EXACT collation please?

Installing a fresh 1.5

I've had problems in the past with foreign chars not displaying correctly in some areas. So from the beginning I want the correct settings. I have customers from around the globe entering addresses the way THEY want to and I need it to record correctly.

During setup it asks me to choose between UTF-8 and Latin1, but when I'm creating the database I get dozens of options. So utf8_general_ci? utf8_bin_ci? utf8_unicode_ci? latin1_bin_ci? latin1_general_ci? latin1_general_cs?

• Steven

If you're using UTF8, then the preferred collation to use when CREATING A NEW DATABASE is utf8_general_ci or utf8_unicode_ci

For Latin1/iso-8859-1 you want latin1_general_ci

Basically you want the "_general_ci" version if possible.

NOTE: The "ci" means "case-insensitive", and if you use "cs" instead then your searches won't work.

Any good hosting company already has that set as the server default so that the user just supplies a database name and goes from there.

Thanks.

However, I was just reading a few articles and postings while I waited for your answer... Several people say that you should use unicode and not general. General is fast but inaccurate and unicode isn't perceivably slower anyway. The suggestion in general being: if you support foreign charsets, you should actually support them correctly.

Any harm in using it?

If so... just a thought that maybe you guys should look at it for a future revision. I'll leave it with you.

• Steven

just a p.s. it look like my host default is utf8_unicode_ci but I have a choice of dozens.

DrByte:

when CREATING A NEW DATABASE is utf8_general_ci

Can I ask, why did you stress that point so strongly??

I am working on a fresh copy with a new DB, but if I were upgrading from one that was using Latin, what's the difference?

s_mack:

Several people say that you should use unicode and not general. General is fast but inaccurate and unicode isn't perceivably slower anyway. The suggestion in general being: if you support foreign charsets, you should actually support them correctly.

Any harm in using it?
Good point. Probably one worth following.

Haven't had any negative run-ins with using general vs unicode myself, so can't comment from experience. But I suspect the advice you're reading about using unicode instead of general for utf8 is correct. Use it.

s_mack:

DrByte:

when CREATING A NEW DATABASE is utf8_general_ciCan I ask, why did you stress that point so strongly??

I am working on a fresh copy with a new DB, but if I were upgrading from one that was using Latin, what's the difference?
Because upgrading is a whole different kettle of fish. And way beyond the topic of this discussion thread.
You can't just change the characterset or collation on a database and have it change everything IN the database.
It's not a simple "flip the switch".

Hence: http://www.zen-cart.com/content.php?313-should-i-convert-my-store-from-iso-8859-1-to-utf8

Thanks!