Question

I have a multilingual project,

the project is fully compatible with HE/EN languages.

In one case it’s going to be uploaded as a subfolder in a big israeli project,

in another it’s going to be used as a full site on it’s own .com domain in english.

I currently have an SVN setup for the israeli project,

I haven’t yet created an SVN for the .com, is this the right way to go, keeping 2 depo’s for this ? I will then have to make sure that the code written on the israeli project is always copied manually to the .com folder,

Can this be managed more efficiently ?

Answer 1

Yes, this can definitely be managed more efficiently. This statement right here is the main problem:

I will then have to make sure that the code written on the israeli project is always copied manually to the .com folder.

That process is not only inefficient, it’s highly error-prone. And it absolutely doesn’t scale at all. Basically you’re doubling your development/testing/deployment/etc. efforts. If you ever need to deploy another instance of the application, you’ll have tripled efforts. And so on.

What you’ll want to do is isolate the components that are different between the instances of the application. Is it just the language content? Is it the whole UI? If it’s the whole UI then you’ll want that UI to be as thin and light-weight as possible. (Well, you want that anyway.) Then all you’d have to double is the UI work. The rest of the logic (business logic, data access, services, etc.) can be constant across all instances of the application.

If it’s just text content, then maybe you can re-factor that text to be pulled from the database. Every label, every article, every menu item, etc. There are CMS systems which do this pretty well, though I have no specific recommendations. But the overall idea is to have the UI be an empty wrapper populated by data specific to that application instance. Then each instance would control its “language” (its content in general) by simply managing that data.

You definitely don’t want multiple code repositories for this. After all, if you have two repositories of code, which one is the correct one? If a change is made in one repository but not the other, which one is right? Your code should have a single “source of truth.” Multiple sources of truth means no truth.