"Could Arduino have been imagined and designed with 8-bit PIC MCUs to begin with, instead of AVR ones?"
Very good question. I've wondered that too, and in my own particular case, it really came down to AVR having a free and libre working development environment I could use on my Linux machines. (I haven't had Windows on any of my machines since 2005 or thereabouts; and between 1997 and 2005, dual-booted only because I had expensive commercial software (Adobe Photoshop, Macromedia Shockwave Internet Studio) without alternatives and could squeeze more value out of generic Intel/AMD hardware than I could of similar investment in Mac hardware.)
I don't know if Linux matters in this picture –– it might, because so many interested in software development outside the Windows environment end up having Linux and/or BSDs installed or dual-bootable –– but probably doesn't. I believe the Arduino developers focused on WinAVR, and don't know their early history to know if any/how many of them used Linux at the time at all.
I suspect that in the high performance computing crowd, the split between CUDA and OpenCL folks is the same 'free' that made the decision between PICs and AVRs for me.
One is a proprietary, closed, single-provider environment; the other is less polished, some ways even 'not as good', but fundamentally not strictly controlled by the hardware manufacturer or vendor.
The mass appeal, however, probably stemmed from the other 'free', i.e. 'no cost'.
So, I propose that the 'free' that directed the developers of the Arduino environment, was different to the 'free' that made Arduino appealing to the masses,
but both are required for such an environment to appear in the first place.
The prices of AVRs and PICs meant either one would have been similarly appealing to the masses, but to the
developers who created the environment in the first place, who wanted the freedom and avoid vendor lock-in and vendor-enforced boundaries, the two were light years separated, with an obvious choice –– and I suspect they made the same choice I made for the same reasons I did.
In particular, even from the beginning, the Arduino developers knew that even if Atmel tried (and their track record showed they wouldn't, but Microchip/PIC might) to force them to reject any competing microcontrollers, it would be completely and even legally safe to ignore such requests, because of the clear legality and provenance of the base projects. (Atmel supported AVR GCC ports, where Microchips attitude is... problematic to say the least, up to straight up lying to their customers what they are legally and contractually allowed to do, namely remove any license-based restrictions on their GCC-based compilers.)
[When Microchip bought Atmel, I was saddened, because I knew I would from there on do stuff
despite the vendor, rather that
with the vendor, because I am just not the type to sit in a walled garden producing officially sanctioned stuff using only officially sanctioned materials. I'm a monkey; either the walls break, or they're covered in poo before the day is over.]
Purely based on observing Microchip business practices, I do not think anyone could have done Arduino based on Microchip products.
This is not based on anything to do with PICs themselves; it is just an extrapolation of observed business behaviour, to attempts of creating an Arduino-like environment around PICs.
Indeed, even now, Microchip is trying pretty damned hard to steer their customers away from any multiplatform solutions, directing them to the Microchip-controlled Walled Garden instead (what with their proprietary compilers and Hardware Abstraction Libraries and so on). Just go look at their
Arduino page. I do believe my characterisation of Microchip business practices is correct and warranted, although they are only based on external observation by a hobbyist; I do not have any of their internal memoes or directives.
I would go so far that if anyone trying to create an Arduino-like environment dedicated to PICs
right now, should be prepared to deal with Microchip lawyers and business folks. (Things like "trademark dilution" is something you'd need to talk about with a lawyer before even considering such a project, because if you cannot publish it without infringing on registered trademarks, it will be pretty hard for even interested people to
find it in the first place.)
If you disagree, let me direct your attention on the authorship on existing Arduino cores for different architectures. Most of them are the work product of individuals without any real links to the hardware vendor, with many having some very nice contributions from engineers working for various hardware vendors but not in any kind of controlling capacity, just submitted like they were by anyone else; with specific cores having a suprisingly tightly controlled authorship... I'll leave it for you to find out which cores that might be, or indeed if any (since it depends on what you consider "surprisingly tight control").
Also, may I ask that instead of stating disagreement based on beliefs based on the scent of jasmine in the wind and calling me a "copyleft zealot commie" or similar, please check if your beliefs are in agreement with the abovementioned, easily verified and observable claims first.
I am
not claiming that free/open source is better, because I do not think so – I personally
like it more, but that is like liking vanilla over chocolate or something, a matter of a personal quirk and nothing more – but that for anyone intending to build something intended to be used by others to build something, the libre-versus-vendor-controlled proposition is, and because of copyrights and software licensing,
has to be important. (Also, basically all software developers whose work product spans multiple operating systems and hardware architectures, has been bitten by vendor lock-in at least once. It ain't fun. So, some of us tend to avoid vendor-locked environments because the risk of that bite at the end makes any temporary gains just not worth it.)
So, I say Arduino came about on AVRs because they were libre, and Arduino itself became popular because it was basically no-cost. Microchip PIC fulfills the latter, but not the former, because of how Microchip the company runs its business.
No value judgments here, only an understanding of
why it happened the way it happened.
(Yes, I'm UTTERLY fed up being labeled a zealot, just because I
very slightly prefer one thing over another, and am completely honest about it and my reasons. Whenever GPL or copyleft licensing comes up, it seems that it so occupies the minds of some that they can no longer fathom that someone could actually just
only slightly prefer it over others and at the same time be completely relaxed and willing to use any other license or business model as long as it makes sound business sense. Put bluntly, I've been happy to work on code licensed under one of at least a dozen different licenses spanning from CC0-1.0 to GPL to official secrets covered by an NDA; and will be happy to do so in future too, regardless of the particular license as long as it makes sense in that particular context – I don't like things being done stupidly, because it is inefficient. Efficiency beats ideology in my opinion. Tux is my mascot, not my idol.)