Closed kent-white closed 8 years ago
It looks to me like buildTable
and buildDiff
are working on equality. Then processDiff
looks for updates by checking identifiedSame
.
Would this whole thing be simpler if that were reversed? If buildTable
and buildDiff
worked off of identifiedSame
and then processDiff
checked equality (perhaps with metadata tracked by buildDiff
)?
Stylistically I believe we prefer a space after comment markers, before the comment.
//not like this
// but like this
Thinking through it, I think we need our extension to be limited to collections whose indices conform to BidirectionalIndexType
, and then we should only be using predecessor()
and successor()
instead of advancedBy()
. Thoughts?
I think maybe after a few more changes we're ready to rebase and merge it in. Probably only worth 1 commit? Thoughts?
processDiff
doesn't handle a combo of insert/delete on the same object at the same index, it will will just leave those steps as is. I commented on the line where that may be handled (163).