Notices by Matthew Garrett (mjg59@nondeterministic.computer), page 2

@lina @nishi Huh you appear confused could you be more explicit about the things you claim

@lina @nishi It's true I'm merely dating four women rather than one woman :(

@lina @nishi Oh I'd say I'm sucked more than I suck

@lina @nishi you have no idea how much I fuck

If anyone is confused about 18F: imagine an in-house consultancy who could show up and solve your problems but who would then be able to use that knowledge to solve someone else's problems without billing them for the prior work, saving everyone money and also making a bunch of it open source. People took pay cuts to work there. Destroying them is a tragedy

@lina k

@lina I'm married to a trans person this isn't new

@lina suddenly? I've had this explicit position for like a decade

Free software is about granting people rights, and as such it is incompatible with racism. It is incompatible with homophobia. It is incompatible with "gender critical" ideology. Those people all reject the idea that all humans have rights based on who they are are ideologically incompatible with free software and deserve to be sidelined. We can replace anyone's contributions, we can't replace all those they drive away.

You see a transphobe in your mentions? You reject them, just like you'd reject a nazi. A bar that accepts transphobes is a transphobe bar.

https://86box.net/2025/02/25/riva128-part-1.html is absolutely the kind of thing that's my bag, baby

FSF voting member ascribes blame for russian invasion of Ukraine to, uh, Europe: https://snac.lx.oliva.nom.br/lxo/p/1740529517.639475 (in response to a tankie meme that kind of fails to take into account that the USSR sphere of influence was RATHER A LOT LARGER than Russia is today)

I would like a list of good accounts (here, bsky, twitter if you *really* have to) on specifically operating system security topics that I can give my students and yes it's fine if they also just post a bunch of Microsoft Flight sim content as well

Many years earlier I'd been trying to download the code referenced in a scientific paper but the URL provided resolved to an HP printer and after a month I just started sending print jobs with big text saying "TAKE THIS TO YOUR IT DEPARTMENT" and small text telling them to fix whichever of their DNS or DHCP was broken

Direct action works

Remembering the time Amazon somehow added an internal address to my account that only showed up through the "Buy now" flow, so I ordered the cheapest thing I could find with free shipping (a $0.69 dog treat) and added a gift note containing a bug report and eventually it went away although I have no idea whether these events are linked

It's been another day of dealing with the fallout of people unconditionally running

eval $(ssh-agent)

in their shell startup. Please don't:

1) If you run it on remote systems, it breaks ssh agent forwarding - everything now talks to the remote agent, not the forwarded one
2) You end up with a per-shell agent, so keys loaded form one shell aren't available to others
3) It overrides any per-session agent started by your desktop

At minimum, check if SSH_AUTH_SOCK is already set before doing it!

We produce some of the finest memory unsafe code in the world. Big code mines just digging up beautiful buffer overflows, factories turning out world-class use after frees. I spoke to someone and he said sir, those commie Europeans want to get rid of their data races, sir, but we make the best data races. Did you know that America made 15 of the top 20 most exploitable codebases in 2000 year? We're going to be bringing that back. No more woke languages. Good honest American C.

(Or the secret third thing of simply not getting involved at all which would be fairly on brand)

…ok putting the FSF where they're going to have to choose between supporting Drew DeVault or Jack Dorsey is *extremely* funny

Hit a *really* corner case bug today - a binary was absolutely present but returned ENOENT whenever you tried to run it. strace showed exevce() returning it, so it was happening when the kernel tried to run it. Normally I'd assume that the rtld (the thing that the kernel delegates loading of dynamic libraries to when running a dynamically linked binary) was missing, but ldd showed nothing missing - everything was resolved, including the rtld. WTF?