| While I definitely agree that Sun, HP, and SGI had amazing engineering this section of your comment is particularly ridiculous: > I cannot name a single GNU/Linux "hacker" inventing anything. Everything they implemented is a shoddy copy of what the engineers at [...] did. If we ignore the obvious counter-arguments (GCC, Apache, nginx, arguably git, and so on), and just look at your statement at face value -- you're saying that nobody from the largest free software community in the world has invented anything new in the past 10-15 years. What possible evidence do you have that this is true? It's not just wrong, it's offensively so. Heck, ZFS-on-Linux is now the repo-of-record for ZFS -- that's a specific example of the GNU/Linux community implementing things alongside the rest of the free software community (and currently ZoL has more features than any other ZFS port). It's also the case that those brilliant engineers did the same thing. Zones are arguably a re-implementation of Jails with some slightly different design goals. IOCP is objectively just a copy of kqueue (to the point where Cantrill said he wished they'd just ported kqueue). And so on. I don't think this is a bad thing at all, but it's quite strange to put them up on a pedestal to the point where you effectively say that any engineer who didn't work at Sun/HP/SGI in 2003 hasn't invented anything. There are obviously plenty of examples of Linux not learning from others (Jails/Zones vs containers, kqueue/IOCP vs epoll, DTrace vs eBPF+bpftrace, and so on). But arguing that nothing innovative has come out of the GNU/Linux community ever is just awful. |
...and Wikipedia, BitTorrent, the WWW (open-source, though developed on NeXTStep), the Objective-C compiler that made it possible, Emacs (and its idea of an extensible self-documenting editor), Perl, bash, the dpkg/apt system the app stores ape poorly, the CTAN and CPAN systems it derived from, most aspects of decentralized source control in the form of arch (and then later Mercurial and Git), rsync, Docker, Nix, GNU make, Python 3, gold (the linker), iptables, Enlightenment, the Hurd, MySQL, PNG, Pango, Numpy, IPython/Jupyter, Ruby, the X11 Render extension, LADSPA, Valgrind, asan/ubsan, basically everything on CPAN, all the inventions in x264, ...
I'm not sure Apache belongs in there, though. Rob McCool might take exception to being called a "GNU/Linux hacker", though I don't think Robert Thau would mind.
> There are obviously plenty of examples of Linux not learning from others (Jails/Zones vs containers, kqueue/IOCP vs epoll, ...
Oh c'mon, is the problem that Linux doesn't innovate enough or that Linux innovates too much? Epoll is an example of Linux arguably innovating too much.