For developers building on Linux, BSD, or Solaris, the Executable and Linkable Format (ELF) is the invisible bedrock. It's the standardized container for executables, object code, and shared libraries that enables interoperability across Unix-like systems. Yet, the history of ELF's specification reveals a surprisingly turbulent journey—one marked by corporate acquisitions, legal warfare, and an ongoing struggle for clear stewardship.

The Birth of an Open Standard (That Wasn't Always Open)

ELF emerged in the late 1980s from Unix System Laboratories (USL) as part of System V Release 4 (SVR4), with Sun Microsystems contributing the dynamic shared library system. Its intent was openness:
* Key papers like "ELF: An Object File to Mitigate Mischievous Misoneism" (1990) detailed its design.
* The 1993 "System V Application Binary Interface" book formalized it.
* The Tool Interface Standard (TIS) Committee (a consortium including Intel and HP) adopted ELF in 1995, publishing the TIS Portable Formats Specification v1.2 under a royalty-free license. This version effectively placed the core spec in the public domain.

This openness fueled widespread adoption. By the mid-90s, Solaris, IRIX, HP-UX, Linux, and FreeBSD had all migrated to ELF, seeking binary compatibility (as highlighted by the 86open Project).

The Tangled Web of Stewardship: SCO, Lawsuits, and Frozen Docs

The TIS Committee dissolved post-1995, leaving ELF's future tied to the volatile ownership of Unix assets:
1. Novell acquired Unix assets in 1993.
2. The Santa Cruz Operation ("old SCO") bought the Unix binary licensing business in 1995 (Novell retained copyrights).
3. Caldera Systems (later "The SCO Group" or "new SCO") acquired SCO's Unix assets in 2001.
4. UnXis (later Xinuos) purchased the assets from SCO Group in 2011.

Maintenance of the System V ABI (and thus the ELF specification within it) passed to each successive owner. Key figures like Dave Prosser (USL -> SCO -> SCO Group) and John Wolfe (SCO Group -> Xinuos) managed updates. However, active development stalled:
* The last significant functional update on the official SCO/Xinuos ABI site was SHF_COMPRESSED in June 2013.
* Wolfe departed Xinuos in 2015, leaving the spec effectively unmaintained.
* Copyright notices listing SCO/Xinuos with "All rights reserved" created confusion and concern, implying control over a collaboratively developed open standard.

This stagnation coincided with The SCO Group's infamous lawsuits against IBM and the Linux ecosystem (2003-), alleging misappropriation of Unix technology. While largely unsuccessful, the litigation cast a long shadow and hindered cooperative evolution.

Innovation Persists: The Community Fills the Void

Despite the frozen official spec, evolution continued via community effort:
* The generic-abi Google Group emerged as a neutral forum for OS and toolchain vendors (primarily GNU, LLVM, and Oracle Solaris) to discuss extensions.
* Key modern features achieved consensus here but weren't formalized on the official site:
* RELR Relative Relocation Format (2018): Significantly reduces relocation overhead.
* ELFCOMPRESS_ZSTD (2022): Adds modern Zstandard compression for sections.
* Processor-Specific ABIs (psABIs) for architectures like AArch64, x86-64, and RISC-V remain actively maintained by their respective communities, independent of the core ELF spec's stagnation.
* OS-Specific Extensions flourished, notably for Linux (e.g., SHT_GNU_HASH), often adopted by others like FreeBSD, blurring strict OS boundaries.

The Road Ahead: Decoupling ELF from its Past?

Efforts to secure a stable future continue:
* A preliminary agreement between toolchain veteran Cary Coutant and Xinuos was reached in 2020.
* In August 2025, Coutant published gabi.xinuos.com and the gabi GitHub repo, signaling a potential revival of centralized documentation.

The fundamental question remains: Should the core ELF specification be decoupled from the broader, historically entangled System V ABI? The success of community-driven psABIs and the generic-abi group suggests a path forward:

"In practice, achieving consensus among major toolchain vendors (GNU and LLVM) may be sufficient... While aligning with Solaris would be ideal... this might not always be feasible due to varying priorities." (Source: maskray.me)

ELF’s resilience lies in its elegant design and the vibrant ecosystem it enabled. Its future relevance depends less on corporate ownership and more on the continued collaboration of the developers, maintainers, and OS vendors who rely on it daily. The frozen specification on SCO's site is a relic; the living standard evolves in compiler commits, psABI documents, and mailing list discussions.

Source: Based on research and analysis from maskray.me's comprehensive overview of the ELF format's evolution.