Several actions are needed for a new language to be supported in &d-i;. The following process may be known as the New Language Process. It describes the minimum required actions and checks that have to be done before starting a new translation effort. Potential translators should be aware that involving themselves into &d-i; translation is not a one-time work. &d-i; is always work in progress and translations will require some attention or some new work from time to time even after they have been completed. After the initial translation effort, which may last for several weeks with 1 hour work per day, the maintenance work needs between 1 hour per week and a few hours per week, depending on the amount of extra translation work done by the new translator(s) for their language in the &debian; project. Translators or translation teams of a not supported language should contact the &d-i; i18n coordination team (see ). They should provide a real name (the Debian Project encourages the use of real names) and possibly a GPG key (this is not mandatory but could help in the future). Such initial contacts are often established by the i18n coordination team members themselves. However, recording the name and contact address of the person who will be the first contact point in initial steps is a very important step to quickly establish a strong relation with identified contacts for the new language. The &i18n-coords; must have the real name and a reliable e-mail address of at least one translator who must be an individual. This individual will be called the language coordinator for the given language. The &i18n-coords; then register the new language in the supported languages list (). For this step to be considered complete, an entry for the future supported language, temporarily with its name as mentioned by the language coordinator, must have been added to the supported languages list XML file (see , with the initial value in the nlp_step field. For technical reasons, only languages listed in the ISO 639-3 standard can be supported directly in software translations. The official ISO 639-3 list can be found on the official ISO 639 Maintenance Agency web site. In case a translation effort is intended for a language not listed in ISO 639-3, the use of one of the reserved codes of ISO 639-3 will be needed. If needed, the &i18n-coords; will search for the ISO 639-3 code for your language if the new translator doesn't already know it, as well as the official English name for the language. Mutual agreement about these important parameters has to be reached at this point. The name and contact address of an existing translation team as well as a backup coordinator are very valuable information to record at this step, but are not considered mandatory. The &i18n-coords; must ask for these information to the language coordinator. For this step to be considered complete, the entry for the future supported language must be completed by its ISO 639-3 code in the supported languages list XML file (see , with the identification value in the nlp_step field. The fields about the backup coordinator and the translation teams are also filled if this information can be provided. code: language ISO 639-3 code english_name: language name in English coord_name: language coordinator name coord_email: language coordinator e-mail supported: should be set to false For a language to be supported completely in &d-i; and later in &debian;, one locale must exist for this language. This document is not aimed at describing locales and which information they should contain. In short, a locale describes specific properties of a language/country combination. For instance, the "de_AT" locale describes specifics of the German language in Austria. Information included in a locale: names of week days (and abbreviations); names of months (and abbreviations); official symbol of the currency; writing style for numbers; writing style for dates; writing style for telephone numbers; usual paper size; collation information for the given language (one of the most complicated parts of locales); .../... The writer of a locale must use the locale files format. The contents of this file must be in UTF-8. Translators shouldn't be afraid by this: this part may be handled by an i18n specialist in Debian, most often one of the &i18n-coords;. The new translator(s) will only need to provide him/her the above information when (s)he asks for it. If the language coordinator is aware of it, (s)he should provide the name of the most used locale(s) for this language. The &i18n-coords; will check whether such locale already exists in Debian. If the new language has no locale, neither in Debian nor from other free sources, writing the new locale will need a lot of work and this step may become very long, especially if it has to be done by one of the &i18n-coords;. For this step to be considered complete, a new locale must either exist or be requested for addition (through a regular bug report) in the locales Debian package. The &i18n-coords; will record this information in the supported languages list by changing the nlp_step field value to locale. The localechooser entry will show both the English official name of the language as well as its translated name. The language coordinator must provide the translation of the new language name (with an initial capital letter if capital letters make sense in this language) and send it to the &i18n-coords; as a single UTF-8 file. UTF-8 is mandatory here. For instance, the French translator will send a file containing the single word: Français. For this step to be considered complete, the new name must be recorded in the appropriate place in the localechooser package. The &i18n-coords; will record this information in the supported languages list by changing the nlp_step field value to language_name. First needed is having the new language added to localechooser (the package responsible for choosing language at the very beginning of a Debian installation). This will be handled by the &i18n-coords;. This entry should mention the &d-i; interfaces which this language is available on (text, latin-1 serial, dialog or graphical interface). This choice depends on the needed rendering mechanism for this language. This defines five categories for languages: Category 0: languages that can be rendered on any ASCII console. This only includes English by a design decision of &d-i;. In short, even if a language can be rendered using only ASCII characters, it will not be placed in category 0; Category 1: languages that can be rendered on any Latin-1 console (several serial terminals). This includes so-called Latin-1 languages; Category 2: languages that can be rendered at the Linux text console, without the need of a framebuffer terminal. Category 3: languages that can be rendered at the Linux text console only by using a framebuffer terminal. The key point for these languages is the availability of their glyphs in the bterm-unifont package. These languages should also not need complex rendering mechanisms such as combination mechanisms. The difference against Category 2 is the possibility to display the language at the Linux console without framebuffer. Category 4: all other languages. Those can only be displayed in a graphical environment. The localechooser entry also mentions a default country choice for this language. This country is often the country where the language is the most spoken (or the official language) and must be covered by an existing locale. If several supported locales exist for this language, this has to be mentioned to the &i18n-coords;. A default locale is also mentioned in this entry. This will be the default choice in case this language is supported in several locales. In case a new locale has to be written in the former initial steps, this default locale can be this new locale, otherwise it must be an already existing locale. With all this information, the &i18n-coords; will write the new entry in localechooser. They will also add the language code to the packages/po/PROSPECTIVE file (see ). For this step to be completed, the new entry must exist in localechooser. It should be commented until the initial steps are completed. The &i18n-coords; will record this information in the supported languages list by changing the nlp_step field value to localechooser_entry. This step is optional for languages that can only be rendered with the graphical version of &d-i; (languages in category 4 at the previous step). The language coordinator must provide a file containing all characters needed for properly displaying the language, not including characters which are part of the ASCII character set (characters in the 32-127 range). This file will be used by the font reduction mechanism (the UTF-8 font used in &d-i; is reduced for size constraints) and is needed for all Latin languages and all non-Latin languages with reasonably sized alphabet. It is allowed to add characters that can be entered directly using a "standard" keyboard for the language and are printed on the keys, for example as displayed on http://en.wikipedia.org/wiki/Keymap. This includes characters accessible through AltGr and characters that can be created using dead keys. Below is an example of the contents of this file for a Latin language, namely French: àâéèëêïîôùûüÿçœÀÂÉÈËÊÏiÎÔÙÛÜŸÇŒ°£µ§ For this step to be completed, the new file must have been committed in the installer/built/needed-characters directory. The file name must be the ISO 639-3 language code with .utf as extension. As long as the language remains a prospective language, that file should be named after the language's ISO 639-3 code and the .prospective file extension. The extension will be changed to .utf when the language is completely supported. The &i18n-coords; will record this information in the supported languages list by changing the nlp_step field value to needed_characters. For font switching support in the bare Linux console, the language coordinator must also provide which console-setup font should be used, to be recorded in its debian/font-switch script. The graphical version of &d-i; will use several different font sets, depending on the language. For each new language, an OpenType font must be available in the main Debian archive. This font must respect the Debian Free Software Guidelines, in short it must be a free font. For this step to be completed, the name of the font and the relevant Debian package must have been recorded by the &i18n-coords; in the DebianInstallerGUIFonts wiki page (&wiki-fonts;). The &i18n-coords; will record this information in the supported languages list by changing the nlp_step field value to ttf_font. This step is meant to decide about the default keymap to propose to users in the console-setup package. This keymap will also become the default keymap on the installed system, both at the console and in graphical environments. For this step to be completed, a bug report mentioning the requested default keyboard mapping for the new language must have been reported by the &i18n-coords; to the console-setup package. The &i18n-coords; will record this information in the supported languages list by changing the nlp_step field value to keymap. All &d-i; work is maintained in a revision control system, using GIT. For translators not familiar with GIT, will give them enough information to be able to commit their translations themselves. This chapter will also explain in details how to get an account on the Salsa server, the server used for centralized development of several co-maintained Debian packages and projects. After your account has been created, please check that you can login with it on Salsa using the chosen password. Then, mention your account name to the &i18n-coords; who will grant you commit access to the needed repositories. For this step to be completed, the language coordinator must have a working &salsa; account and must have sent his/her account name to the &i18n-coords;. The &i18n-coords; will record this information in the supported languages list by changing the nlp_step field value to account and fill in the account field with the account name. After the new translator is notified by the &i18n-coords; that (s)he has been granted commit access to the &d-i; GIT repository, (s)he should check that (s)he's able to checkout the &d-i; repository. For this, will give him/her any needed details. For this step to be completed, the language coordinator must have been able to successfully checkout at least the packages/po directory and add a fake file there. This will be verified by the &i18n-coords;. &i18n-coords; must then add the language code to the packages/po/PROSPECTIVE file in order to avoid the translations to immediately spread out to individual packages. This is motivated by the will of activating languages only after they completed the first two sublevels of translations (). The &i18n-coords; will record this information in the supported languages list by changing the nlp_step field value to commit. They will also check that the language has been added to the list of prospective languages (see ). If the new translator is not used to GNU gettext files (PO files) handling and the use of dedicated PO files editors and tools, (s)he should refer to . At the very minimum, (s)he should choose a gettext file editor and practice with it enough to be familiar with its basic functions. In any case, translators should refer to for information and guidelines about the way they should work. For this step to be completed, the language coordinator must be comfortable with the basics of gettext-based translation. Of course, there is no real way to measure this precisely, so deciding that this step is completed is in general up to the translator him/herself. The &i18n-coords; will record this information in the supported languages list by changing the nlp_step field value to tools. &d-i; translators must subscribe to the &email-debian-i18n-list; where all important announcements and discussions about &d-i; translations are sent. Subscription to this list and all other Debian mailing lists can be done on &url-debian-list-archives;. For this step to be completed, the language coordinator must be subscribed to the &email-debian-i18n-list; mailing list. The &i18n-coords; will record this information in the supported languages list by changing the nlp_step field value to mailing_list. Optionally, translators may subscribe to the &email-debian-boot-list; which is the development mailing list for &d-i;. Topics there are more technical but will allow translators to participate more deeply in &d-i; development. Joining IRC channels is also a good communication method: Debian i18n contributors usually hang around on #debian-i18n on irc.debian.org while &d-i; developers join #debian-boot. When a new translation effort is started, the language coordinator must announce it in the &email-debian-boot-list; mailing list so that all &d-i; developers are aware of the new translation effort. A short personal introduction will be appreciated, even if not mandatory. The &d-i; development team is a real team whose members appreciate working together in a friendly environment. For this step to be completed, the language coordinator must have announced the beginning of the new work to &email-debian-boot-list;. The &i18n-coords; will record this information in the supported languages list by changing the nlp_step field value to prospective. They will also add the language to the /l10n/output-l10n-changes. Finally, they renamed the needed characters and replace the .prospective file extension by a .utf Finally, a new entry will be added to the file used by the scripts which generate the translation status pages (&url-d-i-translation-status;). The new language process officially finishes here. The only work that's left to do is ... translating. Subscription to the &d-i; development list (&email-debian-boot-list;) is also encouraged though not mandatory. This list has a traffic of about 20-30 mails per day, including many administrivia mails (packages uploads or bug reports), but is the best method for being involved in &d-i; development. Many &d-i; developers and translators also use the #debian-boot channel on irc.debian.org for instant communication. Translators are highly welcomed on this channel which is regularly monitored by the &i18n-coords;. A dedicated #debian-i18n channel also exists on irc.oftc.net. Feel free to join it and chat with your fellow colleagues who work on other languages. If translators, and especially language coordinators, happen to not be able anymore to work on the translation of &d-i;, they should notify the &i18n-coords;. When doing so, they should do their best for pointing them to other l10n resources who may be able to continue the work on the given language. Welcome on board: you are now a &d-i; team member! In &d-i; development process, translators are NOT a special kind of developers. I18n and l10n are full part of the development and translators are involved all along this development. Several &d-i; developers started by working on translations and later contributed to other pieces of code. If you didn't already read it, please go and read this documentation chapter for translators (). It contains all the needed information to answer questions such as "Where do I start and how do I work?" As soon as you will have commited your PO file for level 1, even with only a few strings translated, the new language is temporarily called a prospective language. It will be removed from the list of such languages when the level 1 translation (see ) will be about 25% completed (except during the pre-release process), as long as these 25% cover the beginning of the file which contains the most used strings. Don't wait for level 1 to be complete before commiting something. Quick and frequent commits are a good sign for &i18n-coords; that you're really active.