The Unclear Impact

why are our students complaining about the assignment not working out of the box instead of just… doing the assignment?

if the starter project was doing everything we requested out of the box, it wouldn’t have been the starter, it would have been the solution

was playing around with some #StaticAnalysis for #ObjectOriented things a somewhat challenging technical detail is that we can’t fully use recursive datalog queries to drive the analysis, because we sometimes have to take greatest fixed points instead of least fixed points. so we need clever ways to break the recursive knot in some places (and only tie it if an lfp sufficies)

by abuse of terminology, for a given type A, and ommer is a type B which is the direct subtype of a direct or indirect supertype of A, but is not itself a (direct or indirect) supertype of A. we call a direct subtype C of A a good child for the ommer B if squick it is a subtype of B or has a common (direct or indirect) subtype with it. we call ommers B with no corresponding good child are evil

(this ended up completely backwards, because evil ommers can be handled without breaking the recursive knot, and good children gave us considerable problems)

it takes making my internal DNS resolve the development domain to 10.0.2.2 and

-A OUTPUT -d 10.0.2.2 -j DNAT --to-destination

but I can finally debug websites with TLS (to force a secure context) in the #android #emulator.

why can’t it just have normal bridged networking, like the rest of the VMs I run?

and why does it fail to enable host GPU acceleration if I try to run the emulator from the command line (instead of android studio) to add -writable-system so I can add the internal domain to /etc/hosts? i guess it doesn’t like #swayvm much

the language where a dependency can transitively affect the types of completely unrelated global symbols: https://github.com/mobxjs/mobx/issues/3582

I've been using #AnnotatedEquations in my recent papers. I think it really adds to the readability and understanding of the math.

Here are some examples. It uses #tikz in #latex.

Let me know if you like it. Happy for any feedback.

@juliank I think etebase solves this by encrypting everything client-side. the mains downsides are that

1. calendars/contacts/other stuff cannot really be shared with other users (this should be theoretically possible, e2e file sharing services like Tresorit can solve it, but the etesync protocol doesn’t handle it nevertheless)
2. it doesn’t really speak *DAV, but you need to run a bridge locally that actually handles the encryption. but afaik the Protonmail bridge for IMAP works similarly

the phrase ‘Turing-complete financial instruments’ should put horror into anyone who has studied computation, or ever had to debug a program

the phrase ‘financial instruments’ should probably also be terrifying, but for other reasons

turns out i managed to convince the examiners that i can speak french 🇫🇷

so much about this pointless (in CS) requirement (two CEFR accredited language exams) for obtaining a PhD in hungary

@noracodes @nomemory

One way I like to think about programs (I got this brain worm from some summer school on probabilistic programming, I think) is that programs are described by finite terms, but the semantics of terms (defined by structural induction and least fixed points) form some nontrivial \omega-complete partial order when ordered by termination. Which is just abstract nonsense for

1. programs should be described in a finite way
2. you should be able figure out what a program means by looking at their parts and composing them in prescribed ways
3. it should be possible to compare programs by which inputs they terminate on, provided their outputs agree on all inputs they both terminate on
4. composition should be continuous: a program should terminate if all its parts terminate (in parallel, in sequence, or in whatever way the semantics define composition)

after copious amounts of handwaving, this should exclude anything that cannot be executed on a computer (programs with infinite descriptions, or cases when you have to take a non-least fixed point over something infinite – but ltps exist by the Kleene fixed point theorem). but allows expressing the lambda calculus, so it captures all Turing machines

crucially, non-continuous semantics, where a program terminates if a part of it doesn’t terminate is excluded. this encodes the sore lack of a halting oracle

HTML would be excluded because the underlying \omega-cpo is trivial, I think (any HTML page can be rendered in finite time, so there is no meaningful way to compare them by termination). CSS would probably be excluded because the CSS selector language is cleverly crafted to always evaluate in a finite number of steps

the beauty of looking at programs in this way is that this makes clear that the analysis of programs cannot merely depend on their syntax: any sufficiently interesting analysis will have to look at the underlying semantical category, which has much richer structure! handwave, handwave by Turing and Rice we know it is so rich in fact that it’s hopeless to provide a general algorithm for program analysis!

in any other case, where analysis would be possible by just looking at the syntax of terms, we’re facing data (or configuration)

not as if any of this matters for the present topic: humans are rarely practically interested in programs that never terminate in convoluted ways. if it gets stuck in mundane way, like in a loop, we just get bored after a while and figure out enough to just close it

but while creating data or configuration can easily be so complex we’d be better off not to bother, only programs can theoretically provide unfathomably impossible complexity

(nevertheless, I’d expect the complexity faced day-to-day by someone writing “data” in HTML, CSS, Datalog, etc. roughly comparable to someone writing Lisp, C, Java, Haskell, Prolog or anything else outside of mathematical fun times)

@clonejo Hi! It’s relatively simple, you can just configure :mrf_simple in prod.secret.exs. First, I think you must make sure that the policy is actually enabled, that would be something like

config :pleroma, :mrf,
  policies: [
    Pleroma.Web.ActivityPub.MRF.SimplePolicy
  ],
  transparency: false

(set transparency to true to to list the instances you block publicly on /about. include any other MRF policies you want in policies, such as Pleroma.Web.ActivityPub.MRF.ObjectAgePolicy. see https://docs.pleroma.social/backend/configuration/cheatsheet/#mrf for more details)

Then you can configure the policy itself, such as

config :pleroma, :mrf_simple,
  reject: [
    "example.com",
    "example.org"
  ]

to completely reject all activities from the listed instances.

The configuration is relatively straightforward once you get the hang of the Config Elixir module that is used to store instance configuration. Take a look at config.exs to see what are the actual defaults that you’re overriding in prod.secret.exs, the documentation might be a bit spotty on that part.

if space runs out on a btrfs filesystem while pacman is working, it’ll be a bad time, and it’s probably a much better idea to restore a snapshot (which caused the space to run out in the first place) instead of trying to paccheck everything

@rysiek btw, this is a bit silly, but could a service worker just use sendmessage to communicate with the main page which hosts an RTCDataChannel in lieu of instantiating an RTCDataChannel itself?

this idea comes from the fact that the current API (as implemented by Safari) needs a webpage to establish the RTC channel, which then gets sent to the worker (and probably only lives as long as the page is open). so in practice, the worker first has to serve some kind of shim webpage and script, which establishes the channel and sends in to the worker – then the worker could use the channel to answer further requests

so the silly thing to do would be to (in absence of RTC channel transfer support) establish the channel in the shim webpage, and notify the worker of this fact via a message. subsequently, on any fetch request the worker would send a message to the webpage, and the webpage would answer with another message transferring some buffer with the request data. this is of course completely backwards, but FetchEvent.respondWith does take any promise, and Response objects can be created from scratch (albeit with some CORS caveats)…

@midzer @fribbledom

@rysiek looks like the newest webkit is supporting transferring RTCDataChannel to a service worker: https://wpt.fyi/results/webrtc-extensions/transfer-datachannel-service-worker.https.html?label=experimental&label=master&aligned although I imagine that’s not very useful atm, since other browsers completely lack support, and RTC connections cannot be created in the worker itself at any case

you’re probably already familiar with the relevant webrtc issue: https://github.com/w3c/webrtc-pc/issues/230 which explicitly highlights peer-to-peer content delivery as a use-case. interestingly, the last comment says that transferring to a service worker is not yet implemented even in safari, but things seem to have changed since then according to wpt

unfortunately, google seems to have blessed their http3-only webtransport https://web.dev/webtransport/#is-webtransport-an-alternative-to-webrtc-data-channels as a way to open datagram connections in workers, which doesn’t support peer-to-peer connections at all. so I wouldn’t hold my breath for service worker webrtc support in chrome

@fribbledom @midzer

why is chromium / edge text rendering on windows so ugly?

trying to buy a (small, but still larger than most computer monitors) tv without surveillance or ads 2022 challenge

i’m at the point where i have no clue whether the state management approach i tend to use with #reactjs is

1. totally common
2. frowned upon to the point that support for it might be removed from react at any time
3. the scariest thought of it all: both

#react #javascript #mobx