Recently, the software world was rocked by the discovery of a backdoor in XZ Utils. The backdoor represents many stories, but behind all of them is the story of a single Postgres developer, vexed by a relatable problem: "why is it taking so long to login?!"
On the next Oxide and Friends, @ahl and I are thrilled to be joined by that developer, @AndresFreundTec, the discoverer of the XZ backdoor. Join us on April 8th at 5p Pacific to hear this extraordinary story!
But I think that the only thing worse than stack ranking under strict hierarchy would be stack ranking under no hierarchy -- which would be absolute hell.
Bad news: I'm travelling today, so no Oxide and Friends from @ahl and me this week.
Good news: It probably would have just been an episode on gnuplot anyway -- rivaling in popularity only my short-lived idea of an entire episode on Moby Dick.
@hyc The software that makes our offerings work is written -- and open sourced -- by us. For the bedrock on which we depend (e.g., Rust and illumos) we are active and responsible members of the community.
This IEEE Spectrum piece is absolutely unhinged -- and by the same absurd logic, we would have banned BBSs in the 1980s because kids (🙋♂️) were using them to find copies of The Anarchist Cookbook.
@amcasari 💯 These absurd fears are used to distract people from the real (pedestrian!) dangers. I expanded on this in my talk debunking AI-based existential risk:
On @mipsytipsy's excellent piece, the one bit that I would add to it is that there is a deep seated problem among management: they often think of their career in terms of the number of reports in their org. This is dangerous thinking: like a software engineer that thinks of their career solely in terms of the number of lines written, the desire for more reports above all else leads to deeply suboptimal results!
@happyborg@billjings Please watch to the end: I do not, in fact, dismiss unacceptable risks -- and in fact, I think that the "x-risk" fear is problematic exactly because it allows the very real risks of AI to be summarily dismissed. (As for it being a strawman: I wish it were so!)
Still thinking about the time I was debugging a process stopped by mdb with @bcantrill, and he mentioned continuing the process with some command (I think ::cont in mdb). I was like "oh, that's like kill -CONT" and he reacted with "we absolutely DO NOT use signals to do debugging on illumos"
@lanodan@rain They are honestly just different ways for stopping and resuming a process; the reason to separate them is to allow them to co-exist. For more details, I would point you to seminal work of the late Roger Faulkner: https://www.usenix.org/memoriam-roger-faulkner
Two decades ago today, DTrace integrated into the operating system. Much has changed in the last 20 years -- but one thing has remained true: we ourselves still use it on a daily basis. With your forgiveness, a short thread... 1/
Every year on the first day of school, we have the kids pose for a photo and indicate their new grade with their fingers -- but this year we needed some neighborhood help for our high school junior