Article: The Coming Software Apocalypse

[recondas]

Computers had doubled in power every 18 months for the last 40 years. Why hadn’t programming changed?

https://www.theatlantic.com/technology/ ... de/540393/

[HurricaneJames]

Computers had doubled in power every 18 months for the last 40 years. Why hadn’t programming changed?

https://www.theatlantic.com/technology/ ... de/540393/

This article is a wonderful bit of sensationalism written by people who didn't really understand what they were talking about. I love it!

[commodorejohn]

Yep, that is some grade-A "article about programming written by someone who knows nothing about programming" right there.

[pentium]

All I got out of that was that we rely far too much on automation for simple tasks, either to save money or because people are lazy.

I mean, I get that is true in say, IoT.

Otherwise modern programmers suck. Their teaching sucks. Their code sucks. I attended a university programming class where optimization consisted of one day in a semester long class. That isn't automatic when you compile you senile twat. Teach us properly how to make something in 16mb of ram not lose its mind when it realizes it only HAS 16mb total to work in.

*waves for teacher, who comes over to investigate*
*points finger at C code that will not work and says "DUEHH! DUN WORK!"*

[vishnu]

Just pass -o3 to the compiler and bingo, your code is optimized!

[ClassicHasClass]

I have a lot of IoT devices here for home automation, but they live behind a very strict firewall that cannot route to the Internet directly, and I manipulate their APIs directly instead of using whatever client they offer.

[commodorejohn]

I mean, if you want to bitch about the million zillion things being automated that probably shouldn't be, I'm totally on board with that. (Remember, kids: the only difference between a Tesla and a 5,000 lb. 150 mph. crowd-seeking motorized death missile is a clever hacker!) But fucksake, that is not a problem that will be solved by whatever voodoo bullshit New Future Revolution In All Programming Forever these dipshits are peddling.

[Trippynet]

Computers had doubled in power every 18 months for the last 40 years. Why hadn’t programming changed?

I read that one line and instantly disagreed with the article. Computers have not doubled in power every 18 months for about the last 12 years or more. My main PC is from 2009. New PCs are not 32 times more powerful than my system.

Probably not the point the article is making, but when there's such a blatantly incorrect line staring at you...

[ClassicHasClass]

New Future Revolution In All Programming Forever

The problem isn't that the ideas aren't neat - I like a God Mode Mario programming environment as much as the next guy/gal/whatsit. But most of my programming projects do not easily adapt themselves to that metaphor. I could see something like that being useful for certain kinds of domain-specific programming, but not as a general purpose programming language.

[uunix]

Humans are human, we can't think of everything, even lots of us put together, although the first story was rather short sighted on the developers side since they had a no catch. Saying that, that's as far as read, it seemed like doom and gloom I've got hindsight type of story.

[guardian452]

This is a kind of cute story. Almost every word out of his mouth is complete bullshit, but I'm sure he'll get some clicks out of it. It smacks of "software engineer who knows nothing about industrial controls complains web-dev practices are unsuitable for developing industrial controls".

Incompetent developers are just as great at fucking up projects with Mathworks or Dassault as they are with C. In fact, because the worst spaghetti code you can imagine can be multiplied out to multiple drawn flowcharts without any consequence, even your $185/hr genius contract engineer can't understand his own code two hours after he wrote it. Which is about how long his program takes to compile.

There is a saying, "to err is human. To really fuck things up, you need a computer." Model-based programmers just push this same maxim out to the 3rd dimension.

Not saying it isn't a valuable tool, or that there aren't a lot of really smart people out there doing good work with it. But in my experience the same dumb monkeys typing on fancier typewriters isn't going to make anything better. You just get dumber monkeys.


And because I'm sure somebody will eventually ask...

I use a lot of IEC 61131-3 structured text... looks a lot like pascal at first glance but it is extremely verbose and clunky. Our runtime (Codesys) supports some basic model-based development like ladder logic and function blocks, but I don't know anybody that uses anything other than ST.

Our GUI (dashboard) is Java based, and our smaller controllers are just AVRs.

Our control platform is SIL 2 https://en.wikipedia.org/wiki/Safety_integrity_level and we use the required features (watchdog, secure outputs etc)
Almost all automotive controllers are similar. They are certainly not in a race to the bottom, but ICE controllers haven't needed more horsepower (so to speak) since 32-bit chips came out in the 90's. Because we make exclusively EV's, our requirements are much lighter. Admittedly, I don't think you could control an ICE in Codesys. Our RT is 5 ms.

I think you can download Codesys for free and play with it but without a hardware support package ($$) from a controls vendor it is quite useless.

[vishnu]

Probably my all-time favorite book on programming is "The Capability Maturity Model, Guidelines for Improving the Software Process" by a variety of contributors from the Software Engineering Institute at Carnegie Mellon. We're CMMI level 5 appraised at my place of employ, which was a laborious process to achieve but, since our customer required it, we somehow managed to pull it off. However, given the level of ineptitude some of our new-hire programmers exhibit I think our top managers are buying the appraisers some fancy dinners in order to keep our appraisal at that level.

[pentium]

I have a lot of IoT devices here for home automation, but they live behind a very strict firewall that cannot route to the Internet directly, and I manipulate their APIs directly instead of using whatever client they offer.

I've always been a strong opponent to IoT. For as much as I throw around the quote:

Science isn't about "why?", it's about "why not?"

...we already have a strong understanding of the layer that abstracts how IoT actually works as a standard and it's not pretty. The marks of a consumerist product are all over it (if the cloud service that sits your phone and your lightbulb gets switched off the device no longer functions as it should) and the implimentation that tacks it to the standard is riddled with flaws that have the potential to cause real damage that isn't at the fault of the buyer. For example that stint where IoT devices were assimilated into a botnet and proceeded to carry out some of the largest recorded Denial of Service attacks.

Now of course, this is the part where I shamelessly plug X10 however when you look past its total lack of signal encryption and slow unidirectional communications the simple fact that it is not by default an internet connected standard in the 21st century immediately makes it drop off virtually everyone's radar for malicious intent.

[vishnu]

Now of course, this is the part where I shamelessly plug X10 however when you look past its total lack of signal encryption and slow unidirectional communications the simple fact that it is not by default an internet connected standard in the 21st century immediately makes it drop off virtually everyone's radar for malicious intent.

I actually built Steve Ciarcia's X10 controller from his book "Build Your Own Z80 Computer." I wire wrapped the whole thing, I still have the memory module sitting on my desk at work. Built the whole thing on Vector 6" by 4.5" 44 signal bus boards stuffed into a card cage, it worked well...

[guardian452]

The pain I feel when, say, a computer or tablet, etc, suffers a crash or mandatory update is bad enough.

I can't fathom my rage if a light switch or door lock were to stop working.

I've heard the lutron-branded ones are about the best it gets (Caseta) and they are still supposedly unreliable, and come with the bonus of turning on 100% after a power loss. They are better than the garbage from e.g. philips or insteon, but that's not saying much.

I have some dimmers with an RF remote mainly because I'm too lazy to get out of bed to turn off the light. They are just as reliable as the $1.19 single pole switch from home depot.

I have no desire to have this stuff online.

We bought an ecobee tstat at the office and it works great, I'm thinking about getting another for home, tho we don't use much electric on HVAC. The prox sensor per room and humidity sensor per room makes a big difference. One of the rare vendors who's product "exists to be used" rather than "exists to be sold". We got a nest for free from the electric co. years ago and left it at an old apartment because it couldn't even function as a thermostat properly let alone the gee-whiz online stuff. I'm shocked they're still in business, especially after such wonderful sequels like "smoke detector that doesn't detect smoke" and "security camera that isn't secure".