- Today's world is amd64, armv7, and soon aarch64. Everything else is dead, Jim. Noone is investing enough money and brain power in the other architectures. Only a few people actually know about the rest of the ecosystem (mips, power8...), and noone gives a shit anyway.
- Keeping an obsolete platform alive is fun because this reminds you of the '90s, when there was a large choice of hardware platforms, with roughly similar cost/power rations. Eventually, the cheap PC killed almost all competition, and the smartphone market gave ARM an unhealthy market share in the embedded systems world.
- Then, after a while, it's not fun anymore, because noone is writing code with your platform in mind, because it's not deemed powerful enough, because modern compilers no longer support your platform (or they produce broken code for it, which is even worse). Does your platform have a hardware limit of a few hundred MB of physical memory? You won't be able to run a web browser or even a PDF viewer on it. Does someone still run today's gcc's testsuite on your platform? No? Sorry dude, here's a nickel...
- Keeping a platform alive is a real team work. This requires a serious commitment from all the ``building block'' projects: a not-too-bug-ridden toolchain (as/ld/gcc), as well as support in the flagship projects (emacs, python, X11, mozilla, libreoffice...), and accurate, up-to-date documentation available free of charge.
- None of this is true of platforms, except for amd64, armv7 and aarch64.
- Because of this, trying to keep a platform alive is really going against the tide.
- Have a look at all free operating system projects: they really only support amd64, armv7 and aarch64. Gee! Sometimes they pretend to support a few other experimental platforms. Or other platforms which have not been tested in years and are only cross-built because they are not self-hosting anymore.
- Of course, you'll still find a few looneys which will do an incredible amount of work to prevent the decay of their platform of choice, and give the impression that these platforms are still first class citizens. But these guys are the same as ten years ago. And eventually, they get tired and give up. Just like me.
- The worst part in this, is to look back and realize that, after all those years, free software has lost. Companies working on non-free software have been smart enough to get the momentum of the free software developers to work on embedded platforms in order to tremendously shrink they software engineering costs, yet getting positive press.
- So, today, we're building disposable systems, without any consideration to future products, because when the time will come, we'll pick the next trendy SoC. No need to have what makes an architecture: a vision towards the future allowing different hardware generations to share hardware and software; a design allowing today's software to be able to run on tomorrow's hardware as long as the changes are not too drastic; consistency in choices of busses, key chips, address maps; a real firmware, not something utterly disgusting as EFI which can't even compete with > 20 years old Open Firmware (IEEE 1275); and reliably enumerable busses. No, sir, we'll take today's latest SoC, wire a few extra devices wherever we can find some hole in the address map, put a castrated u-boot on it, and today's stable Linux tree, and here you go, here's your ``platform''. And people cheer.
- I've spent 20 years in my life trying to promote a certain view of free software, its values and its ethic, fighting for it, getting involved, trying to lead by example, and on this day, I get the feeling that all I did was wasting my time and that nothing I did has been useful.
- It's a hard pill to swallow.
- Better resign now than keep trying and only get bitterer in the process.
RAW Paste Data