David E. Jones wrote:
On Tue, 2009-05-12 at 18:35 -0700, John D. Hays wrote:
David E Jones wrote:
Could you be more specific? What makes you think OFBiz likes Latin1 as
an encoding character set?
Its in the entity definitions on the SVN code.
I'm not sure what this means... do you mean specifically the
entityengine.xml file?
Yes.
As a random guess of the direction you're going: which database are
you using, is it MySQL?
Yes, MySQL.
This could be the issue. MySQL doesn't handle UTF-8 (or any multi-byte
character set) very well. There are (or used to be) some JDBC driver
issues, but the big problem is that MySQL column sizes are in bytes and
NOT in characters, and a single UTF-8 character takes 3 bytes. In other
words, if you put a 100 character UTF-8 string into the database it will
require 300 bytes, and if it is a size 255 column then BOOM! String too
long error message.
That is why in the entityengine.xml file the default datasource for
MySQL has the char set as a non multi-byte character set.
If you need to do internationalized text OFBiz will handle it great, but
MySQL won't. I'd recommend you use Postgres or something else instead.
-David
In the cases that I am referring to, we are currently using MySQL and
UTF-8. We could switch to a different database such Postgres, but UTF-8
seems to work in MySQL for what we have been doing. (I have had to use
byte[] to pull some UTF-8 strings rather than the getString through jdbc
before, but have made it work.)
Will investigate further.
