I have just upgraded my wordpress installation from the legacy 2.0.11 (which was stable and officially supported until very recently) to 2.8.4 in one fell swoop by following the steps in the wordpress codex. There appeared to be only a few glitches that were relatively quickly ironed out and everything seemed to be running fine. Until I noticed several stray ┬s in some of the older posts. And ├╗s instead of Řs, ├ęs instead of Ús, ├ós instead of Ôs, etc. etc.

I looked in SQL and see that I have two different character sets for the tables in the wordpress database: latin1_swedish_ci collation and utf-8 collation labels

Unfortunately, I am not comfortable to just go ahead and change the tables to utf-8; I only know enough about SQL to get myself in trouble.

Alas, the wordpress "guide" is not written in layman's terms, except for this phrase in the opening paragraph: "character set conversion is not a simple process". And the official wordpress forums are pretty much useless because they are filled with people like me - the blind leading the blind. Luckily, someone has attempted a translation elsewhere:

Quote Originally Posted by mydigitallife.info
[I]n existing WordPress installations that are upgraded from earlier version of WordPress or not explicitly set a Unicode UTF-8 charset collation, the default database character set is normally set up as Latin1 (default on almost all MySQL installation) with latin1_swedish_ci collation. If you run a bilingual or multilingual blog with WordPress, you may face problem on character encoding when your blog posts are written in other foreign languages, or when you export and backup the database and later attempt to re-import the database dump in the event of database failure or server migration and moving. The symptom is obvious, your WordPress posts or pages will contains garbled, weird and funny characters, sometime just lots of ????? (question marks), rendering the WordPress database with your hard work useless and output unreadable. (May cause also by wrong charset collation)

This is EXACTLY what happened to me (thank goodness for copies of database backups kept in separate folders!)

In the Guide to Convert WordPress Database Character Set to UTF8, it says:

Quote Originally Posted by mydigitallife.info
Ensure that your post_content and post title fields on wp_posts table DOES NOT belongs [sic] to any indexes or FULLTEXT indexes. Else the type of the fields may not be converted to BLOB with one of the errors list below. Some plugins, such as related posts tend to add indexes to these fields. In this case, temporarily drop the indexes.
I've asked the following question in mydigitallife.info forum, but the thread is dated 2007, so the question may be overlooked. I am hoping that someone here will help.

What exactly do I look for in phpMyAdmin to find out if fields belong to indexes or FULLTEXT indexes?

Thank you.

EMorris, etherwork.net/blog/ (URL unlinked, in the interests of keeping spammers at bay as much as possible)


WinXP, firefox, SQL 4.1.21-log, PHP 5.2.8, Apache 1.3.37 (Unix)

screenshot of what I see in phpMyAdmin at flickr.com/photos/etherwork/3936558945/sizes/o/

(I am reluctant to simply dive in and hope for the best, having already very recently restored wordpress from a backup after a disastrous result from running the wordpress plugin to Convert WordPress Database Character Set.)