variables.inc.php is an old hack to save having to declare variables
before using them in pages.
Lots of the variables there are unused (leftovers from the old
create-*.php or edit-*.php scripts) or properly initialized in the file
using them, so there's no need to keep them in variables.inc.php.
The remaining variables in this file should also be dropped after
further investigation.
array('') means to include an empty item, and that results in an invalid
remote specification with a trailing "/"
Using an empty array() fixes this.
Reported by oftc_ftw on IRC.
This replaces the hardcoded "LIMIT 10" with "LIMIT <page_size>".
Patch by Dan <dannyro @SF>, https://sourceforge.net/p/postfixadmin/patches/133/
Additional change on top of Dan's patch:
- wrap $CONF['page_size'] in intval() to avoid that a broken config
setting can break or exploit the query
The previous commit changed vacation_notificatoin.notified,
alias_domain.alias_domain and alias_domain.target_domain to latin1, but
did this only in their original upgrade function.
upgrade_1836_mysql() also applies this change to existing databases.
(It's unlikely that these fields are not latin1 - creating them as utf8
or utf8mb4 would break at the index length, but better safe than sorry ;-)
When trying to create a new database with utf8mb4 as default charset,
upgrade.php fails at various places because of too long indexes.
- no longer run upgrade_1_mysql, upgrade_2_mysql and upgrade_3_mysql
which all affect updates from pre-2.1 database layout
- add {LATIN1} to vacation_notificatoin.notified,
alias_domain.alias_domain and alias_domain.target_domain
Thanks to martinx who reported this on IRC and helped to debug it.
* Make it possible by config option that non global admins can send
broadcast messages to their domains.
* Allow the sender to select the domains the broadcast message should be
delivered to
* Allow the sender to decide if the broadcast message should just be
delivered to mailboxes
The MySQL database layout includes a 'cache' column for historical
reasons, the PostgreSQL database never did.
r1883 removed the 'cache' column from VacationHandler to unbreak
PostgreSQL, and at the same time broke MySQL in strict mode.
This patch re-adds the 'cache' column only for MySQL to fix this
regression.
git-svn-id: https://svn.code.sf.net/p/postfixadmin/code/trunk@1890 a1433add-5e2c-0410-b055-b7f2511e0802
Christian Boltz