[xwiki-devs] [VOTE] switching to UTF8 by default in 1.9M2
Hi devs, We've been tackling this proposal for some time, but we never took a final decision on it. So, I propose that we do this switch as soon as possible. Changes: - default encoding UTF-8 in the start_xwiki* scripts, xwiki.cfg and web.xml - remove the LANG= environment variable from the start script, since it's not needed - add the encoding configuration for the mysql jdbc connection URL (http://dev.mysql.com/doc/refman/5.0/en/connector-j-reference-charsets.html) Implications: - hsqldb instances will continue to work fine - for mysql and other DB instances, either the DB will need to be changed to work with the new encoding, or the wiki will have to be changed back to 8859-1. If we append characterEncoding=UTF-8 to the connection URL, then the storage will throw exceptions when trying to store an incompatible character, thus admins will quickly notice the problem and solve the issue, before users report loosing characters after a while (since without this setting the characters are silently transformed to '?' and this won't be noticed until the cache expires). We need to highlight this change in the release notes and upgrade instructions. -- Sergiu Dumitriu http://purl.org/net/sergiu/
+1 On Mon, Apr 27, 2009 at 09:29, Sergiu Dumitriu <[email protected]> wrote:
Hi devs,
We've been tackling this proposal for some time, but we never took a final decision on it. So, I propose that we do this switch as soon as possible.
Changes: - default encoding UTF-8 in the start_xwiki* scripts, xwiki.cfg and web.xml - remove the LANG= environment variable from the start script, since it's not needed - add the encoding configuration for the mysql jdbc connection URL (http://dev.mysql.com/doc/refman/5.0/en/connector-j-reference-charsets.html)
Implications: - hsqldb instances will continue to work fine - for mysql and other DB instances, either the DB will need to be changed to work with the new encoding, or the wiki will have to be changed back to 8859-1. If we append characterEncoding=UTF-8 to the connection URL, then the storage will throw exceptions when trying to store an incompatible character, thus admins will quickly notice the problem and solve the issue, before users report loosing characters after a while (since without this setting the characters are silently transformed to '?' and this won't be noticed until the cache expires).
We need to highlight this change in the release notes and upgrade instructions. -- Sergiu Dumitriu http://purl.org/net/sergiu/ _______________________________________________ devs mailing list [email protected] http://lists.xwiki.org/mailman/listinfo/devs
-- Thomas Mortagne
My +1 for the first if I can be of any weight. I think that LANG= has a few other impacts, e.g. Date.toString()... so I am not sure it's good to remove it. paul Le 27-avr.-09 à 09:29, Sergiu Dumitriu a écrit :
Hi devs,
We've been tackling this proposal for some time, but we never took a final decision on it. So, I propose that we do this switch as soon as possible.
Changes: - default encoding UTF-8 in the start_xwiki* scripts, xwiki.cfg and web.xml - remove the LANG= environment variable from the start script, since it's not needed - add the encoding configuration for the mysql jdbc connection URL (http://dev.mysql.com/doc/refman/5.0/en/connector-j-reference-charsets.html )
Implications: - hsqldb instances will continue to work fine - for mysql and other DB instances, either the DB will need to be changed to work with the new encoding, or the wiki will have to be changed back to 8859-1. If we append characterEncoding=UTF-8 to the connection URL, then the storage will throw exceptions when trying to store an incompatible character, thus admins will quickly notice the problem and solve the issue, before users report loosing characters after a while (since without this setting the characters are silently transformed to '?' and this won't be noticed until the cache expires).
We need to highlight this change in the release notes and upgrade instructions. -- Sergiu Dumitriu http://purl.org/net/sergiu/ _______________________________________________ devs mailing list [email protected] http://lists.xwiki.org/mailman/listinfo/devs
Paul Libbrecht wrote:
I think that LANG= has a few other impacts, e.g. Date.toString()... so I am not sure it's good to remove it.
We're overriding the locale, too, so all UI elements will have the same aspect, regardless of the LANG variable. -- Sergiu Dumitriu http://purl.org/net/sergiu/
+1 Sergiu Dumitriu wrote:
Hi devs,
We've been tackling this proposal for some time, but we never took a final decision on it. So, I propose that we do this switch as soon as possible.
Changes: - default encoding UTF-8 in the start_xwiki* scripts, xwiki.cfg and web.xml - remove the LANG= environment variable from the start script, since it's not needed - add the encoding configuration for the mysql jdbc connection URL (http://dev.mysql.com/doc/refman/5.0/en/connector-j-reference-charsets.html)
Implications: - hsqldb instances will continue to work fine - for mysql and other DB instances, either the DB will need to be changed to work with the new encoding, or the wiki will have to be changed back to 8859-1. If we append characterEncoding=UTF-8 to the connection URL, then the storage will throw exceptions when trying to store an incompatible character, thus admins will quickly notice the problem and solve the issue, before users report loosing characters after a while (since without this setting the characters are silently transformed to '?' and this won't be noticed until the cache expires).
We need to highlight this change in the release notes and upgrade instructions.
+1 (would have preferred to have the first time install wizard which could have warned admins but it's not there yet). -Vincent On Apr 27, 2009, at 9:29 AM, Sergiu Dumitriu wrote:
Hi devs,
We've been tackling this proposal for some time, but we never took a final decision on it. So, I propose that we do this switch as soon as possible.
Changes: - default encoding UTF-8 in the start_xwiki* scripts, xwiki.cfg and web.xml - remove the LANG= environment variable from the start script, since it's not needed - add the encoding configuration for the mysql jdbc connection URL (http://dev.mysql.com/doc/refman/5.0/en/connector-j-reference-charsets.html )
Implications: - hsqldb instances will continue to work fine - for mysql and other DB instances, either the DB will need to be changed to work with the new encoding, or the wiki will have to be changed back to 8859-1. If we append characterEncoding=UTF-8 to the connection URL, then the storage will throw exceptions when trying to store an incompatible character, thus admins will quickly notice the problem and solve the issue, before users report loosing characters after a while (since without this setting the characters are silently transformed to '?' and this won't be noticed until the cache expires).
We need to highlight this change in the release notes and upgrade instructions. -- Sergiu Dumitriu http://purl.org/net/sergiu/
Vincent Massol wrote:
(would have preferred to have the first time install wizard which could have warned admins but it's not there yet).
I was planning to do this as a GSoC project, to have an example for the real students. -- Sergiu Dumitriu http://purl.org/net/sergiu/
Sergiu Dumitriu wrote:
Hi devs,
We've been tackling this proposal for some time, but we never took a final decision on it. So, I propose that we do this switch as soon as possible.
Changes: - default encoding UTF-8 in the start_xwiki* scripts, xwiki.cfg and web.xml - remove the LANG= environment variable from the start script, since it's not needed - add the encoding configuration for the mysql jdbc connection URL (http://dev.mysql.com/doc/refman/5.0/en/connector-j-reference-charsets.html)
Implications: - hsqldb instances will continue to work fine - for mysql and other DB instances, either the DB will need to be changed to work with the new encoding, or the wiki will have to be changed back to 8859-1. If we append characterEncoding=UTF-8 to the connection URL, then the storage will throw exceptions when trying to store an incompatible character, thus admins will quickly notice the problem and solve the issue, before users report loosing characters after a while (since without this setting the characters are silently transformed to '?' and this won't be noticed until the cache expires).
We need to highlight this change in the release notes and upgrade instructions.
4 binding +1, several non-binding ones, done in rev. 19417. Added warning in the release notes. -- Sergiu Dumitriu http://purl.org/net/sergiu/
participants (5)
-
Marius Dumitru Florea -
Paul Libbrecht -
Sergiu Dumitriu -
Thomas Mortagne -
Vincent Massol