In order to keep it small, almost all locales are stripped (original libicudata is about 9MB while the bundled one is about 1.5MB). All the Firebird 2.x versions come with ICU version 3.0, it is located in the firebird/lib directory and consists of three files (libicudata, libiuui18n, libicuuc).īut this is not a full-featured ICU. It means that it carries an ICU library and uses always this one. If you download Firebird from the the project website, it uses so-called bundled ICU. Another good news is that Firebird can use several ICU versions at the same time.
The good news is that GBAK dump and restore always works and this way you can copy data from any server to any server. If you move the database to a server with different ICU, you can connect to the database but every select returns en error You can still move your database freely but only when the same ICU version is used on both servers. The situation changes if you are using national collations. Using default collations (UCS_BASIC, UNICODE, UNICODE_CI) may be safe but I have not tested these cases extensively. For Unicode databases this may not be the case. You can move single-byte databases freely from server to server (plain copy) and they always work. This means that if you generate your collation according to any locale, your Unicode database is bound to the ICU version you have used at the time of creation and cannot be used with another version (older or newer). The thing is that ICU libraries are advancing very fast and different ICU versions generate different keys used for Firebird indexes. This is a good idea since Unicode is a complex topic and ICU libraries are free software developed and maintained by IBM. But if you have been using single-byte databases for some time and you want to migrate to unicode, you better be aware of several conceptual changes.įirebird uses ICU libraries to handle Unicode data. czech).įirebird is a great database server that supports both single-byte and unicode databases and clients. This article describes the principles and catches and gives the examples how to set up a UTF-8 database in Firebird SQL server 2.5, how to setup ICU libraries and how to use national collations (eg. The first encounter with unicode databases and collations on Firebird may be confusing, as it was for me and it may take some time to grasp the concepts and settings.
0 kommentar(er)
