Re: Fixes to index pages

> Bruce Momjian <pgman(at)candle(dot)pha(dot)pa(dot)us> writes:
> >> What happened to our discussion about keeping t_info bit 13 unused??
>
> > I wasn't going to reserve it in the patch. I figured I would make all
> > the items/flags match, and if someone wants to reserve it, it is easy to
> > do in one place. I imagine 7.2 is going to be dump/reload anyway so the
> > decision can be made during development cycle. I basically didn't want
> > to leave a bit gap and leave it unnamed because it could cause
> > confusion.
>
> I object. Strongly. You are making a significant change without
> discussion --- in fact, contrary to what discussion there has been.
> This is not a "trivial cleanup".

OK, attached is the patch. I guess I don't understand how what I am
doing affects anything.

I realize you were commenting about the earlier patch that gave the 13th
bit to the length. I now see the issue you were talking about was this
test:

/*
* Here we make sure that the size will fit in the field reserved for
* it in t_info.
*/
if ((size & INDEX_SIZE_MASK) != size)
elog(ERROR, "index_formtuple: data takes %lu bytes, max is %d",
(unsigned long) size, INDEX_SIZE_MASK);

I originally didn't realize that expanding the _storage_ space for the
index tuples actually allow storage of longer tuples. I see that now,
and this is why I just mark the patch as UNUSED. I will let others
handle it.

--
Bruce Momjian | http://candle.pha.pa.us
pgman(at)candle(dot)pha(dot)pa(dot)us | (610) 853-3000
+ If your life is a hard drive, | 830 Blythe Avenue
+ Christ can be your backup. | Drexel Hill, Pennsylvania 19026