Re: Index vs. table scans in statspack reports

Daniel Morgan <damorgan_at_x.washington.edu> wrote in message news:<1070290026.24951_at_yasure>...
> Howard J. Rogers wrote:
>
> > "Mark Townsend" <markbtownsend_at_attbi.com> wrote in message
> > news:QiDyb.272897$ao4.942367_at_attbi_s51...
> >
> >>>The real issue is why
> >>>such an error is allowed to reside un-corrected on the OTN website in
> >
> > the
> >
> >>>first place. It's Oracle that needs shooting in this case, not Rich.
> >>
> >>
> >>By all means continue to point these types of problems out. I will try
> >>to make sure they are corrected at source. I don't, however, read every
> >>thread, so you are more than welcome to send me a direct email on such
> >>subjects.
> >
> >
> > I wasn't having a go at you, Mark.
> >
> > A more generic point, perhaps: if these sorts of articles could have a
> > 'feedback' link, then these articles would tend to get self-corrected,
> > without some eagle-eyed Oracle staffer having to spot it in the first place.
> >
> > Regards
> > HJR
>
> I'll stand in line with Howard on this one. Every piece of on-line
> documentation shoul have a link that asks readers to rate its content
> and provide feedback on inaccuracies. Oracle Corp. should not be relying
> on you, or any other single person, to act as ad hoc editor.
>
> But, until that happens, I will do my best to get you pointers to
> problems when I find them.
>
> Thanks for be willing to help.

There is a related problem, and that is staleness.

I think first of all, old stuff should be kept online forever, even if for historical reasons only. Users should be able to keep bookmarks of docs they find useful - urls should not be fluid as they are now, with locations changing in short, unpredictable timeframes.

Some old docs are still useful, some become wrong as they age. It can be difficult to tell based just upon the age. For example, the original OFA white paper may be very good, but may have some statements that are just plain wrong given the evolution of semantics and hardware. Such a paper should be published in multiple versions, since it may still be very useful for contemporaneous systems that are still running. The concept has become basic administration, so the paper should be updated to reflect current conditions.

There are a number of things that need to be tracked: accuracy, timeliness, updates, applicability and so on. Simple feedback, electronic sticky-notes or karma-points-ala-slashdot aren't enough, although they might be very useful as an easy short-term improvement. We really need a well-thought-out system that will last through versions (or should I say executives?) and include vetting. And vetting isn't a popularity contest (or a veterinary practice or leaping about in a frisky manner :-) .

None of this is new, there is a whole document-tracking industry full of people who make this their life's work. It's obvious from some doc headers that Oracle corp does some of this, at least to track what is said publicly, but there is room for improvement. I'm curious what Oracle actually does.

jg

--
@home.com is bogus.
http://www.signonsandiego.com/news/uniontrib/wed/business/news_1b3boeing.html