We need a tool to create a dummy version of the l10n files in a test language like pig latin or something. This will allow us to Another possibility is to have a config item that puts the LocalizationService into 'test' mode where it places a distinguishable token around all the strings it knows how to translate, something like: "The quick brown fox jumped over the lazy dog" would be returned as: "%%%The quick brown fox jumped over the lazy dog%%%" so a visual check would show all the strings that weren't within the %%% as being untranslated (from DB, not wrapped in a call to the service, etc...).
I meant to say above: This will allow us to visually inspect the website to search for missing translations.
I added a bit of code to the LocalizationService that wraps the messages in a marker if a specific debug flag is set in the config file: public String getMessage(String messageId, Locale locale, Object[] args) { .... // If we have put the Service into debug mode we // will wrap all the messages in a marker. if (Boolean.valueOf( Config.get().getString("web.l10n_debug")). booleanValue()) { StringBuffer debug = new StringBuffer(); String marker = Config.get().getString("web.l10n_debug_marker"); debug.append(marker); debug.append(mess); debug.append(marker); mess = debug.toString(); } return mess; } I turned it on and immediately noticed that "Logged In :" text on the main page wasn't wrapped in the marker. It provides an easy way to check on a page (or the entire site in general) what has been localized and what hasn't. I'm assuming its OK for me to checkin this modification to the rhn_web.conf file into CVS? Should I put a better guard around that Boolean.valueOf() call in case that config value doesn't exist? Mike
This bug has been closed due to inactivity. Please open a new bug with specific details if this problem is still occurring or if an enhancement is needed.