Cpp tagged strings

[naomijub]

In the book section helpful idiom we present the following raw C++:

(cpp/raw "size_t get_string_size(std::string const &s)
          { return s.size(); }")

My suggestion is limiting embbeded c++ code this to a tagged string with #cpp, similar to what we do in:

• tagged-literals
• #js [..] in clojurescript
• #uuid and #inst in clojure

How it would look like:

(cpp/raw #cpp "size_t get_string_size(std::string const &s)
          { return s.size(); }")

(cpp/raw #cpp "struct person
          { std::string name; };")

The main benefit I see is reducing cpp/raw surface area when creating external cpp code, as well as being easier to handle and manipulate this on the editor level, which would improve the syntax highlight for C++: #24

[jeaye]

Hi! Thanks for the proposal. Others are welcome to join in, but I will share my thoughts.

jank currently supports #cpp strings, such as #cpp "meow". However, this doesn't mean "A string containing C++ code". It means "A C++ literal string." For example:

(let [s #cpp "meow"] ; s is a char const [5]
  (println s))

We also support this for other primitive literals, such as #cpp 5, #cpp true, #cpp 3.14, etc. This allows jank users to work directly with C++ primitive types, rather than boxed jank objects.

So, given that knowledge, using #cpp for C++ strings would be inconsistent with the approach.

However, secondly, I think that #cpp within cpp/raw is also redundant. As we have seen in #24 (which you also linked), Emacs has already added support for highlighting C++ strings within jank's old native/raw. As far as I know, that just needs to be updated to cpp/raw. I think it's very clear, in terms of grammar, that the string coming after cpp/raw contains C++ code.

Altogether, I think this would require more typing for the users, be a frequent mistake for them to forget the #cpp, be inconsistent with our current #cpp usage, and not be necessary for providing good syntax highlighting.