The bit-field algorithm

Exponential stores which languages an object has, and whether it is always available, in one integer: a bit field. This page describes the database fields involved, for developers who query them directly.

Language IDs

Every content language is a row of the table ezcontent_language: its ID, its locale (such as ger-DE) and its name. The IDs are powers of two: 2, 4, 8, 16 and so on. A new language gets the lowest power of two no other language has. The value 1 (bit 0) is not a language: it marks an object as always available.

PHP integers have 64 bits, one of them the sign and one the always-available flag, so an installation holds at most 62 content languages (30 on 32-bit PHP), as eZContentLanguage::maxCount() says.

The fields

The object's mask is updated whenever its languages change, that is when a translation is added or removed or the flag is changed.

Example

With English (ID 2) and French (ID 4):

Filtering

To find the objects in a language, the kernel tests one bit: objects in French are those with language_mask & 4 > 0. To show a siteaccess's content, it combines the IDs of its SiteLanguageList[] into one mask and keeps the objects whose mask shares a bit with it, or that have bit 0 set. Which translation is shown is then decided by the order of the list. This works on every database engine the kernel supports.

In PHP, use the API rather than the bits: eZContentLanguage::fetchByLocale(), eZContentLanguage::maskByLocale(), $object->languages() and $object->attribute( 'always_available' ).

More

• Content languages, section 2, and section 11 of Languages in Exponential on the limit of 62.