I did some testing of the Oracle upgrade last week and I believe this still needs quite a bit of work. I was unable to upgrade a bare instance + file-storage. Upgrading the kernel package only left numerous kernel plsql packages in an invalid state, causing downstream plsql scripts to fail with "existing state of package has been discarded" errors.
I'm pretty sure this can be fixed by more aggressively pasting package body definitions into upgrade-5.2.0d1-5.2.0d2.sql in kernel. I started on this last week.
Why are people still always copying PL/SQL package
definitions into upgrade scripts? AFAICT, most of the time, an
upgrade script should simply source the canonical definitions
of PL/SQL (or plpgsql) code or views, as I discussed
back in 2003.
Is there some good reason I am missing here why OpenACS upgrade
scripts instead so often verbosely copy-and-paste code?