> On Sept. 17, 2014, 4:32 p.m., Dominic Hamon wrote:
> > I'm not convinced that this is necessarily the way we want to go, given the 
> > benefits of move semantics in modern compilers, but for consistency we can 
> > ship this until we have proof that performance favours the alternative.
> 
> Kamil Domanski wrote:
>     Isn't move construction more expensive than const reference?
> 
> Dominic Hamon wrote:
>     
> http://stackoverflow.com/questions/21605579/how-true-is-want-speed-pass-by-value
>     
>     copy, yes. move, no. maybe. ;)

http://www.youtube.com/watch?v=xnqTKD8uD64#t=3145 - the current guidelines. 
Passing by value probably isn't what you want. If you want to get the extra win 
for r-value referenes you should add another overload, or accept via perfect 
forwarding.

Also note GCC libstdc++ doesn't implement the C++11/C++14 std::string, so it is 
Copy on Write (C++89) which means that copies are a little cheaper than they 
would be otherwise.


- Cody


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/25735/#review53688
-----------------------------------------------------------


On Sept. 17, 2014, 2:25 p.m., Kamil Domanski wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/25735/
> -----------------------------------------------------------
> 
> (Updated Sept. 17, 2014, 2:25 p.m.)
> 
> 
> Review request for mesos.
> 
> 
> Bugs: MESOS-1805
>     https://issues.apache.org/jira/browse/MESOS-1805
> 
> 
> Repository: mesos-git
> 
> 
> Description
> -------
> 
> os::shell and an overload of strings::internal::fmt in stout pass a const 
> string parameter by copy instead of reference. This patch fixes that.
> 
> 
> Diffs
> -----
> 
>   3rdparty/libprocess/3rdparty/stout/include/stout/format.hpp 58ab742 
>   3rdparty/libprocess/3rdparty/stout/include/stout/os/shell.hpp 6728ad8 
> 
> Diff: https://reviews.apache.org/r/25735/diff/
> 
> 
> Testing
> -------
> 
> cd 3rdparty/libprocess/3rdparty/ && make check
> 
> 
> Thanks,
> 
> Kamil Domanski
> 
>

Reply via email to