I scanned the version numbers as parsed out of distribution filenames on my minicpan and analyzed the generic format. Since we're talking about distribution version numbers and most of the time those will be generated by M::B, EU::MM or M::I and applied to META.yml and the filename, it's reasonably representative of the problem space.
While there are probably some edge cases that I didn't parse correctly, it's close enough to see what's out there. In the following table "X" is a digit (best viewed monospaced). The full analysis is here: http://echo.dagolden.com/~xdg/format-analysis.txt OVERALL TOP FORMATS X.XX 12986 X.X 1624 X.XXX 1048 X.X.X 754 X.XXXX 359 X.XXXXX 165 # Repository sequence numbers? vX.X.X 138 X.XX_XX 89 X.X.XX 71 X 64 X.XX.X 62 X.XXXXXX 58 X.XX.XX 37 X.XXa 34 X.X.X.X 19 XXXXXXXXXX 19 # YYYYMMDDNN? X_XX 18 XXXXXXXX 18 # YYYYMMDD? X.XXb 17 X.XX_X 16 -- David
