IIRC ZFS 2.4 will do parallel zpool imports, so that should help a bit, especially with a big server.

I think half my boot time on a big backup server is LSI device enumeration - which is totally not parallel - so there's only so much one can do.

Could the mptsas devs parallelize this task? It would certainly make lots of people happy.

It's also possible I could fiddle with systemd dependencies to get me a login shell before all the pools are imported because those pools aren't needed at all for the root filesystem. They'll be used minutes to hours later when a backup is initiated.

On my media machine some services like jellyfin need certain filesystems up before it starts but I don't need jellyfin to be started before sshd comes up. Etc.

We're still a ways from having a good tool to automate these dependency trees and most sysadmins are "meh, I can wait two minutes". It's certainly not the ideal for remarkably fast computers.

Also being starved for PCI lanes on pretty much every system is a pain. I'd take lanes over more cores or more GHz any day of the week. Let the data flow!

Why bother? All you need to do is to DNAT at an intermediate point and identify a nearby doughnut shop as a repair facility. The F35s will then land there and you can reprogram them whilst the pilots are stuffing themselves.

Oh. No. Wait. That's cops. Cops don't use F35s (yet).

Always always always remember that Loki can beat Odin.

It depends. If the cats of the world decide to rise up in revolution (which, given the quality of catfood, is entirely plausible), then being able to control all the vacuum cleaners would be the only defence.

> Why was it removed in the first place,

Per Rene:

IMHO The removal of XAA was a huge cooperate planned obsolesce mission for older GPUs. Rendering everything mostly unusable slow. even for period correct X11 apps. The code should have just been left in peace and only bugs and security patches applied instead of outright deleting it for no good reason.

Here's a commit for XAA support in T2 Linux for others who are interested. I hope Rene has time to push it up to XLibre since it seems like the Xorg people are going to steamroll Wayland if they can and the XLibre fork will be the only surviving X11 server. Obviously it would be best if every distro could run on older hardware and Wayland is likely a poor choice for vintage computing.

I didn't know about T2 Linux and it really looks fantastic - I thought NetBSD was my only choice on some of those machines. Some of the screenshots feature WindowMaker, the spiritual successor to NeXTStep, which ran on an '030 and 2D video so this all makes perfect sense.

Those machines were perfectly usable and we can actually afford, today, the amount of RAM they used.

The claims against the archive owner are wild and would be easily disproved if untrue.

Is this the same operator who would block readers if their ISP used some DNS feature he didn't like, back in the day?

I understand being disagreeable, but, jeeze, this takes it to a whole new level. Way to have people's sympathies and then burn it all to the ground with malice.

Wikipedia was apparently in the position of being forced to amplify the attacks with their links to the archive. Not a supporter of theirs these days but what else were they to do?

The IMF uses the proper definition.

I think we can trust those who actually have access to the data to have represented it accurately. If you have a problem with the results, then that's not a fault with the definition.

Nobody has discussed cold fusion seriously in many decades. Hot fusion is catastrophically under-funded (the total spent on fusion research globally in the lat 60 years is about the same as spent just on subsidies for the fossil fuel industry every three days, to give you a perspective on how expensive energy work actually is even for fuels that are simple and well understood).

If you spent as much on fusion yearly as you spend on fossil fuels yearly, then fusion will be cracked before 2030. If you underfund it, relative to the complexity of the problem, then convergence is guaranteed asymptotic.

Fossil fuels are globally subsidised to the tune of $11 million every minute, according to the International Monetary Fund. That money has to come from somewhere, and there's no way in hell a billionaire is subsidising some soccer mum's SUV. Which mean that the money has to come out of taxes.

So it's not particularly cheap, net. It's just that the total cost is diffused across the food you buy, the house you own, the car tags, the money you earn. All of these different taxes contribute some percentage of the cost of the coal and oil. However, collecting and distributing the money isn't free, which means that you're actually paying MORE than you would if you were paying honestly.

Still, if people want to pay more and get less, and die young as a result, that's really their business. Of course, they're making other people die young, too, but that's a democracy for you.

It wouldn't be so bad, but there are hardly any lumberjacks in the UK.

*runs away and hides from an irate mob of Monty Python fans

Seriously, it very much depends on the area. Rugby, a town-borderline-city, has fewer pubs than the Marple/Mellor collection of villages up in t' norf. This is mostly because Rugby is a run-down dump with a dying town centre and hardly anything left in it, whereas Marple (although it lost its engineering back in the 60s) is a major commuter/retirement town with just enough rational people to keep the businesses vibrant and alive.

And that's what keeps pubs open. Not the economy, but the attitude.

As music evolves, it has tended to become simpler, more repetitive, less original, and basically BORING AS ALL F.

(And those who know me on Slashdot, I think this is the third time I've used that sort of language since the site came into operation, which should tell you something about just how bad modern music is.)

If the only music out there is, honestly, turgid, then having it AI-generated simply eliminates the brain-damage induced by having to memorise and play these excuses for songs.

You cannot blame people for skipping the middleman when the middleman honestly doesn't do any better of a job.

Yeah, I fully understand, not everyone wants to listen to 22-minute metal anthems about the universe (even if it does feature Richard Dawkins and fireworks), or indeed 18 minute songs about exploding air balloons, even if I am of the personal opinion that said people should seek help. And people are going to like what they like.

But if you're going to object to AI music, then the only way that's ever going to work is if you reverse the trend and make songs that have sophistication that AI cannot match.

Personally, I have no problem with electronic music or even music wholly manufactured through complex electronics, and regard Delia Derbyshire and Daphne Oram as polymath geniuses that really should have been respected in their lifetimes, but I'd also argue that they actually made an effort to do precisely what I'm describing. They did not, as a rule, make stuff that was simple, unless ordered by higher-ups to do so. It would not be hard to mix their techniques with modern synthesiser ideas and generative systems to produce much more sophisticated music of decent quality.

The Beatles were capable of producing an album over a weekend.

If they'd had the inclination (and assorted supplies best left undiscussed), they were more than capable of churning out 52 albums a year. Whilst we must be grateful for small mercies (it would likely have had an impact on quality), I would argue that the 60s were not short of new music.

Or time literacy. Permanent!=3 years

I'm tempted to do a recompilation of an old MCC distribution, just to mess with people now. :P (MCC was the very very first Linux distro.)

I found a few that are more recent than the one I saw.

https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fportlandpress.com%2Fbioc...
https://royalsocietypublishing...
https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fwww.mdpi.com%2F2075-1729...