test: ipc: resolve symlinks in which capnp #33749

1. 
   davidgumberg commented at 9:34 pm on October 30, 2025: contributor

   On Fedora, /bin/ and /usr/bin are symlinked, and on one of my boxes (although I could not reproduce this behavior in a docker container), /bin comes before /usr/bin in $PATH, so which capnp reports /bin/capnp, and capnp_dir is set to /include, and the test fails:

   0$ ./build/test/functional/interface_ipc.py
   12025-10-30T20:43:43.753812Z TestFramework (INFO): PRNG seed is: 8370468257027235753
   22025-10-30T20:43:43.754163Z TestFramework (INFO): Initializing test directory /tmp/bitcoin_func_test_b9kjzj2a
   3terminate called after throwing an instance of 'kj::ExceptionImpl'
   4  what():  mp/proxy.capnp:6: failed: Import failed: /capnp/c++.capnp
   5Aborted (core dumped)
   

   This changes the functional test to resolve any symlinks in the capnp binary path reported by which.

2. test: resolve symlinks in which result for capnp 51093d6ae1
3. DrahtBot added the label Tests on Oct 30, 2025
4. 
   DrahtBot commented at 9:34 pm on October 30, 2025: contributor

   The following sections might be updated with supplementary metadata relevant to reviewers and maintainers.

   Code Coverage & Benchmarks

   For details see: https://corecheck.dev/bitcoin/bitcoin/pulls/33749.

   Reviews

   See the guideline for information on the review process.

   Type	Reviewers
   ACK	TheCharlatan, ryanofsky

   If your review is incorrectly listed, please react with 👎 to this comment and the bot will ignore it on the next update.

5. TheCharlatan approved
6. 
   TheCharlatan commented at 8:01 am on October 31, 2025: contributor

   utACK 51093d6ae1d415b8dfa47f11cb634a87bd97e8a9
7. 
   Sjors commented at 11:35 am on October 31, 2025: member

   I tested that it doesn’t interfere with my macOS setup (/opt/homebrew/bin/capnp).
8. 
   fanquake commented at 3:43 pm on October 31, 2025: member

   cc @ryanofsky
9. ryanofsky approved
10. 
    ryanofsky commented at 4:20 pm on October 31, 2025: contributor

    Code review ACK 51093d6ae1d415b8dfa47f11cb634a87bd97e8a9

    I was wondering if this should use .parent.resolve().parent instead of .resolve().parent.parent but I could imagine resolving first being better if someone has a bin directory with symlinks to installs in different prefixes (like nix or some virtual environments have). There probably isn’t a perfect answer but this seems like an improvement.

11. fanquake merged this on Oct 31, 2025
12. fanquake closed this on Oct 31, 2025

    ---

Contributors
davidgumberg DrahtBot TheCharlatan Sjors fanquake ryanofsky

Labels
Tests