I just found out the painful way that if you manually update a column which is normally calculated automatically as an auto_increment it seems to work at first ..

but MySQL doesn't update the auto_increment value for the table - so subsequent inserts can fail.

The solution seems to be to delete and re-insert the row.

http://bugs.mysql.com/bug.php?id=12434

Test code:

mysql> create table test (id int(10) unsigned NOT NULL, col1 varchar(255), PRIMARY KEY (id));
Query OK, 0 rows affected (0.00 sec)

mysql> insert into test (col1) VALUES ('test1');

Most of the time I can get what I want out of SQL using the standard aggregate functions, but today I needed to find the latest timestamp from any one of three columns.

I achieved this using the GREATEST() function, it's not party of the SQL standard but is commonly available.

GREATEST(value [, ...])

I've written a patch for Drupal 7 to enable date and time data types to be defined for the schema API.

http://drupal.org/node/200953

There seems to be some resistance to this idea due to concerns about compatibility between database versions (Drupal now supports MySQL, PostgreSQL and sqlite).

Personally I don't think Drupal can afford to work at lowest common denominator level if the aim is to be a serious option for the enterprise

For most of 2008 I've been working on a large Drupal installation (around 10 developers,  something like 100 custom modules, 17 languages, split over 2 sites).

This all adds up to a lot of updates, and schema changes.

Drupal's hook_install, and hook_update_N functions are pretty good, but not perfect at bringing order to this chaos.

It seems there is a bug in MySQL which is causing restores to run very slowly (hours instead of minutes)

http://bugs.mysql.com/bug.php?id=33057

The bug is fixed in release 5.0.67

But Ubuntu hasn't updated yet....

Fortunately this bug is specific to the command line client. 

Workaround: use the MySQL Administrator GUI instead.

Comments in code are widely used and generally (but not universally) accepted as a good thing.

Databases can also accommodate comments - but these seem to be very rarely used.

I'm of the opinion that database comments are very useful.

It's often hard to determine the exact meaning of a column - maybe it would be ideal to rename the column but you don't want to risk breaking code.