use utf8 actually has fairly little to do with it -- almost no one uses unicode identifiers, and a program can easily be encoding-aware without ever including UTF-8 string literals in the code.
But yes, the best wisdom that I know of for dealing with encodings is this:
- Always know where your data is coming from and how it's formatted, and decode it as soon as possible (unless it's meant to be processed as bytes).
- Always understand the data format you're writing to or what your client is expecting, and encode on output (unless your data is already bytes).
- And when it comes to text, always work with character strings in the "interior" of your program.
The very existence of a million different character sets and a million different encodings should be a detail of the interface as much as possible. There are some things you'll still have to keep in mind -- for example different collations for different languages -- but it's an ideal to strive for anyway, and following it as far as possible should greatly reduce the number of "encoding issues" in your code.
To answer your question more directly, yes -- if you're reading textual data from outside without decoding, or sending data anywhere without encoding, there's a very good chance that you're making a mistake, and that your code will break when someone else uses it in a locale different from yours.