Re: Parallel Index Scans

On Sat, Jan 21, 2017 at 12:23 PM, Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> wrote:
> On Sat, Jan 21, 2017 at 1:15 AM, Robert Haas <robertmhaas(at)gmail(dot)com> wrote:
>>
>> I think it's a good idea to put all three of those functions together
>> in the listing, similar to what we did in
>> 69d34408e5e7adcef8ef2f4e9c4f2919637e9a06 for FDWs. After all they are
>> closely related in purpose, and it may be easiest to understand if
>> they are next to each other in the listing. I suggest that we move
>> them to the end in IndexAmRoutine similar to the way FdwRoutine was
>> done; in other words, my idea is to make the structure consistent with
>> the way that I revised the documentation instead of making the
>> documentation consistent with the order you picked for the structure
>> members. What I like about that is that it gives a good opportunity
>> to include some general remarks on parallel index scans in a central
>> place, as I did in that patch. Also, it makes it easier for people
>> who care about parallel index scans to find all of the related things
>> (since they are together) and for people who don't care about them to
>> ignore it all (for the same reason). What do you think about that
>> approach?
>>
>
> Sounds sensible. Updated patch based on that approach is attached.
>

In spite of being careful, I missed reorganizing the functions in
genam.h which I have done in attached patch.

> I
> will rebase the remaining work based on this patch and send them
> separately.
>

Rebased patches are attached. I have fixed few review comments in
these patches.
parallel_index_scan_v6 - Changed the function btparallelrescan so that
it always expects a valid parallel scan descriptor.
parallel_index_opt_exec_support_v6 - Removed the usage of
GatherSupportsBackwardScan. Expanded the comments in
ExecReScanIndexScan.

--
With Regards,
Amit Kapila.
EnterpriseDB: http://www.enterprisedb.com