Hi there.
We have a website in brazilian portuguese developed using Coldfusion (for the user interface), Hibernate (for the business logic) and Oracle database.
If we consider to support russian, mandarin and japanese languages what concerns do we must have?
Thanks in advance.
The main consideration is to make sure everything (and I mean everything OS,shell,web server, appserver, database, editors) is configured to use utf-8 or unicode by default.
If you expect a lot of asian users its slightly better to use full unicode as most chinese characters fit into a 16 bit UTF-16, but, can take up 24 or 32 bits in utf-8.
With Coldfusion and Oracle this should not present any mojor problems.
The other main consideration is how you plan to handle the internationalisation isssues.
The standard way is to keep langauge/cultural specific items in a "bundle". There are several tools out there to support this, basically you write your app in portuguese making sure all text the user will see is in quoted literals, then run the app through a utility which replaces all literals with a library call and extracts all strings into a "bundle" file. You can then edit the bundle to add other language versions of the strings. The great advantage of this is that these formats are standard and translation agencies will have the tools to edit these files -- so you can easily outsource the translation to specialists.
The other option which requires much more work but IMHO produces a nicer result is to branch of a version of the front end for each language/culture supported. This gets around a lot of problems with text height and string size. Also it handles cultural norms better -- different cultures have differnet ordering and conventions for things like address and title.
A classic example of small differences causing big problems is the Irish Republic and Post Codes, they just dont have them. So if your form validation insists on a Zip code it will annoy your Irish users. The Brits do have post codes but these are two 1 to 4 character alphanumeric strings separated by a space, not the more usual 5 or 7 digit numeric codes.