Repository navigation
few changes for discussion - #1180
Open
haydenflinner wants to merge 4 commits into
Open
haydenflinner wants to merge 4 commits into
haydenflinner wants to merge 4 commits into
Conversation
Generated with [Devin](https://devin.ai) Co-Authored-By: Devin <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Rounds to seconds made per-commit timing useless for tape replay. Renames merge_interval_in_s -> merge_interval_in_ms to match. Generated with [Devin](https://devin.ai) Co-Authored-By: Devin <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Fragments were hard-capped at 256 atoms, so a uniform ~1M-atom span covered ~4k fragments. Each rope leaf split then re-mapped every fragment in the span via update_insert_batch, giving O(splits x span) — ~80M fragment iterations on a 2.2MB doc import (~13s to attach, ~47s to re-export state). Bound fragments by run count instead of atom count — a single-run fragment has one cursor at any size: - Cursor::try_merge joins two adjacent Small insert sets when their combined runs fit SMALL_SET_MAX_LEN, collapsing boundary runs that share a leaf. - IdToCursor::coalesce merges adjacent compatible fragments in the range update_insert/update_insert_batch just touched. Adjacency is required (a.counter_end() == b.counter): the list can carry counter gaps left by other containers' ops, and merging across a gap would change get_insert results inside it. - insert() no longer splits a large uniform single-run cursor into MAX_FRAGMENT_LEN chunks, and merges with the previous adjacent fragment when possible. Result: the 2.2MB production doc attaches in ~9ms (was 13.4s), and perf_import_insert_split_quadratic_e2e (2M atoms, 8192 boundary splits, ~33.5M expected fragment updates before) now runs in ~15ms. State exports verified byte-identical to the pre-fix build.
…acker Adds an off-by-default `tracker-stats` cargo feature (loro + loro-internal) with zero-cost `#[cfg]`-gated counters on the richtext Tracker's hot paths: - op counts: inserts, checkouts, retreat/forward elements - skip_applied forwarding, Fugue in_between scans, leaf splits - id→cursor work: per-fragment update iterations, update_many call totals, dense-array rebuilds, large-set updates, iterator yields, batch call counts, span atoms, and max fragment-list length - stats::dump(tag) prints the totals; called at the end of calc_diff_internal and the richtext tracker rebuild Written to quantify a pathological fragment-remapping case on a 2.2MB document (~80M fragment iterations); kept because it makes the inner loop magnitudes visible without recompiling instrumentation.
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Here are a few changes I've needed while building on Loro. I can submit them separately once we see which ones (if any) make sense to merge here.
undo_span-- I don't want to revert the whole doc, I want to invert only the spans other peers wrote if I decide that I didn't want them to be able to do that. It seems to work, haven't written any torture tests for it though.