I was going to post about my new unmentioned goof-off project, but I got sidetracked trying to reimport my Tumblr posts here, then took some restful goofing off all weekend. Somewhy my import script can't save my original post about space from the short-lived original concept for this blog, just because when I linked to his classic post about the space shuttle I spelled “Maciej Ceglowski” with the proper U+00142 Latin Small Letter L With Stroke.

_mysql_exceptions.Warning: Incorrect string value: '\xC5\x82owsk...' for column 'html' at row 1

Update: Oh, ugh, I just had to actually read the best answer to the top Stack Overflow question about it. Columns can have different character sets, hmm.

It thinks that's an invalid UTF-8 sequence (an “Incorrect string value”), even though I'm following all the appropriate directions for using UTF-8 in MySQL with Django. I had miscreated the databases in the first place, but I converted them (the database, the tables, and I added the default settings to the my.cnf) after the fact, which I thought should work. I'm using MySQL 5.1 so it's not an issue with utf8mb4 character encoding—plus it's only a two byte sequence anyway. I'm was using python-MySQLdb 1.2.3, but downgrading to the Ubuntu package (1.2.2-10build1) didn't help.

A lot of directions seem to think setting DEFAULT_CHARSET has anything to do with the database, weirdly, which I found zero real evidence in the Django code for. Oh, apparently I'd missed where they call this out in the documentation, too. I guess that shows the extent of try-anything handwaving this problem reduces us to.

Maybe that's a reason so many Django people are into Postgres, sigh.