Re: GIN pending list clean up exposure to SQL

On Wed, Jan 27, 2016 at 12:54 PM, Julien Rouhaud
<julien(dot)rouhaud(at)dalibo(dot)com> wrote:
> On 27/01/2016 10:27, Fujii Masao wrote:
>>
>> Thanks for updating the patch! It looks good to me.
>>
>> Based on your patch, I just improved the doc. For example, I added
>> the following note into the doc.
>>
>> + These functions cannot be executed during recovery. + Use
>> of these functions is restricted to superusers and the owner +
>> of the given index.
>>
>> If there is no problem, I'm thinking to commit this version.
>>
>
> Just a detail:
>
> + Note that the cleanup does not happen and the return value is 0
> + if the argument is the GIN index built with <literal>fastupdate</>
> + option disabled because it doesn't have a pending list.
>
> It should be "if the argument is *a* GIN index"
>
> I find this sentence a little confusing, maybe rephrase like this would
> be better:
>
> - Note that the cleanup does not happen and the return value is 0
> - if the argument is the GIN index built with <literal>fastupdate</>
> - option disabled because it doesn't have a pending list.
> + Note that if the argument is a GIN index built with
> <literal>fastupdate</>
> + option disabled, the cleanup does not happen and the return value
> is 0
> + because the index doesn't have a pending list.
>
> Otherwise, I don't see any problem on this version.

This is a corner case that probably does not need to be in the docs,
but I wanted to clarify it here in case you disagree: If the index
ever had fastupdate turned on, when it is turned off the index will
keep whatever pending_list it had until something cleans it up one
last time.

Thanks for the review and changes and commit.

Cheers,

Jeff