How often does one really need to say, version X was this long ago? All that seems to matter to me is to know the current version, and to be able to say whether a version is older than current. I'm not a fan of year.date versioning systems because doesn't allow you to communicate the significance of the changes in a new version, just that it's newer, but dotted decimal versions can do that too. It also keeps our tie to yearly releases, which IMO we should view as a historical coincidence... having huge gaps of time between releases doesn't help us get any better at making releases.
Also, there are actually several issues being discussed here:
1. should we drop the leading 0 in the version to communicate FreeCAD's maturity (and stop communicating its immaturity)
2. should we have a 1.0 release to communicate FreeCAD's readiness for wide usage
3. should we use a major.minor.patch or year.date versioning system
It seems the consensus to 1 is 'yes'. The arguments for 2 are strong--not having a 1.0 release is like not having a graduation ceremony. Deciding on 3 can IMO be deferred after we drop the leading 0 and after we have a 1.0 release--it's a separate issue.
Also, I don't think a year.date versioning system would cause any Debian packaging problems, except that if we switch to it we can't easily switch back, which seems unlikely anyway.