[Chicken-users] Update to the CHICKEN <-> SQLite3 bindings (v1.2.0)

[Thomas Chust]

Hello,

in response to a problem report concerning the sqlite3 egg and SQLite3.3.4 
some larger changes have been made to the egg for CHICKEN:
    - The bindings now handle "valid" NULL statements correctly, which
      allows the execution of statements involving a IF [NOT] EXISTS clause
      without errors in all situations.
    - The bindings perform stricter pointer validity checks for the wrapped
      database and statement handles now, which will hopefully help to
      prevent or detect problems similar to the reported one more easily in
      the future.
    - Statement and database handles are now automatically invalidated when
      they are successfully finalized to prevent accidental reuse.
      Refinalization of successfully finalized handles silently does
      nothing.

All additional checks and magic keep the interface backwards compatible. 
The new code makes use of the TinyCLOS object framework, which will likely 
result in a slight security for performance tradeoff. But in order to 
fully support the strange behaviour of sqlite3_prepare in the SQLite3 C 
library, the changes were necessary.

The new version of the egg and documentation can be found at
   http://www.chust.org/projects/sqlite3.egg
and
   http://www.chust.org/projects/sqlite3.html
as usual.

On a sidenote, please keep in mind, that all prepared statement handles 
become invalid as soon as the database schema changes -- this can 
especially be a pitfall when working with the DROP TABLE IF EXISTS and 
CREATE TABLE IF NOT EXISTS syntax, because these statements may compile to 
NULL statements that *cannot* be magically invalidated when the database 
schema changes. Reexecuting them will thus work, but do nothing, which is 
quite likely not what you want...

I hope my changes haven't broken anything -- at least they still work for 
me ;) But if you find any bugs, don't hesitate to report them.

cu,
Thomas Chust