Fix routerlicious timeouts in compression related tests #3
Conversation
This file contains 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
As a quick mitigation, we have a test that is causing our whole pipeline to run into some serious locks. This performance test has successfully caught a performance regression, we have an item investigating and fixing that [AB#8127](https://dev.azure.com/fluidframework/235294da-091d-4c29-84fc-cdfc3d90890b/_workitems/edit/8127) I ran without this locally, and was able to get these tests to pass.
## Description Add optional "source" to forest cursor creation. In ObjectForest, use it to produce more descriptive errors about unexpected cursors.
These tests have been dead for quite a while, and their existence creates more problems than potential value. We will come back and reconsider options for visual regression testing in the near future. [AB#8120](https://dev.azure.com/fluidframework/235294da-091d-4c29-84fc-cdfc3d90890b/_workitems/edit/8120)
BREAKING CHANGE: A new tree status has been added for SharedTree nodes. `TreeStatus.Created` indicates that a SharedTree node has been constructed but not yet inserted into the tree. Constraints passed to the `runTransaction` API are now marked as `readonly`.
microsoft#20970) Updated the following: client (release group) Dependencies on @fluidframework/eslint-config-fluid updated: @fluidframework/eslint-config-fluid: 5.3.0 There were some code changes needed to adhere to the lint config. They were partially automated then I manually fixed up others.
…1259) Updates the policy config for the npm-package-json-script-dep policy so it checks that `concurrently` is a dep when needed. Also sorts the config alphabetically and adds typing to the config object so we get IntelliSense and type-checking.
…alizeOrUpdateGCState (microsoft#21269) Added peformance test to benchmark the peformance of `initializeOrUpdateGCState` which is known to be expensive when number of unreferenced nodes are high. This is due to the creation and initialization of timer for every unreferenced node. The test added benchmarks the peformance of `initializeOrUpdateGCState` with 5000, 15000 and 30000 unreferenced nodes. Once the timers issue is fixed, the performance of these tests should increase.
…1273) from 5.3.0 @fluidframework/eslint-config-fluid
Policy: - Package "exports" other than those under "./internal" should be linted by api-extractor. - A single "./internal" or alternatively "." export should also be linted by api-extractor with a bundle config. Recently this was "check:release-tags" and now is "check:exports:bundle-release-tags". Unfortunately, api-extractor will only lint a single entrypoint at a time and needs the config file to be local. So, this means lots of files and script entries. Build config: - Add "check:exports" to "lint" tasks. Apply policy to client-utils and container-definitions. [AB#8141](https://dev.azure.com/fluidframework/235294da-091d-4c29-84fc-cdfc3d90890b/_workitems/edit/8141)
…ach release level (except internal) (microsoft#21267) Each package now generates API reports for each release level alpha, beta, and public. These reports correspond with the trimmed roll-ups API-Extractor generates for each release level. This offers convenient insight into which APIs of a given package are available in which entrypoint(s). It also offers a mechanism for tracking API items as they are promoted to higher levels of support guarantees. For now, we are not generating reports for internal-level exports. This is intentional but does not have to be a permanent change. If we decide in the future that it would be valuable to track our internal API surfaces, we can add that level to the set of reports we generate. These changes also exclude the packages under `build-tools`, as those packages don't leverage type-trimming, nor do they manage multiple entrypoints. [AB#7523](https://dev.azure.com/fluidframework/235294da-091d-4c29-84fc-cdfc3d90890b/_workitems/edit/7523) ### Some notes for reviewers 1. Reports for empty package APIs are still generated, and simply don't contain export details. We may wish to file a feature request to make this a bit cleaner. Probably append a notice that there are no exports at the given release level. 2. "Trimmed" reports will contain "untrimmed" imports. This matches the behavior of the corresponding typings rollups, but it's not particularly interesting from a report perspective, and could create noise in automated review enforcement. We should probably file a feature request around this too, possibly just a way to opt out of including imports in the report file (I don't think we generally want them there anyways).
This small change will make it easier for future implementations of "flex trees" to access `indexForAt`.
## Description Unify `IDisposable` interfaces from `@fluidframework/tree` and `@fluidframework/core-interfaces`. ## Breaking Changes Public APIs in `@fluidframework/tree` now use `IDisposable` from `@fluidframework/core-interfaces` replacing `disposeSymbol` with "dispose". `IDisposable` in `@fluidframework/core-interfaces` is now `@sealed` indicating that third parties should not implement it to reserve the ability for Fluid Framework to extend it to include `Symbol.dispose` as a future non-breaking change.
BREAKING CHANGE: `TreeStatus.Created` has been renamed to `TreeStatus.New`.
…es and snapshots (microsoft#21264) Added a document (.md file) to the container runtime's summary folder. it describes the format of summaries and snapshots. This should help developers understand how to look at the data, how to make changes, etc. It should also help users understand the format of their data in summary / snapshot. The documentation goal is to have a README.md in the same folder that describes overall summarization and it links to this document.
## Description Make experimental packages consistent with how they extends the base tsconfigs.
## Description Our default target is es2020. There is no need to explicitly target older than that. ## Breaking Changes Devtool and build-common may now require ES 2020, just like all other client packages.
## Description This adds a new implementation of `FlexTreeNode` (and `FlexTreeField`, etc.) which wraps a tree of `MapTrees` and allows their data to be read via the flex APIs. This will allow us to support the reading of unhydrated SharedTree nodes in the future.
Bumping server dependencies to prerelease version in the client release group to access new [targeted signal capabilities](microsoft#19519) on the client side. Will need confirmation from FF release engineer(s) since they will be the person who will have to do the real server release. I used the following command to complete the bump from the repo root: `flub bump deps server -g client -t greatest --prerelease` I then did `pnpm install` to update the pnpm-lock file.
Change doc references for `AzureUser` as `AzureUser` is internal which prevents `@inheritDoc AzureUser...` resolution. [AB#8135](https://dev.azure.com/fluidframework/internal/_workitems/edit/8135)
[AB#8163](https://dev.azure.com/fluidframework/235294da-091d-4c29-84fc-cdfc3d90890b/_workitems/edit/8163) Reduce e2e pipeline duration for FRS and routerlicious testing (although it may not have an impact as odsp lock tests still will take the longest).
Currently if a local interval from `clients[0]` is rebased over a removal from `clients[1]` which removes the entire contents of the interval, and that interval ends up at one of the extents of the sequence (the beginning or end), that local interval is removed and never sent to other clients. If this causes the interval collection to become empty this creates an inconsistency where only `clients[0]` would see the collection label in `getIntervalCollectionLabels`. If `mergeTreeReferencesCanSlideToEndpoint` is enabled, we should treat intervals at start/end just like any other empty interval and still send it to all the other clients.
microsoft#21135) ## Description We currently store the entire pending message in `savedOps` despite not needing its local op metadata at usage. We noticed this memory leak in some SharedTree usage, which puts a copy of the tree's schema in local op metadata (schema has cheap copy, but still seemed reasonable to fix). Co-authored-by: Abram Sanderson <[email protected]>
…nce pipeline (microsoft#21302) I had added some GC performance benchmark tests via https://github.com/microsoft/FluidFramework/pull/21269/files. In order to run these tests in the performance pipeline, couple more steps are needed. This PR adds these steps: - Added a "postpack" script to the container-runtime package which puts the package's test files (src, lib, and dist) in a tar file, which the perf benchmarks pipeline then consumes - Listed the container-runtime package in the yaml file that should run in the pipeline.
…soft#20852) ## Description Adds the recommendation that users set initial state before attaching any created DDSes to `IFluidContainer.create`, and include a code example. --------- Co-authored-by: Abram Sanderson <[email protected]> Co-authored-by: alex-pardes <[email protected]>
Fix GH microsoft#21252 --------- Co-authored-by: Jason Hartman <[email protected]>
microsoft#21320) Add quorum and protocol handlers to container-loader
…icrosoft#21318) Removes client dependence on 'protocol-base' for utility functions that work with Git summary trees. No suprises / very straightforward.
1. Stop exporting a "defaultWebpackConfig", instead config the examples independently. 2. Update source-map-loader dependency to 5.0.0 throughout the repo (including one usage in server for a test while I'm at it). Remove it as an unused dependency for several packages. 3. Remove bohemiaIntercept, unused. 4. Remove isSynchronized. isSynchronized had a sneaky use through the window object. The test now uses a timeoutPromise strategy instead. 5. Remove webCodeLoader and replace with StaticCodeLoader. 6. Fix stale comments in modelLoader.ts.
The docs pipeline fails because the scope of odsp-client is changed from experimental to fluidframework. This change should fix the pipeline.
…t#21342) ## Description This PR changes unhydrated nodes to be backed by MapTreeNodes to allow them to be read even before insertion.
…icrosoft#21282) #### Description [ADO 8152](https://dev.azure.com/fluidframework/internal/_workitems/edit/8152/) [ADO 5141](https://dev.azure.com/fluidframework/internal/_workitems/edit/5141/) This PR creates telemetry loggers for the `SharedTree` DDS which emit telemetry events for rebase operations of below scenarios: - rebasing the EditManager local branch over new peer edits - rebasing an EditManager peer branch over trunk edits when adjusting the branch to the peer's new ref seq - rebasing a new peer edit to the tip of the trunk - rebasing the inverse of a revertible. - rebasing a user branch The event will include: - parent branch length - child branch length - time duration **Note**: Instead of logging the event for __every__ rebase operation, we should log events in batch (e.g., every 100 operations). --------- Co-authored-by: yann-achard-MS <[email protected]>
github-actions
bot
added
base: main
documentation
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
The timeouts for these tests were being caused by 2 things:
I reduced the size of messages being sent and their respective configs, as well as fixing the logic that tries to force a reconnect based on messages sent/received.
AB#7969
AB#7944