On 1/15/13 6:36 AM, Andres Freund wrote:
> I just think that the price of fixing a single Assert() that hasn't
> changed in years where the variable isn't likely to ever get signed is
> acceptable.
Well, once you get past that one change you proposed, you will also find
pg_standby.c: In function 'SetWALFileNameForCleanup':
pg_standby.c:348:3: error: comparison of unsigned expression >= 0 is
always true [-Werror=type-limits]
(which, curiously, is the only one that clang complains about).
I don't like removing safety checks from code when there is no other
mechanism that could make up for it somehow.
I think the best practice at the moment, as with most gcc -Wextra
warnings, is to manually check them once in a while.