Re: [kbd] [PATCH 5/5] Replace error() by fprintf() and perror()

Alexey Gladkov wrote:
> I prefer to save compatibility with the error() function. It costs
> nothing, but it does not require to keep in mind the difference
> between two implementations.

There is a small cost (which might not be relevant for kbd):
The compiler does not know that kbd_error() will not return in most
cases and might emit warnings about possibly uninitialized variables
because of that.

Just to mentoin it...

Felix