utf8everywhere.org

It doesn't sound like the code was cross platform though right?

Yep, I've read that one too.  It certainly is a complex topic.  It's sad that we still have these issues.  I really need to get to the bottom of this.

If you want to do cross-platform stuff, then since the advice on that site is targeted at Windows it isn't necessarily the best thing to be looking at.

What kind of software are you writing, and what is the functionality of the libraries you say you would need to rewrite?

The advice on that site doesn't seem to be targeted at windows- they appear to be making a real effort to get people to agree on how everyone should be storing strings.

My own libraries are the ones I would need to rewrite.  I'm most interested at the moment in writing cross platform SQLite use the Kompex library, but that's the tip of the iceberg.  I need to build up my libraries again and have something that is simultaneously being developed cross platform.  I never want to touch this code again basically.

I would suggest requesting attention to this question (link at top) to get more input from others, as I have reached the end of my (limited) knowledge on this I am afraid...

I would like to leave this one open.  I do believe it's probably one of the most important topics to advance in computer science.  If we can't truly reuse what we do, we're never advancing as a whole and we're wasting our time with every added function in the future by duplicating work, bugs, etc.  Everything we do should be cross platform.  At least, that's what is true in an ideal world.  To be cross platform, one has to think about just about every aspect of the final product:  the native user-interface which should not be cross-platform for obvious reasons, some common cross-platform business logic, and the "how do they talk to each other" interface.  That separation does not include the encoding logic, so an encoding map needs to be known for each operating system call/API pair between common base code and the layer that separates the specific native code.  All common code uses UTF-8, making conversions only where absolutely necessary.  Each component should always only ever encode their strings in one way, but there are probably libraries out there that can easily produce inconsistent encodings.  I think I may have answered my own question here.  There's probably no other way to do this than to clearly think about all the necessary separations - and even sometimes unfortunately discover that the underlying libraries you've chosen, are also going to be the source of your headaches with unicode or cross-platform"ness".

Thanks everyone