KernelTrap

» see kerneltrap.org as it lived, on the wayback machine

~ the obit ~

Jeremy Andrews spent the 2000s reading a mailing list that most of the people who depended on it never opened. The Linux kernel mailing list was public from the start, which meant anyone with a browser could watch the operating system under half the internet get argued into existence, and almost nobody did, because LKML in a normal week is a few hundred messages of patch review written by people who assume you already know what they are talking about.

That public-but-unreadable quality is what I remember about kernel development from the university years. I was a computer engineering student in Ankara in that window and Linux was already the thing you ran because it was free and because the lab machines had it. The arguments that decided what the scheduler did were sitting in an open archive the whole time, in a register of English that took an afternoon to parse one thread of. Distance here is not geographic. It is that the raw source was available to everyone and useful to about four hundred people.

The Internet Archive's first capture of kerneltrap.org is dated 24 January 2001 and shows a general systems-programming news site with a tagline about going down to the metal and a front page of links about the FCC and the Matrix. The kernel focus came later, and with it the format the site is actually remembered for: a story was a short summary of a thread, followed by the full text of the messages that mattered, with threaded comments underneath.

It ran on Drupal, which Andrews was deep in, and from May 2005 the Oregon State University Open Source Lab hosted it for free, the same lab that kept a lot of open-source infrastructure alive in those years.

~ the rap sheet ~

~ reading lkml so you did not have to ~

The editorial decision that made KernelTrap worth reading was refusing to paraphrase. A summary of a technical argument tells you who won. Reprinting the messages tells you what the argument was about, which is the part that ages well, because kernel threads are where you find out that a design everyone now takes for granted was somebody's unpopular idea in 2004.

It got cited accordingly. Robert Love's Linux Kernel Development referenced it, so did Christopher Negus's Red Hat Linux 9 Bible and a SUSE manual and at least one academic book about measuring software quality. That is a strange bibliography for a one-person Drupal site, and it is the clearest evidence of what the site was for: it was the citable version of a conversation that otherwise only existed as raw mail.

~ the year of silence ~

The site went quiet for a full year and then came back at the end of March 2010. It posted for twelve days and stopped for good on 12 April. There was no closing announcement. Wikipedia's talk page notes that the last entry cached in the archive is dated May 2013, because the site kept serving that final front page for three more years while nothing moved.

~ the Torvalds interview ~

Andrews went on to found Tag1 Consulting, a Drupal and performance shop. In 2020 he published a long interview with Linus Torvalds on the company blog, covering Linux and Git at length, and it reads exactly like a KernelTrap piece. The questions are short and the answers run for paragraphs, with nothing trimmed for a general reader.

Since May 2021 kerneltrap.org points at that interview. The domain of a dead site now serves one piece of the thing the site used to make, on a consultancy's marketing blog, which is a softer ending than a parked lander and a stranger one. The nine years of archives are in the Wayback Machine and nowhere else.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »