Re: localisation in web applications

> How is gettext used? Do you create separate language profile files, that are > loaded up by gettext and passed to PHP? Check out (http://www.gnu.org/software/gettext/gettext.html). That's the home page for the project. Yes, Gettext uses separate files for each language. The files are human-readable, and can be edited with vi, emacs, whatever. There are command-line tools for manipulating the files, for merging change for example. One cool thing is how the English phrase is the key to the translations. You use a function in your code to get a string appropriate for the local language. And if there's no matching phrase, the English phrase is returned. Here's an example (straight from my book): <? //set language to Spanish putenv("LANG=es"); //specify location of translation tables bindtextdomain("error_messages", "./locale"); //choose domain textdomain("error_messages"); //print a test message print(gettext("This book is named Core PHP Programming")); ?> Essentially what's going to happen is that Gettext is going to look for a file named es.po (or similar) in a directory called locale. If it finds "This book is named Core PHP Programming" as a key, it will return whatever's there ("Este libra..."). If not, it will just return they key. That means you won't get weird messages like ERROR_MESSAGE_123, and all the messages are in one language inside the source code. Of course, I should point out that langauges vary in word order. So, when you have a phrase like, "There are %d items in your basket", you have problems. The manual for Gettext talks about this a little bit, and how to deal with it. --- Leon Atkinson <http://www.leonatkinson.com/>