Stack新手构建并发网页抓取工具时的依赖配置问题

Hey there! Let’s work through these Stack dependency hurdles together—building a concurrent web scraper is a fun project, so let’s get those dependencies sorted out.

First: Fix the allow-newer Placement #

You mentioned adding allow-newer: true under the executable section didn’t work—that makes sense! allow-newer is a global/package-level setting, not an executable-specific one. You have two better options:

• Option 1: Add it to your stack.yaml (recommended)
  Open your project’s stack.yaml file and add this line at the top level:

  allow-newer: true
  

  This applies the setting to your entire project, which is more flexible than putting it in the cabal file.

• Option 2: Add it to the top of my-project.cabal
  If you prefer keeping it in the cabal file, place it before the executable block:

  allow-newer: true
  
  executable my-project
    ...
  

Second: Correct Your Dependency Syntax #

Writing http == * triggers a parsing error because Cabal doesn’t allow unconstrained == wildcards. Instead, use one of these approaches for your build-depends section in my-project.cabal:

• Allow any compatible version (simplest for beginners):

  build-depends:
    base >= 4.14 && < 5,
    hxt,
    url,
    http,
    maybet
  

  Stack will automatically pick the version of http that’s compatible with your resolver.

• Specify a minimum version if you need features from a newer release:

  build-depends:
    base >= 4.14 && < 5,
    hxt,
    url,
    http >= 4000.3.9,
    maybet
  

Third: Resolve the "http 4000.3.9 Not Found" Error #

When you specified http == 4000.3.9, Stack complained because that version isn’t included in your current resolver (the set of package versions Stack uses by default). Here’s how to fix this:

1. Check your current resolver by running:

   stack resolver
   

2. See which http versions are available in your resolver:

   stack exec ghc-pkg list http
   

3. If you really need 4000.3.9, add it as an extra-dep in stack.yaml:

   extra-deps:
     - http-4000.3.9
   

   This tells Stack to fetch and use that specific version even if it’s not in the resolver.

Alternatively, you could switch to a resolver that includes http-4000.3.9, but using extra-deps is usually easier for one-off package version overrides.

I remember tripping over these exact dependency issues when I first started with Stack—once you get the hang of resolvers and extra-deps, it gets way smoother. Good luck with your web scraper!

内容的提问来源于stack exchange，提问作者Mark Karavan