Fixes: https://github.com/llvm/llvm-project/issues/51652
Depends On D122060
Differential D122061
[mlir][sparse] Enhancing sparse=>sparse conversion. wrengr on Mar 18 2022, 8:08 PM. Authored by
Details
Diff Detail
Event TimelineComment Actions re-renaming the old appendPointer to appendCurrentPointer (since I've realized the name "finalizeSegment" is better used elsewhere). Comment Actions Lots of reorganization, subsequent to all the recent changes in the parent revisions. Comment Actions Rebasing for D123166. Also removing a bunch of inline keywords, per MLIR style-guide. Comment Actions Rebasing for the latest version of D122060 (2022-05-11T13:09) This also introduces a minor change to the enumerator design. Namely, due to the splitting of SparseTensorEnumeratorBase<V> vs SparseTensorEnumerator<P,I,V>, I've had to add the new SparseTensorStorageBase::newEnumerator methods to allow constructing objects of the abstract base type without reintroducing the "<P,I> vs <Q,J> problem" for direct conversion. Comment Actions Arg! It looks like when I installed the github version of Arcanist to avoid the https://www.mail-archive.com/debian-bugs-dist@lists.debian.org/msg1838310.html bug with Debian's version, the new version isn't automatically applying the formatting script :( So please bear with the overly-long lines for now
|