I'm facing a dilemma. Should I continue to try and support MySQL, when I know full well doing so may cause me a terrible amount of grief -- especially when most any Web host worth its salt also supports the far saner PostgreSQL?
Here's why I say this. It's no secret that MySQL has some of the worst, most legacy-issue-ridden problems of any major database solution out there. A bunch of these have been ameliorated over time; a bunch more have not. Many of the ones that have not been ameliorated impact me directly. I've worked around this by providing an abstraction layer for certain database actions, like creating full-text indexes, but they are still problematic.
On the other hand, not bothering to support MySQL at all severely restricts the number of scenarios I can use, and the flexibility within those scenarios. Using SQLite as the default DB works mainly as an "all-else-fails" fallback, and for a single blog with low traffic it might well work. But for anything larger, most people will opt for MySQL without thinking, so odds are I'm stuck with supporting MySQL as a matter of course and courtesy, not as a preference.
For perspective, I'm using SQLite to run this instance of MeTal. This may not last forever, as I am determined to dogfood a different DB solution whenever possible. But very little of what I've seen with MySQL up close has persuaded me to build anything on top of it.