The comment about Rio vs X Windows would make a lot more sense if you could do the same things in Plan 9 as you could in X Windows. Motif was a giant sack of crap, but you could make an application that was roughly identifiable as a "computer application" of the time to a non-technical user, and even customize/internationalize it (more or less). You'd pretty much be hand-painting every last UI widget in Plan 9. Ridiculous.
Also this comparison was often used as a celebration of not having objects, C++, shared libs, etc. on Plan 9. Again, this would be less ridiculous if any of the large complex things done on other systems that merited shared libs had ever been done at all on Plan 9.
Unfortunately, many of the good ideas in Plan 9 sank alongside this. If they had made "APE" a first-class project they wouldn't have been swivelling back and forth from non-Plan 9 boxes to use the web and the better ideas might have grown into something. :-(
> Motif was a giant sack of crap, but you could make an application that was roughly identifiable as a "computer application" of the time to a non-technical user, and even customize/internationalize it (more or less).
Motif is also not part of X11. If you want something vaguely like it, there's libcontrol which is used by a few applications. Nothing about rio prevents something even thicker from being written (or something like GTK from being ported), other than a strong distaste for that kind of gui.
> Again, this would be less ridiculous if any of the large complex things done on other systems that merited shared libs had ever been done at all on Plan 9.
That's kind of missing the point: The complexity doesn't carry it's weight. I can do most things I care about just fine on plan 9. The major exception is a good web browser, which is an OS on its own, and is best when towed beyond the system and isolated with the birds and fish and tons of crude oil.
> If they had made "APE" a first-class project they wouldn't have been swivelling back and forth from non-Plan 9 boxes to use the web and the better ideas might have grown into something. :-(
The better ideas were largely centered around the simplicity you were just bemoaning. If plan 9 had focused on APE, there would be little of value they could have done.
"other than a strong distaste for that kind of gui"
I would suspect that the process of writing a full-fledged windowing system and graphics layer that people might recognize as behaving roughly like a modern computer (even running a web browser, yet), while holding true to the Plan 9 Received Wisdom, would be a intriguing journey of discovery for someone.
Whether they would emerge as a true believer with their faith strengthened by this trial, or as a Plan 9 heretic, would be an interesting question.
Porting GTK would be similarly doable, all that's really needed is getting a GDK port to do the window management, and porting pixman and cairo for the drawing. I'm not aware of anything that devdraw+rio is missing for a working implementation.
And, if you really want, there were multiple ports of X11 to plan 9 that could be revived -- Equis, for example, , so you could in theory just get did of Gnu-isms, revive one of those unused bits of code, and run it that way.
Again, there are existence proofs -- nothing about plan 9 prevents any of what you're talking, other than the people using plan 9 don't want it, and the people that want it don't want plan 9, or are almost entirely "ideas people" that aren't going to contribute a line of code towards making what they want work.
Web browsers are a different story. The problem with a web browser is that a modern web browser is tens of millions of lines of code to implement a turd that's churning at an incredible rate; Last I looked, Chrome was 20 million lines of code, and needed upwards of 600 substatial out-of-tree patches to even build on BSD. And my bank's website breaks on versions below 70 or so, which means running constantly just to stay with a working version.
We don't have resources to maintain that shit.
For a while, there was 'linuxemu', which emulateds the linux syscall ABI, but even that had a ton of churn, and nobody put in the effort to keep up to date.
Given that plan 9 is all volunteer driven (especially today), if nobody writes the code, nothing happens.
> Whether they would emerge as a true believer with their faith strengthened by this trial, or as a Plan 9 heretic, would be an interesting question.
The question is "what do you want out of this"? To me, this sounds like an awful lot of pointless work that makes the system quite a bit less pleasant, with code that's too big to really understand and easily debug, and a UI model that I don't even use regularly on Unix if I can avoid it. It gives me the same reaction as someone proposing making Linux compatible with MVS (IBM's mainframe operating system), complete with JCL and the data set file system.
Also this comparison was often used as a celebration of not having objects, C++, shared libs, etc. on Plan 9. Again, this would be less ridiculous if any of the large complex things done on other systems that merited shared libs had ever been done at all on Plan 9.
Unfortunately, many of the good ideas in Plan 9 sank alongside this. If they had made "APE" a first-class project they wouldn't have been swivelling back and forth from non-Plan 9 boxes to use the web and the better ideas might have grown into something. :-(