brianosman@ - It's not clear to me whether or not this increase was expected. However, we already have an issue tracking adding noinline to std::string constructors so there might not be anything to do here at this time.

Feel free to close as "Won't Fix" if you think the performance gains are worth the 45 kb while we look at changing std::string itself.

Thanks for the heads up. Yes, the performance wins (within the framework, and elsewhere) were pretty significant. Getting a std::string fix globally would be nice, and we're still looking at follow-up changes to get the same perf wins in other ways. For now, we'd like to keep this in, though.