Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
grim_io
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
17 ms
·
391.
▲
by
grim_io
1y ago
It looks really well done. I'll definitely keep an eye on it until it hopefully matures.
392.
▲
by
grim_io
1y ago
He, who does not produce slop in the first iteration, cast the first stone.
393.
▲
by
grim_io
1y ago
The results would be opposite for me. I need dynamic query building in most places. Doing those inline in SQL would result in a mess of unmaintainable manual string concatenation and parameter interpolation. A code uglier does not exist.
394.
▲
by
grim_io
1y ago
I don’t get this at all, and I tried to understand it. I'm good at SQL. When necessary (maybe once a year), I can drop into pure SQL instead of Django ORM like it's nothing. The thing is, I can't imagine why I would ever want
395.
▲
by
grim_io
1y ago
How? Stored procedures? I find that to be insane. How do you version that code, and how do you reason with the business logic split all over the DB and code?
396.
▲
by
grim_io
1y ago
There is more to it than just a dollar cost per product. Outsourcing is attractive, because then you don't need to care about all the environment protection violations and any other exploitations. Your local customers don't begin
397.
▲
by
grim_io
1y ago
I get it. But this just shows again clearly why people like Trump have any chance at all. I don't share many of his opinions, but nothing in there screams extremist right, sexist or xenophobic.
398.
▲
by
grim_io
1y ago
Nor was my intent to label any of the linked blogs as such. My intent was to say that hobbyists have a different, refreshing approach to programming and it's technologies that I appreciate.
399.
▲
by
grim_io
1y ago
I find it much more interesting to read from people who do programming as a hobby. They focus on the practical solutions much more than on the typical bikeshedding.
400.
▲
by
grim_io
1y ago
woah, got any references to back this up? I've seen a lot of DHH content, and I'd never describe it as radical right wing.
401.
▲
by
grim_io
1y ago
This feels like a scaffold for a badly defined, missing product. Google does it again...
402.
▲
by
grim_io
1y ago
The people in power are convinced that there is nothing the president can't do.
403.
▲
by
grim_io
1y ago
I don't like to compare LLM's to people. There is a real return of investment in co-workers over time, as they get better (most of the time). Now, I don't mind engaging in a bit of Sisyphean endeavor using an LLM, but remembe
404.
▲
by
grim_io
1y ago
I'm not convinced there is any hope for a productive, long-term, burnout-free parallel agent workflow. Not while they need even the slightest amount of supervision/review.
405.
▲
by
grim_io
1y ago
You might be ahead of the US president in understanding how tariffs work.
406.
▲
by
grim_io
1y ago
This is the way. Easy and painless.
407.
▲
by
grim_io
1y ago
Anecdotally, I increased mine to 2. One was not enough weekly quota for heavy codex-cli use.
408.
▲
by
grim_io
1y ago
Wow, worse cars. Finally that stock price will make sense. /s
409.
▲
by
grim_io
1y ago
My argument would be: OSS models are most of the time not competitive on price when the provider does not support caching, and most third party providers don't.
410.
▲
by
grim_io
1y ago
Xai is probably burning cash like crazy now. Who is paying for it when they are giving away the coder and 4-fast model away for free? Seems like a desperate grab for a low-spend part of the customer base. What's the grand plan? Xai ist
411.
▲
by
grim_io
1y ago
I have started "dumbing down" my IDE's and editors, because they are only for the occasional manual editing, code browsing/reviews and commits. I haven't used AI auto complete in months.
412.
▲
by
grim_io
1y ago
There is no hardware format I can imagine that would make sense for something like this. Maybe that's just my limited imagination, but since no one has made anything useful in this space...
413.
▲
by
grim_io
1y ago
Cheap cloud gaming. Those GPU's will finally push pixels again :)
414.
▲
by
grim_io
1y ago
People don't usually buy this much compute for running a mail server or a media transcoder, so... probably AI? :) If it's inference, don't expect great performance or cost-effectiveness. But if you learn a lot during it, I wi
415.
▲
by
grim_io
1y ago
All good, as long as there is self-awareness that it's not built on a "need", but on a "want" basis. It should be obvious to the reader that this is very much overkill, even for the stated goals of expandability and
416.
▲
by
grim_io
1y ago
This exceeds the computing requirements of most IT related companies, by far. Cool, but nothing a single compute machine wouldn't get done with a bunch of VM's if learning and home workload are the focus. This thing probably idles
417.
▲
by
grim_io
1y ago
Sure, it's useful, once. What we need is a way to manage the dynamic part of the context without just starting from zero each time.
418.
▲
by
grim_io
1y ago
A very crude tool. A good start maybe, but it does not give us any information about the message part of the context, the one that matters. We can't really do much with the information that x amount is reserved for MCP, tool calling or
419.
▲
by
grim_io
1y ago
Absolutely ridiculous. My wife is Chinese.
420.
▲
by
grim_io
1y ago
So, jupyter notebooks?
More ›