> pushes the pixels to the X server by calling a function in libX11.
That's how X is used today. The system was designed and implemented completely differently from today's use. The server had fonts and the client asked to get characters drawn. Or arcs. Sending pixels was a big exception for rare cases and it was far too inefficient for the machines of the 1980s and 1990s.
Nowadays the main usage is over a little corner case of the protocol and implementation. The bigger part of code is just collecting dust, bitrotting, a maintenance burden, probaby a vector for security attacks (though not very bad ones, because they typically require local access to the workstation computer). For this reason X is being replaced by Wayland for over 10 years. The speed of adoption seems to be "twice as fast as IPv6"... As IPv4 continues to do the work for the majority of Internet usage, X continues to do the work on the majority of Linux desktops.
> The speed of adoption seems to be "twice as fast as IPv6"...
I don't know. I received an IPv6 address at home since a few years ago and I've been using it and pulling a lot of traffic since then. On the other hand, it's extremely unlikely for me to ever switch to Wayland on my desktop in the foreseeable future because I need some niche and legacy features that are not supported by Wayland. Network transparency and VirtualGL is a big one, it allows me to run an unmodified X application in a 2D-only X server while outsourcing all the 3D graphics to a 3D X server on a separate machine.
> received an IPv6 address at home since a few years ago
Right, and the specs were more or less complete in the first half of the 1990s. So it took 20 years. And you are still a minority, global adoption is below 35% years later. Here in this country you can't really get fixed Internet connections for consumers supporting IPv6 at all. So I don't have it. (Although we have cheaper and better mobile broadband than nearly everywhere else and there the majority of the operators supports IPv6 by default. So when I want to play with IPv6 I just use mobile data.)
Wayland became somehow useable for those who really wanted a bit less than 10 years ago, Fedora bot not many others use it by default now.
That was my reference to twice the speed, 10 instead of 20 years. Of course it's a rough oversimplification, details are much more nuanced. My point was really lack of speed in both cases.
I already saw your point initially, I remember browsing some IPv6 specs with a timestamp of 1999. My point was that at a personal level, I had every reason to use IPv6 and it was only a matter of whether it's provided, thus I started using it immediately after it was made available. However, speaking of the Wayland migration, I have personal reasons for staying on X and not to run Wayland in the foreseeable future, which is why I said my migration to Wayland is probably going to be slower, not faster, than my migration to IPv6.
That's how X is used today. The system was designed and implemented completely differently from today's use. The server had fonts and the client asked to get characters drawn. Or arcs. Sending pixels was a big exception for rare cases and it was far too inefficient for the machines of the 1980s and 1990s.
Nowadays the main usage is over a little corner case of the protocol and implementation. The bigger part of code is just collecting dust, bitrotting, a maintenance burden, probaby a vector for security attacks (though not very bad ones, because they typically require local access to the workstation computer). For this reason X is being replaced by Wayland for over 10 years. The speed of adoption seems to be "twice as fast as IPv6"... As IPv4 continues to do the work for the majority of Internet usage, X continues to do the work on the majority of Linux desktops.