Sources and Method

Every page on this desk rests on documents written by other hands, and the difference between a fact and a rumor is mostly bookkeeping: who stated it, on what page, and on what day that page was read. This page sets the bookkeeping out, including the cases where it gets awkward.

A source is a document, not a project

This desk is not the Overclockix project and serves none of its files. It holds a record of the project, and that record is built from documents by different authors: a version table on DistroWatch, a code repository last pushed on May 31, 2015, and an archive index of this domain. Each is cited as what it is. The release record describes the documents, the documents describe the project, and the distance between the two is kept on purpose.

Why does every figure carry a date?

"Last release 7.8.0" on its own is decoration. "Last release 7.8.0, dated June 1, 2015, stated by DistroWatch, on a page whose own last update is February 8, 2026, read on September 5, 2026" carries handles, and each handle does a different job. The release date belongs to the software, the page update belongs to the publisher, and the read date belongs to this desk. Drop any one of the three and the reader can no longer tell a statement about a project from a statement about the source that described it.

When two sources disagree, the date decides

The LiveCD List called Overclockix active in a card updated December 21, 2013. DistroWatch lists the project as discontinued. Both statements survive because both are dated, and neither has to lose for the other to win. The same version table places 3.79, released August 25, 2005, after 3.8, released April 2, 2005, because it orders by date rather than by number. Reprinting that order without the reason would produce a column that looks wrong and is not.

What a bot check does to a citation

One fetch made while preparing the storage pages came back as a document titled "Making sure you're not a bot!" with a loading notice and no content behind it. That capture says nothing about the tool it was meant to document; it says the fetch was stopped. The handling is plain. Nothing is quoted from a page that said nothing, the gap stays visible, and any claim that was going to lean on it is rewritten to stand without it, or dropped.

What the memory pages borrow from the Memtest86+ site

The memory section leans on the Memtest86+ project site for statements this desk cannot make on its own authority: that the tool is free and open source under the GNU GPL v2.0, that its v6 code base began as PCMemTest by Martin Whitaker on the v5 line of Sam Demeulemeester, and that it is not an edition of the closed-source Memtest86 owned by PassMark since 2013. The same site states that its nightly builds are generated automatically and are fully untested, which is exactly the kind of sentence a stability page needs and cannot invent. What this desk does not repeat is the download table or the changelog. A reader who wants the v8.10 list of fixes reads it there; a reader who wants what a memory pass covers reads it here.

How does a correction happen?

A correction replaces the sentence and keeps the trace. ArchiveOS recorded this domain as not active in a card updated August 23, 2023; the statement stays, held to its day, and is never quietly updated into a description of this site. When a figure turns out to be wrong or a read date is missing, the page changes and the change is dated. A reader who finds a claim this desk cannot trace can write to the desk. The rest of the method fits in one line: date the figure, name the source, and when a document is silent, say so in print.