Marc-André Lureau <marcandre.lur...@gmail.com> writes: > Hi > > On Mon, Dec 10, 2018 at 9:26 PM Markus Armbruster <arm...@redhat.com> wrote: [...] >> Looks good, but I'd like to investigate whether my suggested placement >> of the normalization would simplify things before I give my R-by. >> > > I am not sure I follow, you suggest to have normalization before > check? Then how can we produce an error that reflects the input > accurately? Should we carry additional information (original form etc) > on the source during normalization?
Let me play with it. If my idea works out, showing it to you will be easier than explaining it. If it doesn't, ditching it quietly will also be easier than explaining it :)
