why realsense should code like this? auto&& frame : pipe.wait_for_frames()

[FritzY]

i saw the code example of IntelRealSense/librealsense/examples/save-to-disk/rs-save-to-disk.cpp from github.

link is https://github.com/IntelRealSense/librealsense/blob/master/examples/save-to-disk/rs-save-to-disk.cpp#L34

 line 34  for (auto&& frame : pipe.wait_for_frames())

​why code like this rather "auto frame = pipe.wait_for_frames()"

and line 37 if (auto vf = frame.as<rs2::video_frame>())

rather "auto vf = frame.get_depth_frame()"

is any better to code this way?

[FritzY]

is it in order to minimize the memory cost?

[MartyG]

The link below explains why wait_for_frames is used and the alternatives to it.

https://github.com/IntelRealSense/librealsense/issues/2422#issuecomment-423254709

Using 'video' frame is an appropriate choice when only saving data as a simple PNG, and not a file format with embedded depth matrix information such as .raw. or .bin

[FritzY]

Thanks for reply, it 's really helpful.

But what really disturb me is why it use A reference " auto &&" to define frame, is it a better(tidy) wrriting than using " = " to define an argument(the way we usually do)?and why?

then using "as" methond to change the class of frame. on the contrary, I may use "get_depth_frame" function straightly if my purpose is to get a depth image.

Perhaps it belongs to a c++ question boundary, but i would appriciate it if you give me some suggestions.

[MartyG]

My understanding is that "auto &&" makes it a 'universal reference' that can be bound to anything.

When programming Librealsense, my advice is to use example programs as a base reference and then adjust the code to your own style if you believe that your own method of structuring code is more suitable. Librealsense is open-source exactly so users can mold it to their needs and preferences.

[FritzY]

My opinion is maybe it's a memory saving way to code like this. Anyway, thanks you very much.