Incubating the new Sympa documentation site

Sympa Internationalization

(Work in progress)

Sympa internationalization

(Work in progress)

List internationalization

(Work in progress)

User internationalization

(Work in progress)

Language tag

Sympa uses “language tag” to determine context of language and locale for users and lists. The language tag consists of one or more subtags: language, script, region and variant. Below are some examples.

Language tag Description
ar Arabic language (common)
pt-BR Portuguese language in Brazil
sr-Latn Serbian language with Latin script
ca-ES-valencia Catalan spoken in Valencia
ryu Utinaaguti (around Okinawa island)


Translation catalogs and templates

Sympa is designed to allow easy internationalization of its user interface (service mail messages and web interface). All translations for one language are gathered in a single .po file that can be manipulated by standard GNU gettext tools.

Instructions for translating Sympa are maintained.


Sympa templates refer to translatable strings using the loc TT2 filter.


[%|loc%]User Email[% END %]
[%|loc(,]You have subscribed to list %1 with email address %2[% END %]

Sympa had previously been translated into 15 languages more or less completely. We have automatically extracted the translatable strings from previous templates but this process is awkward and is only seen as a bootstrap for translators. Therefore Sympa distribution will not include previous translations until a skilled translator has reviewed and updated the corresponding .po file.

Defining language-specific templates

The default Sympa templates are language independant, refering to catalogue entries for translations. When customizing either web or mail templates, you can define different templates for different languages. The template should be located in a language subdirectory of web_tt2 or mail_tt2 with the language tag.

Example :


This mechanism also applies to comment.tt2 files used by create list templates.

Web templates can also make use of the lang variable to make templates multi-lingual:

Example :

[% IF lang == 'fr' %]
[% ELSE %]
[% END %]

lang variable is also defined for JavaScript.


Overriding cascading style sheets by language context

web_tt2/css.tt2 for specific languages will override any portion of main css, not fully replacing it. So they may be used for customization for particular language.

For example, default css.tt2 specifies the font families covering Western scripts (Cyrillic, Latin, …). East Asian users may prefer to consistent font family supporting Western along with Eastern scripts (hangeul, hanzi, …). Sympa 6.1.18 and later include css.tt2 specific to these languages (web_tt2/ja/css.tt2, web_tt2/ko/css.tt2, web_tt2/zh-CN/css.tt2 and web_tt2/zh-TW/css.tt2).

For another example, users using languages with right-to-left scripts (Arabic, Hebrew, …) might wish web page layout to be fit in direction of texts. It may be possible to override attributes for alignment by css.tt2 for each language (web_tt2/ar/css.tt2 etc.).

Translating titles of topics, scenarios and description of list creation templates

List topics are defined in a topics.conf file. In this file, each entry can be given a title in different languages. See “List topics”.

Scenarios and comments in list creation templates can have titles by multiple languages. See “Content of scenario file” and “comment.tt2”.

Handling of character sets

Until version 5.3, Sympa web pages were using in each language’s character set (iso-8859-1 for French, utf-8 for Japanese, …) whereas every web page now uses utf-8. Sympa also expects every file to be UTF-8 including : configursation files, templates, authorization scenarios, .po files.

Note that the shared documents (see “Shared document repository”) filenames are Q-encoded to make their storage encoding neutral. This encoding is transparent for end users.

Support for legacy character sets

Sympa fully supports Unicode. By default, all messages sent by Sympa will be encoded by utf-8 character set. However, in some language environments legacy character sets are preferred, for example iso-2022-jp in Japanese language. Messages sent by Sympa may be encoded using legacy character sets spesific to each language. See legacy_character_support_feature parameter.

Note that this setting does not affect messages sent by users but only the messages mailing list robot will send.

CC BY-SA 4.0 Unless otherwise specified, the contents of this document are licensed under the Creative Commons - Attribution - ShareAlike license. For more details see LICENSE and AUTHORS.

Theme originally designed by orderedlist