Changes for version 0.04
- Three good fixes from Robert Mathews <firstname.lastname@example.org>. (Thanks to him for submitting a patch!) In his own words: (1) The first one is nothing big: test 16 fails with BerkeleyDB v1.85 on solaris 5.6. This seems to be due to the fact that we're creating a database (and therefore writing to it), but it's only read-locked. (2) The second is that TIEHASH assumes that SUPER::TIEHASH will succeed. If it doesn't, the lockfile gets left open, and DESTROY is never called to close it. (3) I ran into one other issue: umask isn't restored if sysopen on the lockfile fails. Fixed that too.
- DB_File::Lock - Locking with flock wrapper for DB_File