Homebrew can't compile llama.cpp

[solomatov]

Many people use homebrew to manage and update llama.cpp. It's very convenient to update, and control everything from there.

Unfortunately, recently, build of llama.cpp started to fail.

Here's the run log: https://github.com/Homebrew/homebrew-core/actions/runs/24323950806/job/71015710159

The error is the following:

  /private/tmp/llama.cpp-20260412-6291-7hq766/src/llama.cpp:19:10: fatal error: '../src/ggml-ext.h' file not found
     19 | #include "../src/ggml-ext.h"
        |          ^~~~~~~~~~~~~~~~~~~
  1 error generated.

This is caused by this code:

// TODO: tmp until the ggml meta backend matures and becomes public
#include "../src/ggml-ext.h"

Here's the failing PR: Homebrew/homebrew-core#277280

[chuckwagoncomputing]

The same is true when installing llama.cpp-git from the Archlinux AUR.