Figure 7-4 starts with simply connecting to Oracle. To ensure I didn't lose any critical trace file pieces, I set my dump file size to unlimited. I then created a very simple table, named findme, to make it easy to find the SQL in the massive trace file. Just as I did in Figure 7-1, I executed the same SQL statement but altered an optimizer parameter. This should result in two child cursors. Then I executed a different SQL statement, which references the same findme table. I did this to show that different SQL statements will reference the same table handle. Finally, I performed the library cache dump and disconnected.

Figure 7-5. A library cache diagram based on the SQL run in Figure 7-4, which produced the trace file shown in Figure 7-6. The figure highlights the various library cache object attributes, such as buckets (14778), handles (456dbf90), mutexes (0x456dc044), cursors, child cursors, and their relationships.

Figure 7-5 is nearly identical to Figure 7-3, with the key exception that the contents are filled based on the SQL run in Figure 7-4, which produced the trace file shown in Figures 7-6 and 7-7. (The trace file was simply too large to place in a single figure.)