Re: Add 64-bit XIDs into PostgreSQL 15

On Thu, 3 Nov 2022 at 08:12, Maxim Orlov <orlovmg(at)gmail(dot)com> wrote:
>
> Hi!
>
>> I attach an additional V48-0009 patch as they are just comments, apply it if you want to.
>
> Big thank you for your review. I've applied your addition in the recent patch set below.
>
> Besides, mentioned above, next changes are made:
> - rename HeapTupleCopyBaseFromPage to HeapTupleCopyXidsFromPage, since this old name came from the time when еру "t_xid_base" was stored in tuple,
> and not correspond to recent state of the code;
> - replace ToastTupleHeader* calls with HeapHeader* with the "is_toast" argument. This reduces diff and make the code more readable;
> - put HeapTupleSetZeroXids calls in several places for the sake of redundancy;
> - in heap_tuple_would_freeze add case to reset xmax without reading clog;
> - rename SeqTupleHeaderSetXmax/Xmin to SeqTupleSetXmax/min and refactoring of the function; Now it will set HeapTuple and HeapTupleHeader xmax;
> - add case of int64 values in check_GUC_init;
> - massive refactoring in htup_details.h to use inline functions with type control over macro;
> - reorder code in htup_details.h to reduce overall diff.
>
> As always, reviews and opinions are very welcome!

0008 needs a rebase. heapam.h and catversion.h are failing.

Regards

Thom