Unnecessary copy of request body in multipart message

[bgs99]

When troubleshooting memory usage of our application using this library, I found that crow::multipart::message stores a copy of the uploaded multipart data that is already present in the crow::request.

This seems unnecessary, as (according to my understanding of the spec (https://www.w3.org/Protocols/rfc1341/7_2_Multipart.html) multipart data is not encoded, so it could simply reference the data in the crow::request.

This would significantly lower memory consumption of the request handlers when dealing with huge multipart file uploads.

Would you accept a string_view-based multipart message/parser to replace or complement the current one?

I have a working prototype that uses std::string_view from C++17 STL that I could contribute.
But as I understand, this library promises C++11 compatibility, so it is necessary to use a different implementation. Would you prefer using some existing string_view implementation or directly re-implementing one inside Crow itself?

[gittiver]

sounds not that it was intended that way.
C++17 discussion is a separate thing, I don't know who wants to decide that.
In my opinion in 2024 we could update to C++14 or 17.