> @@ -430,10 +430,10 @@ check_gcc() > -o \( $cc_maj -eq 3 -a $cc_min -lt 2 \) \ > -o \( $cc_maj -eq 4 -a $cc_min -lt 1 -a "$OS" != "darwin" \) \ > -o \( $cc_maj -eq 4 -a $cc_min -gt 9 \) \ > - -o \( $cc_maj -eq 5 -a $cc_min -gt 3 \) \ > - -o \( $cc_maj -eq 6 -a $cc_min -gt 0 \) \ > + -o \( $cc_maj -eq 5 -a $cc_min -gt 4 \) \ > + -o \( $cc_maj -eq 6 -a $cc_min -gt 1 \) \ > -o $cc_maj -gt 6 ]; then > - log_failure "gcc version $cc_maj.$cc_min found, expected gcc 3.x > with x>1 or gcc 4.x with 0<x<10 or gcc 5.3" > + log_failure "gcc version $cc_maj.$cc_min found, expected gcc 3.x > with x>1 or gcc 4.x with 0<x<10, gcc 5.x with x<10 or gcc 6.x with x<2" > fail really
What's the point of checking minor versions on gcc 5 and 6? This isn't done on earlier versions -- after the numbering was flattened, version 5 is equivalent to 4.8 or 4.9. Thus, I understand refusing to build with an untested major version, but FTBFSing on every point release is a bad idea. Meow! -- An imaginary friend squared is a real enemy.
