Yes, the specification will be finalized with version 1.10. Previous versions also include specification changes. Iceberg's implementation of V3 occurs in three stages: Specification Change, Core Implementation, and Spark/Flink Implementation.
So far only Variant is supported in Spark and with 1.10 Spark will support nano timestamp and unknowntype I believe.
What about backups? My greatest fear is self-hosting valuable stuff (like family photos) only for my NAS to fail one day and lose potentially everything.
There are good options. Borg, rsync.net with zfs send/recv. Storage boxes from Hetzner
Hard to make a particular recommendation as backing up to the cloud is a popular option but depends on your upload speed and rate of data change. And depends how much you're willing to spend for what tradeoff
Storage box from hetzner is great. Hetzner is a good company, I'm glad that I'm doing business with them. The service is good, cheap, reliable, and... Un-american which means it would probably stay good and won't enshitify.
Use Hetzner! I'm not affiliated just a happy customer.
i use hetzner too, and I like their current services. But I think we all do good in never assuming or relying on ANY company being our friends. Be vigilant
I keep important data on a zpool that's mirrored across 2 drives, I snapshot it nightly, zfs send/receive the snapshot a drive on a different machine, and run a borgmatic/borg backup to borgbase 3 times a week. I also run a scrub on it quarterly.
I remember one (might have been a hn-er's comment, dunno) about the computer restarting when the toilet was flushed. Turns out it was due to voltage drop when a compressor turned on to refill the reservoir of the toilet.
That's why in rural locales with spotty power it pays to have a UPS on any electronics -- you might not benefit much from 15-30 minutes of extra power in a day long blackout, but it keeps everything happy when the voltage fluctuates.
Analog electronics [0] uses a continuously variable signal while digital electronics interprets the signal with thresholds that define states like 0 and 1.
Here is a simple example: using a few discrete parts, like two transistors (Darlington pair), a LED and resistor, you can create a simple circuit that shows varying brightness of the LED depending on how close you move your hand or an object to an antenna connected to one of the transistors (forming a sort of proximity sensor). No microcontroller, SBC or even a hint of a digital signal involved at all.
> Here is a simple example: using a few discrete parts, like two transistors (Darlington pair), a LED and resistor, you can create a simple circuit that shows varying brightness of the LED depending on how close you move your hand or an object to an antenna connected to one of the transistors (forming a sort of proximity sensor)
Do you have some book/video/etc. recommendations for this "type" of Electrical/Electronics circuit engineering? I only know how to program a MCU :-(
"Digital electronics" communicate using discrete values, 1s and 0s.
"Analog electronics" communicate using voltage/current/temperature/etc levels.
One of the simplest examples is a voltage divider: if you put two resistors across a DC voltage source, like this:
V+-[R1]-¢-[R2]-GND
The voltage at the ¢ point will be:
V+ * (R2 / (R1 + R2))
There are infinite possible values for that voltage, depending on the voltage source and the two resistors. It cannot necessarily be expressed exactly in a digital circuit, and it will fluctuate over time as the environment changes in temperature, humidity, EM noise, and so on.
I usually recommend The Art of Electronics as a well-written, beginner-friendly textbook which covers the basic concepts.
Classifications are messy, but in addition to the other items mentioned already, I would say that some people would break out "power electronics"[1] as its own field.
Probably the wrong way to look at it. Digital electronics doesn't really exist outside of theoretical spaces. It's all analogue underneath and any experienced digital designer will know that and what the consequences for things like signal integrity, noise immunity and latency.
It's better to describe analog and digital electronics as a subset of electronics. For the most part, they look at different domains. Even though they are based upon the same underlying principles, the simplifying assumptions are different. A more dramatic example is with RF electronics. While it may look like you are dealing with the same sort of things as the more common low frequency analog electronics, you are going to have a difficult time coaxing an analog circuit to work in the RF domain.
Contrast that to web developers. They are dealing with very different principles from web browser developers, who are mostly working with different principles than operating system developers, who are working with entirely different principles from those who design hardware. It's not that they are working with a different subset of the same thing because one layer of abstraction is directly on top of the one below it and (ideally) the layers below completely hide how they work from the layers above.
I mean... Analog electronics doesn't exist either, or for that matter, electronics in general.
All of electronics assumes Kirchhoff's Current Law and Kirchoff's Voltage law, which does not truly exist in reality. Electrons often escape a circuit (see antennas, which throw the voltage / current into a wave that is emitted out of your designs). All wires are antennas, so even the most basic circuit doesn't have all the current return in a loop.
The assumptions of KVL and KCL are just over-simplifications of true physics, Maxwell's equations. Because working with Maxwell's equations directly is too much effort in practice.
--------------
Electronics itself is a huge abstraction upon physics. You could, in theory, calculate all the voltages and currents using Maxwell's equations, except this isn't useful at all.
Similarly: most of "Analog Electronics" uses simplifications as well: OpAmps are often assumed to be ideal (aka: infinite gain), which is good enough in most cases.
Digital systems, using relays, predate analog electronics.
There were relays used for railroad signaling in the 19th century. Union Switch and Signal was formed in 1881. The first active electronic device, the deForest Audion, was developed in 1906.
Easy for Bill Gates to advocate for this, as he already got rich partly by people automating away lots of jobs using his operating systems and softwares.
Yes, it would definitely be easier to advocate for good public policy around automation while experiencing the benefits of automation. Everyone else should enjoy the benefits of automation too, and even be more free to effectively advocate for public policy of their choosing as a result.
