Eric Blake wrote:
On 03/29/2010 10:37 AM, Jim Meyering wrote:
> Another one caught by clang:
>
> Note the first test to see if "inst" may be NULL.
> Then, in the following loop, "inst" is unconditionally
> dereferenced via "inst[i]". There are other unprotected
> used of "inst[i]" below, too.
>
> Rather than trying to protect all uses, one by one, I chose
> to return "success" when given an empty list of rules.
ACK that your patch is the minimal fix to avoid a segfault, but we
should probably get Stefan's input on whether returning success on an
empty input is the best course of behavior.
Ok. I've Cc'd him.