> The only thing that I can think of would be if $decompress_program > --test were failing, but actually trying to decompress succeeded. I > would be inclined to dismiss that particular scenario as not important > enough to be worth the additional CPU cycles.
When I wrote this test, it was just to increase coverage for pg_basebackup. As I checked again, it already does that in the pg_verifybackup 008_untar.pl, 009_extrack.pl test you mentioned. Therefore, I agree with your opinion. --- Regards, DongWook Lee.