> both sortCoalesceFL and dropAllInverses should be no-ops from the
> point of view of effect, right?
Yes, that is, assuming effect is only used to apply patches or similar
operations that really only care about the effect of a patch. I hope
this is the case.
> unravel only seems to be used in conflict marking, so I think it's safe
> to change it. Not sure if we should or not though.
I have been looking at the code again and I agree with you that the
change is safe.
Point in favour: the V2 implementation does not use sortCoalesceFL in
equivalent places, in fact the only occurrence of it is in the obscure
remNons function (where I think it could be replaced by dropAllInverses,
though I have no proof of that, of course). Neither does V3 use
sortCoalesceFL anywhere.
Point against: it worked in the past and there seems to be no pressing
reason to make the change.
|