Hacker News Reader

What else would they reasonably default to? Are you proposing requiring collations to always be specified (a breaking change from the SQL spec)?

View on HN

I think the behavior of the "C" collation is what most software developers expect. Plus, there is an advantage to the behavior being the same everywhere, rather than dependent on the system/environment settings.

But those preferences may be specific to my situation -- a software developer running an HA web application.

* If you're not a software developer, you might be surprised by the "C" collation, which puts "Z" before "a".

* If you're not running an HA web application, then it's easy to change the collation later -- just dump/restore the DB.

If you have any non-ASCII characters wouldn’t the C collation be very surprising? I’m having a hard time seeing how the C collation is a sensible default for UTF encoded text. If you want to limit to ASCII the why are you using UTF?

The C collation applied to UTF-8 just orders everything by Unicode codepoint.

Reply on news.ycombinator.com