that to your stdout/in, which didn't impact anything in interactive shells because your shell resets it every time it shows the prompt, but run those same commands in a bash script and later processes will crash with strange error messages because no one has ever tested with stdout being nonblocking. So much for separate processes being independent.

Try buck2

Any thoughts on Databend? I was really impressed by the minimal time to set it up and the 100x faster inserts with parquet upload. But it doesn't seem to be picking up HN hype.