I wish it were easier to just say to someone. Hey I don’t want to hurt your feelings, but I’m feeling to end this conversation now. Is it OK that we bring it to an end?
In my perception ending a conversation is much easier than keeping it alive. People will pick up easily that you are not interested, even in the non verbal part of communication, no?
Just say "Hey, I gotta run. It was nice talking to you." Or something to that effect. Most people aren't going to get upset if you need to wrap up a conversation. And if they do get upset (assuming they aren't having an emergency), it's a red flag.
Alternatives: "can we talk later, I'm really busy with ___."
Or in a work setting, "I'm in the middle of something, do you mind sending an email?"
Social settings, excuse yourself to the bathroom, they'll move on.
If you don't like the conversation, "do you mind if we talk about something else?".
There are tons of ways to wrap up or avoid conversation. The more honest you are the better. And take note of how others gracefully end conversations with you and use those phrases too.
Preloading authorization data into memory does not, by itself, provide the specific security guarantee (consistency) that defines Zanzibar.
The Zanzibar model is famous not just because it is fast, but because it solves the "New Enemy" problem (or causal consistency). Simple in-memory caching (preloading) often fails this test unless it is paired with complex invalidation logic that mimics Zanzibar's Zookies.
It is actually slightly worse than even that: while New Enemy [1] is the primary concern, caching like this can also introduce a staleness issue from the other direction: let's say a user adds a new row or document, and immediately sends the link to their coworker... who tries to load that piece of data, but the (stale) access control dataset is cached and they are not in it... they get a "no access" error. While certainly fail safe (vs fail dangerous for New Enemy), it can be a fairly important UX concern as well.
Generally, the solution is to keep a timestamp of when the data changed (Zookies as you mentioned) or you can proactively reload or recompute the cache when the underlying data changes (sometimes in very smart ways), but yeah: it adds significant complications over a "simplified" approach to Zanzibar.
Disclaimer: I'm the cofounder and CTO of AuthZed and we develop the SpiceDB [2] and Materialize [3], which have quite a bit of logic around these exact problems
I'd start with reading the Zanzibar Paper. We built an annotated version [1] that provides additional guidance on some of the denser sections and how we interpreted them.
Then, I'd take a look at the history of SpiceDB [2] for how we developed the system over time.
Finally, if you have any questions, feel free to jump into our Discord [3] and ask: we're happy to answer!
On Tinder, after autoswiping 100k to 200K profiles :')
I wish I had the exact stats but I just got some crappy JS code from some Github website, edited it a bit and within 10 minutes the swiper was swiping. I'd then go meticulously go through every "match" (looking at photo's, reading the bio's, all of it) because if you only have around 250 "matches" each month (for each 50K women I swiped right on auto pilot), you can actually take the time. I unmatched a lot of them and was left with my real matches.
I met her in Valencia. She was on vacation and I was digital nomadding at the time. She asked me on a date and I was really tired and really didn't want to go because of it. But I remember thinking "you have to shoot all your shots and take all your chances." I'm so happy I did, she made me forget I was tired in a heartbeat and we were goofing around for the whole 3 hours, to the point where I realized I hadn't even asked one normal question and didn't know much about her haha (and vice versa). I love conversations like that, just fun vibes.
Cool! Who is this for? Can you share some potential use cases? Do you have a clear understanding of your target audience? Or is this more of a “build it and they will come” approach?
Great questions. It's targeted at dev teams wanting to build their own custom experiences on the frontend without having to deal with the complications of a scheduling backend. This could be anything from calendar functionality in their application to a platform that handles bookings for an office, etc.
While assisting several startups get their product built, we ran into a need for something like this which drove us to build this as a standalone service.
Figured others might find it useful as well!
A DDoS attack is often used to distract a company's security team. While the security staff is scrambling to get the website back online, the attackers use the chaos to conduct a more serious, stealthy attack.
I don't doubt there will have been sporadic examples of this, but what points to this "often" being the case? It seems like a tactic that wouldn't often pay off, since DDoS mitigation rarely involves relaxing security systems
Mistakes can be made during reconfigurations but you'd have to catch those while the issue is still live. Sounds like an advanced threat actor and not the run of the mill ransomware people (not that they're necessarily unsophisticated, but why'd they bother with these odds when there's low-hanging fruit to reliably exploit)
It was interesting to read that the record breaking attack caused no glitch whatsoever in the service MS provides. Which is so slow normally that I start to wonder if that is a strategy, having headroom for these kind of situations, no-one realizes slowdown when it is already slow. ;)
This is just a crazy thought, tangential to what are happening during an attack.
or rather the slowness problems of MS has nothing to do with hardware or infrastructure limitations. You cannot just throw infra at a problem to mask poorly written code beyond a point.
I wonder if you would get better results if you tell the LLM there's a token limit in the prompt.
something like "You only have 1000 tokens. Generate an analog clock showing ${time}, with a CSS animated second hand. Make it responsive and use a white background. Return ONLY the HTML/CSS code with no markdown formatting"
I got a ~1600 character reply from gpt, including spaces and it worked first shot dumping into an html doc. I think that probably fits ok in the limit? (If I missed something obvious feel free to tell me I'm an idiot)
On the second minute I had the AI World Clocks site open the GPT-5 generated version displayed a perfect clock. Its clock before and every clock from it since has had very apparent issues though.
If you could get a perfect clock several times for the identical prompt in fresh contexts with the same model then it'd be a better comparison. Potentially the ChatGPT site you're using though is doing some adjustments that the API fed version isn't.
reply