Yeah, my experience with Aqara has been like that: flawless out of the box, pairs quickly with no issues, reports state correctly, all good. Then after a random while (at least a few hours, but within a day) they just drop off the network never to be seen again until I re-pair them.

I even got an Aqara smart plug just to act as a router and pair the Aqara stuff through it, it was better but this time it dropped off after a few weeks. Might’ve been the battery though, it was a temperature sensor in -20°C.

Running a bunch of services here on a i3 PC I built for my wife back in 2010. I’ve since upgraded the RAM to 16GB, added as many hard drives as there are SATA ports on the mobo, re-bedded the heatsink, etc.

It’s pretty much always ran on Debian, but all services are on Docker these days so the base distro doesn’t matter as much as it used to.

I’d like to get a good backup solution going for it so I can actually use it for important data, but realistically I’m probably just going to replace it with a NAS at some point.

For self-hosting purposes, Docker = lightweight disposable VMs that are configured via docker-compose.yml. All important data should be in “volumes”, which are just shared folders between the host and the container.

The end result is that you can delete and re-create containers at any time and they should just pick up where they left off from the data that’s in these volumes.

Each individual published image has some paths they want to use for that; everything is usually specified in their example docker-compose files.

If you’re not a dev, don’t even try to understand Dockerfiles, it’s not for you.

It’s not quite as point-and-click, but I’m using Docker for that because Yunohost kept messing up updates. Most server apps will have some instructions on how to run them in docker, especially a docker-compose.yml file, so you don’t have to rely on the Yunohost team to package said app.

The way I do it is that I put each suggested compose file in their own file, and import them in my main docker-compose.yml file like this:

version:  '3'
include:
    - syncthing.yml

Then just run docker compose pull && docker compose up -d every time you change something or want to update your apps, and you’re good to go.

Software updates in particular are waaaaaayyy easier on Docker than Yunohost.

Docker’s secret that most “getting started” tutorials seem to miss is docker-compose.yml. Who wants to type these long-ass commands to start containers? I always just create a compose file, and then docker compose up -d.

Dockerfile is for developers, you shouldn’t need more than a docker-compose.yml for self-hosting stuff.

That’s exactly what I was saying a few comments up. There’s a use-case for cars, and there needs to be a way to get into the city when you arrive by car. Any good solution will probably involve parking the car somewhere outside city limits, otherwise it’ll probably be co-opted by suburbanites.

I’m not talking about commutes, I’m talking about going to the city for an appointment/shopping/conference/concert/sightseeing/etc.

But yeah, cycling the last mile works in the Netherlands between cities or suburbs because they are relatively well served by inter-city transit, but what about places like this random dairy farm . Can this guy just take his bike to downtown Amsterdam?

Something something dining philosophers.

Bikes ain’t gonna work for people coming from far outside the city. I’m not talking about commuting distance, I’m talking about people who live in rural areas 2+ hours away from a city that need to come in occasionally. Having them make the whole trip by car necessitates maintaining car infrastructure in the city center, which will soon be co-opted by suburbanites. This use-case needs a bi-modal strategy.

What usually works better for moving people in and out of cities is park-n-ride setups where you setup a giant parking lot in the suburbs next to a metro station. People can just ditch their car outside the city and proceed using public transit. I often do this in Montreal, for example.

For goods, it’s a similar setup but with big trucks transferring cargo to smaller trucks; this is already pretty common.

The lining in question is very thin (akin to a layer of paint) and just burns up when the cans are re-melted.

Recycling beer bottles is indeed pretty easy once you get them to the processing center intact, but it’s getting there that’s the hard part. They’re fragile, pretty heavy and don’t stack well unless you put them in some form of packaging.

Once they’re broken, they’re basically useless; glass isn’t recycled much except as grit material for sandpaper; re-melting it is resource-intensive and sensitive to impurities.

It’s true that you can easily fall into analysis paralysis when you start learning JS, but honestly things have somewhat stabilized in recent years. 10 years ago everybody was switching frameworks every 6 months, but these days we’re going on 8+ years of absolute React dominance. So I guess that’s it for the view layer.

The data layer has seen some movement in more recent years with Flux then GraphQL / Relay, but I think most people have settled on either Apollo or react-query now (depending on your backend).

On the backend there was basically only express.js, and I think it’s still the king if you only want to write a backend.

Static websites came back in fashion with Jekyll and Github Pages so Gatsby solved that problem in js-land for a while, but nowadays Next also fulfills that niche, along with the more fullstack-oriented apps.

Svelte, Vue, Aurelia and Mithril are mostly niche frameworks. They have a dedicated, vocal fanbase (see the Svelte guy as sibling to your comment) but most of the industry has settled along the lines I’ve mentioned.

Honestly I think the main thing that the JS ecosystem does well is dependency / package management (npm). The standard library is very small so everything has to be added as a dependency in package.json, but it mostly works without any of the issues you often see in other languages.

Yeah, it’s not perfect, but it’s better than anything else I’ve tried:

• Python’s approach is pretty terrible (pip, easy_install, etc.) and global vs local packages
• Ruby has its own hell with bundler and where stuff goes
• PHP has had a few phases like python (composer and whatnot) and left everyone confused
• Java needs things somewhere in its $PATH but it’s never clear where (altough it’s better with Gradle and Maven)
• C needs root access because the only form of dependency management is apt-get

In contrast, NPM is pretty simple: it creates a node_modules and puts everything there. No conflicts because project A uses left-pad 1.5 and project B uses left-pad 2.1. They can both have their own versions, thank you very much.

The only people who managed to mess this up are Linux distributions, who insist on putting things in folders owned by root.

C is crazy. While you are learning it you are learning Make and gcc without your consent.

Java is crazy. While you are learning Spring you are learning Maven or Gradle even without your consent.

To any non-js dev taking this too seriously: A good half of the technologies mentioned in this meme are redundant, you only need to learn one of them (in addition to the language). It’s like complaining that there are too many Linux distributions to learn: you don’t, you just pick one and go with it.

Haven’t watched the video, but what do you think circularization is? If you’re “just a circulization away from orbit”, you are indeed going a bit slower than orbital velocity. There’s no point to going orbital velocity if your trajectory still brings you back inside the atmosphere. To get to orbit you want to raise your periapsis outside the atmosphere, and you do that by doing a burn at the apoapsis, which is what we commonly call “circularization”.

I use famous programmers. First Linux server was Torvalds, first mac was Woz, currently in service I have Kernighan (one of the inventors of C), KJohnson (Katherine Johnson was a programmer for NASA) and Shamir (The S in RSA).

Treason is a crime you can only commit against your own country. The US can’t accuse a non-US citizen of treason…

By that same logic, can Russia ask Japan to extradite a US citizen because they advocated for LGBTQ+ rights while they were in South Korea? Because that’s basically what’s happening here, I just swapped the offence and the countries involved.

Dude isn’t a US person, wasn’t in the US when he committed the alleged crime, and said alleged crime isn’t a crime where he allegedly committed it. US law isn’t world law.

EVEN IF the guy might’ve been rapist asshole (allegations were fishy as heck), this extradition proceeding is a gross overreach by the US, and the UK should have laughed it out of court. If a country has any leg to stand on regarding extradition, it’s Sweden (I think that’s where he was when he committed all the alleged crimes, both the sexual ones and the wikileaks ones).

Yeah, but sadly they’re looking to replace it with a “universal” pattern too