Localization of programs and applications
An overview of how different languages are identified is at W3C Language-tags
Additionally, some i18n tips may help to localize your software.
Localization of compiler messages
FPC was created in a way that allows easy localization of compiler messages to different languages. Quite a few translations are already available, but taking into account the number of languages used around the world, there's always room for improvement. ;-)
How to start
First of all, check which languages are already supported. In the installed tree, the message files are stored in a subdirectory called "msg". In SVN, these files are stored under /compiler/msg/.
Every file contains all message in a particular language with a particular character encoding or "code page" (note that different platforms use different ways of encoding characters). As of now, the compiler has no restrictions regarding how the individual files are named, but the convention is that all files start with "error", followed with one or two letters identifying the particular language and then optionally followed with another letter used to distinguish the supported encoding (e.g. "w" for Windows, "i" for ISO character sets used for Unix platforms, or possibly "d" for the so called IBM or OEM codepages used with DOS and OS/2). All message files have an extension ".msg".
If you don't find a message file for your language, or if you find that the existing one looks outdated, you can send an e-mail to fpc-devel mailing list stating that you are interested in providing translation to a new language. This e-mail is recommended to avoid having two people working on translation to the same language in parallel without knowing about each other. Once you're finished, you can send another e-mail to fpc-devel announcing so. You should be contacted by one of the FPC core team members shortly afterwards. He will provide you with instructions where to send the finished file (the mailing list has limits for maximum size of attachments) and then add the file to the SVN repository.
The primary language is English, which means that errore.msg is always the most up to date message file. Although you could use a different language as your basis, it isn't recommended (you can still use some other language as a reference if a particular message in English isn't clear to you and you know some other language better, of course). In addition, it's recommended to always start with the latest revision of errore.msg from trunk branch in SVN repository rather than e.g. the message file from some previous FPC release (which might be already several months old at that time).
If you cannot use an SVN client for accessing our SVN repository for some reason, you can always use ViewCVS WWW interface as an alternative. Some hints for using this interface follow:
- List of message files
- You can see the existing message files and information about their most recent update (revision number and age) here. Clicking on the message file name provides you with the history of changes for this file. Clicking on its revision number shows the latest version and provides a link for downloading it.
- Differences between two revisions
- This link shows all changes performed in errore.msg between SVN revision 1 (roughly corresponding to FPC release 2.0.0) and revision 3940 (the latest SVN revision as of writing this text). Changing the revision numbers and/or file name in the link allows you to display changes for other revisions/files. As you can see, it's very useful to know the SVN revision number of the errore.msg file used as basis for your translation (and preferably put it to comments on the top of the translated file together with your contact information) - when updating a previously created message file, one can easily check changes in the English version since that revision and translate just the modified parts.
Message file structure
Every message file consists of three types of lines:
- All lines starting with "#", these are obviously not expected to be translated at all.
- Lines starting with "%" provide description for individual messages for FPC documentation (immediately following the description). There's no need to translate these lines unless you plan to translate the complete documentation too (probably not - see below).
- All other lines are the real messages. These start with an identifier (not translated) followed with "=", message number, "_", optionally one letter specifying the verbosity level (error, warning, hint, note, information, etc.) and another "_" and then the real text. Special case are multi-line messages, these start with "[", end with "]" and all text in between belongs to that message (used e.g. for help).
Additional information regarding the format can be found as comments directly within the message files.
As an alternative to file based translation of messages, you can use the online translation tool. There are several gotchas, though:
- If you don't have access to that module, you need to ask for it in fpc-devel list first.
- You need to make sure that the latest errore.msg from SVN is loaded as a template.
- If a message file for your language already exists in SVN, you should check that it doesn't include additional updates (e.g. by importing it into the online translation tool and checking for conflicts in the administration module).
- When importing existing message file, you need to make sure that it is in character set/codepage supported by the translation tool (e.g. ISO 8859-1 for "western" languages). You can use e.g. GNU recode tool for conversion if needed. Similarly, you might need to convert to the target character set after downloading the message file from the translation tool before it can be committed to SVN.
Localization of other resources
There are other texts related to FPC which can be translated to other languages: