Live data from Hacker News

Ask HN: Cursor or Windsurf?

news.ycombinator.com

401–410 of 410 posts

Re: Ask HN: Cursor or Windsurf?

#401

Sometimes I feel like I'm the only one sitting here with vim enjoying myself. Letting this whole AI wave float away.

I don't mind having to learn these new tools, but I don't see any drawbacks in waiting a year or more until it stabilizes. Same as in the crazy times of frontend libraries when it was a new one every week. Just don't jump on anything, and learn the winner in the end. Sure, I may not be state of the art. But I can pick up whatever fast. Let someone else do all the experiments.

What was the winner in frontend that you learned?

Re: Ask HN: Cursor or Windsurf?

#402
post #393

You need none of these fancy tools if you iterate over specs instead of iterating over code. I explain it all in here: https://www.cleverthinkingsoftware.com/spec-first-developmen...

I played around with your suggestion for a day or two now. While I'm intrigued, there are some real-world issues with this approach: - The same spec is processed by the same LLM differently when implementing from scratch. This can maybe mitigated somewhat by adjusting the temperature slider. But generally speaking, the same spec won't give the same result unless you are very specific. - Same if you use different LLMs…

Why would you expect, or even need, the exact same methods used every time,

when you do not expect this from two human devs (junior OR senior) both given the same task?

Re: Ask HN: Cursor or Windsurf?

#403

Zed. They've upped their game in the AI integration and so far it's the best one I've seen (external from work). Cursor and VSCode+Copilot always felt slow and janky, Zed is much less janky feels like pretty mature software, and I can just plug in my Gemini API key and use that for free/cheap instead of paying for the editor's own integration.

Why are the Zeds guys so hung up on UI rendering times....? I don't care that the UI can render at 120FPS if it takes 3 seconds to get input from an LLM. I do like the clean UI though.

Re: Ask HN: Cursor or Windsurf?

#404

Claude code is the best so far, I am using the 200$ plan. in terms of feature matrix all tools are almost same with some hits and misses but speed is something which claude code wins.

Do you think you use more than just 200$ worth of API credits in a month? I've used both Claude Code and Cursor, and I find myself liking the Terminal CLI, but the price is much more than 20$ per month for me.

I was spending around 8 - 10$ in 2-4 hours easily and I had to be conscious of the credits being deducted, so 200$ for a month unlimited usage didn't seemed too bad, and now i just throw a lot of code generated by claude if a don't like and retry from scratch

Re: Ask HN: Cursor or Windsurf?

#405
post #197

Earlier quoted context omitted.

It still works

Do they actually improve the model you can get without cursor? Or in reality all the development goes to cursor's autocomplete without making this available to supermaven subscribers? It is hard to make sure of that from their website and lack of info online.

Probably not

Re: Ask HN: Cursor or Windsurf?

#406
I'm currently trying to use Windsurf at work since we have a license. The problem I have is that I find it's auto complete distracting from what I want to write. And when I do not know how to do a specific task, It will usually hallucinate an answer because it is not in it's training data. I mostly do staff level android tooling work which likely means it's not likely to be a common coding task.

Re: Ask HN: Cursor or Windsurf?

#407
post #61

Earlier quoted context omitted.

I suspect that if you're a vim user those friction points are a bit different. For me, Aider's git auto commit and /undo command are what sells it for me at this current junction of technology. OpenHands looks promising, though rather complex.

The (relative) simplicity is what sells aider for me (it also helps that I use neovim in tmux). It was easy to figure out exactly what it's sending to the LLM, and I like that it does one thing at a time. I want to babysit my LLMs and those "agentic" tools that go off and do dozens of things in a loop make me feel out of control.

I like to be the human in the loop and everytime it does something I don't like, I will add a rule in conventions.md. Overtime, I watch it less and less.

Re: Ask HN: Cursor or Windsurf?

#408

If you have any intellectual property worth protecting or need to comply with HIPAA, a completely local installation of Cline or Aider or Codeium with LM Studio with Qwen or DeepSeek Coder works well. If you'd rather not bother, I don't see any option to GitHub Copilot for Business. Sure, they're slower to catch up to Cursor, but catch up they will. https://github.com/features/copilot/plans?cft=copilot_li.fea...

Which entity is going to steal your IP? Cursor / Windsurf or OpenAI / Anthropic?

Individual employees. Although, I wouldn't be surprised if companies steal ideas, either.

Re: Ask HN: Cursor or Windsurf?

#409
post #156

Earlier quoted context omitted.

This is very inexpensive. What is your workflow and savings techniques! I can spend $10/h or more with very short sessions and few files.

Huh, I didn't configure anything for saving, honestly. I just add the whole repo and do my stuff. How do you get to $10/h? I probably couldn't even provoke this. I assume we have a very different workflow.

What is your workflow? Mine is add minimal files (2-5), keep sessions short, and ask for specific tasks generally in 1 file.

Re: Ask HN: Cursor or Windsurf?

#410

For a time windsurf was way ahead of cursor in full agentic coding, but now I hear cursor has caught up. I have yet to switch back to try out cursor again but starting to get frustrated with Windsurf being restricted to gathering context only 100-200 lines at a time. So many of the bugs and poor results that it can introduce are simply due to improper context. When forcibly giving it the necessary context you can cle…

Wait, are these 800 lines of code? Am I the only one seeing that as a major code smell? Assuming these are code files, the issue is not AI processing power but rather bread and butter coding practices related to file organisation and modularisation.

The people editing 800+ line files often didn't write them, legacy codebases often stink!

I've dealt with a few over the years with 30k+ line long files, always aiming to refactor that into something more sensible, but that's only possible over a long time.

Post reply on HN