Evil Empire
Microsoft has always followed directly in IBM's footsteps, with perhaps a 20 year delay. So the sooner IBM goes out of business...
284 publicly visible posts • joined 27 Jul 2018
AWS CEO Andy Jassy recently stated with respect to K8s: "I don't believe in one tool to rule the world," hoping instead to keep customers on its AWS-specific container platform
Yes he does. As long as it's *their* tool. AWS is the new Evil Empire, as Microsoft before it and IBM before that.
(BTW, consider that my speaker cable is very visually similar to HOUSEHOLD LIGHT CABLE but it's really high-grade speaker cable HONEST, it's just disguised like that so no one steals it....)
That 100m roll of lamp cord from the home improvement store is the very very very best value for speaker cable, bar none. Anything labeled "Monster" is a ripoff.
In a new build I'd say put in 'dumb' speakers and wire them back to a central hub point.
And for those who can't or won't haul cable through their homes ... a simple FM radio transmitter dongle will transmit to any receiver made in the last 80 years, even those from manufacturers that are long gone. And it costs a lot less than a matrix switch!
Frankly I am skeptical that anyone really needs to have the same signal synced to every speaker in the house anyway. My home has built-in bluetooth speakers all over, and not once have any of us ever wanted to play the same source in more than one area of the house.
I'll see your BA monitors and raise you six NS-10s, driven by Tripath Class D amps. Fed from a custom Linux box with a low-noise D/A, natch.
I'll see your NS-10s and raise you a pair of Minimus-7 speakers, purchased at Radio Shack in the 1980's. Because they'll *still* be sounding great in 30 years when Sonos is a distant memory and none of their products can be made to work anymore.
A colleague installed a bunch of Sonos speakers in a building that already has a centrally managed wifi network. The speakers refuse to log in to the building network; they insist on creating their own mesh. Predictably, they work fine when the building is empty, but when the building is full of people and the main network is active, they stop working. The speakers have been binned. They don't work.
I just built a new PC from all-new parts ... something I haven't done in 10-15 years. So it won't count towards Lenovo or HP's bottom line. It won't even count towards Microsoft's bottom line because it runs Kubuntu. And it replaced *two* older computers, so there goes a small shrink in the PC fleet...
At this point it should be obvious to Mac users that your binaries aren't going to run forever. From M68K to PowerPC to Intel and now to 64 bit only, Apple has made it clear that they can and will change the underlying architecture at any time, and they're only going to provide backwards compatibility for a short time. Are we going to be good for a while now? Of course not -- everyone knows by now that the next move is to ARM; the only question is when.
After living through decades of IBM and then decades of Microsoft, do we really want a third monopoly era, this time of AWS? Anyone in charge of IT at their organization would be foolish to trust their computing to Amazon. I'm not saying don't use cloud; I'm saying don't use Amazon.
Hybrid cloud vendors are starting to point out that moving your IT workloads into AWS is pretty much like moving them into the old mainframe data center. One big dominant vendor who doesn't care about anyone but themselves. You have to use their tools and you have zero leverage with them, and very little interoperability with other vendors unless you layer it on from the outside. Welcome to the new mainframe.
The way to make IPv6 "backwards compatible" would have been to make addresses variable-length. Initially, all addresses would just happen to be 32 bits in length. But after everyone's converted over, then you start handing out addresses that are longer than 32 bits. In 20 years we'd have gotten there by now.
Variable length addresses would have been compatible with existing routing algorithms that look for the "most specific route" and would give *every* node the ability to split its own address into a network. For example, if your ISP assigns you 100.64.10.10, you could subnet it into 100.64.10.10.1, 100.64.10.10.2, etc. and then downstream you could do the same thing with 100.64.10.10.2.1, 100.64.10.10.2.2, etc. And when someone on the other side of the world sends a packet deep into your network, the most specific address on the public Internet is still 100.64.10.10, and it goes to your router.
Think that's silly? Keep in mind that SNMP OID's are assigned *exactly* this way. You get your *one* number, and you can put as many levels underneath it as you want.
The place to begin is by incenting consumer ISPs to move to IPv6-first architecture. For example, if you run Android on T-Mobile you get a native IPv6 address and a fake IPv4 address that is tunneled to a CGN somewhere on their edge. All consumer ISPs should do this. This will incent web service designers to build IPv6-native applications.
The biggest problem with OpenStack is that no one ever provided a *usable* out-of-box experience. To get anything done with OpenStack you need a team of engineers to build it for your environment.
Kubernetes is definitely a better move. Yes, I know that Kubernetes does not provision virtual machines, and those who think that containers are just "a poor man's virtual machine" are missing the point. When you can run an application in a container, you eliminate the *reason* why most shops dedicate a virtual machine to a single application. With the sandbox effect of containers, many workloads can coexist on the same host again.
You'd be surprised how many of the most obnoxious ads can be blocked simply by setting all of Facebook's domains to 0.0.0.0 in your hosts file. Seriously, try it without any ad blocker on and you'll be impressed.
This has the side effect of making your computer unable to access Facebook itself, but that's a feature, not a bug.
Wow. I'm not sure how to feel about this. Stallman has been a tumor on the open source community for decades, definitely outlived his usefulness and has been a liability for a long time now ... but on the other hand, I never like seeing the outrage mob claiming another scalp.
I wonder if the mob realizes how much of an extreme anti-capitalist they took down this time.
Honestly, if a job is so simple that it can be shipped from the first world to the third world, eventually it's going to be simple enough to be automated. Looking at it that way, I'd rather see call center bods have some of their functions moved to automation (yes, SOME of them ... you're not going to take the humans out of the equation, no matter how hard you "digitally transform") instead of having all of them offshored.
Why does anyone still listen to Gartner? For decades, pointy-haired types have been paying big bucks for Gartner to say (in way more words) "current trends will continue". When a real shift is afoot, Gartner is the last to recognize it. It seems the real purpose of Gartner is to give people CYA when their decision-making causes local oblivion - they can say "Gartner advised us that this trend would continue so we invested in it".
If LetsEncrypt can do that, then why can't DigiCert and the rest of them ? Because they're in it for the money, security is just the cow they milk.
Actually, the protocol used by LetsEncrypt can be used by DigiCert and the rest of them. It's an open standard and all open source software. There's nothing stopping DigiCert from selling you 5 years of service and then providing unattended, automatic certificate updates every 90 days, using the exact same software.
Our IT folks are already force-marching us to Teams. Without the Skype for Business (fka Lync) marked for extinction, we had to go to something else. I wish it wasn't Teams, though. I really don't like it. And the fact that there's no on-prem option makes it even worse.
I wonder if they'll eventually do the same to Exchange.
Tried it, wrote a few hundred LoC in it ... gotta say I'm not impressed. For systems-level stuff, C is still king. For portability, garbage collection, type flexibility, etc... why would one want to write in anything other than Python. Go doesn't seem to solve a problem that actually exists.
Back when Andrew Tridgell was building the first version of Samba, he didn't even know that there was such a protocol. He had a copy of DEC Pathworks running, which was a product that allowed DEC machines to act as file servers to DOS and Windows clients. And he basically reverse engineered it, byte by byte, until he had an equivalent server running. There were opaque blobs of bytes that he just reproduced without knowing what they did.
So ... hats off to an excellent hacker who produced something very useful, even before he understood the protocol he was implementing.
This is basically Google's play to move their spyware directly into the operating system kernel (or, since it's a microkernel, a "userspace module" ... hey, how's HURD doing?). But they'll abandon it once they have their own chip fab and can move their spyware directly onto the silicon.
Red Hat seems blissfully unaware that India Business Machines is just going to shutter most of the Raleigh operation and ship the jobs offshore. I spoke with some of our reps and they seem to think that it's Red Hat who will change the culture of IBM, not the other way around. I'd love it if that were true, but somehow I doubt it.
There is no reason to do this. In fact, there is every reason to do the opposite. Chrome should be using libcurl as its networking library. I reworked a substantially sized application to use libcurl instead of its own protocol implementations, and was able to shed tens of thousands of LoC. And at the same time, the networking stack got more reliable (because more people have tested it) and we didn't have to maintain any of that stuff anymore.
The thing that made WSL 1 attractive is that it *isn't* a virtual machine. It's a bona fide API translation layer that does a pretty good job at transparently accessing the host's filesystem, networking, and other service. If WSL 2 is just a virtual machine ... then it doesn't offer any more value than simply running an actual copy of Linux under Hyper-V or VirtualBox or VMware. We don't just want it to run on the same computer. We want it integrated. We want it to be so integrated that future versions of Windows would be able to simply run Linux software out of the box.
If WSL 2 is just a VM with a VHD ... there's no value anymore. I'll just uninstall it and run my own VM.
You can't really support two competing sewage treatment plants because how are you going to route the waste to the one you choose without a whole separate network of pipes? Likewise with water, electricity, or gas.
We're doing it now with electricity. The electricity you pay for isn't the same electricity you get. For every kilowatthour you consume from the grid, your ESCO is producing a kilowatthour *into* the grid. That's why there are separate production and transport costs on your electric bill.
There is an easy solution to wireline monopolies. Simply pass regulation stating that the companies which provide wireline connectivity from a central office to customer premises, CANNOT provide ANY services over those wires. And these companies of course can be heavily regulated, since there is limited space on the poles and in the conduits. Then, the carriers who provide services (voice, video, data, etc) simply build out to the central offices and purchase the last-mile from a company they are no longer competing with. Problem solved, with regulations placed only on the part that needs it.
Hold on there, cowboy. There's a difference between "doesn't know Exchange" and "doesn't hold an MCSE". Holding a Minesweeper Consultant Solitaire Expert certificate simply means that someone knows how to take tests well. It has bugger all to do with the person's actual skill set.