Jabdoa2

By running vllm at work I learned that you will usually outperform benchmarks which you will find on the internet. Mostly by running proper parameter tuning. So you get impressive numbers. Still in reality those numbers do not tell the whole truth. We found that real usage is more 60k-140k prompt with a few k of output for realistic agentic coding workloads. Most of the time that will give you half to a third of decode throughput vs the benchmark. It also becomes much more about cache aware routing and kv cache management (how to offload and restore). Because prefilling 60-140k will simply take too long for interactive work.

Now you just have to mod your webserver to send the image chunk by chunk (with waits in-between). That way network latency does not matter. Also it probably reduces artefacts as bytes from one frame most likely are received in one network packet.

You can login with an OpenStreetmaps.org (osm) account and edit it yourself. This app is used to add those objects. There is also an app "StreetComplete".

You can also "just" set the 802.11k entries manually. Add 802.11r and you should be mostly good. Usteer makes it slightly better by moving clients to the best AP when they stay stationary for longer whiles.

You can insulate the chamber. That works fine. There is a vent on top which is open in case the printer needs lower temps. For everything else it will turn on the chamber fan. The parameters are tunable in the menu (or via G-Code).

Sounds like a very useful service. I just signed up. Did you consider to inform new users that the service is financed by donations? You seem to focus on "always free" which is nice but I guess most people are also ok to pay some money if your service is useful to them. I guess this might drive some more donations if people knew that you need money. Also a banner like Wikipedia (maybe a bit less annoying) would work.

I guess according to GDPR this counts as tracking nontheless. GDPR does not specifically mention cookies or anything technical. An identifier is enough (does not have to be a uuid). IP, location, browser etc already counts. This probably would count as storing something like a cookie on the client.

Is there a way to specify a specific version of a language in devenv? I checked to docs and even read the nix tutorial but could not figure this out.

I like to idea of having this in every repository in our codebase. Would make bootstrapping easier for new developers. However, you often want a specific golang or python version. Once you update some tool or language everybody gets the new env. That would be neat. Is that possible somehow?

I guess initially we would roll this out just for dev and keep Dockerfiles how they are but eventually we could then use it as builder in docker.

Bonus question: Can I use devenv/nix together with Bazel? We use that in quite a few newer projects and it also suffers from the local dev env issue.

Guest Login in Microsoft Teams on Linux seems to be broken for everybody now and Microsoft is not even working on a fix yet.

Any chance you could enable search for countries such as Germany? I.e. searching for "Hannover, Germany" (a city with 500k people) just shows the whole globe without machines.