Page MenuHomePhabricator

Alternative compression API design for illustration
AbandonedPublic

Authored by dblaikie on Aug 10 2022, 6:07 PM.

Details

Reviewers
jhenderson
Summary

An alternative/hybrid between D130516 and D130506

Diff Detail

Event Timeline

dblaikie created this revision.Aug 10 2022, 6:07 PM
Herald added a project: Restricted Project. · View Herald Transcript
dblaikie requested review of this revision.Aug 10 2022, 6:07 PM
Herald added a project: Restricted Project. · View Herald TranscriptAug 10 2022, 6:07 PM
MaskRay added a comment.EditedAug 10 2022, 6:17 PM

The base probably should be set to the current state of llvm/lib/Support/Compression.cpp for comparison, instead of a rebase on D130516

The base probably should be set to the current state of llvm/lib/Support/Compression.cpp for comparison, instead of a rebase on D130516

Oh, right ,sorry - wasn't my intent that it be based on that - thought that was just an implementation detail of how I'd got here. I'll see about getting that fixed up.

dblaikie updated this revision to Diff 452003.Aug 11 2022, 3:25 PM
dblaikie retitled this revision from Alternative compression API design for illustration An alternative/hybrid between D130516 and D130506 to Alternative compression API design for illustration.
dblaikie edited the summary of this revision. (Show Details)

rebase on main instead of D130516

Is there a demonstrating how this performs better than D130506 with an example like llvm-objcopy --compress-debug-sections=zstd?

Is there a demonstrating how this performs better than D130506 with an example like llvm-objcopy --compress-debug-sections=zstd?

Ping this question.

dblaikie abandoned this revision.Aug 25 2022, 1:09 AM

Is there a demonstrating how this performs better than D130506 with an example like llvm-objcopy --compress-debug-sections=zstd?

Ping this question.

After various design discussions I think D131992 is, for my design preferences, more well fleshed out than this patch, so I'm abandoning this one in favor of that one. I'll follow up on D131992 with more thoughts there.