I can't offer any more than Godmy already has for this thread, but I will say this:
From a comparative/historical perspective, the 'European' view that only
vowel quantity makes a difference is facile. It pretends that the written language exists in a vacuum, that the speakers aimed at perfection in writing when really all they wanted was to understand one another (utility), & that the literary standard accurately reflects the *spoken standard.
When we consider that these vowel sounds are very often the result of an
earlier sound (or sounds) undergoing some fundamental change, it seems hard to believe that the resulting symbols would be a perfect fit. Consider the 'u' vowel:
The Italic family evidences a considerable amount of vowel reduction, just as does
our language, and a distinctive example is to be found in the '2nd declension' (o-stems). Compare the Greek χόρτος with its Latin equivalent
hortus (and more impressively Oscan
húrz, where 'z' = /t͡s/ as in German). The accented vowel retains (ostensibly) its quality, where the unaccented (atonic) vowel
changes quality.
On the other hand, the long 'ū' sound is most often the result of a merging of two (or more) earlier vowels (monophthongization), as in
dūcō < Old Latin
douco.
My point is that it is equally as unlikely that the reduced, short 'u' of the o-stem nom. should have so distinct a quality as /u/, given that it vanishes entirely in some words (e.g.
puer) and regularly in sister dialects, as it is unlikely that the merged, long 'ū' should have so
indistinct a quality as we suppose the short 'u' did (that is /ʊ/). In short, there's no equivalence.
Part of the problem is the (mistaken) thought that Greek
also had minimal pairs like ε:η, which were only supposed to be distinguished by
quality quantity. The idea being that if
Greek is like this (and it's not), then Latin must be also (and it's not).
*A language that strove a little harder in the other direction was Sanscrit, but even the pundits had troubles of their own!