Dear pghackers,
During the last two weeks, I have enjoyed a total of 113 hours of
development of pov - PostgreSQL Object Version control system.
Version 1.0 is now finally released!
POV will not touch any of your "data" (tables, sequences, indexes) -
it will only keep track of your "logics" (functions, constraints,
defaults, triggers, aggregates and views).
The possibilities are hopefully end-less, imagine a world where you
never have to worrie about cumbersome reverting.
If you screw-up your logics, you can simply travel back in time
without losing any of your data.
I just managed to start out with a clean database, load pov, take a
snapshot: select pov(), import my company's entire production database
(consisting of 2159 objects), take a new snapshot: select pov(), then
restore to the original state: select pov(1), i.e. removing all the
functions/constraints/defaults/triggers/aggregates/views back to the
orignal state, then restoring the database again: select pov(2),
without losing any of the data nor touching any of the "data
containers" (i.e. tables, sequences, indexes).
5 minute teaser demo video: http://screencast.com/t/mYFlvcI5H38X
To install:
git clone git@github.com:gluefinance/pov.git
cd pov
psql -f $PGSRC/postgresql-9.1alpha3/contrib/pgcrypto/pgcrypto.sql
psql -f sql/install.sql
To take a snapshot:
postgres=# select pov();
To restore a snapshot:
postgres=# select pov(1); -- replace "1" with the snapshot #
Potential future features: merging schemas, diffing, etc...
I would highly appreciate if you hackers could test to play around
with this, using real-life database schemas, to detect any bugs.
To all of you, have a great weekend! Happy hacking!
--
Best regards,
Joel Jacobson
Glue Finance
