Thread: Question about change in page/tuple header (v4)

I'm comparing now different version of page layer, and I have two questions:

1) We now store only low 16bits TLI, but name in structure stays same. 
Maybe pg_tli_lo could be better.

2) HASOID has been moved in infomask and original place is unused. Is 
there some reason for that? This change little bit complicate tuple 
header upgrade. If there is not real reason to have it in current place, 
I recommend to move it back.

    thanks Zdenek

Zdenek Kotala <Zdenek.Kotala@Sun.COM> writes:
> 2) HASOID has been moved in infomask and original place is unused. Is 
> there some reason for that?

To keep it next to the other bits that are about tuple content rather
than transactional behavior.

> This change little bit complicate tuple 
> header upgrade. If there is not real reason to have it in current place, 
> I recommend to move it back.

There are enough other changes there that you're going to have to
manipulate the infomask anyway.
        regards, tom lane