Extract PullIndexJournal from ImportClient - #468
Open
adamziel wants to merge 5 commits into
Open
Conversation
Contributor
Pull pipeline performance —
|
| Stage | PR | trunk | Δ | Status | Details |
|---|---|---|---|---|---|
playground-sqlite-db-pull |
7.63 s | 7.54 s | ⚪ +81 ms (+1.1%) | ✓ | condition=db-pull in PHP.wasm runtime=php.wasm 8.3 wp_mysql_parser=enabled mode=lexer native_lexer=verified native_token_stream=WP_MySQL_Native_Token_Stream native_token_count=18 native_parser=selected trunk: condition=db-pull in PHP.wasm runtime=php.wasm 8.3 wp_mysql_parser=enabled mode=lexer native_lexer=verified native_token_stream=WP_MySQL_Native_Token_Stream native_token_count=18 native_parser=selected |
playground-sqlite-db-apply |
3.07 s | 3.07 s | ⚪ -4 ms (-0.1%) | ✓ | condition=db-apply to SQLite in PHP.wasm runtime=php.wasm 8.3 wp_mysql_parser=enabled mode=parser native_lexer=verified native_token_stream=WP_MySQL_Native_Token_Stream native_token_count=18 native_parser=verified native_ast=WP_MySQL_Native_Parser_Node sqlite_driver_parser=verified trunk: condition=db-apply to SQLite in PHP.wasm runtime=php.wasm 8.3 wp_mysql_parser=enabled mode=parser native_lexer=verified native_token_stream=WP_MySQL_Native_Token_Stream native_token_count=18 native_parser=verified native_ast=WP_MySQL_Native_Parser_Node sqlite_driver_parser=verified |
| Total | 10.70 s | 10.62 s | ⚪ +77 ms (+0.7%) |
Numbers carry runner noise; treat single-run deltas as directional, not authoritative.
📈 Trunk performance history — commit-by-commit timeline.
adamziel
force-pushed
the
adamziel/design-import-write-class
branch
from
August 7, 2026 00:55
ead787c to
4030e6e
Compare
This was referenced Aug 7, 2026
adamziel
force-pushed
the
adamziel/design-import-write-class
branch
from
August 10, 2026 10:43
4030e6e to
84a87f3
Compare
adamziel
changed the base branch from
trunk
to
codex/file-index-diff-processor
August 10, 2026 10:45
adamziel
force-pushed
the
adamziel/design-import-write-class
branch
8 times, most recently
from
August 11, 2026 00:15
39c8c30 to
e069996
Compare
adamziel
added a commit
that referenced
this pull request
Aug 11, 2026
… between push and pull (#541) Adds `FileIndexDiffProcessor`, a single-pass, resumable traversal over an old and new path-sorted filesystem index. It labels each current path as `added`, `modified`, `deleted`, or `unchanged`, so callers choose an operation without rebuilding the index comparison. `PushPlan` now uses the same processor. ## Usage Create the comparison and select its first path: ```php $index_diff = FileIndexDiffProcessor::create( $old_index_file, $new_index_file ); $has_path = $index_diff->next_path(); ``` `get_path_transition()` describes the difference between the old and new index records. A path present in both indexes is `modified` when its type, size, or ctime differs: ```php while ($has_path) { $path = $index_diff->get_path(); switch ($index_diff->get_path_transition()) { case "added": handle_added_path($path); break; case "modified": handle_modified_path($path); break; case "deleted": handle_deleted_path($path); break; case "unchanged": break; } $has_path = $index_diff->next_path(); save_cursor($index_diff->get_cursor()); } $index_diff->close(); ``` The per-index getters expose record details when an operation needs them: ```php if ($index_diff->get_path_transition() === "modified") { compare_sizes( $index_diff->get_size_in_old_index(), $index_diff->get_size_in_new_index() ); } ``` When a path is absent from an index, the neighboring-path getters describe the position where it would occur. For a path deleted from the new index: ```php if ($index_diff->get_path_transition() === "deleted") { $preceding_path = $index_diff->get_preceding_path_in_new_index(); $following_path = $index_diff->get_following_path_in_new_index(); } ``` For a path added to the new index, inspect the following path in the old index: ```php if ($index_diff->get_path_transition() === "added") { $following_path = $index_diff->get_following_path_in_old_index(); } ``` A following-path getter requires the current path to be absent from that index. Otherwise the processor has not read the entry after the current path and rejects the call. Resume from the cursor stored after `next_path()` advances past a processed path: ```php $index_diff = FileIndexDiffProcessor::resume( $old_index_file, $new_index_file, $stored_cursor ); ``` A selected path is not part of the cursor until the following `next_path()` call advances past it. Closing before that call leaves the path available for replay after resume. Both index files must remain unchanged while a cursor may be resumed. The processor holds at most one unread entry from each index. Its dedicated tests cover all four transition labels, decoded-path ordering, preceding and following paths, cursor movement, resume, missing indexes, EOF, and close. #468 follows this PR in the stack. ## Testing ``` cd tests ../vendor/bin/phpunit Import/FileIndexDiffProcessorTest.php Import/PushPlanTest.php ../vendor/bin/phpcs ../packages/reprint-client/src/lib/index/class-file-index-diff-processor.php Import/FileIndexDiffProcessorTest.php cd .. vendor/bin/phpstan analyze --memory-limit=1G git diff --check ```
Move the pull/index.wal writer and the merge which publishes its records into the remote and local indexes out of ImportClient into a dedicated PullIndexJournal class. ImportClient keeps the cursor and state ordering; the journal only guarantees that flushed records are durable before each cursor checkpoint.
Move sequential remote-index parsing, blank-line skipping, and byte-offset seeking into a read-only reader. Use it for diffing, journal application, prefix lookup, and statistics while leaving raw record scans untouched.
adamziel
force-pushed
the
adamziel/design-import-write-class
branch
from
August 11, 2026 00:28
e069996 to
644e58f
Compare
adamziel
changed the base branch from
codex/file-index-diff-processor
to
trunk
August 11, 2026 00:28
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.
files-pullnow keeps pull-index publication inPullIndexJournaland sequential remote-index entry reads inRemoteIndexReader, instead of mixing both responsibilities intoImportClient.PullIndexJournalownspull/index.wal: appending completed mutations, flushing them before cursor checkpoints, merging complete records into the remote and local indexes, and removing the empty lifecycle marker. Publication remains restart-safe: the remote index and any required local-index replacement are published before the records are cleared, so a stopped application replays the same batch.RemoteIndexReaderis read-only. It covers entry-shaped reads ofremote-index.jsonlandremote-index.next.jsonl: decoding validated paths, skipping blank lines, and exposing byte offsets for diff resume. It does not readlocal_index.jsonl; that relative-path schema includes the optionalemptyfield and remains owned byPushPlanand the local-index merge helpers. Raw line counters and symlink-record scans also remain inImportClientbecause they either do not parse entries or needtargetandintermediate, which are outside the reader's entry shape.Merge ordering, selection, preserve-local behavior, deletion policy, writers, and raw record handling remain with their existing owners.
Testing
The journal tests exercise its public API. The reader tests cover sequential reads, blank lines, a missing index, byte-offset resume, and continuing after an invalid line.