关于将Xorg现代化为仅作为协议的图形层的提案
Linux桌面已经花费了十多年时间在新的图形栈上进行过渡。Wayland带来了许多优势，尤其是在移动设备风格的安全性和简洁性方面。但在这个过程中，我们悄然失去了一些宝贵的东西：曾经使Linux图形模型独特的分布式、基于协议、与传输无关的理念。
这并不是怀旧。这些能力对于远程工作、自动化、多机器工作流、瘦客户端、云桌面以及未来的分布式系统至关重要。它们并不是“遗留特性”；而是可能再次变得重要的架构优势。
问题不在于Wayland本身，而在于它从未被设计来支持这些用例。它的理念是故意局限于本地、单用户和以合成器为中心。这对于移动设备来说是完全合理的，但对于桌面和分布式环境来说却留下了空白。
另一方面，Xorg面临的是一个老化的实现，而不是过时的理念。它的核心思想——基于协议的渲染、远程执行、可组合性和传输独立性——依然相关。我们缺乏的是一个现代的、简约的、仅基于协议的继任者，能够保留这些优势而不带有Xorg的历史包袱。
这样的项目不需要复制Xorg的全部功能集。它不需要服务器端渲染、字体、输入法、窗口管理或安全策略。它只需定义一个干净、现代的协议和一个稳定的抽象层。现有的合成器可以实现它，现有的驱动程序无需更改，Mesa也不需要进行重大重设计。工程工作量远小于重写整个图形栈。
这并不是呼吁取代Wayland，而是呼吁承认Linux桌面可能需要不止一种图形模型。一个以协议为先、与实现无关的层可以与Wayland共存，互为补充，并保留那些否则会消失的能力。
如果没有人开始这项工作，行业将自然趋向于移动风格的图形架构，而过去的分布式能力可能会被遗忘很长时间。但如果有人开始一个现代的、仅基于协议的Xorg继任者，社区或许最终会找到一种平衡简洁性与桌面曾经拥有的灵活性——而这种灵活性可能再次变得必要。
查看原文
The Linux desktop has spent more than a decade transitioning toward a new graphics stack. Wayland brings many advantages, especially for mobile-style security and simplicity. But in this process, we are quietly losing something valuable: the distributed, protocol‑driven, transport‑agnostic ideas that once made the Linux graphics model unique.<p>This is not nostalgia. These capabilities matter for remote work, automation, multi‑machine workflows, thin clients, cloud desktops, and future distributed systems. They are not “legacy features”; they are architectural strengths that may become important again.<p>The problem is not Wayland itself, but the fact that it was never designed to support these use cases. Its philosophy is intentionally local, single‑user, and compositor‑centric. That is perfectly valid for mobile devices, but it leaves a gap for desktop and distributed environments.<p>Xorg, on the other hand, suffers from an aging implementation, not an outdated philosophy. Its core ideas—protocol‑based rendering, remote execution, composability, and transport independence—remain relevant. What we lack is a modern, minimal, protocol‑only successor that preserves these strengths without carrying Xorg’s historical baggage.<p>Such a project would not need to replicate Xorg’s entire feature set. It would not need server‑side rendering, fonts, input methods, window management, or security policy. It would simply define a clean, modern protocol and a stable abstraction layer. Existing compositors could implement it. Existing drivers would not need to change. Mesa would not need major redesign. The engineering effort is far smaller than rewriting a full graphics stack.<p>This is not a call to replace Wayland. It is a call to acknowledge that the Linux desktop may need more than one graphics model. A protocol‑first, implementation‑agnostic layer could coexist with Wayland, complement it, and preserve capabilities that would otherwise disappear.<p>If no one starts this work, the industry will naturally converge on mobile‑style graphics architectures, and the distributed capabilities of the past may be forgotten for a long time. But if someone begins a modern protocol‑only successor to Xorg, the community may finally have a path that balances simplicity with the flexibility the desktop once had—and may need again.