Tomer Gabel

Interim investigation shows that, as one would expect, [this line](https://github.com/flipkart-incubator/zjsonpatch/blob/master/src/main/java/com/flipkart/zjsonpatch/JsonDiff.java#L103) turned out to be a really bad idea :-) ``` // Hack to fix broken COPY operation, need better handling...

As far as I can tell the copy calculation mechanism is fundamentally broken; it finds nodes in the document that stay the same (position/value) after applying the patch, and makes...

That is actually correct behavior; patches are applied on the state of the target object that they operate on. This means the two options are actually equivalent ([1,2,3] -> [1,2]...

@goldsam, can you please provide an example? From what I can tell, the RFCs stipulate that erroneous operations *SHOULD* terminate processing (see [RFC 6902 section 5](https://tools.ietf.org/html/rfc6902#section-5)), whereas the HTTP JSON...

Picking this up, this would essentially be providing a "fuzzy processor" instead of the default one, which suggests this could easily be done outside of the mainline codebase (which I...

@goldsam Still interested? Would appreciate a followup, otherwise I believe we can close this @vishwakarma

Eugh. You're absolutely correct, my only concern is that once these interfaces are made public it'll be hard to modify them (as I did in #95), and they'll need some...

It's a possibility, but I'm not sure it's actually the right strategy. To start with, the API is somewhat limited in what it can offer (e.g. it's possible to write...

To clarify, the result is `[{"op":"replace","path":"","value":"[2, 1, 3]"}]`, which _is_ a valid diff. As things stand we don't have a clear-cut strategy for diff generation (e.g. smallest, fastest-to-diff, fastest-to-apply...), so...

I'm pretty comfortable closing this, unless anyone wants to pick up the discussion and try to formalize the different strategies and/or how to implement them...?