The #line 1 "xdev.c" part indicates that the next line originally
came from line 1 of xdev.c . The empty line after that is what
was originally the first line of xdev.c but which got emptied because
it was a comment. The line after that is where stdio.h got included,
so the expansion of stdio.h begins with a marker #file 1 "stdio.h"
saying that the next line originally came from line 1 of stdio.h .
Usually the first line of stdio.h is a preprocessor directive, so
I show an empty line where the preprocessor directive got processed
and converted to an empty line (to keep the line counts consistant).
The compiler keeps count relative to the last #line, so it would know
that the "... some lines of actual stdio.h content" occured
starting at line 2 of stdio.h . Eventually stdio.h finishes and
another #line is emitted to indicate that the next source line
was originally line 3 of xdev.c . Again that was a comment that got
blanked out. Keeping count again from the last #line, the compiler
knows that the next line must have originally been from line 4 of
xdev.c . And that's how the compiler knows, when it prints out
the syntax error message for that line, that it should be complaining
about line 4 of xdev.c . If #line did not exist, then after preprocessing
had expanded stdio.h and everything that it pulls in, the compiler
might think that it's on line 2500 or whatever (stdio.h expands a lot!)
and it wouldn't make much sense to the user to complain about
an error on line 2503 of a 5 line source file!
--
   I was very young in those days, but I was also rather dim.
   -- Christopher Priest