I realize that we have given up on setting a complete system for that aspect
of XLIFF, but I just want to add one additional issue we haven't talked
about yet:
When specifying a maxwidth in bytes or a maxbytes values, tools will not be
really able to use efficiently such metadata if they don't know the encoding
that will be used in the final output. For example, if a tool calculate a
text size in Shift-JIS during translation, but the target platform is Unix
and the encoding ultimately used for the translated text is EUC-JP, the
validation done by the tool will be irrelevant as those two encoding may
give different byte length for the same text. The same goes for Latin 1 text
and UTF-8 vs windows-1252, etc. We miay need an extra attribute to specify
the "target encoding", or something like that. Just something to keep in
mind in your thoughts for 2.0...