When is a triplet not actually played as three equal-duration note lengths?
If I may use ratio notation -- I have always understood the only acceptable interpretation of a triplet notation to be the following:
1:1:1
In particular, when the rhythm is misinterpreted as a sort of clave:
3:3:2
... that's always been a good indication to me that the person executing the triplet is not processing it correctly.
However, I have recently seen it claimed that Baroque-era triplets (like the alternate version of Bach's 1st Invention) would correctly be interpreted (when making use of historical performance practice) in one of two duple rhythms:
2:1:1
1:1:2
I would like to see a summary with references of the prevailing wisdom on this issue -- some preliminary searching seems to indicate it is at least not so simplistic as claimed.