Oracle FAQ | Your Portal to the Oracle Knowledge Grid |
Home -> Community -> Usenet -> c.d.o.server -> Re: Data Dictionary: relative_fno question
"Richard Foote" <richard.foote_at_bigpond.com> wrote in message
news:lCtV9.25253$jM5.67091_at_newsfeeds.bigpond.com...
> "Howard J. Rogers" <howardjr2000_at_yahoo.com.au> wrote in message
> news:JUoV9.25118$jM5.66456_at_newsfeeds.bigpond.com...
>
> > Which just goes to prove that sometimes I can't remember things the
right
> > way around, and sometimes I just get them plain wrong: 102*3* is the
right
> > number of data files, not 1022.
> >
>
> Hi again Howard,
>
> I 'think' the 1022 is O/S dependent.
Oh, well that might explain the confusion then. It says in the 8.0 DBA courseware 1023. Taidg corrected that many moons ago to 1022. In the 8i courseware I believe on one page it said 1022 and on another 1023. The self-contradictory nature of the beast caused me to revert to believing Taidg.
>
> I've a question I hope you might be able to answer. Do you have any ideas
on
> why Oracle chose to add the object id portion to the rowid and make the
file
> number "relative" rather than simply allowing the file number portion bit
to
> be bigger.
Well, as I posted... my understanding was that increasing the size of the file number component would simply have meant there was a huge incompatibility between 7 and 8.0. Migration from 7 to 8 would have had to adjust every single rowid stored in indexes, for example. And can you imagine the number of migrations that would have failed half-way through when the database suddenly ran out of index tablespace space? or hit maxextents for an index segment? Not pretty. Besides which, certain badly-behaved applications (mentioning no names but SAP springs to mind) which hard-coded the old-fashioned rowids into their imposing demonstrations of how-not-to-code-an-app would simply have failed to work after the upgrade. And they couldn't take that chance.
>
> I believe the answer has to do with efficiencies in how it deals with
> partitioned objects but I've yet to be convinced that it's merits or
> otherwise.
>
Could be, I suppose. Partitioning came in at the same time as the new rowid format, so maybe there's a relationship after all.
What got me was why they simply didn't put TS# as part of the rowid instead of the object#, which is effectively only a proxy for the tablespace number in the first place. Would have been much simpler all round, I'd have thought.
Life's Little Mystery 1022.
Regards
HJR
> Cheers
>
> Richard
>
>
>
>
Received on Thu Jan 16 2003 - 02:23:27 CST