Re: Why there is a union in HeapTupleHeaderData struct

Hello,

I think the comment just above the HeapTupleFields struct definition
has the related details.

/*
 * Heap tuple header.  To avoid wasting space, the fields should be
 * laid out in such a way as to avoid structure padding.
 *
 * Datums of composite types (row types) share the same general structure
 * as on-disk tuples, so that the same routines can be used to build and
 * examine them.  However the requirements are slightly different: a Datum
 * does not need any transaction visibility information, and it does need
 * a length word and some embedded type information.  We can achieve this
 * by overlaying the xmin/cmin/xmax/cmax/xvac fields of a heap tuple
 * with the fields needed in the Datum case.  Typically, all tuples built
 * in-memory will be initialized with the Datum fields; but when a tuple is
 * about to be inserted in a table, the transaction fields will be filled,
 * overwriting the datum fields.


especially the last line points as to what roles each of them plays,
though, I would like to hear more about additional details from others
who might reply.

On Mon, May 20, 2013 at 4:28 PM, Soroosh Sardari
<soroosh.sardari@gmail.com> wrote:
> Dear Hackers
>
> In fix part oh HeapTuple, there is a union that is named t_choice,
> union
>     {
>         HeapTupleFields t_heap;
>         DatumTupleFields t_datum;
>     }            t_choice;
>
> I can't find out why we need t_datum, actually there is no comment about
> DatumTupleFields.
>
> Regards
> Soroosh
>
>

--
Amit Langote