Re: [PATCH] Tests for reloptions

В письме от 19 октября 2017 14:20:52 Вы написали:

> I'm hesitant to hardcode things like the number of bits in bloom, as you
> had in the original. If I understand correctly, that number could
> change with compile options (different blocksize?), so I removed that
> part.

#define MAX_BLOOM_LENGTH (256 * SIGNWORDBITS)

#define SIGNWORDBITS ((int) (BITS_PER_BYTE * sizeof(BloomSignatureWord)))

typedef uint16 BloomSignatureWord;

Here everything is based on uint16, and it is platform independent, as far as
I can get.

But this is not really important now...

> I also fixed a few spelling errors.
Thank you. I am not so good with natural languages :-)

> And pushed.
Thanx!

> Let's see what the buildfarm says about this.
It seems to me that it is quite happy about these tests :-)

> Oh, one more thing: be careful when editing parallel_schedule. There
> are constraints on the number of entries
Oh, I did not payed attention to this issue, through it it mentioned in
parallel_schedule comments. I've added it to the wiki
https://wiki.postgresql.org/wiki/Regression_test_authoring So it was all
described in one place.

> in each group; you had added a
> 20th entry after the comment that the group can only have 19.
Oups it was definitely my mistake. I should be more attentive... :-(

--
Do code for fun. Can do it for money (Perl & C/C++ ~10h/week)