On Thu, Jul 22, 2021 at 01:19:41AM +1200, David Rowley wrote:
> On Thu, 22 Jul 2021 at 00:44, Michael Paquier <michael@paquier.xyz> wrote:
>>
>> On Thu, Jul 22, 2021 at 12:32:39AM +1200, David Rowley wrote:
>> > I see both of these are limited to 64 on windows. Won't those fail on Windows?
>>
>> Yes, thanks, they would. I would just cut the range numbers from the
>> expected output here. This does not matter in terms of coverage
>> either.
>
> Sounds good.
>
>> x> I also wondered if it would be worth doing #define MAX_JOBS somewhere
>> > away from the option parsing code. This part is pretty ugly:
>>
>> Agreed as well. pg_dump and pg_restore have their own idea of
>> parallelism in parallel.{c.h}. What about putting MAX_JOBS in
>> parallel.h then?
>
> parallel.h looks ok to me.
Okay, done those parts as per the attached. While on it, I noticed an
extra one for pg_dump --rows-per-insert. I am counting 25 translated
strings cut in total.
Any objections to this first step?
--
Michael
