Yes I'm definitiely asking for single-user systems, no real servers - and within that frame, there ought to be something in the 100 bucks price range that's not total crap? I mean, it's not rocket science. It seems ridiculous that I buy a damn glorified HDD controller for more than double or triple of what my actual main computer costs.

What you pay for is getting a system that works reliably and that holds your data with redundancy. This is something that matters for single-user system as much as for systems that serve a thousand users. Losing data sucks big time. Restoring from backup sucks as well, even if you don't lose much.

[...]

Harddisks start making noise? Pull them out one-by-one, plug in new ones, and worry no more. Data is still there and you need not interrupt your work for one minute. That's what you pay for.

Yes, yes, you sold me already on the synology, ordered one a few hours ago ;-)
There might be a plethora of hackier options, but one thing that I really don't have is time to mess around and things still not working after a weekend.
Now if you were literal with a few clicks to have a "remote" GIT / SVN server, that would be a very nice plus :-D

Okay, so I got the Synology DS214se.

And noticed that I overlooked some nasty detail: its support for file systems other than ext4 is only for external drives, so one is stuck with ext4.

Now, I have read several seemingly knowledgeable people explain that ext4 is quite sensitive to power loss - so badly that it may trash a partition unrecoverably. (I am not quite an expert on file systems or linux, but that scenario was explicitly outlined)
This quite upsets me - since it makes having a darn expensive RAID-1 setup rather pointless, doesn't it? I convinced myself to spend over 400 EUR for some more storage "because data loss gets a lot more unlikely", so now this is kinda disappointing, to put it mildly.

I guess I'm still searching for a NAS then, one that supports a file system which lends itself better to data safety.

Well, have to research that subject more, when I got the time (TM) *sigh*

You probably refer to this? That is somewhat outdated (2009) info. Unluckily, the internet keeps everything forever. The reason for this problem was trying to optimize for speed. Which, of course, did more harm than good.

See https://www.kernel.org/doc/Documentation/filesystems/ext4.txt for more recent (2013) info -- note the data=ordered and commit=nrsec fields. The first is (*) enabled by default, and the second defaults to the exact value that is present in ext3, too and is exactly as resilient to power failure (and much more resilient than e.g. NTFS).

Yes, you will probably (or at least possibly) lose the last 5 seconds worth of writes if power fails. And yes, in theory it is possible to trash the complete disk when power fails. That's in theory always possible no matter what you do, since you don't have any guarantees what goes on inside the harddisk while power cuts off, nor what the harddisk will make of it. In theory, the disk controller could go frenzy and write out complete garbage during the last 1-2 milliseconds. Practically, no bad things should happen (a no-junk harddisk should have safeguards against that scenario, and since writing takes more power than sustaining the controller, this is very unlikely to begin with, anyway).

Trashing the disk is -- however theoretically possible -- not very likely to happen, however, but even so it is easily fixed with an UPS. Which anyone with a reasonable chance of power failure (say, might occur once per year) should have anyway.

Ironically, the likelihood of power failures has increased with technology advancing. I remember one 5-second power failure back when I was a child. That was a "Huh, what is this? You mean, power can fail?" moment. I've been using UPS out of power failure paranoia for 15 years with zero power failures during 13 years, and 3 during the last 2 years (one of them being 35 minutes -- but that's still lucky compared to what you see on TV sometimes).