I do not believe we are handling the problem of what to
do if the user inserts characters into the editor that
can't be represented in the original source file.
The simplest case is if you are using WinMergeU.exe and
editing 8-bit source files, and you add some Greek,
some Telugu, and some Arabic -- they can't possibly be
converted into the source file codepage.
What do we want to do in this case ?
Then, what about plugin unpackers and packers -- do
they report if they did lossy conversion ? How do we
handle that ?
What if the unpacker wasn't lossy, but the packer is ?
(I don't have a proposal for all this, so I'm hoping
someone else will.)