I don't normally continue with these type "debates", but you make some pretty ridiculous claims ...
.. one of their C8051F04's (with fancy adc) ... it turns out most of these bugs are in the hardware. Randomly corrupting sfrpage registers, corrupt can data, missing adc cycles.
I write my own kernels, and code exclusively in assembler. Then I run trace points on DSOs and monitor memory/IRQs/processing times.
To date, after 35+ yrs, I have rarely had any products exhibit the problems you have, including the Silabs range from their origins.
Maybe you (or your programmers) should learn to program better and not rely on free C-compilers. Heck, even expensive C-compilers aren't perfect.
And that goes for ANY MPUs free/otherwise C-compiler. YES, they have a LOT of internal hardware and a "smart" programmer would know how to
keep that in mind and work with it.
Bottom line - If user1 has NO issues and user2 does have issues, well .. work it out. Don't trash cores just because you can't get it to work, it's
not very professional. And YES, the free Keil compilers do have "bugs".
..... pushing the envelope on these "antique" cores.
Nothing wrong with the cores. Learn assembler IF you want to PUSH the MPU. Same with ANY MPU. That's supposedly your job to figure out.
... there is always something better...
What the HECK does "better" mean. You chose an MPU for a job. Should I use a P7 cause it's better?? That's just a plain dumb statement. Sorry.
... They do not like relays. If you have a (din-rail) relay near, it'll crash the debugger ....
A lot of dev systems don't like arcing relays, or bad earths or noise or .. or .. YES, I can get the debugger to crash, as I could with ANY of my dozen
or so $10K - $20K emulators / dev systems. YES, some were more robust, but I paid a shltload for them. On some projects where I do get a few
more crashes than I'd like, I put better filtering / grounding .. a bit of rewiring and all is acceptable again. Again, that's "our" job to figure out.
All yours.