Re: Nitpicking: unnecessary NULL-pointer check in pg_upgrade's controldata.c

On Fri, Jun 26, 2015 at 9:06 AM, Andres Freund <andres(at)anarazel(dot)de> wrote:
> On 2015-06-26 22:03:05 +0900, Michael Paquier wrote:
>> Hi,
>>
>> Coverity is nitpickingly pointing out the following thing:
>> --- a/src/bin/pg_upgrade/controldata.c
>> +++ b/src/bin/pg_upgrade/controldata.c
>> @@ -402,8 +402,7 @@ get_control_data(ClusterInfo *cluster, bool live_check)
>> }
>> }
>>
>> - if (output)
>> - pclose(output);
>> + pclose(output);
>> The thing is that output can never be NULL, pg_upgrade leaving with
>> pg_fatal before coming to this code path. Hence this NULL check could
>> be simply removed.
>
> FWIW, I think these type of coverity items should just be discarded with
> prejudice. Fixing them doesn't seem to buy anything and there's enough
> actually worthwhile stuff going on.

I don't mind committing patches for this kind of thing if it makes the
Coverity reports easier to deal with, which I gather that it does.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company