"Venerable" does not quite mean "decrepit". The meaning is more like respect for an elder. "Official standard" or not, it has been around as a real language since the early 90s (Borland?), which definitely makes it an elder in computer land. I mean, Windows is written in C++ and it definitely gets the "venerable" title by now.

I'm apparently obsolete in my thinking :)

My familiarity is mostly with older installs and the problems in keeping older infrastructure working. There is a whole cottage industry built around that very problem... lots of things out there like serial to Ethernet converters. You can slowly modernize this kind of a system, but at some point you have to start moving things over to a more modern controller.

If we are talking control systems, then the best bet is to go browse the popular offerings from Schneider or Siemens. Those will be around and supported for a very long time, and you won't be the only idiot out there building a factory with it. You naturally want to update things - that's the whole point of this question - but it is not like you are going to be continuously upgrading the core architecture over the 25 year life. I would not be surprised at all if the same architecture is in place for 25 years at a single installation. Sure, you'll swap out parts and maybe even the controller a few times - but it will all look very familiar to the guy who originally spec'd it, and new people will roll their eyes and laugh at it.

If it is something in IT land, that is out of my field. There are many suggestions on here - but I'm betting that anything popular (Linux, Windows, BSDs, mainframes) will run in emulation going forward. I'd probably avoid Oracle, Apple, or other proprietary hardware - even though I told you to go with proprietary stuff for the factory :)

I think we're talking past one another. The original poster is asking about future-proofing his development, not freezing development. Naturally you want the system to be adaptable and as easy to update as is possible - but the basic architecture is probably going to be pretty much frozen in time. "Upgraded" equipment is usually compatible with the same interface, and they tend to make popular interfaces for a very long time. Even when the interface becomes obsolete, there is usually a very long period with transitional hardware. At some point - unless you are lucky - you are looking at replacing the controller, and that can be quite painful so it makes sense to (a) put that off as long as possible, (b) spend time up front picking a popular and flexible platform.

Hey, I hang out with a lot of creative people. Not Elon Musk, but Steve Jobs for more than a decade, and lots of people at least as smart that you don't know. They can be really brilliant, and successful, and they can still make really stupid mistakes and sell them to the rest of us pretty well because they believe in themselves completely and they have a track record. I've done that too.

That's the hyperloop. Something Elon never meant to stand behind (and still really isn't), just put out there to torpedo a worthy project that he didn't believe in.

Anyone who looks at the hyperloop design can see it's not a no-brainer. It has safety issues up the wazoo :-) It's going to take a long time to get right.

Meanwhile, little Switzerland can have incredible trains everywhere and the United States can't get it together, and unlike with rockets and cars Elon's not helping this time. And I am not sure that the "lease" part of his solar business is a great thing for the world either.

My young kids don't even ask for the TV - they only want the tablets. We sometimes reflexively put something on TV for them, and they often ask if they can't watch it on the tablets instead.

Heh, don't buy stock in that company.

I'm doing a very minor thing at work right now in the same vein. We have some ancient equipment that would cost $60-70k to replace. It still works, but the data collection PC just died. It had a workflow involving macros, Windows 2000, and several serial ports. I need to get it all working with the Win7 replacement - but at least it is all simple serial communication and the equipment seems amenable to USB adapters, and the communication is well-documented. There are two stations, so it isn't exactly time-critical - but I totally get why people would not want to spend money until absolutely necessary.

Although the hyperloop is possible and might even be practical someday, let's please be honest about the reason it was created. Elon Musk just wanted to kill the California high-speed rail.

That might have been OK if there was a hope that we could actually replace it practically with a hyperloop. But given the history of bleeding-edge rail - ride any maglevs lately? We haven't even had much success with monorails outside of theme parks and Las Vegas - we don't really have any working system to replace high-speed rail. Hyperloop should really be called "Pipes that carry People" and we need decades of work on it before considering intercity lines.

I don't think many people would wait until their last spare to start retrofitting their system. At the same time, you want to stretch your investment as long as you can get away with it.

In the case of old style PLCs, there have been a number of transitional technologies, since so many people were in the same situation.

This is good advice. You want a whole bunch of other people to be in the same pickle as yourself. That means there is a market, and where there is a market there will be a vendor catering to very old system support.

That depends on the application. If he's making an industrial control system, then no, he probably will not be maintaining it organically. It will get built, qualified, and then expected to run for the life of the process. Think nuclear plant... what is more painful, re-qualification or running obsolete tools? Plants built in the 80s (power, sewer, etc) are still running DOS control systems with ancient serial PLCs.

That's brilliant, actually.

Linux is at the heart of many embedded devices, most smartphones, and a whole crapload of servers. Given the staying power of golden oldies like COBOL running on mainframes (or virtualized mainframes), I don't think that there is any doubt that the Linux kernel will be around.