It looks cool, but I think more examples of where normal diff fails would be good.
Ignoring date fields is a great example you show off, and and made me realise why this is needed over regular diff.
Does it work if you shuffle the lines?
Ignoring date fields is a great example you show off, and and made me realise why this is needed over regular diff.
Does it work if you shuffle the lines?
Yes, it does work if you shuffle lines. There is a --primary-key option that lets you choose what columns uniquely identify a row, it can be compound too. Its assumed to be the first column by default.