[BUG] Resource::Create() throws bad_variant_access if process.executable.name isn't a string

[brhenc]

Describe your environment

Reproduced by source read + build on main at 11fa0db0 (also present in v1.28.0, the latest release -- not a regression). sdk/src/resource/resource.cc, Resource::Create.

Steps to reproduce

Resource Resource::Create(const ResourceAttributes &attributes, const std::string &schema_url)
{
  ...
  if (resource.attributes_.find(semconv::service::kServiceName) == resource.attributes_.end())
  {
    std::string default_service_name = "unknown_service";
    auto it_process_executable_name =
        resource.attributes_.find(semconv::process::kProcessExecutableName);
    if (it_process_executable_name != resource.attributes_.end())
    {
      default_service_name += ":" + nostd::get<std::string>(it_process_executable_name->second);
    }
    resource.attributes_[semconv::service::kServiceName] = default_service_name;
  }
  return resource;
}

it_process_executable_name->second is an AttributeValue (nostd::variant). nostd::get<std::string> throws if the variant doesn't currently hold a std::string. process.executable.name can end up holding a different alternative than std::string in more than one way: a custom ResourceDetector setting it directly, an environment/config-driven attribute source, or simply a caller passing it in attributes as const char*/string_view/an integer by mistake -- none of those are prevented by the ResourceAttributes type itself.

To trigger: call Resource::Create(attributes, schema_url) with attributes containing {"process.executable.name", <anything that isn't a std::string>} and no service.name already set.

What is the expected behavior?

If process.executable.name is present but not a string, Resource::Create should fall back to just "unknown_service" (or otherwise degrade gracefully) rather than crash.

What is the actual behavior?

nostd::get<std::string> throws nostd::bad_variant_access. Since Resource::Create (and its callers up through TracerProvider/LoggerProvider/MeterProvider construction) isn't inside any exception handling, this propagates out and terminates initialization -- often during process/provider startup or a reload, i.e. exactly when an application least wants an unhandled crash.

Additional context

Suggested fix -- use nostd::get_if instead of the throwing nostd::get, and simply skip the suffix if the attribute isn't a string:

     if (it_process_executable_name != resource.attributes_.end())
     {
-      default_service_name += ":" + nostd::get<std::string>(it_process_executable_name->second);
+      if (const auto *executable_name =
+              nostd::get_if<std::string>(&it_process_executable_name->second))
+      {
+        default_service_name += ":" + *executable_name;
+      }
     }

Compile-checked against a clean build of this file -- no warnings or errors. Happy to open a PR with this if useful.

[github-actions]

This issue is available for anyone to work on. Make sure to reference this issue in your pull request.
✨ Thank you for your contribution! ✨