rockot, which labels cause mojo triage?

Component :)

I agree String should be movable. I think our plan is to replace any use of mojo::String with either std::string or WTF::String depending on whom we're generating bindings for. Will leave this bug open at least until that's sorted though.

There is an issue in the current mojo bindings generator that strings in callback parameter lists are mapped to mojo::String passed-by-value incorrectly.

I am planning to change that to const&, so that mojo::String is always passed by const& in mojom generated code. (In this pending CL, I have fixed the behavior for WTF::String and added a TODO for mojo::String: https://codereview.chromium.org/1751563002/)

It doesn't hurt to add move support for mojo::String. I will do that.

Ken: Now that we are not considering to replace mojo::Array/Map with std::vector/map (or scoped_ptr<std::vector>/scoped_ptr<std::map> in the nullable case), I think we may want to continue using mojo::String, rather than switching to std::string/scoped_ptr<std::string>? That way it is more consistent. I think mojo::String is pretty convenient for users and shouldn't have any perf impact. WDYT? Thanks!

Ah SGTM

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/0b3918953d5aac5a179fca9370a17a801f45d874

commit 0b3918953d5aac5a179fca9370a17a801f45d874
Author: yzshen <yzshen@chromium.org>
Date: Mon May 23 22:56:30 2016

Mojo C++ bindings: move support for mojo::String.

BUG= 593220 

Review-Url: https://codereview.chromium.org/2006823002
Cr-Commit-Position: refs/heads/master@{#395448}

[modify] https://crrev.com/0b3918953d5aac5a179fca9370a17a801f45d874/mojo/public/cpp/bindings/string.h
[modify] https://crrev.com/0b3918953d5aac5a179fca9370a17a801f45d874/mojo/public/cpp/bindings/tests/string_unittest.cc

Here's a vote for fewer string types. Unless you have very very convincing data that proves that using std::string cannot possibly work for you, please use std::string, like the rest of the codebase.