Обсуждение: AW: [HACKERS] Volunteer: Large Tuples / Tuple chaining

> Hmm,we have discussed about LONG.
> Change by LONG is transparent to users and would resolve
> the big tuple problem mostly.
> I'm suspicious that tuple chaining is worth the work now.

All commercial db's I know allow at least 32kb tuples,
they all do it with chaining, because they usually have a 
smaller (often configurable) pagesize. 
Imho it is definitely worth it.

Andreas

Zeugswetter Andreas SB wrote:

> > Hmm,we have discussed about LONG.
> > Change by LONG is transparent to users and would resolve
> > the big tuple problem mostly.
> > I'm suspicious that tuple chaining is worth the work now.
>
> All commercial db's I know allow at least 32kb tuples,
> they all do it with chaining, because they usually have a
> smaller (often configurable) pagesize.
> Imho it is definitely worth it.
>

There would be few cases > 8K tuples after LONG was implemented.
And tuple chaining is much difficult to implement than LONG.
If it is badly designed it would be a disaster.
Is it still worth doing now ?

At least the design must be verified sufficiently before going.

Regards.

Hiroshi Inoue
Inoue@tpf.co.jp