Re: [HACKERS] pgrowlocks relkind check

On 6/12/17 21:10, Amit Langote wrote:
> On 2017/06/13 0:29, Peter Eisentraut wrote:
>> On 4/24/17 21:22, Amit Langote wrote:
>>>>> create extension pgrowlocks;
>>>>> create view one as select 1;
>>>>> select pgrowlocks('one');
>>>>> -- ERROR:  could not open file "base/68730/68748": No such file or directory
>>>>>
>>>>> With the attached patch:
>>>>>
>>>>> select pgrowlocks('one');
>>>>> ERROR:  "one" is not a table, index, materialized view, sequence, or TOAST
>>>>> table

> FWIW, patch seems simple enough to be committed into 10, unless I am
> missing something.
> 
> Rebased one attached.

According to CheckValidRowMarkRel() in execMain.c, we don't allow row
locking in sequences, toast tables, and materialized views.  This is not
quite the same as what your patch wants to do.  I suppose we could stillallow reading the relation, and it won't ever
showanything
 
interesting.  What do you think?

-- 
Peter Eisentraut              http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services