Re: Surely ARM is the problem?
ARM is not a problem unless you have x86 binaries you want to run.
MS's C/C++ compilers can cross-compile ARM-native code. The problem is actually DOING it.
If you restrict your shipped applications to x86 amd64 and "CLR", *THAT* is the problem!
Unless WIn-10-nic on these ARM devices is being limited to "store only" there should be no issue, except that ARM code is currently a bit slower due to 3Ghz RISC and not 4Ghz CISC.
[CISC uses fewer instructions, so in theory SHOULD be faster, but will run hotter and eat battery more quickly]
NOT being able to easily change the battery means that it's expected to only last 2-3 years. For the typical investment in a Surface, that's a DEAL BREAKER with me, other than the Win-10-nic and any proprietary stuff that tries to STOP me from running Linux or FreeBSD on it.
/me does FBSD and Linux on RPi, no problems there. Assuming similar ARM core, maybe slightly better, OBVIOUSLY faster, you'd think you'd just have a better performing version, right?
Underneath the hood, the ARM and MIPS and other non-x86 systems typically use a different method of figuring out what hardware they have [OFW/FDT for RPi]. ACPI has ruled x86 and amd64 for decades, but is occasionally a problem with non-windows OSs. I do not know what is happening with ARM/Surface but I expect it's ACPI and/or similar, secure boot, locked down, cannot run THE OS OF YOUR CHOICE on it without a 'jailbreak', which is (in MY bombastic opinion) *THE* *BIGGEST* *PROBLEM* *WITH* *THESE* *MACHINES* !!!!!
Oh and the battery needs to be REPLACEABLE. The cost of a 2-pin inline connector is TRIVIAL and a properly designed one would NOT take up enough physical space to matter. MS: Just void the warranty for opening the thing up and replacing the battery, and nobody will mind. Seriously. We _WANT_ to be able to replace the fastest wearing out most commonly replaced component.