Symphony.

Symphony requires that your database collation is set to UFT-8 (e.g. utf8_general_ci), but this is not mentioned anywhere in the README file.

Even if you install in compatibility mode?

I dont know what compatibility mode is supposed to do. So that’s another missing point in the install instructions, isn’t it?

Well, there is some help text in the installer itself. And it’s mentioned in the install tutorial.

Suggestions on how to handle this better?

And do we even know if that was your issue?

Attachments:

Selection_001.png

We often see this as an issue in the forum. Typically MySQL defaults to latin1_swedish_CI collation when creating new databases, and many users run into problems with extended character sets.

I don’t see that the installer help text tells people how to do it right (without compatibility mode). So maybe that is the real problem. Should the UX team take a look?

Yeah maybe. This is a tricky one because it’s not really a Symphony issue, but it does affect our users. Curious to know what the UX guys think.

I don’t believe this is an issue anymore. I’ve changed the MySQL class in a previous commit to set the correct encoding on all database activities so Symphony will never create a table that is not in the default charset.

By default in Symphony this is UTF-8 unless it is changed by a developer, or ‘high compatibility’ is set on installation (in which case it’ll use whatever the database was created in).

Can you confirm that this does not happen in your case?

“Compatibility mode” confused the hell out of me, and I’m still not sure what it means. It conjures up memories of trying to run legacy .NET applications in “Windows XP Compatibility Mode”.

Does it mean Symphony will not prepend any UTF8 strings to queries and will always inherit the server’s setting?

Could it be relabeled as “Always use UTF-8”, and have it checked by default. The explanation copy could say:

Use UTF-8 character encoding (recommended). If disabled, Symphony will use your server’s default encoding (latin1_swedish_CI) instead of UTF-8.

Can we populate with the user’s actual server setting?

Should this option only be visible if the server isn’t set to use UTF-8?

I believe yes to all nick, but I’d prefer to wait until 2.2.1 to add that (which I sure your implying anyway :))