Re: @Vic @Ledswinger
[re: obsolescence: project development lifecycle longer than component production lifecycle]
"I would point out that the situation you describe is entirely normal in aerospace projects..."
It may be common practice but it is neither necessary nor universal.
In thirty years or so I've worked on design and support of flying products based on everything from 16bit embedded micros to 16/32bit micros to full 64bit flyable workstations built around COTS technologies.
If the right purchasing decisions and agreements are made early enough, the bought-in stuff can match the lifetime of the project. Obviously that is likely to cost more than picking a cheap and cheerful component with a limited lifetime, but with outfits like (say) Raytheon (to pick a relatively well known one) able to militarise a commercial workstation for use as (say) a flying radar operator station with a multi-decade support lifetime, lots of obsolescence problems go away. The budget goes up, but hey... Not sure if they still do now that PCs have supplanted SPARCs and what have you, and are frequently regarded as consumables, mind you. But that's relatively recent.
Similarly, 68000-family chips were still around last time I looked, as were the software and hardware tools to support them. Z8000, less so. Z8000 fell off the gcc supported list many years ago (and rightly so, Z8001/Z8002 were an abomination to program).
Yes it's a challenge. It needs thought, and planning, and budgets.
But it's not always impossible.