Re: redundant check of msg in does_not_exist_skipping

On Thu, 17 Nov 2022 at 23:06, Japin Li <japinli@hotmail.com> wrote:
> On Thu, 17 Nov 2022 at 20:12, Ted Yu <yuzhihong@gmail.com> wrote:
>> Hi,
>> I was looking at commit aca992040951c7665f1701cd25d48808eda7a809
>>
>> I think the check of msg after the switch statement is not necessary. The
>> variable msg is used afterward.
>> If there is (potential) missing case in switch statement, the compiler
>> would warn.
>>
>> How about removing the check ?
>>
>
> I think we cannot remove the check, for example, if objtype is OBJECT_OPFAMILY,
> and schema_does_not_exist_skipping() returns true, the so the msg keeps NULL,
> if we remove this check, a sigfault might be occurd in ereport().
>
>         case OBJECT_OPFAMILY:
>             {
>                 List       *opfname = list_copy_tail(castNode(List, object), 1);
>
>                 if (!schema_does_not_exist_skipping(opfname, &msg, &name))
>                 {
>                     msg = gettext_noop("operator family \"%s\" does not exist for access method \"%s\", skipping");
>                     name = NameListToString(opfname);
>                     args = strVal(linitial(castNode(List, object)));
>                 }
>             }
>             break;

Sorry, I didn't look into schema_does_not_exist_skipping(), and after look
into schema_does_not_exist_skipping function and others, all paths that go
out switch branch has non-NULL for msg, so we can remove this check safely.

-- 
Regrads,
Japin Li.
ChengDu WenWu Information Technology Co.,Ltd.