> The patch is a bit "ugly" (in the sense that it's a bit too ad-hoc)

Yes, I feel the same way. What seems more appropriate is if there was
something like process-output-p which would indicate to Semantic that
process output is ready to accept, and thus that Semantic needs to
unwind out of its save-excursion form before allowing
accept-process-output. Does the while-no-input you mentioned exit in
that circumstance?

> But the main problem I see is that you already have a few "trivial"
> patches installed, so this is summing up to something non-trivial
> which requires copyright paperwork.

That's already done and you granted me commit privs to Emacs and
ELPA. Let me know if you want me to forward the docs again confirming
the processing is complete.