This is a breaking change but only for code that uses turbofish syntax on the diff method, the DiffIter type, or the DiffItem type.
Motivation:
We use diff in an algorithm that merges two maps similar to union_with_key, but tries to maximize sharing of internal nodes between similar maps. See "Stage 2" in https://github.com/bodil/im-rs/issues/166. This works by collecting the results of diff into two Vecs of (key, value) pairs that would need to be inserted in either input map in order to turn it into the desired merge/union. After that is done, we pick the smaller set of updates in order to decide which of the input maps to mutate.
At the moment these Vec contain cloned keys and cloned values. With this PR, they could instead be borrowed with each Vec only borrowing from one of the two input maps. After we’ve decided which Vec to use, only its contents needs to be cloned for inserting into the mutated map.
This is a breaking change but only for code that uses turbofish syntax on the
diff
method, theDiffIter
type, or theDiffItem
type.Motivation:
We use
diff
in an algorithm that merges two maps similar tounion_with_key
, but tries to maximize sharing of internal nodes between similar maps. See "Stage 2" in https://github.com/bodil/im-rs/issues/166. This works by collecting the results ofdiff
into twoVec
s of(key, value)
pairs that would need to be inserted in either input map in order to turn it into the desired merge/union. After that is done, we pick the smaller set of updates in order to decide which of the input maps to mutate.At the moment these
Vec
contain cloned keys and cloned values. With this PR, they could instead be borrowed with eachVec
only borrowing from one of the two input maps. After we’ve decided whichVec
to use, only its contents needs to be cloned for inserting into the mutated map.The full algorithm is at https://foss.heptapod.net/octobus/mercurial-devel/-/commit/53e9ccfaacd492852419cb9dca1abae1765a7474#1e7b6882ee05cf3bc9882ed81b2a0a8dac61ded6_179_182