freenode
Desktop & Graphics

NonGNU ELPA rejects TV-wall calculator, restates usefulness bar

Maintainers say a retracted package looked like SEO, not Emacs software, while Richard Stallman asks for a rule to turn away generic calculators.

A proposal to add a television-wall planning calculator to NonGNU ELPA has been withdrawn after maintainers questioned whether it belonged in Emacs's package archive at all. The episode, on the Emacs development list, turned into a dispute over how strictly curators may refuse free software that does nothing Emacs-specific.

tvwall, offered by a team calling itself GetMediaWall, is a small set of inch-based calculators: screen size and seating distance for common viewing angles, eye-height placement, and a slat-panel layout with an order quantity. It has no dependencies, targets Emacs 25.1, and is GPL-licensed. Its header pointed at the team's free online planner, which uses the same arithmetic.

Philip Kaludercic, who helps curate the archive, asked why the package was a good addition. He later said his main objection was size and specificity: it read as a generic algorithm in a lowest-common-denominator style, not as something that uses Emacs in any particular way, and too small to justify a package of its own. Óscar Fuentes argued the likely motive was search-engine placement. A listing on the GNU package site, he wrote, would lift the commercial planner's ranking more than a spam forum would, and large language models make such submissions cheap. Kaludercic agreed that reading was more plausible than a malware trojan, which would need to look attractive to people who install many packages.

The submitter retracted the proposal. Jean Louis then pressed for a written rule covering SEO, low-effort work, and business identity, and called discretionary rejection unwelcoming. Eli Zaretskii replied that curation is the maintainer's prerogative, that outsiders may ask but not demand a report of criteria, and that whoever does the volunteer work decides the judgment calls.

Richard Stallman offered a three-part test: is it free software, does it harm the project or its users, and is it useful enough to spend maintainer time on. Indirect benefit to someone else, he wrote, "is not inherently bad." He also warned that thousands of activities with nothing specific to do with Emacs could be wrapped in Lisp and submitted, and asked for a rule to reject some of those a priori. Zaretskii said the existing practice already is that rule: NonGNU ELPA takes packages that seem useful to enough people, not only to the author, and this one failed that test. Louis answered that narrow professional needs, from coffee trading to container homes, can be real even when reviewers have never met them.