From this SO answer it is my understanding that the following C code is bad practice because declaring c as a char will make a char like 0xFF and the EOF constant indistinguishable.

char c;

while ((c = getchar()) != EOF) {   
    ...
}

However in this SO answer detailing the type conversion process when two integral operands of different types are being operated on (in this case with !=), the two types are converted to a common type. Typically to the type with more bits.

Question 1:
Why then if the char 0xFF was found in a file would it not be converted to type int, and then compared to EOF?

Question 2:
In a case where the char 0xFF, or, in binary, 1111 1111 is converted to an int (let’s say 16 bits for simplicity), does it then become 0000 0000 1111 1111? If so, would a signed char lose its sign and change in decimal value from -1 to 255?