erchache2000 wrote:
Brion Vibber escribió:
anders.nygard wrote:
- Mediawiki is supposed to convert old ISO format article text to
UTF-8 on the fly, right. How is this triggered, Do I need to set $wgUseLatin1 to get it to work or what.
$wgLegacyEncoding.
Brion, with this i dont need to upgrade to utf-8? Install on a blank site and only change database?
When a text blob is loaded from the 'text' table which has an old_flags value not including the 'utf-8' tag, the text will be transcoded from the setting in $wgLegacyEncoding to UTF-8 during loading.
This effects only text coming out of 'text' (or 'archive' for deleted revisions). It does not apply to usernames, page titles, comments, and other database fields which are used directly.
To successfully upgrade an old database which contained ISO-8859-1 data, you must transcode all these other fields to UTF-8. You can either find your own way to do this (iconv, playing around with encoding support in MySQL 4.1+, etc), or you can very carefully try using the upgrade1_5.php script which we used to upgrade from 1.4 to 1.5 schema.
This script was made for internal use and is not documented. Please read through it to confirm that it does what you need and that you have configured things correctly for it before considering its use.
-- brion vibber (brion @ pobox.com)