<fill in SMERGE and NMERGE notes here>
Some snippet of IRC which might be woven into this Notes file:
<lm> ob, the whole point of pluggable is that you don’t. It seems like the focus is on some mythical right answer for automerge; I think looking at how you handle the non-automerge case is the point <ob> and if you automerge wrong? <lm> you deal, as has been noted anything will automerge wrong. <ob> who are you and what have you done with lm? <lm> I never advocated for a perfect automerge, I’ve been trying to move focus to the conflicts for years <lm> The old approach w/ smerge was we had all these algs that were disabled, we turned them off one by one as we found cases where they did the wrong thing. But that also means we are leaving any benefits from those algs on the table. The place you realize those benefits is in fm3tool and you need a way to cycle through the algs to do so. That’s what I’ve been pushing for for years. <lm> I do not believe I’ve ever argued we should turn on the stuff that works sometimes by default, you need to look at those.
Guess
The smerge works by first line up sequence numbers (order that line appears in weave) as a quick way to anchor diff region. Then run a set of heuristics over the remaining hunks to reduce, split, join, and generally make for a more pleasant conflict resolving experience.