Swift/Xcode build errors since #4935: `spm-headers/llama.h:235:9 Unknown type name 'ggml_backend_sched_eval_callback'`

[brittlewis12]

Since b1901 (rev 44a1a4a #4935), I've been unable to build llama.cpp with the provided Package.swift. I bisected back to this PR as the first breaking commit — b1900 (rev c918fe8) seems to work well.

I'm not terribly comfortable with C++, so to my eye I don't see anything wrong with the patch. I noticed the lack of GGML_API prefix like other types in the file, but I tried changing that locally and didn't notice a difference, though it's very possible I applied said change incorrectly/incompletely. EDIT: it seems committing this change indeed resolved the visibility issue I experienced as a result. Proposed a patch: #5030

I would be more than happy to share any other relevant information or test any proposed fixes to the extent that would be helpful!

---

Here's the full error log:

<module-includes>:1:9: note: in file included from <module-includes>:1:
#import "/Users/tito/code/llama.cpp/spm-headers/llama.h"
        ^
/Users/tito/code/llama.cpp/spm-headers/llama.h:235:9: error: unknown type name 'ggml_backend_sched_eval_callback'
        ggml_backend_sched_eval_callback cb_eval;

[ggerganov]

It should work without GGML_API - I think your changes coincided with a ggml package update. If you init cleanly the Swift project it should work without modifications

[brittlewis12]

That does indeed resolve the build errors I was seeing — thanks very much!

I wonder if this might also have been compounded by the way the ggml dependency is declared in this repo's Package.swift — it appears to use a branch specifier for "release", rather than mirroring the latest synced SHA in the repo. Could that have an effect if they get out of sync temporarily?

EDIT: I've read thru #4867 and appreciate the current design is to enable whisper.cpp & this repo to share the ggml dep.

[ggerganov]

Well, it's pretty much the same thing, because I manually update the release branch when I have to update the Swift package. Overall, the current process is not optimal, but I think we are converging to a stable solution and such kind of problems will happen less often

[brittlewis12]

I can very much appreciate the underlying complexity you're wrangling, and I have no doubt it will continue to improve with time. Thanks again for all your hard work on these fantastic projects!