🚀 go-pugleaf

> [...]
>> And, it _does_
>> require modifications to the loader (at least to the extent of
>> recognising all the long suffixes to the procedure names which
>> describe which combination of aliasing is allowed in each object
>> code version).
> 
> You are grasping at straws. The number of the partition chosen takes at
> most a few bits to encode.

The number of different _possible_ partitions in your scheme varies
as 2^N where N is the number of procedure arguments.  This ignores the
possible aliasing of globals which multiplies in another big factor.
Since you recommend allowing the user to assert aliasing information,
any of these partitions are possibly generated as a result (assuming
that you don't always automatically generate all of them).

Well, you have an exponential naming problem (at least), which is more
than a few bits.

J. Giles