What I often want to do when reading user data is not treat it as a "encoded string", but just as a stream of bytes. Most data I work with (HTML files, output of other programs) can't be treated as anything but bytes, because people put junk in files / output of programs.
If you are representing strings as bytes, you are intrinsically using an encoding.
> What I often want to do when reading user data is not treat it as a "encoded string", but just as a stream of bytes. Most data I work with (HTML files, output of other programs) can't be treated as anything but bytes, because people put junk in files / output of programs.
Yes, it makes a mockery of the notion that "human readable data is easy". In many cases, you don't want to work with the actual strings in the data anyway, so bytes is the right thing to do.
But yes, this strategy largely avoids encoding issues... until it doesn't.
It's just binary data that might resemble a string. No encoding necessary.
oh, actually there was (either us-ascii or more likely iso-8859-1) the bytes are just values 0-255 what these values mean is the encoding. You're confused because the encoding was implicit, rather than explicit.
It would perhaps be clearer to see it if you for example had to chose if you use ASCII or legacy EBCDIC encoding.
