On Tue, Dec 15, 2020 at 02:11:45PM -0600, Nicholas Geovanis wrote: > I smile at this discussion, having gone through this same thought-cycle so > many times. > I claim this is motivation for the separation of a graphical client > workstation, simpler and separate from the server, having ONLY a > single-user identity at a time. In other words, re-implementation of the > XWindows approach. > I think that identity, access and security friction as we see in this > discussion will eventually push the pendulum back in that other direction. > Except for those who dont mind the in/decrement in complexity and > responsiveness in an installation that provides both. In the data center > this separation is enforced for many years; X-Windows is not installed on > servers. There's almost never a need for it there and it presents another > attack surface.
Agree, in general. One nitpick: it's "X", "X11" or "the X Window System". Calling it "X Windows" is giving in to The Reptile's influence. [Sorry. I'm a bit of a language nerd. I've already asked my doctor, she says she can't do anything about it] I think what happened (somewhere between 1990 and early 2000s) is, that with everyone having a computer on her desk, the idea of separate users became blurry. Desktop environments had one "real user", and she was sitting right there, next to the graphics hardware. Unix got many things right from the beginning, because this multi-user paradigm was deeply rooted in there from day one: battle-tested by students keen on playing pranks, with people using the system /at the same time/ (*GASP!*) from different time zones. Whoever has seen time handling in Windows during its period from Windows 3.1 to Windows 95 and XP has had a lot to laugh. And to weep. So the desktop environments started aping Windows more and more, and, although the underlying system was inherently multi-user, they became kind of single-user. Watch all that "multiseat" nonsense of the time (promoted, afaik, by some bigger Linux "vendors" [1]). Then, with the take-off of the browser and Active-X 3.0 [2] (aka Javascript), multi-user becomes interesting again, but for different reasons. If you do some archaeology, you'll find that many things have been reinvented in some strange, sometimes perverted way [3]... and slapped on top of the existing layer. Cgroups vs. process groups; systemd's socket activation vs. (x)inetd; and so on. The future? Usually I'm called an optimist, but this time... sorry. I think we'll be back to that terminal thingie: people will have a browser on their desktops/pockets/retinas; that browser will be written in node.js and be totally controlled by Google, and will be running on dedicated hardware. It will be "open source", mind you, but it will be so fiendishly complex that you'll need some AI-driven SDK whose training data will be Google's and Palantir's trade secrets to even stand a chance to hack at it. Know what? I'm glad I passed my 2^6 th year. It was my last power of two ;-P But then, I meet so incredibly smart young people, who surprise me in (well, duh!) unforeseen ways... this, again, keeps me alive and curious :-D Cheers [1] It might seem I'm bashing here early Linux vendors. Not at all: Red Hat and Suse did and do a lot to keep the lights on for many a cool hacker. OTOH they are capitalist enterprises in the classical sense (i.e. the money dictates where things go), so they pick up some of the antipatterns of the trade. Things are as they are. [2] Active-X 2.0 being Flash. Java applets never got an own version, sorry ;-) [3] Not saying that there wasn't an itch to scratch, mind you. There sure was. Often, though, you'll also witness some amount of "second system syndrome" [5] [5] https://en.wikipedia.org/wiki/Second_system_syndrome - t
signature.asc
Description: Digital signature
