Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

My experience with Claude Code after two weeks of adventures

sankalp.bearblog.dev

Post reply on HN
Print view View original

301–310 of 388 posts

Re: My experience with Claude Code after two weeks of adventures

#301

Post by tom_m »

ianberdin wrote: post #232

Reading all these glowing reviews of Claude Code, I still get the feeling that either everyone’s been paid off or it’s just the die-hard fans of terminal windows and editors like Emacs and Vim. Using the terminal is right up their alley—it’s in their DNA. Every time I read comments saying Claude Code is far better than Cursor, I fire it up, pay for a subscription, and run it on a large, complex TypeScript codebase. F…

Die hard fans. I tried so many Claude models. Some are ok. Really. Surprisingly some versions of Sonnet are better than Opus. So the latest isn't always the greatest. Models are specialized sometimes (intentionally designed that way or not).

I think people just need to try a bunch to be honest. I think that we will get to a point (or are already there) where some models just resonate better with you and your communication (prompt) style.

But after using a bunch...I have to say that Gemini 2.5 Pro is consistently the best one I've used yet. It's pricey, but it just works.

I'm still looking for a good local model, but it's just going to be a while until anything local can match Gemini. It will eventually I think, but it'll take some time.

I probably won't be using Claude Code, but I'm glad it works for some folks.

Re: My experience with Claude Code after two weeks of adventures

#302

Post by Wowfunhappy »

nico wrote: post #281
Wowfunhappy wrote: post #254

Earlier quoted context omitted.

> Personally, I also like that usually it just presents to me only one change/file at a time, so it's easier for me to review This is interesting. I haven't used Cursor, but one of my frustrations with Claude Code is that some of the individual changes it asks me to approve are too small for me to make a decision. There are cases where I almost denied a change initially, then realized Claude's approach made sense onc…

> some of the individual changes it asks me to approve are too small for me to make a decision. There are cases where I almost denied a change initially, then realized Claude's approach made sense once I saw the full change set Yes, I've adapted to just review quickly, then if it makes sense as part of the task, let it keep going until it's done with the whole thing. Most of the times, by the end it does the right th…

> I love that it doesn't auto-commit everything, ala aider, so it's pretty painless to undo stuff.

Yeah, I'm definitely glad it doesn't commit for me. The main issue I have is that I'm never sure how granular to make my commits. Sometimes I make them very granular because I'm experimenting with Claude and I want to be able to revert to any point in the conversation—but now I have to write a message each time to keep track of which is which. Conversely, when I don't make the commits as granular I loose the ability to roll back, and sometimes regret it.

Also, sometimes Claude gets a bit too smart! Let's say I decide I want Claude to try again with a slightly different prompt. I save my current changes in a branch, roll back, to the previous state, and ask Claude to try again. Sometimes Claude will say "I see this is already implemented in the XX branch. Let me continue to build on that implementation."

Re: My experience with Claude Code after two weeks of adventures

#303

Post by leptons »

crinkly wrote: post #193
jasoncartwright wrote: post #191

Earlier quoted context omitted.

I enjoy the idea that when programming before AI I wasn't 'merely a machine operator'

I did most of my programming and thinking on paper before I committed to the machine. I still do today.

The only time I ever wrote a program on paper was when I was 10 years old, a few months before my 11th birthday when I got my first computer. And that was in the late 1970s.

Re: My experience with Claude Code after two weeks of adventures

#304

Post by mitchitized »

ianberdin wrote: post #232

Reading all these glowing reviews of Claude Code, I still get the feeling that either everyone’s been paid off or it’s just the die-hard fans of terminal windows and editors like Emacs and Vim. Using the terminal is right up their alley—it’s in their DNA. Every time I read comments saying Claude Code is far better than Cursor, I fire it up, pay for a subscription, and run it on a large, complex TypeScript codebase. F…

Each use case, which for many is a project-by-project thing, likely determines the right tool for the job.

For new projects, I find Claude Code extremely helpful, as I start out with a business document, a top-level requirements document, and go from there. In the end (and with not a lot of work or time spent) I have a README, implementation plan, high-level architecture, milestones, and oftentimes a swagger spec, pipeline setup and a test harness.

IMHO pointing CC at a folder of a big typescript project is going to waste a ton of compute and tokens, for minimal value. That is not a good use of this tool. I also have a pretty strong opinion that a large, complex typescript codebase is a bad idea for humans too.

Point CC at a python or go repo and it is a whole 'nother experience. Also, starting out is where CC really shines as stated above.

For a big complex typescript repo I would want very specific, targeted help as opposed to agentic big-picture stuff. But that also minimizes the very reason I'd be reaching for help in the first place.

Re: My experience with Claude Code after two weeks of adventures

#305

Post by leptons »

danielbln wrote: post #211
vitaflo wrote: post #206

Earlier quoted context omitted.

It’s always telling when people don’t show their work. I’m not saying LLMs can’t do a good job but if you’re not even explaining the steps you used or showing the code that was generated or fixed then I have to assume what was really produced was unmaintainable spaghetti code that just happened to compile.

I literally cannot share my code, as that belongs to the company or the client. So what would I show? My prompts? Pseudo code? Unless you're an open source developer or build personal projects, requests for "show your code, bro" are hard to satisfy.

That's fine, but then don't write an article about it if you can't show the code. The vagueness just makes the article look unsupported by facts.

Re: My experience with Claude Code after two weeks of adventures

#306

Post by danielbln »

moomoo11 wrote: post #149
gjsman-1000 wrote: post #30

Earlier quoted context omitted.

> It’s the first time it really does feel like working with a junior engineer to me. I have mixed feelings; because this means there’s really no business reason to ever hire a junior; but it also (I think) threatens the stability of senior level jobs long term, especially as seniors slowly lose their knowledge and let Claude take care of things. The result is basically: When did you get into this field, by year? I’m…

We are using probabilistic generators to output what should be deterministic solutions.

You know what else is probabilistic? You and me. That's why we have tooling in place to mitigate that and to constrain our variable outputs into more reliable, deterministic results. And luckily, a lot of that tooling can be used for probabilistic machines as well.

Re: My experience with Claude Code after two weeks of adventures

#307

Post by leptons »

aniforprez wrote: post #130
TheRoque wrote: post #88

Earlier quoted context omitted.

100% agree, I have been looking for a YouTube video or stream of someone leveraging AI to get their productivity boost, but I haven't found anything that made me think "okay, that really speeds up things"

It was extremely telling to me that the Zed editor team put out a video about using their AI interface and I don't remember what model they used but they asked it to add a new toggle for a feature and then spent half they video demonstrating their admittedly excellent review workflow to accept or reject the AI generated code and you could directly see how useless it was down to adding completely superfluous new lines…

I asked the "AI" to do something relatively simple - fix when a button was enabled or disabled. And it then rewrote the entire 200+ line file, which then did not run, it was completely broken. Nice job "AI"!

Re: My experience with Claude Code after two weeks of adventures

#308

Post by pxc »

Myrmornis wrote: post #249
james_marks wrote: post #238

Earlier quoted context omitted.

Some of it is taste. Personally, I dislike VSCode and its descendants; it feels bloated and like I’m always being advertised to inside the product. I want my editor to disappear from thought, and Sublime does this for me. In that context, CC is amazing.

A lot of people don't realize that you can completely change the vscode UI. All the side bars in particular - just get rid of them. You're left with a blank screen on which syntax-highlighted code appears with LSP and either the command palette or if you want to be more vim/emacs-y then basically every operation is key-bindable.

I assume that on subsequent runs, it's less offensive, but every time I start a blank VSCode setup, I find myself repulsed by a series of pop-ups, persistent tooltips, extra tabs, etc., that are at best uninteresting and unwarranted. It definitely has strong Windows-y my-computing-environment-is-an-obstacle-course vibe.

Re: My experience with Claude Code after two weeks of adventures

#309

Post by leptons »

filoeleven wrote: post #264
ookblah wrote: post #121

Earlier quoted context omitted.

this sounds like a cop out, but honestly you will probably see the most vocal on both sides of this here while vast majority are just quietly doing working and doing their stuff (ironic i'm writing this). i feel like some kind of shill, but honestly i'm anywhere from 1.5x to 10x on certain tasks. the main benefit is that i can reduce a lot of cognitive load on tasks where they are either 1) exploratory 2) throwaway 3…

> i'm anywhere from 1.5x to 10x on certain tasks By your own reckoning. There was that recent study showing how senior devs using AI thought they were 20% faster, when they were objectively 20% slower.

They may spend 20% less time writing code but 20% more time explaining to the AI what they want and fixing the output. People are not good at basic math or time accounting, even programmers.

Re: My experience with Claude Code after two weeks of adventures

#310

Post by ruszki »

scrollaway wrote: post #233
crinkly wrote: post #229

Earlier quoted context omitted.

People pay me to do quality work not look busy. Thinking happens before you start.

People most likely pay you to do valuable work, not quality work. One doesn’t necessarily mean the other. If your coworkers can get into the groove faster than you (for whatever reason), you’ll continue to do quality work, not look busy, and no longer be paid for it.

Looking at my salary, which is at least twice as much as my coworkers in the past 15 years at various companies, because I ship quality solutions, companies seem to think that quality is quite valuable.
Post reply on HN