As someone else pointed out, it just feels so unnecessary. So much overhead, security risk, and added expense. It’s difficult to believe there isn’t a company that specializes in kiosk systems running on some stripped down Linux distribution.
A question to consider seriously: name a company that has a full OS that supports modern tooling/development environments with consistent graphical fidelity across a wide range of hardware that a manufacturer can pay to maintain the host OS, provides guarantees to OS LTS/security patching and has a proven track record in deploying, supporting and delivering kiosk support.
The only serious answer is Microsoft, and maybe Canonical… But Canonical hasn’t been around for as long as most of these kiosks have.
There are a couple of huge blockers for manufacturers looking at companies that provide Linux support:
Industry track record. Red Hat, Canonical, Google and Oracle are basically the only large scale players in the enterprise Linux support. Red Hat basically only provides support for server/backend infrastructure. Has Google had anything other than Gmail and maps last for more than five years? So that leaves us with Canonical. What’s the longest release Canonical has? 4 years now? Microsoft has 15 year support contracts. The only other player in the market that even comes close is Oracle (Oracle still supports Java 1.4 for example: 22years)
Consistent graphical performance: until the last 5 years graphical fidelity on Linux has been a shit show. A decade ago, getting even the largest players to support Linux was a huge undertaking. Basically the only consistent graphics support was the result of android and that is basically only mediatek.
Development environments. Windows wins this hands down without even a question. Go back 15-20 years and it’s even more obviously in Microsoft’s favor. NET gui apps are brain dead easy to make, super consistent and stupid easy to maintain. This drastically decreases development time and cost allowing companies to pay for the crazy expensive support contracts.
The numbers these companies deal with isn’t thousands or even hundreds of thousands of dollars. It’s tens or hundreds of millions. There is no way in hell a manufacturer is going to give an untested bespoke Linux distro maintainer 25 million to keep that Linux distro running for the next 10-20 years. There isn’t a feasible way for a small company to even support at that price for that length of time.
Oracle and RedHat are the only truly feasible options, and it costs more to develop GUI apps on either platform when there isn’t a 20 year track record of known success. It’s obvious why companies pick Microsoft.
Many used to (pre windows ce), but writing the whole stack was more expensive than license+support costs.
Many still do, but they aren’t full fledged kiosks. By the time you get to full HD screens, the cost of the chips needed to refresh the screen in a reliable way outpace the cost of going standard consumer electronics. Cost for parts/replacement is also lower that way. This dovetails into needing an OS that supports those chips, which suddenly we are into a full OS.
Who are they going to pay to maintain FLTK? There are still companies that are adverse to using Linux because they don’t know what is going to happen when Linus dies. That might sound strange to us, but companies need legal protections that they can enforce through contracts and support contracts make that happen.
The laggy bit can be explained this way: all of these decisions are made because in theory this all sounds “right” (to the company) but then they get their prototype out with a medium level hardware solution and they look for places to squeeze. Oh, you mean I can take this half price min spec machine and it works 98% of the time? Sold.
Im not trying to say these are good practices, I am trying to explain the decisions that are made.
There is probably an opportunity in this space to provide ultra low cost single board SPA/elctron serving applications. But getting it adopted is going to be an issue.
A good industrial engineer is going to look at it kinda suspiciously. Kinda like how Tesla got rightfully raked over the coals for trying to use consumer grade electronics in cars and their screens started melting as well.
Why would MC Donald’s need a windows PC. I would have imagined these were on some free software like Linux and their own custom wrapper or app on top. When you’re deploying potentially millions, it would seem like quite a saving.
Wonder how much they pay per kiosk. I doubt they wrote this GUI as system-dependent to be bothered with Win. Probably just doesn’t trust the word ‘free’.
Don’t quote me on that but I think I remember a McDonalds menu screen where the taskbar was visible and the menu was just a webapp opened in Google Chrome. I imagine the same goes for the ordering terminals.
I don’t know about those in big franchises with payments processing, but small individual projects like interactive sceens in exhibitions usually do, as they don’t need anything but a couple of pages, galleries, a simple update mechanism etc. Maybe this method was used there too.
More than likely though the activation system just shat itself. If you’ve ever managed a large farm of windows machines with legitimate KMS activation you know shit breaks for no reason at all way too damn often.
Shit, that happened with my home pc, and they acted like they were doing me a favor by giving me some code to make the software I had already paid for work.
Add comment