MySQL 4.1 character set problem

This still applies to cvs. INSTALL.txt and database.mysql.inc do not reflect these changes.

Until all the modules use utf8-compliant functions, we need to use "binary" collation on several columns, at least in tables "cache" and "search_*". That's because in a utf8 connection, not all character strings are valid.

Until this is fixed, it's probably better to use a latin1 charset connection, and to modifiy the database creation script to use "latin1" collation.

damz: are those two the only tables that "matter"?

Besides the connection string that anonymous mentions (of which I know little about it), we can convert the default encoding of the database and tables using the following commands in MySQL 4.1.

ALTER DATABASE database CHARACTER SET utf8 COLLATE utf8_general_ci;
ALTER TABLE {each table name} CONVERT TO CHARACTER SET utf8 COLLATE utf8_general_ci;

More talk about this issue here: http://www.cesspit.net/drupal/node/897.

Since this particular bug can corrupt backups, I'm setting it to critical. A broken backup is very very bad.

Changing back to normal. Further testing is needed.

Here are results of my experiment, but i am drupal newbie, correct me if i am wrong
1. My ISP has MySQL 4.1 set to default values:
collation_database=cp1250_general_ci
collation_server=cp1250_general_ci
character_set_database=cp1250
character_set_server=cp1250
2. I can say Drupal 4.6.3 works with slovak language, but
3. cron.php completes successfully but the search_ tables are empty, therefore search module does not work.
4. trip_search works good, as well with slovak characters
5. ALTER DATABASE database CHARACTER SET utf8 COLLATE utf8_general_ci;
solved the cron troubles, but
6. it is no more possible input slovak characters into content (page, story, ...)
7. RESUME: i will use 8bit encoding and trip_search with MySQL 4.1

I agree with #4 and because nobody else complained about this particular problem I'm closing it.
It's critical to have an usable and recoverable database, but it seems it was one person case.