Re: [COMMITTERS] pgsql: Use a bitmask to represent role attributes

* Tom Lane (tgl(at)sss(dot)pgh(dot)pa(dot)us) wrote:
> Stephen Frost <sfrost(at)snowman(dot)net> writes:
> > * Tom Lane (tgl(at)sss(dot)pgh(dot)pa(dot)us) wrote:
> >> I'd have gone with just adding more bool columns as needed.
>
> > I don't think I was the only one concerned that adding a bunch of new
> > columns would bloat the size of pg_authid and the C structure behind it,
> > but I'm not remembering offhand who else considered it.
>
> Lessee, as of 9.4 pg_authid required 76 bytes per row, plus row header
> overhead that'd have probably pushed it to 104 bytes per row (more if
> you had non-null rolpassword or rolvaliduntil). If we add as many as 20
> more booleans we'd be at 124 bytes per row, whereas with this approach
> we'd have, well, 104 bytes per row. I'm not seeing much benefit to
> justify such a drastic change of approach.

I suppose. I didn't consider it to be a terribly drastic change but
rather simply using a better representation for a mostly-internal bit of
data. It also lended itself pretty nicely to maniuplation (at least,
imv, the code is a lot cleaner with the bitmask, but it's not a huge
deal). Guess I had been expecting concerns to be raised around adding
many more bytes where there wouldn't have been.

Thanks,

Stephen