sebastiangrill

I agree. Still better than the native elements. The biggest downside is 100kb js elements because they try to handle everything. I used basecoat and that worked ok, even if some essentials are missing. Currently I am using my own webcomponents created with rocket (from datastar)

I think this is completely wrong. The reason why nkt a lot of people use the built in elements is because sooner or later you will hit a limitation that you cannot fix. If it is a javascript element you can do whatever you want if it doesn'fit.

Take dialog as an example. If you have a server side rendered app and you want to show an open dialog without js. You are out of luck, you can't just render it as open because that opens a dialog that does not correctly work. Or take a multiselect input. Just unusable as a normal browser element. Also most elements miss something essential like no search function in an input. So also unusable for big lists. I could make a list of 100 things that are wrong with native browser elements. It is just easier to use a js framework/library and have a clean view and state.

I still find plans to be useful for big changes. I first ask to create a plan for the specific feature i have in mind. Then i ask for a second round of research with the current project in mind. Then i ask if it can make improvements on the plan. Then i read the plan and note issues down and let it correct them.

In my experience the problem with plans is that llms (at least gpt and claude) are really lazy and will stop at the first solution without even understanding your project. I had plans that wanted to add a few thousand lines of code just to support a very minimal part of a feature. After asking for a second round of research with the project specific limitations in mind it recommended a very small change to the existing codebase that did the same thing.

This is probably also the reason why most llm generated code stacks thousands of lines of code.

What tasks? Creating a extraordinarily mundane bomb?

I think so too. The only way to redeem this theft would be to force all AI companies to open source their models if they cannot prove that copyrighted material was not used to train them.

Good, H1B is just modern servitude. But there should be an incentive to hire local workers instead of just hiring foreigners in their own country.

The problem i see with this approach is that nearly nobody uses dumps as backups. I consulted Postgres for a few hundred customers and the amount of times that I saw dumps as Backup can probably be counted on two hands.

Does not help that the description reads like AI slop.