Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
stephenr
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
5 ms
·
31.
▲
by
stephenr
5mo ago
You're seriously suggesting that an electron app you vibe coded in two days is more performant than native code that's been developed and improved and tweaked over literal decades?
32.
▲
by
stephenr
5mo ago
Who knew that the killer task for LLMs would be "recreate existing things that exist without me mentioning them by name, based on a worse alternative as the example". Kaleidoscope exists. File Merge exists.
33.
▲
by
stephenr
5mo ago
You can leave your laptop anywhere on the planet. Just put it down and walk away. The difference between the places is the amount of time before it starts being someone else's laptop.
34.
▲
by
stephenr
5mo ago
> And IAM is not a wholly independent software stack: they rely on DynamoDB and a few other services, which in turn have a circular dependency on IAM. When you dogfood your own Rube Goldberg machine.
35.
▲
by
stephenr
5mo ago
.... that an IDE providing a suggestion about what comes next as you type is not new, and the entire basis of how an LLM works is "what word probably comes next". I'd have thought someone who's so enamoured with the te
36.
▲
by
stephenr
5mo ago
There's a reason I call it spicy autocomplete.
37.
▲
by
stephenr
5mo ago
> If you are letting your LLM look at StackOverflow, you are doing it wrong So you've evaluated all the sources that the model was trained on initially have you? How long did that take you? > I'm shipping quality software
38.
▲
by
stephenr
5mo ago
> more so the attitude of people heavily invested in it Exactly. LLM bros are just the new blockchain/crypto bros, but they aren't necessarily even writing their own spruiking comments any more.
39.
▲
by
stephenr
5mo ago
Your views might carry more weight if the crux of your rebuttal wasn't manufactured outrage that I used a laughably accurate nickname for a type of software.
40.
▲
by
stephenr
5mo ago
I've posited for a while now that the people who find spicy autocomplete to be exciting are the people who can't really do what it does. I played with Image Playground last year some time. It was really fun. You know why? I can&
41.
▲
by
stephenr
5mo ago
> successfully adopting agentic engineering practices What's your definition of "successfully"? More LOC committed per day is probably the only one that's guaranteed when you let spicy autocomplete take the wheel. I d
42.
▲
by
stephenr
5mo ago
> LLMs get more intelligent The Spicy Autocomplete koolaid club is out in force today I see. We clearly have different ideas of what the word "intelligent" means.
43.
▲
by
stephenr
5mo ago
Fair enough. Any plans to have a "serve" mode to serve up a web issues board?
44.
▲
by
stephenr
5mo ago
This seems like it's possibly a spiritual successor to BugsEverywhere (which is relatively vcs agnostic). Is there anything about this that is actually git specific or could it work with eg Mercurial?
45.
▲
by
stephenr
6mo ago
For me? Infrastructure simulation. Why buy a second extra machine to do testing of multi machine infrastructure configurations when my workstation can run the VMs locally? For others? I don't know that's why I think it's ridi
46.
▲
by
stephenr
6mo ago
Maybe I should have used the same dismissive tone. Imagine thinking everyone who buys a Mac and runs VMs wants to run heaps of macOS VMs.
47.
▲
by
stephenr
6mo ago
You realise you can run VMs for any other os right? It's a limit on running macOS not a limit on running VMs.
48.
▲
by
stephenr
6mo ago
I run up to a dozen Linux VMs at once on my Macs. I've never hit the referenced limit because it isn't a limit on running VMs it's a limit on running macOS, and I hardly ever run macOS VMs. I'm not sure why people don&#x
49.
▲
by
stephenr
6mo ago
It's wild that these companies have convinced you to pay to be a beta (at best; arguably much of this is pre alpha quality shit) tester and you're perfectly happy with that scenario.
50.
▲
by
stephenr
7mo ago
The lengths people go to not to solve the actual issue never cease to amaze me. Currently "I don't know how to configure xdebug so I wrote a convoluted alternative" is tied for top place with "I don't know how MySQL
51.
▲
by
stephenr
7mo ago
Why would you use a phone number for 2FA. It's like saying you only use md5 hashing for passwords.
52.
▲
by
stephenr
7mo ago
What PII? You store a TOTP secret on your <device>.... It's less PII than an ssh public key because it's literally just a random string, that *they* generated, and you only need it for the web UI. So please tell me how the A
53.
▲
by
stephenr
7mo ago
I've seen https://codefloe.com mentioned, can't say I've used it myself yet though.
54.
▲
by
stephenr
7mo ago
s/bash/hash/
55.
▲
by
stephenr
7mo ago
If by need you mean, can choose to use, and if by push you mean, login to the GitHub web ui, then sure.
56.
▲
by
stephenr
7mo ago
> adds risk (what if you lose it) Lose what exactly? Decent 2FA setups make you confirm you've recorded a set of backup codes somewhere (they often recommend print and store in a safe, I find a secure note in a password manager wo
57.
▲
by
stephenr
7mo ago
You're very welcome.
58.
▲
by
stephenr
7mo ago
It probably also doesn't make much sense to me because I see external services as something to use when we have to, not as default choice. When your application runs on VMs you control and just uses a payment gateway and an email gate
59.
▲
by
stephenr
7mo ago
In the era of enshittification I can't really see the logic in tying a bunch of your infrastructure/services to the likes of stripe. Then again I also don't see the logic in asking spicy autocomplete to write code or provisio
60.
▲
by
stephenr
7mo ago
Are you using an existing forge package (like eg Forgejo which codeberg is built on) or something custom?
More ›