Yeah, I noticed this too - and fixed it too - and ended up forgetting I
did it. :)

Let me see - in a perfect world (my perfect world) - this email (yours)
would more or less be sent to me since I am a co-maintainer of
VersionNumber on SM (using some nice mapping between PI and SM etc that
actually is in place but not in use) and I would put the fix in and
publish a new version of that darn package (which btw is a quite poor
class all in all, when you look really, really close you realize that
you can only branch ONCE per version number, ouch!!).

Eh, ok, so I just tried and... hmmph. Ok, so VersionNumber is an old .st
file on SM. I made a new release of that. But... hmmm, so I notice two
methods on VersionNumber are class extensions from SMBase - which could
be moved into VersionNumber now that I am messing with it (haven't done
that earlier).

  (ok, so I tried using an old 3.8-6665 image and did "SMSqueakMap
bootStrap" in it and then upgrade all - seems to end up fine with the
fix in place)

But SMBase... has diverted off in 3.10 with changes that I am fairly
sure are not backwards compatible with 3.8 (ChangeSorter vs
ChangeOrganizer).

Ok, I give up there. I can't say I understand how SM is supposed to be
maintained anymore - feel free to enlighten me. In fact - feel free to
explain to us all how the new process/model for bug fixing, maintenance,
packages etc is meant to work (IIRC that was the purpose of 3.10).

And no, not being sarcastic - possibly a bit tired.

regards, Göran