On Mon, Dec 10, 2018 at 1:49 PM Hendrik Boom hendrik@topoi.pooq.com wrote:
On Mon, Dec 10, 2018 at 12:45:42AM +0100, Paul Boddie wrote:
Of course, one could look more closely at microkernel-based systems for a possible remedy. Sadly, ever since the famous Torvalds versus Tanenbaum discussion, plenty of people cling to the remarks of the former as he sought to ridicule the work of the latter, oblivious to the fact that...
- Microkernel performance was always a tradeoff (acknowledged by the DMERT work done by Bell Labs in the 1970s and in other contemporary work).
- Performance has improved substantially over the years and in some cases wasn't that bad to begin with, either.
- Billions of devices have shipped with microkernels.
Some people also probably cling to the idea that Torvalds "won" his debate. Now that MINIX 3 runs in every Intel CPU supporting Management Engine functionality, it is clear who actually won, at least in terms of the "bottoms on seats" measure of success that the Linux kernel developers tend to emphasise over things like GPL compliance by vendors (some of those vendors being Linux Foundation members, of course).
Just curious -- what microkernel systems are available to run on modern home computers just in case one is tired of Linux and wanting to try something else?
SE/L4. one research group actually created a complete minimum-compliant POSIX subsystem on top of SE/L4, absolutely nothing to do with any operating system "per se", and then successfully ported an entire qt-based webkit browser *and all its dependencies* to run on it.
the "filesystem" was entirely flat. no subdirectories. so when i say "minimally compliant" it really really was "minimally compliant".
l.