fix: adapt to llama.cpp breaking changes

Author: giladgd

src/bindings/utils/compileLLamaCpp.ts

@@ -131,9 +131,14 @@ export async function compileLlamaCpp(buildOptions: BuildOptions, compileOptions
                 if (!cmakeCustomOptions.has("GGML_CCACHE"))
                     cmakeCustomOptions.set("GGML_CCACHE", "OFF");
 
-                if (!cmakeCustomOptions.has("LLAMA_CURL"))
+                if (!cmakeCustomOptions.has("LLAMA_CURL") || isCmakeValueOff(cmakeCustomOptions.get("LLAMA_CURL"))) {
                     cmakeCustomOptions.set("LLAMA_CURL", "OFF");
 
+                    // avoid linking to extra libraries that we don't use
+                    if (!cmakeCustomOptions.has("LLAMA_OPENSSL"))
+                        cmakeCustomOptions.set("LLAMA_OPENSSL", "OFF");
+                }
+
                 if (buildOptions.platform === "win" && buildOptions.arch === "arm64" && !cmakeCustomOptions.has("GGML_OPENMP"))
                     cmakeCustomOptions.set("GGML_OPENMP", "OFF");