Failure after updating OS: can the error log be better?

[tomjaguarpaw]

I updated my OS, and my linker changed. I knew this meant I had to reinstall GHC so I did so (through GHCup). After all these updates, HLS wasn't working properly. Symptom: no red/green squiggly lines on the various errors. I also thought I should reinstall HLS, which I did. No improvement.

To diagnose I looked at the buffer that Emacs calls *lsp-haskell::stderr*. Here are its contents: https://gist.github.com/tomjaguarpaw/dc103fd01784e0fbf1004523bb65b83c

I saw a lot of lines like this:

2025-12-27T19:18:46.269422Z | Debug | pragmas-suggest: Rule Failed: GhcSession
2025-12-27T19:18:46.269492Z | Debug | retrie: Rule Failed: TypeCheck
2025-12-27T19:18:46.269556Z | Debug | explicit-fields: Rule Failed: CollectRecords
2025-12-27T19:18:46.269620Z | Debug | changeTypeSignature: Rule Failed: GetParsedModule
2025-12-27T19:18:46.269684Z | Debug | overloaded-record-dot: Rule Failed: CollectRecordSelectors
2025-12-27T19:18:46.269755Z | Debug | gadt: Rule Failed: GetParsedModuleWithComments
2025-12-27T19:18:46.269818Z | Debug | importLens: Rule Failed: ImportActions
2025-12-27T19:18:46.269881Z | Debug | qualifyImportedNames: Rule Failed: TypeCheck
2025-12-27T19:18:46.269942Z | Debug | alternateNumberFormat: Rule Failed: CollectLiterals
2025-12-27T19:18:46.270010Z | Debug | pragmas-disable: Rule Failed: GhcSession

Searching for "Rule failed" I found #4034 (comment), and so deleted ~/.cache/ghcide. That didn't help. Luckily ChatGPT gave me something to try

haskell-language-server-wrapper --debug .

That showed an obvious error:

  Failed to run
  ["cabal","v2-repl","/home/tom/Haskell/bluefin/bluefin-examples/src/Bluefin/Examples/Stream/InsideAndOut.hs"]
  in directory "/home/tom/Haskell/bluefin". Consult the logs for full command and error.
  Failed command: cabal
  --builddir=/home/tom/.cache/hie-bios/dist-bluefin-e0a07456644d29ba7cac91a1bc6b2c92 v2-repl
  --with-compiler /home/tom/.cache/hie-bios/wrapper-b54f81dea4c0e6d1626911c526bc4e36 --with-hc-pkg
  /home/tom/.cache/hie-bios/ghc-pkg-144f893acdeed51210386e543bf1db9a
  /home/tom/Haskell/bluefin/bluefin-examples/src/Bluefin/Examples/Stream/InsideAndOut.hs
...
  Cannot find the program 'ld'. User-specified path '/usr/bin/ld.gold' does not refer to an
  executable and the program is not on the system path.

Ah! So the problem was the ~/.cache/hie-bios was stale too. If I'd looked closely enough I would have found #4583 (comment) that would have told me the same thing. Deleting that cache directory fixed the problem 🥳

You can even see that ld.gold is mentioned in Emacs' error log: https://gist.github.com/tomjaguarpaw/dc103fd01784e0fbf1004523bb65b83c#file-lsp-haskell-stderr-L567

---

My question I want to ask with this issue is: could the error log be better? I found the error log, I looked through it, but it was so verbose I missed the exact root cause that was contained explicitly in that log. There are around 500 lines that occur after the root cause line. Why? If HLS can't build the project shouldn't it just stop there and then, and say so?

[fendor]

Hi! Thank you for the bug report!

I agree that the logs are often too verbose and not informational enough at the same time.
So, yes, there is definitely room for improvement!
For your particular issue, I have the following thoughts:

• You used Debug verbosity, of course you get flooded by messages. Do you recall why you changed that? There would have been fewer messages in between.
• Setup issues, such as the one you observe should have been bubbled up to the editor, i.e. in one of your editor buffers, you have been able to see roughly that error message as a diagnostic on the first line of the buffer.

It is generally hard to improve this particular situation. HLS cant tell what the cause of a cabal setup error is, whether it is transient or permanent.
We do not want to just stop, because if the error is caused by the former, then, for example, changing the .cabal file should eventually cause HLS to reload correctly. If we stopped, it would be quite annoying to fix these transient issues.

Things we could easily improve, we could offer a way to clean cache directories from the HLS CLI and generally improve our FAQs so you don't need to consult chatgpt in these kinds of scenarios to figure out what is going on.
Note, this would still not work if you poisoned your .cabal/store directory, which can easily happen on a system upgrade.

[tomjaguarpaw]

You used Debug verbosity, of course you get flooded by messages. Do you recall why you changed that? There would have been fewer messages in between.

Interesting, I have no idea how that is set, and I'm not aware I set it myself. Is that configured through the editor? Perhaps it is just the default for Emacs's lsp mode.

Setup issues, such as the one you observe should have been bubbled up to the editor

Hmm, I looked at all the Emacs buffers that seemed relevant but I didn't see this error message specifically.

---

I'd be interested to know what VS Code, say, reports in this situation. Perhaps it's just an Emacs/lsp mode problem (or a "@tomjaguarpaw doesn't know how to use Emacs/lsp problem").

[fendor]

Perhaps it is an emacs issue, I dont know :) In VSCode and zed, you have to opt into the verbosity.

With zed it looks like this:

VSCode (hover message)

While the logs in zeditor look arguably even worse, there are very few logs between the last action and the actual error:

---

// Send:
{"jsonrpc":"2.0","id":86,"method":"textDocument/codeAction","params":{"textDocument":{"uri":"file:///home/hugin/Documents/haskell/aoc24/app/App3.hs"},"range":{"start":{"line":0,"character":0},"end":{"line":0,"character":0}},"context":{"diagnostics":[{"range":{"start":{"line":0,"character":0},"end":{"line":1,"character":0}},"severity":1,"source":"cradle","message":"Failed to run cabal '--builddir=/home/hugin/.cache/hie-bios/dist-aoc24-162a37c5da0b065a4c3b2b26a3fa8017' v2-repl --with-compiler /home/hugin/.cache/hie-bios/wrapper-b54f81dea4c0e6d1626911c526bc4e36 --with-hc-pkg /home/hugin/.cache/hie-bios/ghc-pkg-43915caf88959ca8151890c62f324d5a --keep-temp-files --enable-multi-repl /home/hugin/Documents/haskell/aoc24/app/App3.hs in directory \"/home/hugin/Documents/haskell/aoc24\". Consult the logs for full command and error.\nFailed command: cabal '--builddir=/home/hugin/.cache/hie-bios/dist-aoc24-162a37c5da0b065a4c3b2b26a3fa8017' v2-repl --with-compiler /home/hugin/.cache/hie-bios/wrapper-b54f81dea4c0e6d1626911c526bc4e36 --with-hc-pkg /home/hugin/.cache/hie-bios/ghc-pkg-43915caf88959ca8151890c62f324d5a --keep-temp-files --enable-multi-repl /home/hugin/Documents/haskell/aoc24/app/App3.hs\nResolving dependencies...\n\nConfiguration is affected by the following files:\n- cabal.project\n- cabal.project.local\nWarning: The package list for 'hackage.haskell.org' is 15 days old.\nRun 'cabal update' to get the latest list of available packages.\nError: [Cabal-7107]\nCould not resolve dependencies:\n[__0] trying: aoc24-0.1.0.0 (user goal)\n[__1] next goal: base (dependency of aoc24)\n[__1] rejecting: base-4.20.2.0/installed-127e (conflict: aoc24 => base>=4.14 && <4.16)\n[__1] skipping: base; 4.21.0.0, 4.20.2.0, 4.20.1.0, 4.20.0.1, 4.20.0.0, 4.19.2.0, 4.19.1.0, 4.19.0.0, 4.18.3.0, 4.18.2.1, 4.18.2.0, 4.18.1.0, 4.18.0.0, 4.17.2.1, 4.17.2.0, 4.17.1.0, 4.17.0.0, 4.16.4.0, 4.16.3.0, 4.16.2.0, 4.16.1.0, 4.16.0.0 (has the same characteristics that caused the previous version to fail: excluded by constraint '>=4.14 && <4.16' from 'aoc24')\n[__1] rejecting: base; 4.15.1.0, 4.15.0.0, 4.14.3.0, 4.14.2.0, 4.14.1.0, 4.14.0.0, 4.13.0.0, 4.12.0.0, 4.11.1.0, 4.11.0.0, 4.10.1.0, 4.10.0.0, 4.9.1.0, 4.9.0.0, 4.8.2.0, 4.8.1.0, 4.8.0.0, 4.7.0.2, 4.7.0.1, 4.7.0.0, 4.6.0.1, 4.6.0.0, 4.5.1.0, 4.5.0.0, 4.4.1.0, 4.4.0.0, 4.3.1.0, 4.3.0.0, 4.2.0.2, 4.2.0.1, 4.2.0.0, 4.1.0.0, 4.0.0.0, 3.0.3.2, 3.0.3.1 (constraint from non-reinstallable package requires installed instance)\n[__1] fail (backjumping, conflict set: aoc24, base)\nAfter searching the rest of the dependency tree exhaustively, these were the goals I've had most trouble fulfilling: base, aoc24\n\n\n\nProcess Environment:\nHIE_BIOS_GHC: /home/hugin/.ghcup/bin/ghc\nHIE_BIOS_GHC_ARGS: -B/home/hugin/.ghcup/ghc/9.10.3/lib/ghc-9.10.3/lib","data":{"cabalProjectFiles":["/home/hugin/Documents/haskell/aoc24/aoc24.cabal","/home/hugin/Documents/haskell/aoc24/cabal.project","/home/hugin/Documents/haskell/aoc24/cabal.project.local"]}}],"only":["","quickfix","refactor","refactor.extract","source"]}}}


// Send:
{"jsonrpc":"2.0","id":87,"method":"textDocument/codeAction","params":{"textDocument":{"uri":"file:///home/hugin/Documents/haskell/aoc24/app/App3.hs"},"range":{"start":{"line":0,"character":0},"end":{"line":0,"character":0}},"context":{"diagnostics":[{"range":{"start":{"line":0,"character":0},"end":{"line":1,"character":0}},"severity":1,"source":"cradle","message":"Failed to run cabal '--builddir=/home/hugin/.cache/hie-bios/dist-aoc24-162a37c5da0b065a4c3b2b26a3fa8017' v2-repl --with-compiler /home/hugin/.cache/hie-bios/wrapper-b54f81dea4c0e6d1626911c526bc4e36 --with-hc-pkg /home/hugin/.cache/hie-bios/ghc-pkg-43915caf88959ca8151890c62f324d5a --keep-temp-files --enable-multi-repl /home/hugin/Documents/haskell/aoc24/app/App3.hs in directory \"/home/hugin/Documents/haskell/aoc24\". Consult the logs for full command and error.\nFailed command: cabal '--builddir=/home/hugin/.cache/hie-bios/dist-aoc24-162a37c5da0b065a4c3b2b26a3fa8017' v2-repl --with-compiler /home/hugin/.cache/hie-bios/wrapper-b54f81dea4c0e6d1626911c526bc4e36 --with-hc-pkg /home/hugin/.cache/hie-bios/ghc-pkg-43915caf88959ca8151890c62f324d5a --keep-temp-files --enable-multi-repl /home/hugin/Documents/haskell/aoc24/app/App3.hs\nResolving dependencies...\n\nConfiguration is affected by the following files:\n- cabal.project\n- cabal.project.local\nWarning: The package list for 'hackage.haskell.org' is 15 days old.\nRun 'cabal update' to get the latest list of available packages.\nError: [Cabal-7107]\nCould not resolve dependencies:\n[__0] trying: aoc24-0.1.0.0 (user goal)\n[__1] next goal: base (dependency of aoc24)\n[__1] rejecting: base-4.20.2.0/installed-127e (conflict: aoc24 => base>=4.14 && <4.16)\n[__1] skipping: base; 4.21.0.0, 4.20.2.0, 4.20.1.0, 4.20.0.1, 4.20.0.0, 4.19.2.0, 4.19.1.0, 4.19.0.0, 4.18.3.0, 4.18.2.1, 4.18.2.0, 4.18.1.0, 4.18.0.0, 4.17.2.1, 4.17.2.0, 4.17.1.0, 4.17.0.0, 4.16.4.0, 4.16.3.0, 4.16.2.0, 4.16.1.0, 4.16.0.0 (has the same characteristics that caused the previous version to fail: excluded by constraint '>=4.14 && <4.16' from 'aoc24')\n[__1] rejecting: base; 4.15.1.0, 4.15.0.0, 4.14.3.0, 4.14.2.0, 4.14.1.0, 4.14.0.0, 4.13.0.0, 4.12.0.0, 4.11.1.0, 4.11.0.0, 4.10.1.0, 4.10.0.0, 4.9.1.0, 4.9.0.0, 4.8.2.0, 4.8.1.0, 4.8.0.0, 4.7.0.2, 4.7.0.1, 4.7.0.0, 4.6.0.1, 4.6.0.0, 4.5.1.0, 4.5.0.0, 4.4.1.0, 4.4.0.0, 4.3.1.0, 4.3.0.0, 4.2.0.2, 4.2.0.1, 4.2.0.0, 4.1.0.0, 4.0.0.0, 3.0.3.2, 3.0.3.1 (constraint from non-reinstallable package requires installed instance)\n[__1] fail (backjumping, conflict set: aoc24, base)\nAfter searching the rest of the dependency tree exhaustively, these were the goals I've had most trouble fulfilling: base, aoc24\n\n\n\nProcess Environment:\nHIE_BIOS_GHC: /home/hugin/.ghcup/bin/ghc\nHIE_BIOS_GHC_ARGS: -B/home/hugin/.ghcup/ghc/9.10.3/lib/ghc-9.10.3/lib","data":{"cabalProjectFiles":["/home/hugin/Documents/haskell/aoc24/aoc24.cabal","/home/hugin/Documents/haskell/aoc24/cabal.project","/home/hugin/Documents/haskell/aoc24/cabal.project.local"]}}],"only":["","quickfix","refactor","refactor.extract","source"]}}}

// Send:
{"jsonrpc":"2.0","id":88,"method":"textDocument/codeAction","params":{"textDocument":{"uri":"file:///home/hugin/Documents/haskell/aoc24/aoc24.cabal"},"range":{"start":{"line":27,"character":37},"end":{"line":27,"character":37}},"context":{"diagnostics":[],"only":["","quickfix","refactor","refactor.extract","source"]}}}

// Receive:
{"id":88,"jsonrpc":"2.0","result":[]}

// Receive:
{"id":86,"jsonrpc":"2.0","result":[]}

// Receive:
{"id":87,"jsonrpc":"2.0","result":[]}

// Send:
{"jsonrpc":"2.0","id":89,"method":"textDocument/codeLens","params":{"textDocument":{"uri":"file:///home/hugin/Documents/haskell/aoc24/app/App3.hs"}}}

// Send:
{"jsonrpc":"2.0","id":90,"method":"textDocument/codeLens","params":{"textDocument":{"uri":"file:///home/hugin/Documents/haskell/aoc24/app/App3.hs"}}}

// Send:
{"jsonrpc":"2.0","id":91,"method":"textDocument/codeLens","params":{"textDocument":{"uri":"file:///home/hugin/Documents/haskell/aoc24/aoc24.cabal"}}}

// Receive:
{"error":{"code":-32601,"message":"No plugins are available to handle this SMethod_TextDocumentCodeLens request.\n Plugins installed for this method, but not available to handle this request are:\neval does not support .cabal filetypes).\nghcide-type-lenses does not support .cabal filetypes).\nimportLens does not support .cabal filetypes).\nclass does not support .cabal filetypes).\nmoduleName does not support .cabal filetypes)."},"id":91,"jsonrpc":"2.0"}

// Receive:
{"error":{"code":-32803,"message":"moduleName: Rule Failed: GhcSession"},"id":89,"jsonrpc":"2.0"}

// Receive:
{"error":{"code":-32803,"message":"moduleName: Rule Failed: GhcSession"},"id":90,"jsonrpc":"2.0"}

Just as an example.

[tomjaguarpaw]

Ah thanks. It seems that lsp is much worse at error reporting then. There was no reported error in the main buffer at all. I had to go looking for it. I guess this is a not an HLS problem then. Shall we close?

[fendor]

I think we should expand parts of the documentation or FAQ to include at least the cache directories...
So, perhaps lets not close it just yet.

[tomjaguarpaw]

Yes, actually it would be good to have a clear way to explain how to purge cache (and maybe have a tool for it, if it's complicated?).

[dschrempf]

@tomjaguarpaw I am also an Emacs user. Do you think we should open a ticket upstream (lsp-mode, eglot, what are you using?). Could you provide more detail on how we could improve the Emacs packages?

[tomjaguarpaw]

Hi @dschrempf, I am using lsp-mode. I don't really know much about the emacs community or how its packages are developed. I'm reluctant to open an issue because I don't know if it would be welcome, or if it would be welcome, exactly what the best form of feedback for me would be.

The biggest problem I have when using HLS is that when it doesn't work, or when bits of it seem not to work, I have no idea how to track down the root cause. I don't know if that's an HLS problem, but from the discussion on this ticket it sounds like a lot of the responsibility is on lsp-mode.