I’ve read this thread: Merge conflict markers for easy merge? - #2 by pmeunier but I think I still got some doubts on it and I’d like to be wrong if I’m missing something:
I can understand the scenario where two authors make the identical change for different underlying reasons, leading to a ‘want both’ intent (e.g., two people incrementing a counter separately).
However, I view the decision to treat identical changes from different authors as a conflict as more of a design choice than a mathematical necessity. Even if we merged such changes automatically, the category-theoretic structure remains intact and a pushout can still be computed.
Arguing that it ‘might cause semantic issues’ feels a bit overly restrictive—after all, even changes in entirely different files can lead to semantic breakage or compilation errors. IMO, it’s more common that the intent from different author is the same than different. I think each time encountering this issue I have to resolving conflicts or reset really produces some friction
As for the requirement that every vertex must originate from a specific change, I think maybe we could simply allow a vertex’s internal structure to reference multiple source changes simultaneously.
I’m really looking forward to hearing your thoughts on this and would appreciate any insights that could help clear up my confusion.