Jim Hall Posts the PD-DOS Announcement That Became FreeDOS

On June 29, 1994, Jim Hall posted “ANNOUNCEMENT OF PD-DOS PROJECT” to the Usenet group comp.os.msdos.apps. The message proposed a public-domain DOS-compatible system, referred to a manifest and implementation task list, invited direct mail from prospective developers, and asked to be connected with any existing effort. That post is the founding event from which FreeDOS dates its anniversaries.

The announcement followed discussion, not a finished code drop

Hall wrote that he had raised the idea months earlier and received strong support, including the challenge to “start writing.” He did not announce a complete kernel or claim to have solved DOS compatibility. He announced an organized attempt and a list of work that needed contributors.

The message separated implementation from debate. Hall invited developers and people with concrete suggestions to contact him directly, while leaving broader arguments about the merits and morality of cloning DOS to the public discussion. That was a practical way to turn a polarized newsgroup topic into coordinated work.

The post also included a fallback: if such a project already existed, Hall wanted its leader identified so he could contribute. PD-DOS was therefore not framed as a personal brand that had to displace every alternative. Its stated goal was getting a freely available DOS effort moving.

Microsoft’s direction supplied urgency

Hall had begun using Linux but still returned to MS-DOS for word processing, spreadsheets, games, and other applications. In his later first-person histories, he recalls magazine interviews saying the next Windows release would do away with standalone DOS. He wanted to retain the DOS command-line environment and reasoned that Internet collaborators could build one, just as Linux contributors were building a Unix-like system.

This account should not be converted into the false claim that DOS vanished when Windows 95 shipped. Windows 95 retained an MS-DOS layer for startup and compatibility. The issue in 1994 was Microsoft’s product direction and the future of DOS as an independently maintained environment, not an immediate inability to boot existing software.

PD meant public domain—and carried a PC-DOS echo

Hall used PD-DOS because he initially expected the software to be in the public domain; he also says the abbreviation deliberately resembled IBM’s PC-DOS name. He soon learned that public domain and free software under a license were not equivalent governance models.

Public-domain dedication would allow reuse but would not require modified distributions to preserve source freedoms. Hall preferred the GNU General Public License used by Linux because its copyleft terms would prevent someone from taking the shared code, distributing a modified binary, and withholding the corresponding source under GPL conditions.

Within weeks—described in Hall’s accounts as another week or so, or by late July—the project became Free-DOS under the GPL. Later it dropped the hyphen and became FreeDOS. The historical evidence supports that sequence; an exact July 26 rename should not be asserted without the corresponding primary follow-up message.

Contributors answered with existing utilities and new components

Hall shared DOS command-line utilities he had already written, including replacements or extensions for familiar commands. Other participants brought their own tools or searched public FTP archives for code whose redistribution terms fit the project.

Tim Norman volunteered to write the command interpreter that became FreeCOM. Kernel development eventually centered on DOS-C, originally written by Pat Villani and continued by other maintainers. Those components demonstrate why the announcement was effective: it supplied a shared target around which independent programs could become an operating-system distribution.

Hall’s 25th-anniversary history says Alpha 1 arrived less than three months after the announcement. By Alpha 5, the distribution included more than 60 utilities, and by October 1998 it had a working kernel. The route to 1.0 in September 2006 was long, but public releases began in 1994.

Compatibility required interfaces, not copied source

The project aimed to run software written for MS-DOS, which requires matching documented behavior such as interrupt services, program loading, filesystems, device handling, command syntax, and batch conventions. It did not have Microsoft’s MS-DOS source code available under an open-source license; that historical material reached GitHub under MIT only in 2018.

Reimplementing an interface does not guarantee identical behavior. Programs may depend on undocumented quirks, specific memory managers, proprietary drivers, timing, or a particular DOS version. FreeDOS compatibility has therefore always depended on testing real applications and hardware as well as implementing the public API.

The manifest and task list made the proposal actionable

A broad declaration that “someone should write DOS” is easy to endorse and difficult to execute. Hall’s announcement said he had divided the objective into a manifest, outline, and task list showing what needed to be written. That created entry points for contributors who could build one command or subsystem without first designing the entire OS.

The packaging model that emerged reflects that modular start. FreeDOS distributions contain a kernel, command shell, installer, drivers, utilities, applications, games, help, and development tools with separate maintainers and versions. Integration work converts those pieces into bootable media; no single repository is the whole user experience.

Why June 29 remains the founding date

FreeDOS could plausibly commemorate its first alpha, first kernel boot, FreeDOS 1.0, or a later stable release. Instead the project marks June 29 because the public announcement created the collaboration. Its 25th anniversary in 2019 and 30th anniversary in 2024 both used that date.

That choice distinguishes founding from completion. On June 29, 1994, PD-DOS was a proposal with early utilities and a work plan. Within weeks it had a license-informed name; within months it had an alpha distribution; over later years it acquired the kernel, shell, packages, and installers associated with FreeDOS.

The lasting news is therefore specific: one Usenet post turned scattered interest and personal DOS tools into an identifiable development project. The technical system came afterward, but the coordination mechanism began that day.

Related:

Sources: