Twenty Old X.Org Components See New Updates

[cj.wijtmans]

Another Phoronix thread, in which old men are shaking fist to sky because someone is trying to replace their beloved ancient software with something that meets today's needs.

Or rather grown men that never grew up from their angsty rebellion teenage phase trying to fix something that was never broken and ending up with something broken by design anyway.

[Myownfriend]

Or rather grown men that never grew up from their angsty rebellion teenage phase trying to fix something that was never broken and ending up with something broken by design anyway.

You can't really called X11 something that "isn't broken" and call Wayland something that's "broken by design". When the design of X11 prevents it from properly supporting multiple displays with different refresh rates and DPI scales, that kind of infers that X11 is broken by design, at least for the kind of base-line features expected today.

[Hibbelharry]

Or rather grown men that never grew up from their angsty rebellion teenage phase trying to fix something that was never broken and ending up with something broken by design anyway.

Except.... x.org is broken in many ways. A codebase deemed hardly maintainable, nulling any further progress. Large parts of it's core functionality being unused for years (no one uses x11 primitives and stuff like that anymore...), usable and really used parts being glued together extensions. Problems with mixed DPI, problems with hotplugging displays, problems with tearing, security unknown for bigger parts of it. Even the praised X11 forwarding is slow and inefficient as hell. Yes, it works on broadband low latency links, but fear any lesser connection....

[ryao]

Another Phoronix thread, in which old men are shaking fist to sky because someone is trying to replace their beloved ancient software with something that meets today's needs.

You have it backward. The problem is that people want to replace something that meet’s today’s needs with something that does not.

[ryao]

Except.... x.org is broken in many ways. A codebase deemed hardly maintainable, nulling any further progress. Large parts of it's core functionality being unused for years (no one uses x11 primitives and stuff like that anymore...), usable and really used parts being glued together extensions.

This could be said about any mature codebase. It might even be true if said about glibc and gcc. In GCC’s case:

https://gcc.gnu.org/backends.html

In glibc’s case:

https://www.etalabs.net/compare_libcs.html

That is before even reading the source code. glibc suffers from many levels of indirection that make it very difficult to find the code for library functions to understand why your program is crashing in localtime(). With GCC, it is amazing if you can find your way around that source tree and understand the code.

Problems with mixed DPI, problems with hotplugging displays, problems with tearing, security unknown for bigger parts of it. Even the praised X11 forwarding is slow and inefficient as hell. Yes, it works on broadband low latency links, but fear any lesser connection....

Half of these are not problems as far as I know. As for the half that is, you can find a list of limitations for anything, including wayland. To use the GCC analogy, GCC has recently annoyed me by being terrible at compiling GNU C vector code into SIMD instructions when the vectors are a multiple of the hardware SIMD width. It also cannot generate a vpmovzxdq on its own needed to load 32-bit values into 64-bit AVX2 register components. The result of that alone was a 40% reduction in fletcher4 performance. It suffered even more from the other issue (likely a performance collapse due to it emitting several instructions for every 1 instruction LLVM emitted), but I did not benchmark it.

Life would be easier if the world switched to LLVM/Clang. Oh wait… LLVM cannot turn byteswaps into vector permutations without __builtin_convertvector when using a 64x8 vector in GNU C, but GCC can.

[luno]

which Distro has iso without any X11 bits which Wayland doesn't need. like there are distro without systemd ?

[Myownfriend]

You have it backward. The problem is that people want to replace something that meet’s today’s needs with something that does not.

What are you smoking? lol

Half of these are not problems as far as I know.

In what ways are they not problems?

[kpedersen]

Another Phoronix thread, in which old men are shaking fist to sky because someone is trying to replace their beloved ancient software with something that meets today's needs.

Not at all. Very happy to see continual updates to appear in Xorg to keep it meeting today's needs. i.e Big changes such as replacing the old setuid and TCP code with a more secure alternative (UNIX domain sockets) has historically been very welcome. As is removing XDMCP from xdm.

Poor Solaris and BSD guys having to keep a dead horse walking...

Having a solid cleaned up Xorg (Xenocara) is one of the best things about OpenBSD! Linux guys have never experienced that. Matthieu Herrb is a prominant figure in Xorg so has made sure the focus on his platform of choice has been good.

Solaris is similar. Alan Coopersmith from Oracle is a lead Xorg developer. He keeps things working very smoothly.

Again, I can see why Linux wants to move to their own display system (and I agree with it). They have always had the real shite end of Xorg for a long time now. Being a third class citizen isn't nice. Perhaps the reason why Android went with their own niche display system (SurfaceFlinger) is similar, generic Linux is no different.

I think the only non-Linux platform that is caught up in the mess is FreeBSD. Their Xorg implementation is full of random Linux-style dependencies and shim layers. Hopefully once Linux fscks off on their own tangeant; Xorg can finally be cleaned up for FreeBSD and the Linux hacks stripped out.

[asriel]

I've tried several times to use wayland. And every time returned back to Xorg. Wayland still looks like a piece of crap. it works more or less stable only with gnome3 - which I hate by look feel and design, and if you try to use it with something different from gnome3 - get ready for bugs. And absolutely no benefits. None. Zero. Null. Wayland creators promised performance, power efficiency etc - nothing is there.
With all the money RedHat invested in this teenage mutant technology it should fly us to the Moon for free already - but it still can not catch up with old software where nobody invest a penny. How can that be? The only reason I see logical - that is is so bad by design that even with all the money invested in it is is still in such poor condition.

[dreich]

I think the only non-Linux platform that is caught up in the mess is FreeBSD. Their Xorg implementation is full of random Linux-style dependencies and shim layers. Hopefully once Linux fscks off on their own tangeant; Xorg can finally be cleaned up for FreeBSD and the Linux hacks stripped out.

The kernel code for FreeBSD's input driver evdev is a Linux port too, no? So it is not just their Xorg implementation that will need to be cleaned up.

But regardless, Linux today is at its best as far as open source graphics go, supporting both X11 and Wayland. With more work maintaining X11 and improvements in store for Wayland, the situation for Linux will keep improving.