https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124163
--- Comment #7 from Dirk Herrmann <dirk.herrmann-privat at gmx dot de> --- > I agree that this is more subtle, but the static semantics of a type > conversion do not say that the result is a machine number; they only say that > the result belongs to (the range of) the target subtype. I see - and although I don't understand the reasoning, this seems intentional, looking at ARM 22+25, paragraph 4.6:31 compared to 4.6:32. They have even made a distinction between decimal fixed point and ordinary fixed point: ordinary fixed point values are handled like Float, conversion to decimal fixed point values requires mapping to the machine representation. Thanks a lot again for the explanation - I have learned a lot about Ada arithmetics within a very short time.
