[castler]

Summary

Fixes all 5 readability-redundant-string-cstr findings (shared_memory_resource.cpp x2, configuration_json_parsing_strategy.cpp, tracing_filter_config.cpp x2) by passing the std::string directly instead of .c_str()/.data().

Verification

• clang-tidy: no remaining findings for this check
• bazel build //score/... and bazel test //score/memory/... //score/mw/com/impl/... pass

[castler]

I personally think, we should rather deactivate the check...as it does not provide any value to me.

[limdor]

I personally think, we should rather deactivate the check...as it does not provide any value to me.

I would agree with clang tidy change, but the I do not think it is well explain and the reason might be different than the checker claim.
The problem is that if you call .data() or .c_str() you loose any length information. The string needs to be reconstructed with a new lookup for a null termination character. If this would be a string_view it would be worst, because there is no guarantee on being null terminated. For this second reason, in some other projects I was even flagging every call to .data() and c_str because most of the time there is a better safer alternative.
I would just accept the change and leave the checker.