It wasn't even for sale in my area until the USB problem had been fixed. Funny thing is I got a replacement one "new" for a friend whose kid wrecked the keyboard in late 2013 - the repair centre still had some unused.
The dismal battery life is with WiFi turned on, so it only lasts uncharged about a day at work with WiFi on or three days at home with WiFi off.
Something with the same form-factor and more modern (less hungry) hardware would be nice, even if it is android and not real linux.

went from over three minutes to down to less

Something is definitely seriously wrong in that case. Are you sure it wasn't a change to autofs to mount them on demand instead of boot or using more correct NFS mount options that made the difference?

The other bits are interesting and make your point but that final one is a somewhat pathological case.

As for my example - a eeepc with an SSD but not a very quick CPU. Boot time with the stock distro (xandros) was about 15 seconds, around 45 with a cut down recent Fedora with hardly anything starting. It now has FreeBSD10 on it.

Yes - the correct behavior as Lennart sees it is to halt and wait for the user to insert a rescue CD.
I don't see that as correct myself but he has a desktop perspective inspired by growing up after Win95 and not paying much attention to server environments.

Considering how it blocks (with no output as to why) if a filesystem is not present is it REALLY doing enough in parallel or just enough to have race conditions in the binary logs?

most importantly, things are improving.

They certainly have to continue to improve before systemd becomes a more worthwhile option than the things it is replacing.
The only problem systemd solves is to replace things so old that they are maintained by people that have been coding for longer than Lennart Poettering.

The systemd suite provides features such as faster boot times

I haven't seen any sign of that anywhere and I saw the opposite on a eeepc by about half a minute when I put a newer distro with systemd on it. Is there any proof or are the faster boot times just on the wish list?

It got stuck and needed a reboot two or maybe three years ago and has been rock solid since. Still nothing viable to replace it without having to use two separate devices.

Can you imagine millennial parents giving their precious offspring pocket knives?

I was about to reply that I've seen a few young kids with pocket knives but their parents grew up in places like China and have associated more with older generations in the west than people of their own age.

I had my own .22 rifle by the time I was 10.

Maybe if we had more of that now people would see the things properly as tools instead of the NRA insanity of it being an external sign of manhood, patriotism and being ready to overthrow the USA in a minute.

From mucking about with it professionally (foundry sand packing test - pump mercury under a little bit of pressure through a sand sample) and reading a lot about mercury safety at the time it's the fumes that are the problem. Don't breath in mercury fumes and you'll be as fine as the gold miners working outdoors that used to stick their hands in the stuff and far better off than the hatters indoors that were poisoned by the fumes from heating the stuff up.
Washing it down the drain to where it can end up in small organisms then concentrated into top level predators that people eat is also very bad news.

Only if you do something as insane in hindsight as put lead acetate in the wine as a cheap sweetener. Lead pipes give you tiny trace amounts. Guzzling down cheap vino with a lead based sweetener like the Romans did is a few orders of magnitude more.
The lead pipes myth came from someone who knew about the poisoning but not about the wine so made a bit of a guess - lucky for us a wrong one since there's still some lead plumbing around.

Fair enough, but applying a reasonable degree of reading comprehension instead of just skimming key words before replying is normally assumed.

Sometimes it turns up on servers with a relatively lean RHEL/Centos install and starts running when you don't even have X going - I have killed it on servers when they were running short on memory.

As an aside, a really weird side effect of pulseaudio is that if you block the port it uses it can really speed up remote X - got no idea what Lennart was thinking with that bug/feature. It's things like that which make me wonder why he's being allowed near an init system.

I used system V init on embedded system since late 1990 and I just delivered my first system using systemd this week.

You are a brave man to go in blind. I've been using systemd stuff on test boxes for more than six months and I've seen and worked around far too many fuckups to want to use it on an important production system, but at least I'm prepared to do some workarounds if I actually do.

Since some of them in high positions have been working a lot less time than Slashdot has been around that is far less of a joke than you think.

You say that, but why have nearly all distros moved to systemd?

RedHat staffroom politics and Gnome club politics. It's addressing the non-problem of a bunch of things being under the control of a lot of different people instead of just under Lennart's control. The "faster boot" never happened and was never a big deal outside of systems too small to sanely consider systemd anyway - the old init not doing much is faster than starting the systemd "cathedral" to not do much.