Re: referencing objects

I don't think I have missed the point, I have only asked you to demonstrate that the point you claimed is both true and significant.

1. I presented an argument based on quantity of recursive SQL which suggests the 1:2:4 ratio you claimed is incorrect.
2. I quoted an example of parsing 4,000 statements in 14 seconds which appeared to do considerably better than the 1:2:4 ratio that you claimed, and incidentally suggests that Oracle is doing better at keeping information about public synonyms than you imply - hence your point may be true but largely irrelevant.

Your comment in another post about the 'dependency tables' is not only relevant to public synonyms - queries involving (particularly) OR-expansion of large IN-lists have always had to face the extra burden of potentially large dependency tables, a problem addressed by setting an event in later versions of 7.3. and removed by the IN-list Iterator operation in the 8.1 optimiser.

Of course, my example was running on 8.1 - so it is possible that the problem you describe is relevant only to an earlier version of Oracle. Either way, I'd be interested to see details of an experiment that proves your point.

--

Jonathan Lewis
Yet another Oracle-related web site:  http://www.jlcomp.demon.co.uk

Howard J. Rogers wrote in message <38e5d34b_at_news.iprimus.com.au>...


>

>"Jonathan Lewis" <jonathan_at_jlcomp.demon.co.uk> wrote in message

>news:954575805.19620.0.nnrp-11.9e984b29_at_news.demon.co.uk...

>

>You seem to have missed the point that the structures required to make

>public synonyms work cannot be 'kept' in the library cache, and that there

>is therefore a tendency for public-synonym-dependent applications to be

>doing lots more re-parsing than would be 'desirable'.  That is precisely

 the


>point, of course.

>