Copyright and removal

Release notes, changelog entries and the text of security advisories stay the property of whoever wrote them. Collecting a note here changes nothing about who owns it, and Traceary claims no ownership over any of it. Product names and marks belong to their owners, and are used here to say which product a page is about.

What Traceary does own is its own work around that text: the writing on pages like this one, the design, the collection code, and the computed layer sitting over the archive, which is the index, the version bands, the counts and the matches between an advisory and the release that fixed it.

An archive of release notes is worth having only when the notes are the ones that shipped. Summaries drift, rewording introduces claims the publisher never made, and a version page that paraphrases cannot be cited. So the wording is left exactly as its author published it, and nothing is summarised, reworded or translated.

Every page names the publisher and links back to the original entry, 36 products carry a link to the source they were collected from, and the source itself is named on each product page. Nothing here is placed behind a paywall, and no advertising is sold against it.

A rights holder, or someone authorised to act for one, can have an entry taken down. There is no form to fill in and no argument to win. Write to [email protected] with:

• the address of what should come down, either a product page or a single version page
• the project or company the material belongs to
• whether you want one version removed, or the product removed from the archive entirely
• a line saying you hold the rights or act for whoever does

A reason helps but is not required. A request from the publisher of the material is acted on rather than assessed.

The entry goes into a register that both halves of the system read. The collector stops fetching it and deletes the copy it already holds, and the site drops it from the pages, the feeds, the catalog, the security page and the sitemap. The address then answers as not found.

That register matters more than it sounds. The collector re-reads every public source every five minutes, so deleting a file on its own would put the entry back before the day was out. Recording the removal is what makes it hold, and it holds while the publisher keeps publishing the same note.

Removal reaches the live site on the next deployment, which is minutes rather than days. What cannot be promised is anything outside this domain: copies already taken by search engines, archives and caches elsewhere fall to whoever runs them.

Nothing has been withheld so far, so the register is empty. It is checked on every collection run either way.

Removals are reversible. If an entry came down on a request that turned out to be wrong, or the publisher later wants it back, write to the same address and it is restored on the next collection run.

Nothing is collected without a public, machine readable feed that the publisher chose to publish, and no page is scraped. A publisher who would rather not be in the archive at all can say so and never be added, which is the same request handled before it becomes one.

The reverse works too. How the data is collected covers what qualifies a product, and suggestions go to the same address.

The terms of use cover what the archive claims to be, the privacy page covers what it collects about readers, which is nothing, and the imprint says who publishes it.