While the article highlights interface upgrades, it overlooks potential downsides like increased complexity in debugging. Anyone experienced issues with this?
Could you provide an example of how you would implement this approach to store Credit/Debit events for an account? Additionally, how would you handle a scenario where there are 30,000 events on the account, and you need to calculate the balance to prevent overdraft?
Heh, you found the hard case. You want to add a denormalized table (point #6) specifically for locking on the balance, just cause Postgres/MySQL `serializable` mode is way too slow to rely on. You still keep baseline insert-only credits/debits table(s) that you insert into in the same xact, and all the usual rules apply there.
You can also do this without making such an exception. I used to keep a separate "pending" table that I'd insert into, commit, then check the balance with the pending row included before moving it to non-pending. So two transactions. That worked, problem is it was annoying. Though it was a good solution for debits/credits that involved an async external step that could fail or time out; simply ignore the pending rows that are too old and never got resolved. 30K rows is still small enough to query quickly.
Tbh I had to just look up event sourcing, but thanks, I'll add that to my vocabulary. Seems like my approach includes a form of event sourcing, though I'm not always storing the ordering of events, only in cases where it's needed. And that concept isn't specific to relational DBs.
Hi HN! I'm an indie founder introducing SupportBuddy, a humble project aiming to enhance website experiences. It's a no-code AI chatbot that adapts to your site's content, offering real-time, AI-powered responses.
Key features:
* Easy setup: Just add a <script> tag
* Personalized chatbot: Learns from your content
* Budget-friendly plans: Suitable for small websites
I built this tool to help creators like me improve user engagement and satisfaction on their sites. I'd love your feedback and suggestions! Give it a try at supportbuddy.io.
As a customer, this sounds incredibly dystopian. Even the current "if-else" chatbots are frustrating to the point of rage. Having an LLM spout lies at me isn't making the customer experience better.
Funfact; AROS is also the name of a Danish art museum http://aros.dk/. AROS means River mouth in old Danish and is the old name of the second largest city in Denmark, Aarhus.
I am in awe of the museum's website. With so much of animations I expected it to be a terrible experience but it is implemented brilliantly and functions so smooth.
> I'm not seeing anything about pricing. How do they make money?
Right next to the "Apply for enterprise access" button, there is a giant block of text saying "Our enterprise solutions are customized with predictable pricing".
Basically, they make money, because that API access isn't free. They don't list the actual price there because, as it is the case with most enterprise sales, there is no fixed pricing, and they work out individual deals on a case-by-case basis (which is based on the number of requests you expect your API to make per day/hour, volume discounts, long-term contract discounts, etc.).
Long story short, there is a price on it, but to find out what it is, you will have to apply. And the price you will get will probably not be representative of the price other people/enterprises get, but at least you will know what it will cost for your specific company/project.
You know, just like a population of blue whales who're hiding in the ocean, NYtimes article contents are hiding in somewhere in browser's local storage too
In form, it was a pretty simple discussion site. But it had a lot of insider info and an enormous amount of snark, and it really captured the spirit of the popping of Bubble 1.0 (and the post 9/11 downturn).
I remember one where a CEO made a big deal about the tech staff working through the holidays to finish vital new products. And then the CEO went on vacation to Hawai'i, leaving the nerds to grind it out. The memo and vacaction leaked to Fucked Company, where comments were spirited, and I think it eventually caused a big to-do.
We all got hooked on the forum when waiting for news of the inevitable for our company, which ironically didn't fail, and we actually ended up cashing out in an acquisition.
IIRC, some of Howard Stern's wack pack people were regular posters on the forum as well. It was a pretty awesome message board for that era, until ruined by nazis and other lowlifes.
I remember a suggestion to place all startups in a death pool which they were only allowed to leave if they managed not to buy any Aeron chairs (those were all the rage at that time) after one year in business. I loved that idea.
The other side of this, is the company I worked for about that time, purchased the entire set of office furniture from a defunct company to setup our office when we moved. Turns out we didn't need it all, and sold a lot of it. At the end of the year, the majority of our profit was the furniture hussle.
At the end of the first dot-com boom, there was a place in San Jose called Consolidated Office Supply which bought the furniture of failed dot-coms. They had a warehouse which covered an entire city block, full of used furniture. Looked like the warehouse from Raiders of the Lost Ark.
I bought furniture there for something I was starting up. Very cheap, and they delivered.
I am currently sitting in an Aeron chair from one of the dot com bust furniture auctions...
I had a friend who rescued servers just abandoned in their racks when one company just folded mid-day, the last remaining guy just walked out of the office.
Crazy times.
It allowed employees to leak when their company was laying off employees. There was a sister site called luckedcompany.com which allowed employees to leak when their company got investment (IIRC)
I think they just allowed anonymous posts. Different era where they were actually sort of reliable. I think they got sued a lot which forced them to shut down. In the end Fucked Company was fucked, too.
