Re: [PoC] Improve dead tuple storage for lazy vacuum

On Wed, Jan 17, 2024 at 12:32 PM John Naylor <johncnaylorls(at)gmail(dot)com> wrote:
>
> I wrote:
>
> > > Hmm, I wonder if that's a side-effect of the "create" functions doing
> > > their own allocations and returning a pointer. Would it be less tricky
> > > if the structs were declared where we need them and passed to "init"
> > > functions?
>
> If this is a possibility, I thought I'd first send the last (I hope)
> large-ish set of radix tree cleanups to avoid rebasing issues. I'm not
> including tidstore/vacuum here, because recent discussion has some
> up-in-the-air work.

Thank you for updating the patches! These updates look good to me.

>
> Should be self-explanatory, but some thing are worth calling out:
> 0012 and 0013: Some time ago I started passing insertpos as a
> parameter, but now see that is not ideal -- when growing from node16
> to node48 we don't need it at all, so it's a wasted calculation. While
> reverting that, I found that this also allows passing constants in
> some cases.
> 0014 makes a cleaner separation between adding a child and growing a
> node, resulting in more compact-looking functions.
> 0019 is a bit unpolished, but I realized that it's pointless to assign
> a zero child when further up the call stack we overwrite it anyway
> with the actual value. With this, that assignment is skipped. This
> makes some comments and names strange, so needs a bit of polish, but
> wanted to get it out there anyway.

Cool.

I'll merge these patches in the next version v54 patch set.

Regards,

--
Masahiko Sawada
Amazon Web Services: https://aws.amazon.com