@GergelyOrosz Waste or not, I would say it got a lot of AI skeptics to jump onboard and many now happily and willingly use it for their work every day.
English
Steven Luu
2.8K posts
Steven Luu
@stnluu
Doing some stuff at Facebook. Microsoft alum. Opinions expressed here are definitely NOT mine.
Seattle, WA Katılım Mart 2011
114 Takip Edilen139 Takipçiler
Devs game everything and anything seen as a target for more bonus or promos. This was no different.
Talked w devs at Meta and this was SO MUCH waste - just to get on the leaderboard!!
Meta removed the leaderboard now. Source: me
(Next up I expect to do the same: Microsoft)
Jyoti Mann@jyoti_mann1
Exclusive: Meta employees are “tokenmaxxing” and competing on an internal leaderboard called “Claudeonomics” for status as a token legend. Over a recent 30-day period, total usage on the dashboard topped 60 trillion tokens.
English
Good engineers will say B is better. Great engineers will say it depends…
Captain-EO 👨🏾💻@EOEboh
Interviewer: Which of these two queries performs better and why? -- Query A SELECT * FROM orders WHERE YEAR(created_at) = 2024 -- Query B SELECT * FROM orders WHERE created_at >= '2024-01-01' AND created_at < '2025-01-01'
English
@EOEboh Not enough information to answer. What are the indexes?
Say, is there an index like this:
KEY `year` ( (YEAR(created_at)) )
And is this even a SQL query or some Presto?
English
Interviewer:
Which of these two queries
performs better and why?
-- Query A
SELECT * FROM orders
WHERE YEAR(created_at) = 2024
-- Query B
SELECT * FROM orders
WHERE created_at >= '2024-01-01'
AND created_at < '2025-01-01'
English
Believe it or not we still don't know what to officially call this panel.
Ideas?
English
@geekyAyushi There's no need to over-engineer a URL shortener redd.it/1kjcjzd
English
Designing a URL shortener sounds trivial until you're tasked with generating 100K unique URLs per second, with low latency, strict correctness, and multi-tenant isolation.
Over the weekend, I explored a brilliant breakdown of how Rebrandly engineered for this scale 🧵
English
English
@paulg we have considered this--we could do it nicely on 100 meters square.
it would be a cool monument.
(we also thought about printing GPT-2 out in books and lining our new office with it!)
English
goodbye, GPT-4. you kicked off a revolution.
we will proudly keep your weights on a special hard drive to give to some historians in the future.
English
@mariioalt @OrdinaryInds @Doctorthe113 "One command"? Sure, which one? Apt or snap or flatpak or yum...? And where did it install?
English
@OrdinaryInds @Doctorthe113 I mean.. on Linux it’s one command. Yeah, it’s different depending on what you’re running, but it’s pretty much always one command to install anything.
Or an App Store as an alternative if you’re into that.
English
Yeah let’s switch to next > next > next > next > this user > next > next > next > accept.
You’ve been given the simplest installer that tells you exactly where the app lives and you want to complain. Fascinating.
nizzy@nizzyabi
WHY ARE WE STILL DOING THIS IN 2025
English
@neogoose_btw @oisyn i meant do the equivalent like the Java side where you wrap the unsafe inside a safe function, then do .expect(...) instead of just .unwrap()
English
English
on the left you can see a sane person FFI, on the right it's android
java was is and will be a mistake
English
@VoxelPrismatic @neogoose_btw the C side also uses unwrap() while the java side [intentionally] uses a long line for expect(...)?
English
![PRIZ ;]](https://pbs.twimg.com/profile_images/1815124531935068160/1xbxJQ_q.png)
@neogoose_btw >java was a mistake
>complains that rust bindings for java has more boilerplate than rust bindings for c
>the C example uses unsafe and the java one doesn't
English
@neogoose_btw @oisyn and cool. now do it without unsafe on the C side
English
@oisyn The strings on the left side are C strings owned by the caller, on the right side it makes “something” to get a new string which is getting allocated on every call for every parameter and needs to be manually dropped + the stdout issue
English
@gregmushen @DougChampion what about redundancy? MR/DR? your ISP down? power outage? basement flooding? I self-host what i can, but i hope you realize in addition to hardware, you're paying for the SLA & the up time.
English
@DougChampion I've been in tech for 30 years. Started out as a UNIX SA, then network engineer. I've worked on almost every aspect of a data center.
It's not willful ignorance at all, it's just knowing what I need, and engineering for that. I don't need all of AWS in my garage.
English
To give you an idea of how expensive cloud has gotten, I picked up a server off FB marketplace today for $500. 128Gb of RAM and a 1Tb SSD.
It will cost about $6/month in electricity to run it.
An equivalent server in AWS would cost $950/mo. so I'll have payback in two weeks.
I'm going to buy a GPU for $600, and the cost of that in AWS would be about $350/mo.
I think if someone created a distributed garage hosting network it would kill it.
English
I'm 61 and over the past two months, my 401(k) has lost 25 percent of its value. I'm always happy to contribute to the greater good, but can someone please explain to me how this helps anyone?
English
All the replies in here saying "you shouldn't sort by ID" doesn't realize the database itself often uses the primary key to order how it stores data. This industry is doomed.
Pontus Abrahamsson — oss/acc@pontusab
UUIDs are not great, and they are not sortable. Learned the hard way.
English
@davepl1968 c++ makes it easy to abuse.
besides, rust has a better type system, fewer to no UB.
most importantly, Cargo. c++ build and dependency management is still a mess.
English
No, Rust exists because people treat C++ as "C with Classes" and continue to abuse it.
If you stick to unique_ptr and shared_ptr and so on, you can write entirely memory-safe code trivially.
It's 2025, so it's time to learn C17 at least...
trish@TrisH0x2A
Rust exists because C++ refused to admit it has problems.
English
@LewisCTech @davepl1968 don't forget cargo. dependency and build management in c++ is still a mess
English
@davepl1968 I can't go back to C++ after Rust. It's not even a safety thing, it's an ergonomics thing. Real sum types, real option types, hygienic macros, faster compilation times, more expressive code, less language complexity, no template errors... C++ is like going back to the dark age.
English
@davepl1968 @Shinsego @ZacksJerryRig Disagree with "not of their own making." It's these people's own doing when they choose to support Elon, that's part of the risk they take. And it is their own making for not assessing the risks associated with whom they're investing in.
English
@Shinsego @ZacksJerryRig Generally, yes, one should feel empathy for others who undergo hardship not of their own making.
Now, if you're so full of hate that you rationalize it as "Well, that's what they get for supporting Elon", then I probably can't help you, and stand by my post.
English
@CountTwoOne @veerpy @im_roy_lee sounds like you have never worked in big tech. I won’t continue this conversation further as we don’t have the same knowledge baseline about this subject
English
Yes, I am aware of the big tech strategy of Zerg-rushing problem domains with fresh Stanford grads; it is why they are unable to innovate and can barely maintain the well-heeled stack.
Why not just replace them with LLMs and focus your energy on hiring experts to use those?
We all know why; same reason the US truck drivers needed to lower their standards or the (documented reason) the US university system went abroad:
The remaining employees would have too much bargaining power.
English
something that @im_roy_lee and other twitter grifters don't understand (because they've never worked at a real company ever) is that large companies prioritize risk aversion. if you get rid of leetcode, you force companies' hand to exclusively interview from top universities.
English
@CountTwoOne @veerpy @im_roy_lee And “implementing an interesting feature” is just half a step away from leetcode questions, depending on how “interesting” the question is.
English
@CountTwoOne @veerpy @im_roy_lee Doing the code review style interviews DO provide some signals, but again it still comes back down to having the candidate write code, as you suggested, in order to see their ability to solve problems and how quickly they can implement an idea.
English