The Unclear Impact

Kristóf Marussy ferdi |

I'm a PhD student working on the extra-functional requirements and formal verification of cyber-physical system architectures.
I also like free (as in liberty) software, privacy enhancing technologies, and cryptography.

re: firefox

it’s a pity that the changelog for developer-visible changes has no mention of these changes. maybe they don’t want theming to be a public API any more?


Me after upgrading to firefox 92: there’s something weird about how my url bar looks blobfoxconfused I remember setting up userChrome.css to look more like the rest of my system

fiddling around with userChrome.css for a while with no avail

ah, so 92 renamed a bunch of css variables and reddit comes to the rescue

--lwt-toolbar-field-background-color -> --toolbar-field-background-color
--lwt-toolbar-field-focus -> --toolbar-field-focus-background-color

at least the -focus version is now consistent with the rest, and my browser looks comfy again blobfoxcomfytea

(see also browser.urlbar.groupLabels.enabled and set it to false to hide the annoying Firefox Suggest label which is good for absolutely nothing since I disable search suggestions)

i am saying nothing new here but it is beyond disgusting that judith butler, one of the most important and influential thinkers in the world, can be censured by a newspaper of record for upsetting a bunch of english women with bad haircuts


@hazel That sounds even more terrifying, then.

@hazel Ah, hashing… blobfoxteaterrified

gender, media, shitpost

The whole censoring of the Judith Butler interview is so utterly bizarre… if some rag doesn’t want to publish material that slams TERFs, then maybe they shouldn’t schedule an interview with the person who came up with the term gender performativity. Other courses of action, such as hiding in a hole, seem to be much more conductive towards that effect. blobfoxupsidedown

Re: cryptography, moderation, whatsapp

@meisam @sotolf Isn’t that kinda the point of the double ratchet protocol?

I presume you’d only share the message keys for the messages you want to expose to the moderators, who can then verify that the keys indeed decrypt the ciphertexts stored on the server. It shouldn’t be possible to calculate keys for earlier messages from the exposed keys, and later messages are protected by pfs. (Usually, this is formulated for accidental keys exposure, but should hold for deliberate ones, too.)

So exposing the full session to someone who already has a record of the ciphertexts seems to be unnecessary, or I’m misunderstanding something.

vax, plague

3rd vax dose hit me quite heavily after 12 hours. I guess this means it’s doing… something useful? blobcatsick

(Pfizer after 2 doses of Sputnik-V, unapproved, but widely recommended protocol in Hungary)

Ours was now a country in which the cost of replacing a broken machine with a newer model was typically lower than the cost of having it fixed by an expert, which itself was typically lower than the cost of sourcing the parts and figuring out how to fix it yourself. This fact alone virtually guaranteed technological tyranny, which was perpetuated not by the technology itself but by the ignorance of everyone who used it daily and yet failed to understand it. To refuse to inform yourself about the basic operation and maintenance of the equipment you depended on was to passively accept that tyranny and agree to its terms: when your equipment works, you'll work, but when your equipment breaks down you'll break down, too. Your possessions would possess you.
- Edward Snowden, Permanent Record

these quotes are from Blessed is the Flame, a book about anarchism, nihilism, and the resistance movements inside the nazi concentration camps. it's a heavy read but very good

Re: linux desktop performance/experiments

@jookia I’m sleepy, fail to scroll back in threads, and make bad jokes!

peeks out of corner

that said, setting up cgroups + niceness according to a config file legitimately sounds like systemd (user) units…

(I’m a bit surprised that firejail doesn’t have similar functionality, since it’s also nearly a kitchen sink in terms of linux process management gizmos, and stuffing all child processes into a cgroup wouldn’t be much of a stretch before stuffing them into a process namespace, anyways)

Re: linux desktop performance/experiments
@jookia that sounds like systemd

_ducks in corner_

Why is Teams suggesting me to have a meeting with my student at 1 AM? blobcatsleepless

(Okay, if a student would want me to look at some work at 1 AM I’d likely do that, especially if research related, but having it as a suggested time slot is a bit extreme.)

@hazel It’s a joke. Like, our uni spends like 50x as much for journal access every year as it would cost to host the whole Sci-Hub catalog. And Sci-Hub still ends up much nicer to use.


One way to find out the name of a weird mathematical structure is to look at the source code of related Haskell libraries.

I just learned from Algebra.Lattice.Wide that adding a top and bottom element to a set to form a lattice

data Wide a = Top | Middle a | Bottom

is called the flat lattice. Which is not the name of this thing in the code, but at least the commit history had the sane name that I could look up elsewhere. Apparently, the corresponding abstract interpretation is the constant abstraction.

(Of course, such obscure names are useless and are just impressive-sounding word salad. Luckily, I’m forced to write some impressive-sounding word salad right now. blobfoxhyper2)

Firefox takes 30 seconds to start for me, which is not ideal. blobfoxbongo

After applying it can finally start with gfx.webrender.compositor.force-enabled, which is pretty cool and fast, but it’s unfortunately also a glitchy mess.

Ah, here it is: – hopefully the fix will be released soon

My Matrix homeserver setup has just been broken by podman 3.3.0 starting a container with a new IP instead of joining the pod’s network upon receiving the --pod-id-file argument to join a pod. blobfoxangry

Context, someone sent me this:

… things like modern web sites, [are] not going to be pretty on a single core, single threaded CPU.

If your documents (and that’s all a website is) is so bloated that it is “not going to be pretty” rendered by a single core CPU, that website is objectively inferior than one which would “be pretty” on the same hardware. Simple as that.