Peter Eisentraut <peter_e@gmx.net> writes:
> Storing char(n) takes 4 bytes plus as many bytes are are required to
> store the actual string, plus n - length(value) bytes for padding
> spaces. This is at least "n" bytes.
Peter omitted one critical point that I think David hasn't absorbed
yet: char(N) measures N in characters, not bytes. When using a
multibyte encoding, N characters may require more than N bytes.
Only in single-byte encodings can you make any simple statements
about the number of bytes occupied by char(N). This is why the docs
are a bit vague.
regards, tom lane