hacker-news-custom-logo

Hackr News App

14 comments

  • jasonthorsness

     

    2 days ago

    prev

    next

    [ - ]

    Instead of adding a TODO or filing a ticket to find again later agent mode/background tasks/agents let you just spin off the actual work for doing the task. So you can keep your focus on the main thing you are working on but have something else go fix the bug you noticed, write the missing test, or even just fix a spelling mistake in a comment. It improves code quality by lowering effort required to prepare a change and work through lint/CI etc. In this way it’s great even though it can’t handle much complexity yet.

    reply
  • ComplexSystems

     

    2 days ago

    prev

    next

    [ - ]

    I have been using this with mostly decent results. I am curious how it compares to other IDEs though - Cursor, Windsurf, Roocode, etc. Any thoughts?

    reply

    CharlesW

     

    2 days ago

    parent

    next

    [ - ]

    [ x ]

    <@ComplexSystems> I personally like Windsurf a bit more than Cursor, but recently I've been far more productive using Claude Code with an IDE than I was using a VSC-derived AIDE.

    reply

    runekaagaard

     

    2 days ago

    root

    parent

    next

    [ - ]

    [ x ]

    <@CharlesW> My sweet spot at the moment is Claude Desktop with mcp servers for editing and aider --watch for quick fixes. Claude Code uses way, way, way too many tokens on the large project i work most on.

    reply

    himeexcelanta

     

    2 days ago

    root

    parent

    next

    [ - ]

    [ x ]

    <@runekaagaard> Get one of the max plans! It pays for itself.

    reply

    CharlesW

     

    2 days ago

    root

    parent

    prev

    next

    [ - ]

    [ x ]

    <@runekaagaard> > Claude Code uses way, way, way too many tokens on the large project i work most on.

    That's a very fair critique, and it makes the pay-as-you-go pricing model (vs. one of their subscription options) a completely unrealistic option for doing anything serious with Claude Code.

    reply

    dist-epoch

     

    2 days ago

    root

    parent

    next

    [ - ]

    [ x ]

    <@CharlesW> You don't get paid for programming?

    reply

    runekaagaard

     

    15 hours ago

    root

    parent

    prev

    next

    [ - ]

    [ x ]

    <@dist-epoch> Luckily i do, but i mean it triggers the api limit in 10 minutes amount of tokens

    reply

    Terretta

     

    17 hours ago

    root

    parent

    prev

    next

    [ - ]

    [ x ]

    <@CharlesW> > * I've been far more productive using Claude Code with an IDE*

    Curious why proportionately few seem to find this a better way of working.

    Does it have to do with familiarity with XP or pair programming? What had you trying in IDE first, and what prompted you to try switching?

    Have you used https://aider.chat to compare? If so, thoughts?

    reply
  • recursinging

     

    2 days ago

    prev

    next

    [ - ]

    This is cool, but still lagging behind VS Code. I still can't the enterprise/premium models in VS (I can in VS Code) so no Claude 4 Sonnet. I even went so far as to try some C# dev in VS Code just to see how the Claude 4 deals with C# - Meh. In comparison, I just wrapped up a C++ microcontroller project in VS Code using agent mode and it was amazing. In retrospect I recognize that Claude will resort to grep and other text based tools. This works well when all the sources are included, less so with Nuget packages. If the models can aim better at intellisense tooling, they'd be more effective. IMO Copilot agents in VS are not quite there, but not far off.

    reply
  • bionhoward

     

    2 days ago

    prev

    next

    [ - ]

    Who can take em seriously when cursor has privacy mode while copilot implicitly trains AI on copilot interactions with customers’ private repo code unless you get some enterprise tier of service?

    The core of GitHub is about version control, including private repos, the fact they would violate that trust relationship just to chase AI capabilities means they’ve totally lost the plot as far as I’m concerned.

    Maybe Copilot is a nice toy for open source projects or noobs working on private codebases who don’t take 5 minutes to read their website, but cursor just seems way ahead in terms of basic respect for customers

    reply

    voxic11

     

    2 days ago

    parent

    next

    [ - ]

    [ x ]

    <@bionhoward> > copilot implicitly trains AI on copilot interactions with customers’ private repo code unless you get some enterprise tier of service?

    Seems incorrect to me, from the copilot individual license docs:

    > By default, GitHub, its affiliates, and third parties will not use your data, including prompts, suggestions, and code snippets, for AI model training. This is reflected in your personal settings for GitHub Copilot and cannot be enabled.

    https://docs.github.com/en/copilot/managing-copilot/managing...

    reply

    dist-epoch

     

    2 days ago

    parent

    prev

    next

    [ - ]

    [ x ]

    <@bionhoward> Are you sure about that? I pay $10 for Copilot, that's it, no other subscription to GitHub, and it says my data is not used for training:

    Allow GitHub to use my data for product improvements: Unchecked

    Allow GitHub to use my data for AI model training: Disabled

    reply