Allowing that this is probably a minority position, but could I highlight a use case and ask whether any change relating to namespaces etc. would cause problems.
For some years I have used a program organisation where a library- perhaps a backend to interface to a specific physical instrument- can be linked either statically (ordinary "uses" clause) or dynamically (run-time dynamic loading of a .so or .dll).
If the linkage is static, the library (e.g. Dso112a) appears to be a unit and functions it exports may be accessed as e.g. Dso112a.MagicNumber().
If the linkage is dynamic, the library is an instance of a class wrapping the .so or .dll, loading the binary and finding entry points at first use. Exported functions/methods may again be accessed as e.g. Dso112a.MagicNumber() where in this case Dso112a is the name of an instance.
Could the introduction of namespaces mess up this convention? If so, is there an elegant workaround?
MarkMLl