can command line utilities be faster than C++? [closed]
I have a project where I want to manipulate certain output files.
This can be accomplished using a combination of grep and sed and piping with |
Alternatively, I can also write a C++ program to do the same thing.
Is there a conclusive answer on which method will be faster since grep and sed should already be fairly well optimised?
From a technical standpoint, a well-written self-contained C++ program that does everything you need will be faster than using two (or more) shell commands interconnected with a pipe, simply because there will be no IPC overhead, and they can be tailor-made and optimized for your exact needs.
But unless you're writing a program that will be run 24/7 for years, you'll never notice enough gain to be worth the effort.
And the standard rules for pre-optimization apply...
If I were you, use what is already out there as these have likely been around a long time and have been tested and tried. Writing a new program yourself to do the same thing seems like a reinventing the wheel type action and is prone to error.
If you really need faster performance than you'll get with piping, you can download the source for grep and sed and tailor it to your needs in one application (be wary of licenses if you plan on distributing your code). I'd be highly surprised if you'd even notice the overhead of piping (like Flimzy mentioned), so if things are really that slow I'd start profiling your app.
It is likely that if you are a very good C/C++ programmer and spend a lot of time, that you will be able to write a program that's faster than the pipeline you're thinking of. But unless performance is so critical in this case that you absolutely must do it this way you should use the pipeline.
精彩评论