I'll soon be working on a large c# project and would like to build in multi-language support from the start. I've had a play around and can get it working using a separate resource file for each language, then use a resource manager to load up the strings.
Are there any other good approaches that I could look into?
I've seen projects implemented using a number of different approaches, each have their merits and drawbacks.
I'd say the resource method you've chosen makes a lot of sense. It would be interesting to see other people's answers too as I often wonder if there's a better way of doing things like this. I've seen numerous resources that all point to the using resources method, including one right here on SO.
I don't think there is a "best way". It really will depend on the technologies and type of application you are building.
Webapps can store the information in the database as other posters have suggested, but I recommend using seperate resource files. That is resource files seperate from your source. Seperate resource files reduces contention for the same files and as your project grows you may find localization will be done seperatly from business logic. (Programmers and Translators).
Microsoft WinForm and WPF gurus recommend using seperate resource assemblies customized to each locale.
WPF's ability to size UI elements to content lowers the layout work required eg: (japanese words are much shorter than english).
If you are considering WPF: I suggest reading this msdn article To be truthful I found the WPF localization tools: msbuild, locbaml, (and maybe an excel spreadsheet) tedious to use, but it does work.
Something only slightly related: A common problem I face is integrating legacy systems that send error messages (usually in english), not error codes. This forces either changes to legacy systems, or mapping backend strings to my own error codes and then to localized strings...yech. Error codes are localizations friend
+1 Database
Forms in your app can even re-translate themselves on the fly if corrections are made to the database.
We used a system where all the controls were mapped in an XML file (one per form) to language resource IDs, but all the IDs were in the database.
Basically, instead of having each control hold the ID (implementing an interface, or using the tag property in VB6), we used the fact that in .NET, the control tree was easily discoverable through reflection. A process when the form loaded would build the XML file if it was missing. The XML file would map the controls to their resource IDs, so this simply needed to be filled in and mapped to the database. This meant that there was no need to change the compiled binary if something was not tagged, or if it needed to be split to another ID (some words in English which might be used as both nouns and verbs might need to translate to two different words in the dictionary and not be re-used, but you might not discover this during initial assignment of IDs). But the fact is that the whole translation process becomes completely independent of your binary (every form has to inherit from a base form which knows how to translate itself and all its controls).
The only ones where the app gets more involved is when a phase with insertion points is used.
The database translation software was your basic CRUD maintenance screen with various workflow options to facilitate going through the missing translations, etc.
I'd go with the multiple resource files. It shouldn't be that hard to configure. In fact I recently answered a similar question on setting a global language based resource files in conjunction with form language resource files.
Localization in Visual Studio 2008
I would consider that the best approach at least for WinForm development.
You can use commercial tools like Sisulizer. It will create satellite assembly for each language. Only thing you should pay attention is not to obfuscate form class names (if you use obfuscator).
I highly recommend ResXManager tool which works with ResourceDirectories (resx) and {x:static} markup.
ResXManager works as addin for visual studio or standalone application.
Most opensource projects use GetText for this purpose. I don't know how and if it's ever been used on a .Net project before.
Standard resource files are easier. However, if you have any language dependent data such as lookup tables then you will have to manage two resource sets.
I haven't done it, but in my next project I would implement a database resource provider. I found how to do it on MSDN:
http://msdn.microsoft.com/en-us/library/aa905797.aspx
I also found this implementation:
We use a custom provider for multi language support and put all texts in a database table. It works well except we sometimes face caching problems when updating texts in the database without updating the web application.
